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:: Passing definition names as strings
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: Tokens for Words
321:
322: * Execution token:: represents execution/interpretation semantics
323: * Compilation token:: represents compilation semantics
324: * Name token:: represents named words
325:
326: The Text Interpreter
327:
328: * Input Sources::
329: * Number Conversion::
330: * Interpret/Compile states::
331: * Literals::
332: * Interpreter Directives::
333:
334: Word Lists
335:
336: * Why use word lists?::
337: * Word list examples::
338:
339: Files
340:
341: * Forth source files::
342: * General files::
343: * Search Paths::
344:
345: Search Paths
346:
347: * Forth Search Paths::
348: * General Search Paths::
349:
350: Other I/O
351:
352: * Simple numeric output:: Predefined formats
353: * Formatted numeric output:: Formatted (pictured) output
354: * String Formats:: How Forth stores strings in memory
355: * Displaying characters and strings:: Other stuff
356: * Input:: Input
357:
358: Programming Tools
359:
360: * Debugging:: Simple and quick.
361: * Assertions:: Making your programs self-checking.
362: * Singlestep Debugger:: Executing your program word by word.
363:
364: Assembler and Code Words
365:
366: * Code and ;code::
367: * Common Assembler:: Assembler Syntax
368: * Common Disassembler::
369: * 386 Assembler:: Deviations and special cases
370: * Alpha Assembler:: Deviations and special cases
371: * MIPS assembler:: Deviations and special cases
372: * Other assemblers:: How to write them
373:
374: Locals
375:
376: * Gforth locals::
377: * ANS Forth locals::
378:
379: Gforth locals
380:
381: * Where are locals visible by name?::
382: * How long do locals live?::
383: * Programming Style::
384: * Implementation::
385:
386: Structures
387:
388: * Why explicit structure support?::
389: * Structure Usage::
390: * Structure Naming Convention::
391: * Structure Implementation::
392: * Structure Glossary::
393:
394: Object-oriented Forth
395:
396: * Why object-oriented programming?::
397: * Object-Oriented Terminology::
398: * Objects::
399: * OOF::
400: * Mini-OOF::
401: * Comparison with other object models::
402:
403: The @file{objects.fs} model
404:
405: * Properties of the Objects model::
406: * Basic Objects Usage::
407: * The Objects base class::
408: * Creating objects::
409: * Object-Oriented Programming Style::
410: * Class Binding::
411: * Method conveniences::
412: * Classes and Scoping::
413: * Dividing classes::
414: * Object Interfaces::
415: * Objects Implementation::
416: * Objects Glossary::
417:
418: The @file{oof.fs} model
419:
420: * Properties of the OOF model::
421: * Basic OOF Usage::
422: * The OOF base class::
423: * Class Declaration::
424: * Class Implementation::
425:
426: The @file{mini-oof.fs} model
427:
428: * Basic Mini-OOF Usage::
429: * Mini-OOF Example::
430: * Mini-OOF Implementation::
431:
432: Tools
433:
434: * ANS Report:: Report the words used, sorted by wordset.
435:
436: ANS conformance
437:
438: * The Core Words::
439: * The optional Block word set::
440: * The optional Double Number word set::
441: * The optional Exception word set::
442: * The optional Facility word set::
443: * The optional File-Access word set::
444: * The optional Floating-Point word set::
445: * The optional Locals word set::
446: * The optional Memory-Allocation word set::
447: * The optional Programming-Tools word set::
448: * The optional Search-Order word set::
449:
450: The Core Words
451:
452: * core-idef:: Implementation Defined Options
453: * core-ambcond:: Ambiguous Conditions
454: * core-other:: Other System Documentation
455:
456: The optional Block word set
457:
458: * block-idef:: Implementation Defined Options
459: * block-ambcond:: Ambiguous Conditions
460: * block-other:: Other System Documentation
461:
462: The optional Double Number word set
463:
464: * double-ambcond:: Ambiguous Conditions
465:
466: The optional Exception word set
467:
468: * exception-idef:: Implementation Defined Options
469:
470: The optional Facility word set
471:
472: * facility-idef:: Implementation Defined Options
473: * facility-ambcond:: Ambiguous Conditions
474:
475: The optional File-Access word set
476:
477: * file-idef:: Implementation Defined Options
478: * file-ambcond:: Ambiguous Conditions
479:
480: The optional Floating-Point word set
481:
482: * floating-idef:: Implementation Defined Options
483: * floating-ambcond:: Ambiguous Conditions
484:
485: The optional Locals word set
486:
487: * locals-idef:: Implementation Defined Options
488: * locals-ambcond:: Ambiguous Conditions
489:
490: The optional Memory-Allocation word set
491:
492: * memory-idef:: Implementation Defined Options
493:
494: The optional Programming-Tools word set
495:
496: * programming-idef:: Implementation Defined Options
497: * programming-ambcond:: Ambiguous Conditions
498:
499: The optional Search-Order word set
500:
501: * search-idef:: Implementation Defined Options
502: * search-ambcond:: Ambiguous Conditions
503:
504: Image Files
505:
506: * Image Licensing Issues:: Distribution terms for images.
507: * Image File Background:: Why have image files?
508: * Non-Relocatable Image Files:: don't always work.
509: * Data-Relocatable Image Files:: are better.
510: * Fully Relocatable Image Files:: better yet.
511: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
512: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
513: * Modifying the Startup Sequence:: and turnkey applications.
514:
515: Fully Relocatable Image Files
516:
517: * gforthmi:: The normal way
518: * cross.fs:: The hard way
519:
520: Engine
521:
522: * Portability::
523: * Threading::
524: * Primitives::
525: * Performance::
526:
527: Threading
528:
529: * Scheduling::
530: * Direct or Indirect Threaded?::
531: * DOES>::
532:
533: Primitives
534:
535: * Automatic Generation::
536: * TOS Optimization::
537: * Produced code::
538:
539: Cross Compiler
540:
541: * Using the Cross Compiler::
542: * How the Cross Compiler Works::
543:
544: Other Forth-related information
545:
546: * Internet resources::
547: * Books::
548: * The Forth Interest Group::
549: * Conferences::
550:
551: @end detailmenu
552: @end menu
553:
554: @node License, Goals, Top, Top
555: @unnumbered GNU GENERAL PUBLIC LICENSE
556: @center Version 2, June 1991
557:
558: @display
559: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
560: 675 Mass Ave, Cambridge, MA 02139, USA
561:
562: Everyone is permitted to copy and distribute verbatim copies
563: of this license document, but changing it is not allowed.
564: @end display
565:
566: @unnumberedsec Preamble
567:
568: The licenses for most software are designed to take away your
569: freedom to share and change it. By contrast, the GNU General Public
570: License is intended to guarantee your freedom to share and change free
571: software---to make sure the software is free for all its users. This
572: General Public License applies to most of the Free Software
573: Foundation's software and to any other program whose authors commit to
574: using it. (Some other Free Software Foundation software is covered by
575: the GNU Library General Public License instead.) You can apply it to
576: your programs, too.
577:
578: When we speak of free software, we are referring to freedom, not
579: price. Our General Public Licenses are designed to make sure that you
580: have the freedom to distribute copies of free software (and charge for
581: this service if you wish), that you receive source code or can get it
582: if you want it, that you can change the software or use pieces of it
583: in new free programs; and that you know you can do these things.
584:
585: To protect your rights, we need to make restrictions that forbid
586: anyone to deny you these rights or to ask you to surrender the rights.
587: These restrictions translate to certain responsibilities for you if you
588: distribute copies of the software, or if you modify it.
589:
590: For example, if you distribute copies of such a program, whether
591: gratis or for a fee, you must give the recipients all the rights that
592: you have. You must make sure that they, too, receive or can get the
593: source code. And you must show them these terms so they know their
594: rights.
595:
596: We protect your rights with two steps: (1) copyright the software, and
597: (2) offer you this license which gives you legal permission to copy,
598: distribute and/or modify the software.
599:
600: Also, for each author's protection and ours, we want to make certain
601: that everyone understands that there is no warranty for this free
602: software. If the software is modified by someone else and passed on, we
603: want its recipients to know that what they have is not the original, so
604: that any problems introduced by others will not reflect on the original
605: authors' reputations.
606:
607: Finally, any free program is threatened constantly by software
608: patents. We wish to avoid the danger that redistributors of a free
609: program will individually obtain patent licenses, in effect making the
610: program proprietary. To prevent this, we have made it clear that any
611: patent must be licensed for everyone's free use or not licensed at all.
612:
613: The precise terms and conditions for copying, distribution and
614: modification follow.
615:
616: @iftex
617: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
618: @end iftex
619: @ifnottex
620: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
621: @end ifnottex
622:
623: @enumerate 0
624: @item
625: This License applies to any program or other work which contains
626: a notice placed by the copyright holder saying it may be distributed
627: under the terms of this General Public License. The ``Program'', below,
628: refers to any such program or work, and a ``work based on the Program''
629: means either the Program or any derivative work under copyright law:
630: that is to say, a work containing the Program or a portion of it,
631: either verbatim or with modifications and/or translated into another
632: language. (Hereinafter, translation is included without limitation in
633: the term ``modification''.) Each licensee is addressed as ``you''.
634:
635: Activities other than copying, distribution and modification are not
636: covered by this License; they are outside its scope. The act of
637: running the Program is not restricted, and the output from the Program
638: is covered only if its contents constitute a work based on the
639: Program (independent of having been made by running the Program).
640: Whether that is true depends on what the Program does.
641:
642: @item
643: You may copy and distribute verbatim copies of the Program's
644: source code as you receive it, in any medium, provided that you
645: conspicuously and appropriately publish on each copy an appropriate
646: copyright notice and disclaimer of warranty; keep intact all the
647: notices that refer to this License and to the absence of any warranty;
648: and give any other recipients of the Program a copy of this License
649: along with the Program.
650:
651: You may charge a fee for the physical act of transferring a copy, and
652: you may at your option offer warranty protection in exchange for a fee.
653:
654: @item
655: You may modify your copy or copies of the Program or any portion
656: of it, thus forming a work based on the Program, and copy and
657: distribute such modifications or work under the terms of Section 1
658: above, provided that you also meet all of these conditions:
659:
660: @enumerate a
661: @item
662: You must cause the modified files to carry prominent notices
663: stating that you changed the files and the date of any change.
664:
665: @item
666: You must cause any work that you distribute or publish, that in
667: whole or in part contains or is derived from the Program or any
668: part thereof, to be licensed as a whole at no charge to all third
669: parties under the terms of this License.
670:
671: @item
672: If the modified program normally reads commands interactively
673: when run, you must cause it, when started running for such
674: interactive use in the most ordinary way, to print or display an
675: announcement including an appropriate copyright notice and a
676: notice that there is no warranty (or else, saying that you provide
677: a warranty) and that users may redistribute the program under
678: these conditions, and telling the user how to view a copy of this
679: License. (Exception: if the Program itself is interactive but
680: does not normally print such an announcement, your work based on
681: the Program is not required to print an announcement.)
682: @end enumerate
683:
684: These requirements apply to the modified work as a whole. If
685: identifiable sections of that work are not derived from the Program,
686: and can be reasonably considered independent and separate works in
687: themselves, then this License, and its terms, do not apply to those
688: sections when you distribute them as separate works. But when you
689: distribute the same sections as part of a whole which is a work based
690: on the Program, the distribution of the whole must be on the terms of
691: this License, whose permissions for other licensees extend to the
692: entire whole, and thus to each and every part regardless of who wrote it.
693:
694: Thus, it is not the intent of this section to claim rights or contest
695: your rights to work written entirely by you; rather, the intent is to
696: exercise the right to control the distribution of derivative or
697: collective works based on the Program.
698:
699: In addition, mere aggregation of another work not based on the Program
700: with the Program (or with a work based on the Program) on a volume of
701: a storage or distribution medium does not bring the other work under
702: the scope of this License.
703:
704: @item
705: You may copy and distribute the Program (or a work based on it,
706: under Section 2) in object code or executable form under the terms of
707: Sections 1 and 2 above provided that you also do one of the following:
708:
709: @enumerate a
710: @item
711: Accompany it with the complete corresponding machine-readable
712: source code, which must be distributed under the terms of Sections
713: 1 and 2 above on a medium customarily used for software interchange; or,
714:
715: @item
716: Accompany it with a written offer, valid for at least three
717: years, to give any third party, for a charge no more than your
718: cost of physically performing source distribution, a complete
719: machine-readable copy of the corresponding source code, to be
720: distributed under the terms of Sections 1 and 2 above on a medium
721: customarily used for software interchange; or,
722:
723: @item
724: Accompany it with the information you received as to the offer
725: to distribute corresponding source code. (This alternative is
726: allowed only for noncommercial distribution and only if you
727: received the program in object code or executable form with such
728: an offer, in accord with Subsection b above.)
729: @end enumerate
730:
731: The source code for a work means the preferred form of the work for
732: making modifications to it. For an executable work, complete source
733: code means all the source code for all modules it contains, plus any
734: associated interface definition files, plus the scripts used to
735: control compilation and installation of the executable. However, as a
736: special exception, the source code distributed need not include
737: anything that is normally distributed (in either source or binary
738: form) with the major components (compiler, kernel, and so on) of the
739: operating system on which the executable runs, unless that component
740: itself accompanies the executable.
741:
742: If distribution of executable or object code is made by offering
743: access to copy from a designated place, then offering equivalent
744: access to copy the source code from the same place counts as
745: distribution of the source code, even though third parties are not
746: compelled to copy the source along with the object code.
747:
748: @item
749: You may not copy, modify, sublicense, or distribute the Program
750: except as expressly provided under this License. Any attempt
751: otherwise to copy, modify, sublicense or distribute the Program is
752: void, and will automatically terminate your rights under this License.
753: However, parties who have received copies, or rights, from you under
754: this License will not have their licenses terminated so long as such
755: parties remain in full compliance.
756:
757: @item
758: You are not required to accept this License, since you have not
759: signed it. However, nothing else grants you permission to modify or
760: distribute the Program or its derivative works. These actions are
761: prohibited by law if you do not accept this License. Therefore, by
762: modifying or distributing the Program (or any work based on the
763: Program), you indicate your acceptance of this License to do so, and
764: all its terms and conditions for copying, distributing or modifying
765: the Program or works based on it.
766:
767: @item
768: Each time you redistribute the Program (or any work based on the
769: Program), the recipient automatically receives a license from the
770: original licensor to copy, distribute or modify the Program subject to
771: these terms and conditions. You may not impose any further
772: restrictions on the recipients' exercise of the rights granted herein.
773: You are not responsible for enforcing compliance by third parties to
774: this License.
775:
776: @item
777: If, as a consequence of a court judgment or allegation of patent
778: infringement or for any other reason (not limited to patent issues),
779: conditions are imposed on you (whether by court order, agreement or
780: otherwise) that contradict the conditions of this License, they do not
781: excuse you from the conditions of this License. If you cannot
782: distribute so as to satisfy simultaneously your obligations under this
783: License and any other pertinent obligations, then as a consequence you
784: may not distribute the Program at all. For example, if a patent
785: license would not permit royalty-free redistribution of the Program by
786: all those who receive copies directly or indirectly through you, then
787: the only way you could satisfy both it and this License would be to
788: refrain entirely from distribution of the Program.
789:
790: If any portion of this section is held invalid or unenforceable under
791: any particular circumstance, the balance of the section is intended to
792: apply and the section as a whole is intended to apply in other
793: circumstances.
794:
795: It is not the purpose of this section to induce you to infringe any
796: patents or other property right claims or to contest validity of any
797: such claims; this section has the sole purpose of protecting the
798: integrity of the free software distribution system, which is
799: implemented by public license practices. Many people have made
800: generous contributions to the wide range of software distributed
801: through that system in reliance on consistent application of that
802: system; it is up to the author/donor to decide if he or she is willing
803: to distribute software through any other system and a licensee cannot
804: impose that choice.
805:
806: This section is intended to make thoroughly clear what is believed to
807: be a consequence of the rest of this License.
808:
809: @item
810: If the distribution and/or use of the Program is restricted in
811: certain countries either by patents or by copyrighted interfaces, the
812: original copyright holder who places the Program under this License
813: may add an explicit geographical distribution limitation excluding
814: those countries, so that distribution is permitted only in or among
815: countries not thus excluded. In such case, this License incorporates
816: the limitation as if written in the body of this License.
817:
818: @item
819: The Free Software Foundation may publish revised and/or new versions
820: of the General Public License from time to time. Such new versions will
821: be similar in spirit to the present version, but may differ in detail to
822: address new problems or concerns.
823:
824: Each version is given a distinguishing version number. If the Program
825: specifies a version number of this License which applies to it and ``any
826: later version'', you have the option of following the terms and conditions
827: either of that version or of any later version published by the Free
828: Software Foundation. If the Program does not specify a version number of
829: this License, you may choose any version ever published by the Free Software
830: Foundation.
831:
832: @item
833: If you wish to incorporate parts of the Program into other free
834: programs whose distribution conditions are different, write to the author
835: to ask for permission. For software which is copyrighted by the Free
836: Software Foundation, write to the Free Software Foundation; we sometimes
837: make exceptions for this. Our decision will be guided by the two goals
838: of preserving the free status of all derivatives of our free software and
839: of promoting the sharing and reuse of software generally.
840:
841: @iftex
842: @heading NO WARRANTY
843: @end iftex
844: @ifnottex
845: @center NO WARRANTY
846: @end ifnottex
847:
848: @item
849: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
850: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
851: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
852: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
853: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
854: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
855: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
856: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
857: REPAIR OR CORRECTION.
858:
859: @item
860: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
861: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
862: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
863: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
864: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
865: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
866: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
867: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
868: POSSIBILITY OF SUCH DAMAGES.
869: @end enumerate
870:
871: @iftex
872: @heading END OF TERMS AND CONDITIONS
873: @end iftex
874: @ifnottex
875: @center END OF TERMS AND CONDITIONS
876: @end ifnottex
877:
878: @page
879: @unnumberedsec How to Apply These Terms to Your New Programs
880:
881: If you develop a new program, and you want it to be of the greatest
882: possible use to the public, the best way to achieve this is to make it
883: free software which everyone can redistribute and change under these terms.
884:
885: To do so, attach the following notices to the program. It is safest
886: to attach them to the start of each source file to most effectively
887: convey the exclusion of warranty; and each file should have at least
888: the ``copyright'' line and a pointer to where the full notice is found.
889:
890: @smallexample
891: @var{one line to give the program's name and a brief idea of what it does.}
892: Copyright (C) 19@var{yy} @var{name of author}
893:
894: This program is free software; you can redistribute it and/or modify
895: it under the terms of the GNU General Public License as published by
896: the Free Software Foundation; either version 2 of the License, or
897: (at your option) any later version.
898:
899: This program is distributed in the hope that it will be useful,
900: but WITHOUT ANY WARRANTY; without even the implied warranty of
901: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
902: GNU General Public License for more details.
903:
904: You should have received a copy of the GNU General Public License
905: along with this program; if not, write to the Free Software
906: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
907: @end smallexample
908:
909: Also add information on how to contact you by electronic and paper mail.
910:
911: If the program is interactive, make it output a short notice like this
912: when it starts in an interactive mode:
913:
914: @smallexample
915: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
916: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
917: type `show w'.
918: This is free software, and you are welcome to redistribute it
919: under certain conditions; type `show c' for details.
920: @end smallexample
921:
922: The hypothetical commands @samp{show w} and @samp{show c} should show
923: the appropriate parts of the General Public License. Of course, the
924: commands you use may be called something other than @samp{show w} and
925: @samp{show c}; they could even be mouse-clicks or menu items---whatever
926: suits your program.
927:
928: You should also get your employer (if you work as a programmer) or your
929: school, if any, to sign a ``copyright disclaimer'' for the program, if
930: necessary. Here is a sample; alter the names:
931:
932: @smallexample
933: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
934: `Gnomovision' (which makes passes at compilers) written by James Hacker.
935:
936: @var{signature of Ty Coon}, 1 April 1989
937: Ty Coon, President of Vice
938: @end smallexample
939:
940: This General Public License does not permit incorporating your program into
941: proprietary programs. If your program is a subroutine library, you may
942: consider it more useful to permit linking proprietary applications with the
943: library. If this is what you want to do, use the GNU Library General
944: Public License instead of this License.
945:
946: @iftex
947: @unnumbered Preface
948: @cindex Preface
949: This manual documents Gforth. Some introductory material is provided for
950: readers who are unfamiliar with Forth or who are migrating to Gforth
951: from other Forth compilers. However, this manual is primarily a
952: reference manual.
953: @end iftex
954:
955: @comment TODO much more blurb here.
956:
957: @c ******************************************************************
958: @node Goals, Gforth Environment, License, Top
959: @comment node-name, next, previous, up
960: @chapter Goals of Gforth
961: @cindex goals of the Gforth project
962: The goal of the Gforth Project is to develop a standard model for
963: ANS Forth. This can be split into several subgoals:
964:
965: @itemize @bullet
966: @item
967: Gforth should conform to the ANS Forth Standard.
968: @item
969: It should be a model, i.e. it should define all the
970: implementation-dependent things.
971: @item
972: It should become standard, i.e. widely accepted and used. This goal
973: is the most difficult one.
974: @end itemize
975:
976: To achieve these goals Gforth should be
977: @itemize @bullet
978: @item
979: Similar to previous models (fig-Forth, F83)
980: @item
981: Powerful. It should provide for all the things that are considered
982: necessary today and even some that are not yet considered necessary.
983: @item
984: Efficient. It should not get the reputation of being exceptionally
985: slow.
986: @item
987: Free.
988: @item
989: Available on many machines/easy to port.
990: @end itemize
991:
992: Have we achieved these goals? Gforth conforms to the ANS Forth
993: standard. It may be considered a model, but we have not yet documented
994: which parts of the model are stable and which parts we are likely to
995: change. It certainly has not yet become a de facto standard, but it
996: appears to be quite popular. It has some similarities to and some
997: differences from previous models. It has some powerful features, but not
998: yet everything that we envisioned. We certainly have achieved our
999: execution speed goals (@pxref{Performance})@footnote{However, in 1998
1000: the bar was raised when the major commercial Forth vendors switched to
1001: native code compilers.}. It is free and available on many machines.
1002:
1003: @c ******************************************************************
1004: @node Gforth Environment, Tutorial, Goals, Top
1005: @chapter Gforth Environment
1006: @cindex Gforth environment
1007:
1008: Note: ultimately, the Gforth man page will be auto-generated from the
1009: material in this chapter.
1010:
1011: @menu
1012: * Invoking Gforth:: Getting in
1013: * Leaving Gforth:: Getting out
1014: * Command-line editing::
1015: * Environment variables:: that affect how Gforth starts up
1016: * Gforth Files:: What gets installed and where
1017: * Startup speed:: When 35ms is not fast enough ...
1018: @end menu
1019:
1020: For related information about the creation of images see @ref{Image Files}.
1021:
1022: @comment ----------------------------------------------
1023: @node Invoking Gforth, Leaving Gforth, Gforth Environment, Gforth Environment
1024: @section Invoking Gforth
1025: @cindex invoking Gforth
1026: @cindex running Gforth
1027: @cindex command-line options
1028: @cindex options on the command line
1029: @cindex flags on the command line
1030:
1031: Gforth is made up of two parts; an executable ``engine'' (named
1032: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
1033: will usually just say @code{gforth} -- this automatically loads the
1034: default image file @file{gforth.fi}. In many other cases the default
1035: Gforth image will be invoked like this:
1036: @example
1037: gforth [file | -e forth-code] ...
1038: @end example
1039: @noindent
1040: This interprets the contents of the files and the Forth code in the order they
1041: are given.
1042:
1043: In addition to the @file{gforth} engine, there is also an engine called
1044: @file{gforth-fast}, which is faster, but gives less informative error
1045: messages (@pxref{Error messages}).
1046:
1047: In general, the command line looks like this:
1048:
1049: @example
1050: gforth[-fast] [engine options] [image options]
1051: @end example
1052:
1053: The engine options must come before the rest of the command
1054: line. They are:
1055:
1056: @table @code
1057: @cindex -i, command-line option
1058: @cindex --image-file, command-line option
1059: @item --image-file @i{file}
1060: @itemx -i @i{file}
1061: Loads the Forth image @i{file} instead of the default
1062: @file{gforth.fi} (@pxref{Image Files}).
1063:
1064: @cindex --appl-image, command-line option
1065: @item --appl-image @i{file}
1066: Loads the image @i{file} and leaves all further command-line arguments
1067: to the image (instead of processing them as engine options). This is
1068: useful for building executable application images on Unix, built with
1069: @code{gforthmi --application ...}.
1070:
1071: @cindex --path, command-line option
1072: @cindex -p, command-line option
1073: @item --path @i{path}
1074: @itemx -p @i{path}
1075: Uses @i{path} for searching the image file and Forth source code files
1076: instead of the default in the environment variable @code{GFORTHPATH} or
1077: the path specified at installation time (e.g.,
1078: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1079: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1080:
1081: @cindex --dictionary-size, command-line option
1082: @cindex -m, command-line option
1083: @cindex @i{size} parameters for command-line options
1084: @cindex size of the dictionary and the stacks
1085: @item --dictionary-size @i{size}
1086: @itemx -m @i{size}
1087: Allocate @i{size} space for the Forth dictionary space instead of
1088: using the default specified in the image (typically 256K). The
1089: @i{size} specification for this and subsequent options consists of
1090: an integer and a unit (e.g.,
1091: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1092: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1093: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1094: @code{e} is used.
1095:
1096: @cindex --data-stack-size, command-line option
1097: @cindex -d, command-line option
1098: @item --data-stack-size @i{size}
1099: @itemx -d @i{size}
1100: Allocate @i{size} space for the data stack instead of using the
1101: default specified in the image (typically 16K).
1102:
1103: @cindex --return-stack-size, command-line option
1104: @cindex -r, command-line option
1105: @item --return-stack-size @i{size}
1106: @itemx -r @i{size}
1107: Allocate @i{size} space for the return stack instead of using the
1108: default specified in the image (typically 15K).
1109:
1110: @cindex --fp-stack-size, command-line option
1111: @cindex -f, command-line option
1112: @item --fp-stack-size @i{size}
1113: @itemx -f @i{size}
1114: Allocate @i{size} space for the floating point stack instead of
1115: using the default specified in the image (typically 15.5K). In this case
1116: the unit specifier @code{e} refers to floating point numbers.
1117:
1118: @cindex --locals-stack-size, command-line option
1119: @cindex -l, command-line option
1120: @item --locals-stack-size @i{size}
1121: @itemx -l @i{size}
1122: Allocate @i{size} space for the locals stack instead of using the
1123: default specified in the image (typically 14.5K).
1124:
1125: @cindex -h, command-line option
1126: @cindex --help, command-line option
1127: @item --help
1128: @itemx -h
1129: Print a message about the command-line options
1130:
1131: @cindex -v, command-line option
1132: @cindex --version, command-line option
1133: @item --version
1134: @itemx -v
1135: Print version and exit
1136:
1137: @cindex --debug, command-line option
1138: @item --debug
1139: Print some information useful for debugging on startup.
1140:
1141: @cindex --offset-image, command-line option
1142: @item --offset-image
1143: Start the dictionary at a slightly different position than would be used
1144: otherwise (useful for creating data-relocatable images,
1145: @pxref{Data-Relocatable Image Files}).
1146:
1147: @cindex --no-offset-im, command-line option
1148: @item --no-offset-im
1149: Start the dictionary at the normal position.
1150:
1151: @cindex --clear-dictionary, command-line option
1152: @item --clear-dictionary
1153: Initialize all bytes in the dictionary to 0 before loading the image
1154: (@pxref{Data-Relocatable Image Files}).
1155:
1156: @cindex --die-on-signal, command-line-option
1157: @item --die-on-signal
1158: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1159: or the segmentation violation SIGSEGV) by translating it into a Forth
1160: @code{THROW}. With this option, Gforth exits if it receives such a
1161: signal. This option is useful when the engine and/or the image might be
1162: severely broken (such that it causes another signal before recovering
1163: from the first); this option avoids endless loops in such cases.
1164: @end table
1165:
1166: @cindex loading files at startup
1167: @cindex executing code on startup
1168: @cindex batch processing with Gforth
1169: As explained above, the image-specific command-line arguments for the
1170: default image @file{gforth.fi} consist of a sequence of filenames and
1171: @code{-e @var{forth-code}} options that are interpreted in the sequence
1172: in which they are given. The @code{-e @var{forth-code}} or
1173: @code{--evaluate @var{forth-code}} option evaluates the Forth
1174: code. This option takes only one argument; if you want to evaluate more
1175: Forth words, you have to quote them or use @code{-e} several times. To exit
1176: after processing the command line (instead of entering interactive mode)
1177: append @code{-e bye} to the command line.
1178:
1179: @cindex versions, invoking other versions of Gforth
1180: If you have several versions of Gforth installed, @code{gforth} will
1181: invoke the version that was installed last. @code{gforth-@i{version}}
1182: invokes a specific version. If your environment contains the variable
1183: @code{GFORTHPATH}, you may want to override it by using the
1184: @code{--path} option.
1185:
1186: Not yet implemented:
1187: On startup the system first executes the system initialization file
1188: (unless the option @code{--no-init-file} is given; note that the system
1189: resulting from using this option may not be ANS Forth conformant). Then
1190: the user initialization file @file{.gforth.fs} is executed, unless the
1191: option @code{--no-rc} is given; this file is searched for in @file{.},
1192: then in @file{~}, then in the normal path (see above).
1193:
1194:
1195:
1196: @comment ----------------------------------------------
1197: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1198: @section Leaving Gforth
1199: @cindex Gforth - leaving
1200: @cindex leaving Gforth
1201:
1202: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1203: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1204: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1205: data are discarded. For ways of saving the state of the system before
1206: leaving Gforth see @ref{Image Files}.
1207:
1208: doc-bye
1209:
1210:
1211: @comment ----------------------------------------------
1212: @node Command-line editing, Environment variables, Leaving Gforth, Gforth Environment
1213: @section Command-line editing
1214: @cindex command-line editing
1215:
1216: Gforth maintains a history file that records every line that you type to
1217: the text interpreter. This file is preserved between sessions, and is
1218: used to provide a command-line recall facility; if you type @kbd{Ctrl-P}
1219: repeatedly you can recall successively older commands from this (or
1220: previous) session(s). The full list of command-line editing facilities is:
1221:
1222: @itemize @bullet
1223: @item
1224: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1225: commands from the history buffer.
1226: @item
1227: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1228: from the history buffer.
1229: @item
1230: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1231: @item
1232: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1233: @item
1234: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1235: closing up the line.
1236: @item
1237: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1238: @item
1239: @kbd{Ctrl-a} to move the cursor to the start of the line.
1240: @item
1241: @kbd{Ctrl-e} to move the cursor to the end of the line.
1242: @item
1243: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1244: line.
1245: @item
1246: @key{TAB} to step through all possible full-word completions of the word
1247: currently being typed.
1248: @item
1249: @kbd{Ctrl-d} on an empty line line to terminate Gforth (gracefully,
1250: using @code{bye}).
1251: @item
1252: @kbd{Ctrl-x} (or @code{Ctrl-d} on a non-empty line) to delete the
1253: character under the cursor.
1254: @end itemize
1255:
1256: When editing, displayable characters are inserted to the left of the
1257: cursor position; the line is always in ``insert'' (as opposed to
1258: ``overstrike'') mode.
1259:
1260: @cindex history file
1261: @cindex @file{.gforth-history}
1262: On Unix systems, the history file is @file{~/.gforth-history} by
1263: default@footnote{i.e. it is stored in the user's home directory.}. You
1264: can find out the name and location of your history file using:
1265:
1266: @example
1267: history-file type \ Unix-class systems
1268:
1269: history-file type \ Other systems
1270: history-dir type
1271: @end example
1272:
1273: If you enter long definitions by hand, you can use a text editor to
1274: paste them out of the history file into a Forth source file for reuse at
1275: a later time.
1276:
1277: Gforth never trims the size of the history file, so you should do this
1278: periodically, if necessary.
1279:
1280: @comment this is all defined in history.fs
1281: @comment NAC TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
1282: @comment chosen?
1283:
1284:
1285: @comment ----------------------------------------------
1286: @node Environment variables, Gforth Files, Command-line editing, Gforth Environment
1287: @section Environment variables
1288: @cindex environment variables
1289:
1290: Gforth uses these environment variables:
1291:
1292: @itemize @bullet
1293: @item
1294: @cindex @code{GFORTHHIST} -- environment variable
1295: @code{GFORTHHIST} -- (Unix systems only) specifies the directory in which to
1296: open/create the history file, @file{.gforth-history}. Default:
1297: @code{$HOME}.
1298:
1299: @item
1300: @cindex @code{GFORTHPATH} -- environment variable
1301: @code{GFORTHPATH} -- specifies the path used when searching for the gforth image file and
1302: for Forth source-code files.
1303:
1304: @item
1305: @cindex @code{GFORTH} -- environment variable
1306: @code{GFORTH} -- used by @file{gforthmi}, @xref{gforthmi}.
1307:
1308: @item
1309: @cindex @code{GFORTHD} -- environment variable
1310: @code{GFORTHD} -- used by @file{gforthmi}, @xref{gforthmi}.
1311:
1312: @item
1313: @cindex @code{TMP}, @code{TEMP} - environment variable
1314: @code{TMP}, @code{TEMP} - (non-Unix systems only) used as a potential
1315: location for the history file.
1316: @end itemize
1317:
1318: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1319: @comment mentioning these.
1320:
1321: All the Gforth environment variables default to sensible values if they
1322: are not set.
1323:
1324:
1325: @comment ----------------------------------------------
1326: @node Gforth Files, Startup speed, Environment variables, Gforth Environment
1327: @section Gforth files
1328: @cindex Gforth files
1329:
1330: When you install Gforth on a Unix system, it installs files in these
1331: locations by default:
1332:
1333: @itemize @bullet
1334: @item
1335: @file{/usr/local/bin/gforth}
1336: @item
1337: @file{/usr/local/bin/gforthmi}
1338: @item
1339: @file{/usr/local/man/man1/gforth.1} - man page.
1340: @item
1341: @file{/usr/local/info} - the Info version of this manual.
1342: @item
1343: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1344: @item
1345: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1346: @item
1347: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1348: @item
1349: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1350: @end itemize
1351:
1352: You can select different places for installation by using
1353: @code{configure} options (listed with @code{configure --help}).
1354:
1355: @comment ----------------------------------------------
1356: @node Startup speed, , Gforth Files, Gforth Environment
1357: @section Startup speed
1358: @cindex Startup speed
1359: @cindex speed, startup
1360:
1361: If Gforth is used for CGI scripts or in shell scripts, its startup
1362: speed may become a problem. On a 300MHz 21064a under Linux-2.2.13 with
1363: glibc-2.0.7, @code{gforth -e bye} takes about 24.6ms user and 11.3ms
1364: system time.
1365:
1366: If startup speed is a problem, you may consider the following ways to
1367: improve it; or you may consider ways to reduce the number of startups
1368: (for example, by using Fast-CGI).
1369:
1370: The first step to improve startup speed is to statically link Gforth, by
1371: building it with @code{XLDFLAGS=-static}. This requires more memory for
1372: the code and will therefore slow down the first invocation, but
1373: subsequent invocations avoid the dynamic linking overhead. Another
1374: disadvantage is that Gforth won't profit from library upgrades. As a
1375: result, @code{gforth-static -e bye} takes about 17.1ms user and
1376: 8.2ms system time.
1377:
1378: The next step to improve startup speed is to use a non-relocatable image
1379: (@pxref{Non-Relocatable Image Files}). You can create this image with
1380: @code{gforth -e "savesystem gforthnr.fi bye"} and later use it with
1381: @code{gforth -i gforthnr.fi ...}. This avoids the relocation overhead
1382: and a part of the copy-on-write overhead. The disadvantage is that the
1383: non-relocatable image does not work if the OS gives Gforth a different
1384: address for the dictionary, for whatever reason; so you better provide a
1385: fallback on a relocatable image. @code{gforth-static -i gforthnr.fi -e
1386: bye} takes about 15.3ms user and 7.5ms system time.
1387:
1388: The final step is to disable dictionary hashing in Gforth. Gforth
1389: builds the hash table on startup, which takes much of the startup
1390: overhead. You can do this by commenting out the @code{include hash.fs}
1391: in @file{startup.fs} and everything that requires @file{hash.fs} (at the
1392: moment @file{table.fs} and @file{ekey.fs}) and then doing @code{make}.
1393: The disadvantages are that functionality like @code{table} and
1394: @code{ekey} is missing and that text interpretation (e.g., compiling)
1395: now takes much longer. So, you should only use this method if there is
1396: no significant text interpretation to perform (the script should be
1397: compiled into the image, amongst other things). @code{gforth-static -i
1398: gforthnrnh.fi -e bye} takes about 2.1ms user and 6.1ms system time.
1399:
1400: @c ******************************************************************
1401: @node Tutorial, Introduction, Gforth Environment, Top
1402: @chapter Forth Tutorial
1403: @cindex Tutorial
1404: @cindex Forth Tutorial
1405:
1406: @c Topics from nac's Introduction that could be mentioned:
1407: @c press <ret> after each line
1408: @c Prompt
1409: @c numbers vs. words in dictionary on text interpretation
1410: @c what happens on redefinition
1411: @c parsing words (in particular, defining words)
1412:
1413: This tutorial can be used with any ANS-compliant Forth; any
1414: Gforth-specific features are marked as such and you can skip them if you
1415: work with another Forth. This tutorial does not explain all features of
1416: Forth, just enough to get you started and give you some ideas about the
1417: facilities available in Forth. Read the rest of the manual and the
1418: standard when you are through this.
1419:
1420: The intended way to use this tutorial is that you work through it while
1421: sitting in front of the console, take a look at the examples and predict
1422: what they will do, then try them out; if the outcome is not as expected,
1423: find out why (e.g., by trying out variations of the example), so you
1424: understand what's going on. There are also some assignments that you
1425: should solve.
1426:
1427: This tutorial assumes that you have programmed before and know what,
1428: e.g., a loop is.
1429:
1430: @c !! explain compat library
1431:
1432: @menu
1433: * Starting Gforth Tutorial::
1434: * Syntax Tutorial::
1435: * Crash Course Tutorial::
1436: * Stack Tutorial::
1437: * Arithmetics Tutorial::
1438: * Stack Manipulation Tutorial::
1439: * Using files for Forth code Tutorial::
1440: * Comments Tutorial::
1441: * Colon Definitions Tutorial::
1442: * Decompilation Tutorial::
1443: * Stack-Effect Comments Tutorial::
1444: * Types Tutorial::
1445: * Factoring Tutorial::
1446: * Designing the stack effect Tutorial::
1447: * Local Variables Tutorial::
1448: * Conditional execution Tutorial::
1449: * Flags and Comparisons Tutorial::
1450: * General Loops Tutorial::
1451: * Counted loops Tutorial::
1452: * Recursion Tutorial::
1453: * Leaving definitions or loops Tutorial::
1454: * Return Stack Tutorial::
1455: * Memory Tutorial::
1456: * Characters and Strings Tutorial::
1457: * Alignment Tutorial::
1458: * Interpretation and Compilation Semantics and Immediacy Tutorial::
1459: * Execution Tokens Tutorial::
1460: * Exceptions Tutorial::
1461: * Defining Words Tutorial::
1462: * Arrays and Records Tutorial::
1463: * POSTPONE Tutorial::
1464: * Literal Tutorial::
1465: * Advanced macros Tutorial::
1466: * Compilation Tokens Tutorial::
1467: * Wordlists and Search Order Tutorial::
1468: @end menu
1469:
1470: @node Starting Gforth Tutorial, Syntax Tutorial, Tutorial, Tutorial
1471: @section Starting Gforth
1472: @cindex starting Gforth tutorial
1473: You can start Gforth by typing its name:
1474:
1475: @example
1476: gforth
1477: @end example
1478:
1479: That puts you into interactive mode; you can leave Gforth by typing
1480: @code{bye}. While in Gforth, you can edit the command line and access
1481: the command line history with cursor keys, similar to bash.
1482:
1483:
1484: @node Syntax Tutorial, Crash Course Tutorial, Starting Gforth Tutorial, Tutorial
1485: @section Syntax
1486: @cindex syntax tutorial
1487:
1488: A @dfn{word} is a sequence of arbitrary characters (expcept white
1489: space). Words are separated by white space. E.g., each of the
1490: following lines contains exactly one word:
1491:
1492: @example
1493: word
1494: !@@#$%^&*()
1495: 1234567890
1496: 5!a
1497: @end example
1498:
1499: A frequent beginner's error is to leave away necessary white space,
1500: resulting in an error like @samp{Undefined word}; so if you see such an
1501: error, check if you have put spaces wherever necessary.
1502:
1503: @example
1504: ." hello, world" \ correct
1505: ."hello, world" \ gives an "Undefined word" error
1506: @end example
1507:
1508: Gforth and most other Forth systems ignore differences in case (they are
1509: case-insensitive), i.e., @samp{word} is the same as @samp{Word}. If
1510: your system is case-sensitive, you may have to type all the examples
1511: given here in upper case.
1512:
1513:
1514: @node Crash Course Tutorial, Stack Tutorial, Syntax Tutorial, Tutorial
1515: @section Crash Course
1516:
1517: Type
1518:
1519: @example
1520: 0 0 !
1521: here execute
1522: ' catch >body 20 erase abort
1523: ' (quit) >body 20 erase
1524: @end example
1525:
1526: The last two examples are guaranteed to destroy parts of Gforth (and
1527: most other systems), so you better leave Gforth afterwards (if it has
1528: not finished by itself). On some systems you may have to kill gforth
1529: from outside (e.g., in Unix with @code{kill}).
1530:
1531: Now that you know how to produce crashes (and that there's not much to
1532: them), let's learn how to produce meaningful programs.
1533:
1534:
1535: @node Stack Tutorial, Arithmetics Tutorial, Crash Course Tutorial, Tutorial
1536: @section Stack
1537: @cindex stack tutorial
1538:
1539: The most obvious feature of Forth is the stack. When you type in a
1540: number, it is pushed on the stack. You can display the content of the
1541: stack with @code{.s}.
1542:
1543: @example
1544: 1 2 .s
1545: 3 .s
1546: @end example
1547:
1548: @code{.s} displays the top-of-stack to the right, i.e., the numbers
1549: appear in @code{.s} output as they appeared in the input.
1550:
1551: You can print the top of stack element with @code{.}.
1552:
1553: @example
1554: 1 2 3 . . .
1555: @end example
1556:
1557: In general, words consume their stack arguments (@code{.s} is an
1558: exception).
1559:
1560: @assignment
1561: What does the stack contain after @code{5 6 7 .}?
1562: @endassignment
1563:
1564:
1565: @node Arithmetics Tutorial, Stack Manipulation Tutorial, Stack Tutorial, Tutorial
1566: @section Arithmetics
1567: @cindex arithmetics tutorial
1568:
1569: The words @code{+}, @code{-}, @code{*}, @code{/}, and @code{mod} always
1570: operate on the top two stack items:
1571:
1572: @example
1573: 2 2 .s
1574: + .s
1575: .
1576: 2 1 - .
1577: 7 3 mod .
1578: @end example
1579:
1580: The operands of @code{-}, @code{/}, and @code{mod} are in the same order
1581: as in the corresponding infix expression (this is generally the case in
1582: Forth).
1583:
1584: Parentheses are superfluous (and not available), because the order of
1585: the words unambiguously determines the order of evaluation and the
1586: operands:
1587:
1588: @example
1589: 3 4 + 5 * .
1590: 3 4 5 * + .
1591: @end example
1592:
1593: @assignment
1594: What are the infix expressions corresponding to the Forth code above?
1595: Write @code{6-7*8+9} in Forth notation@footnote{This notation is also
1596: known as Postfix or RPN (Reverse Polish Notation).}.
1597: @endassignment
1598:
1599: To change the sign, use @code{negate}:
1600:
1601: @example
1602: 2 negate .
1603: @end example
1604:
1605: @assignment
1606: Convert -(-3)*4-5 to Forth.
1607: @endassignment
1608:
1609: @code{/mod} performs both @code{/} and @code{mod}.
1610:
1611: @example
1612: 7 3 /mod . .
1613: @end example
1614:
1615: Reference: @ref{Arithmetic}.
1616:
1617:
1618: @node Stack Manipulation Tutorial, Using files for Forth code Tutorial, Arithmetics Tutorial, Tutorial
1619: @section Stack Manipulation
1620: @cindex stack manipulation tutorial
1621:
1622: Stack manipulation words rearrange the data on the stack.
1623:
1624: @example
1625: 1 .s drop .s
1626: 1 .s dup .s drop drop .s
1627: 1 2 .s over .s drop drop drop
1628: 1 2 .s swap .s drop drop
1629: 1 2 3 .s rot .s drop drop drop
1630: @end example
1631:
1632: These are the most important stack manipulation words. There are also
1633: variants that manipulate twice as many stack items:
1634:
1635: @example
1636: 1 2 3 4 .s 2swap .s 2drop 2drop
1637: @end example
1638:
1639: Two more stack manipulation words are:
1640:
1641: @example
1642: 1 2 .s nip .s drop
1643: 1 2 .s tuck .s 2drop drop
1644: @end example
1645:
1646: @assignment
1647: Replace @code{nip} and @code{tuck} with combinations of other stack
1648: manipulation words.
1649:
1650: @example
1651: Given: How do you get:
1652: 1 2 3 3 2 1
1653: 1 2 3 1 2 3 2
1654: 1 2 3 1 2 3 3
1655: 1 2 3 1 3 3
1656: 1 2 3 2 1 3
1657: 1 2 3 4 4 3 2 1
1658: 1 2 3 1 2 3 1 2 3
1659: 1 2 3 4 1 2 3 4 1 2
1660: 1 2 3
1661: 1 2 3 1 2 3 4
1662: 1 2 3 1 3
1663: @end example
1664: @endassignment
1665:
1666: @example
1667: 5 dup * .
1668: @end example
1669:
1670: @assignment
1671: Write 17^3 and 17^4 in Forth, without writing @code{17} more than once.
1672: Write a piece of Forth code that expects two numbers on the stack
1673: (@var{a} and @var{b}, with @var{b} on top) and computes
1674: @code{(a-b)(a+1)}.
1675: @endassignment
1676:
1677: Reference: @ref{Stack Manipulation}.
1678:
1679:
1680: @node Using files for Forth code Tutorial, Comments Tutorial, Stack Manipulation Tutorial, Tutorial
1681: @section Using files for Forth code
1682: @cindex loading Forth code, tutorial
1683: @cindex files containing Forth code, tutorial
1684:
1685: While working at the Forth command line is convenient for one-line
1686: examples and short one-off code, you probably want to store your source
1687: code in files for convenient editing and persistence. You can use your
1688: favourite editor (Gforth includes Emacs support, @pxref{Emacs and
1689: Gforth}) to create @var{file} and use
1690:
1691: @example
1692: s" @var{file}" included
1693: @end example
1694:
1695: to load it into your Forth system. The file name extension I use for
1696: Forth files is @samp{.fs}.
1697:
1698: You can easily start Gforth with some files loaded like this:
1699:
1700: @example
1701: gforth @var{file1} @var{file2}
1702: @end example
1703:
1704: If an error occurs during loading these files, Gforth terminates,
1705: whereas an error during @code{INCLUDED} within Gforth usually gives you
1706: a Gforth command line. Starting the Forth system every time gives you a
1707: clean start every time, without interference from the results of earlier
1708: tries.
1709:
1710: I often put all the tests in a file, then load the code and run the
1711: tests with
1712:
1713: @example
1714: gforth @var{code} @var{tests} -e bye
1715: @end example
1716:
1717: (often by performing this command with @kbd{C-x C-e} in Emacs). The
1718: @code{-e bye} ensures that Gforth terminates afterwards so that I can
1719: restart this command without ado.
1720:
1721: The advantage of this approach is that the tests can be repeated easily
1722: every time the program ist changed, making it easy to catch bugs
1723: introduced by the change.
1724:
1725: Reference: @ref{Forth source files}.
1726:
1727:
1728: @node Comments Tutorial, Colon Definitions Tutorial, Using files for Forth code Tutorial, Tutorial
1729: @section Comments
1730: @cindex comments tutorial
1731:
1732: @example
1733: \ That's a comment; it ends at the end of the line
1734: ( Another comment; it ends here: ) .s
1735: @end example
1736:
1737: @code{\} and @code{(} are ordinary Forth words and therefore have to be
1738: separated with white space from the following text.
1739:
1740: @example
1741: \This gives an "Undefined word" error
1742: @end example
1743:
1744: The first @code{)} ends a comment started with @code{(}, so you cannot
1745: nest @code{(}-comments; and you cannot comment out text containing a
1746: @code{)} with @code{( ... )}@footnote{therefore it's a good idea to
1747: avoid @code{)} in word names.}.
1748:
1749: I use @code{\}-comments for descriptive text and for commenting out code
1750: of one or more line; I use @code{(}-comments for describing the stack
1751: effect, the stack contents, or for commenting out sub-line pieces of
1752: code.
1753:
1754: The Emacs mode @file{gforth.el} (@pxref{Emacs and Gforth}) supports
1755: these uses by commenting out a region with @kbd{C-x \}, uncommenting a
1756: region with @kbd{C-u C-x \}, and filling a @code{\}-commented region
1757: with @kbd{M-q}.
1758:
1759: Reference: @ref{Comments}.
1760:
1761:
1762: @node Colon Definitions Tutorial, Decompilation Tutorial, Comments Tutorial, Tutorial
1763: @section Colon Definitions
1764: @cindex colon definitions, tutorial
1765: @cindex definitions, tutorial
1766: @cindex procedures, tutorial
1767: @cindex functions, tutorial
1768:
1769: are similar to procedures and functions in other programming languages.
1770:
1771: @example
1772: : squared ( n -- n^2 )
1773: dup * ;
1774: 5 squared .
1775: 7 squared .
1776: @end example
1777:
1778: @code{:} starts the colon definition; its name is @code{squared}. The
1779: following comment describes its stack effect. The words @code{dup *}
1780: are not executed, but compiled into the definition. @code{;} ends the
1781: colon definition.
1782:
1783: The newly-defined word can be used like any other word, including using
1784: it in other definitions:
1785:
1786: @example
1787: : cubed ( n -- n^3 )
1788: dup squared * ;
1789: -5 cubed .
1790: : fourth-power ( n -- n^4 )
1791: squared squared ;
1792: 3 fourth-power .
1793: @end example
1794:
1795: @assignment
1796: Write colon definitions for @code{nip}, @code{tuck}, @code{negate}, and
1797: @code{/mod} in terms of other Forth words, and check if they work (hint:
1798: test your tests on the originals first). Don't let the
1799: @samp{redefined}-Messages spook you, they are just warnings.
1800: @endassignment
1801:
1802: Reference: @ref{Colon Definitions}.
1803:
1804:
1805: @node Decompilation Tutorial, Stack-Effect Comments Tutorial, Colon Definitions Tutorial, Tutorial
1806: @section Decompilation
1807: @cindex decompilation tutorial
1808: @cindex see tutorial
1809:
1810: You can decompile colon definitions with @code{see}:
1811:
1812: @example
1813: see squared
1814: see cubed
1815: @end example
1816:
1817: In Gforth @code{see} shows you a reconstruction of the source code from
1818: the executable code. Informations that were present in the source, but
1819: not in the executable code, are lost (e.g., comments).
1820:
1821: You can also decompile the predefined words:
1822:
1823: @example
1824: see .
1825: see +
1826: @end example
1827:
1828:
1829: @node Stack-Effect Comments Tutorial, Types Tutorial, Decompilation Tutorial, Tutorial
1830: @section Stack-Effect Comments
1831: @cindex stack-effect comments, tutorial
1832: @cindex --, tutorial
1833: By convention the comment after the name of a definition describes the
1834: stack effect: The part in from of the @samp{--} describes the state of
1835: the stack before the execution of the definition, i.e., the parameters
1836: that are passed into the colon definition; the part behind the @samp{--}
1837: is the state of the stack after the execution of the definition, i.e.,
1838: the results of the definition. The stack comment only shows the top
1839: stack items that the definition accesses and/or changes.
1840:
1841: You should put a correct stack effect on every definition, even if it is
1842: just @code{( -- )}. You should also add some descriptive comment to
1843: more complicated words (I usually do this in the lines following
1844: @code{:}). If you don't do this, your code becomes unreadable (because
1845: you have to work through every definition before you can undertsand
1846: any).
1847:
1848: @assignment
1849: The stack effect of @code{swap} can be written like this: @code{x1 x2 --
1850: x2 x1}. Describe the stack effect of @code{-}, @code{drop}, @code{dup},
1851: @code{over}, @code{rot}, @code{nip}, and @code{tuck}. Hint: When you
1852: are done, you can compare your stack effects to those in this manual
1853: (@pxref{Word Index}).
1854: @endassignment
1855:
1856: Sometimes programmers put comments at various places in colon
1857: definitions that describe the contents of the stack at that place (stack
1858: comments); i.e., they are like the first part of a stack-effect
1859: comment. E.g.,
1860:
1861: @example
1862: : cubed ( n -- n^3 )
1863: dup squared ( n n^2 ) * ;
1864: @end example
1865:
1866: In this case the stack comment is pretty superfluous, because the word
1867: is simple enough. If you think it would be a good idea to add such a
1868: comment to increase readability, you should also consider factoring the
1869: word into several simpler words (@pxref{Factoring Tutorial,,
1870: Factoring}), which typically eliminates the need for the stack comment;
1871: however, if you decide not to refactor it, then having such a comment is
1872: better than not having it.
1873:
1874: The names of the stack items in stack-effect and stack comments in the
1875: standard, in this manual, and in many programs specify the type through
1876: a type prefix, similar to Fortran and Hungarian notation. The most
1877: frequent prefixes are:
1878:
1879: @table @code
1880: @item n
1881: signed integer
1882: @item u
1883: unsigned integer
1884: @item c
1885: character
1886: @item f
1887: Boolean flags, i.e. @code{false} or @code{true}.
1888: @item a-addr,a-
1889: Cell-aligned address
1890: @item c-addr,c-
1891: Char-aligned address (note that a Char may have two bytes in Windows NT)
1892: @item xt
1893: Execution token, same size as Cell
1894: @item w,x
1895: Cell, can contain an integer or an address. It usually takes 32, 64 or
1896: 16 bits (depending on your platform and Forth system). A cell is more
1897: commonly known as machine word, but the term @emph{word} already means
1898: something different in Forth.
1899: @item d
1900: signed double-cell integer
1901: @item ud
1902: unsigned double-cell integer
1903: @item r
1904: Float (on the FP stack)
1905: @end table
1906:
1907: You can find a more complete list in @ref{Notation}.
1908:
1909: @assignment
1910: Write stack-effect comments for all definitions you have written up to
1911: now.
1912: @endassignment
1913:
1914:
1915: @node Types Tutorial, Factoring Tutorial, Stack-Effect Comments Tutorial, Tutorial
1916: @section Types
1917: @cindex types tutorial
1918:
1919: In Forth the names of the operations are not overloaded; so similar
1920: operations on different types need different names; e.g., @code{+} adds
1921: integers, and you have to use @code{f+} to add floating-point numbers.
1922: The following prefixes are often used for related operations on
1923: different types:
1924:
1925: @table @code
1926: @item (none)
1927: signed integer
1928: @item u
1929: unsigned integer
1930: @item c
1931: character
1932: @item d
1933: signed double-cell integer
1934: @item ud, du
1935: unsigned double-cell integer
1936: @item 2
1937: two cells (not-necessarily double-cell numbers)
1938: @item m, um
1939: mixed single-cell and double-cell operations
1940: @item f
1941: floating-point (note that in stack comments @samp{f} represents flags,
1942: and @samp{r} represents FP numbers).
1943: @end table
1944:
1945: If there are no differences between the signed and the unsigned variant
1946: (e.g., for @code{+}), there is only the prefix-less variant.
1947:
1948: Forth does not perform type checking, neither at compile time, nor at
1949: run time. If you use the wrong oeration, the data are interpreted
1950: incorrectly:
1951:
1952: @example
1953: -1 u.
1954: @end example
1955:
1956: If you have only experience with type-checked languages until now, and
1957: have heard how important type-checking is, don't panic! In my
1958: experience (and that of other Forthers), type errors in Forth code are
1959: usually easy to find (once you get used to it), the increased vigilance
1960: of the programmer tends to catch some harder errors in addition to most
1961: type errors, and you never have to work around the type system, so in
1962: most situations the lack of type-checking seems to be a win (projects to
1963: add type checking to Forth have not caught on).
1964:
1965:
1966: @node Factoring Tutorial, Designing the stack effect Tutorial, Types Tutorial, Tutorial
1967: @section Factoring
1968: @cindex factoring tutorial
1969:
1970: If you try to write longer definitions, you will soon find it hard to
1971: keep track of the stack contents. Therefore, good Forth programmers
1972: tend to write only short definitions (e.g., three lines). The art of
1973: finding meaningful short definitions is known as factoring (as in
1974: factoring polynomials).
1975:
1976: Well-factored programs offer additional advantages: smaller, more
1977: general words, are easier to test and debug and can be reused more and
1978: better than larger, specialized words.
1979:
1980: So, if you run into difficulties with stack management, when writing
1981: code, try to define meaningful factors for the word, and define the word
1982: in terms of those. Even if a factor contains only two words, it is
1983: often helpful.
1984:
1985: Good factoring is not easy, and it takes some practice to get the knack
1986: for it; but even experienced Forth programmers often don't find the
1987: right solution right away, but only when rewriting the program. So, if
1988: you don't come up with a good solution immediately, keep trying, don't
1989: despair.
1990:
1991: @c example !!
1992:
1993:
1994: @node Designing the stack effect Tutorial, Local Variables Tutorial, Factoring Tutorial, Tutorial
1995: @section Designing the stack effect
1996: @cindex Stack effect design, tutorial
1997: @cindex design of stack effects, tutorial
1998:
1999: In other languages you can use an arbitrary order of parameters for a
2000: function; and since there is only one result, you don't have to deal with
2001: the order of results, either.
2002:
2003: In Forth (and other stack-based languages, e.g., Postscript) the
2004: parameter and result order of a definition is important and should be
2005: designed well. The general guideline is to design the stack effect such
2006: that the word is simple to use in most cases, even if that complicates
2007: the implementation of the word. Some concrete rules are:
2008:
2009: @itemize @bullet
2010:
2011: @item
2012: Words consume all of their parameters (e.g., @code{.}).
2013:
2014: @item
2015: If there is a convention on the order of parameters (e.g., from
2016: mathematics or another programming language), stick with it (e.g.,
2017: @code{-}).
2018:
2019: @item
2020: If one parameter usually requires only a short computation (e.g., it is
2021: a constant), pass it on the top of the stack. Conversely, parameters
2022: that usually require a long sequence of code to compute should be passed
2023: as the bottom (i.e., first) parameter. This makes the code easier to
2024: read, because reader does not need to keep track of the bottom item
2025: through a long sequence of code (or, alternatively, through stack
2026: manipulations). E.g., @code{!} (store, @pxref{Memory}) expects the
2027: address on top of the stack because it is usually simpler to compute
2028: than the stored value (often the address is just a variable).
2029:
2030: @item
2031: Similarly, results that are usually consumed quickly should be returned
2032: on the top of stack, whereas a result that is often used in long
2033: computations should be passed as bottom result. E.g., the file words
2034: like @code{open-file} return the error code on the top of stack, because
2035: it is usually consumed quickly by @code{throw}; moreover, the error code
2036: has to be checked before doing anything with the other results.
2037:
2038: @end itemize
2039:
2040: These rules are just general guidelines, don't lose sight of the overall
2041: goal to make the words easy to use. E.g., if the convention rule
2042: conflicts with the computation-length rule, you might decide in favour
2043: of the convention if the word will be used rarely, and in favour of the
2044: computation-length rule if the word will be used frequently (because
2045: with frequent use the cost of breaking the computation-length rule would
2046: be quite high, and frequent use makes it easier to remember an
2047: unconventional order).
2048:
2049: @c example !! structure package
2050:
2051:
2052: @node Local Variables Tutorial, Conditional execution Tutorial, Designing the stack effect Tutorial, Tutorial
2053: @section Local Variables
2054: @cindex local variables, tutorial
2055:
2056: You can define local variables (@emph{locals}) in a colon definition:
2057:
2058: @example
2059: : swap @{ a b -- b a @}
2060: b a ;
2061: 1 2 swap .s 2drop
2062: @end example
2063:
2064: (If your Forth system does not support this syntax, include
2065: @file{compat/anslocals.fs} first).
2066:
2067: In this example @code{@{ a b -- b a @}} is the locals definition; it
2068: takes two cells from the stack, puts the top of stack in @code{b} and
2069: the next stack element in @code{a}. @code{--} starts a comment ending
2070: with @code{@}}. After the locals definition, using the name of the
2071: local will push its value on the stack. You can leave the comment
2072: part (@code{-- b a}) away:
2073:
2074: @example
2075: : swap ( x1 x2 -- x2 x1 )
2076: @{ a b @} b a ;
2077: @end example
2078:
2079: In Gforth you can have several locals definitions, anywhere in a colon
2080: definition; in contrast, in a standard program you can have only one
2081: locals definition per colon definition, and that locals definition must
2082: be outside any controll structure.
2083:
2084: With locals you can write slightly longer definitions without running
2085: into stack trouble. However, I recommend trying to write colon
2086: definitions without locals for exercise purposes to help you gain the
2087: essential factoring skills.
2088:
2089: @assignment
2090: Rewrite your definitions until now with locals
2091: @endassignment
2092:
2093: Reference: @ref{Locals}.
2094:
2095:
2096: @node Conditional execution Tutorial, Flags and Comparisons Tutorial, Local Variables Tutorial, Tutorial
2097: @section Conditional execution
2098: @cindex conditionals, tutorial
2099: @cindex if, tutorial
2100:
2101: In Forth you can use control structures only inside colon definitions.
2102: An @code{if}-structure looks like this:
2103:
2104: @example
2105: : abs ( n1 -- +n2 )
2106: dup 0 < if
2107: negate
2108: endif ;
2109: 5 abs .
2110: -5 abs .
2111: @end example
2112:
2113: @code{if} takes a flag from the stack. If the flag is non-zero (true),
2114: the following code is performed, otherwise execution continues after the
2115: @code{endif} (or @code{else}). @code{<} compares the top two stack
2116: elements and prioduces a flag:
2117:
2118: @example
2119: 1 2 < .
2120: 2 1 < .
2121: 1 1 < .
2122: @end example
2123:
2124: Actually the standard name for @code{endif} is @code{then}. This
2125: tutorial presents the examples using @code{endif}, because this is often
2126: less confusing for people familiar with other programming languages
2127: where @code{then} has a different meaning. If your system does not have
2128: @code{endif}, define it with
2129:
2130: @example
2131: : endif postpone then ; immediate
2132: @end example
2133:
2134: You can optionally use an @code{else}-part:
2135:
2136: @example
2137: : min ( n1 n2 -- n )
2138: 2dup < if
2139: drop
2140: else
2141: nip
2142: endif ;
2143: 2 3 min .
2144: 3 2 min .
2145: @end example
2146:
2147: @assignment
2148: Write @code{min} without @code{else}-part (hint: what's the definition
2149: of @code{nip}?).
2150: @endassignment
2151:
2152: Reference: @ref{Selection}.
2153:
2154:
2155: @node Flags and Comparisons Tutorial, General Loops Tutorial, Conditional execution Tutorial, Tutorial
2156: @section Flags and Comparisons
2157: @cindex flags tutorial
2158: @cindex comparison tutorial
2159:
2160: In a false-flag all bits are clear (0 when interpreted as integer). In
2161: a canonical true-flag all bits are set (-1 as a twos-complement signed
2162: integer); in many contexts (e.g., @code{if}) any non-zero value is
2163: treated as true flag.
2164:
2165: @example
2166: false .
2167: true .
2168: true hex u. decimal
2169: @end example
2170:
2171: Comparison words produce canonical flags:
2172:
2173: @example
2174: 1 1 = .
2175: 1 0= .
2176: 0 1 < .
2177: 0 0 < .
2178: -1 1 u< . \ type error, u< interprets -1 as large unsigned number
2179: -1 1 < .
2180: @end example
2181:
2182: Gforth supports all combinations of the prefixes @code{0 u d d0 du f f0}
2183: (or none) and the comparisons @code{= <> < > <= >=}. Only a part of
2184: these combinations are standard (for details see the standard,
2185: @ref{Numeric comparison}, @ref{Floating Point} or @ref{Word Index}).
2186:
2187: You can use @code{and or xor invert} can be used as operations on
2188: canonical flags. Actually they are bitwise operations:
2189:
2190: @example
2191: 1 2 and .
2192: 1 2 or .
2193: 1 3 xor .
2194: 1 invert .
2195: @end example
2196:
2197: You can convert a zero/non-zero flag into a canonical flag with
2198: @code{0<>} (and complement it on the way with @code{0=}).
2199:
2200: @example
2201: 1 0= .
2202: 1 0<> .
2203: @end example
2204:
2205: You can use the all-bits-set feature of canonical flags and the bitwise
2206: operation of the Boolean operations to avoid @code{if}s:
2207:
2208: @example
2209: : foo ( n1 -- n2 )
2210: 0= if
2211: 14
2212: else
2213: 0
2214: endif ;
2215: 0 foo .
2216: 1 foo .
2217:
2218: : foo ( n1 -- n2 )
2219: 0= 14 and ;
2220: 0 foo .
2221: 1 foo .
2222: @end example
2223:
2224: @assignment
2225: Write @code{min} without @code{if}.
2226: @endassignment
2227:
2228: For reference, see @ref{Boolean Flags}, @ref{Numeric comparison}, and
2229: @ref{Bitwise operations}.
2230:
2231:
2232: @node General Loops Tutorial, Counted loops Tutorial, Flags and Comparisons Tutorial, Tutorial
2233: @section General Loops
2234: @cindex loops, indefinite, tutorial
2235:
2236: The endless loop is the most simple one:
2237:
2238: @example
2239: : endless ( -- )
2240: 0 begin
2241: dup . 1+
2242: again ;
2243: endless
2244: @end example
2245:
2246: Terminate this loop by pressing @kbd{Ctrl-C} (in Gforth). @code{begin}
2247: does nothing at run-time, @code{again} jumps back to @code{begin}.
2248:
2249: A loop with one exit at any place looks like this:
2250:
2251: @example
2252: : log2 ( +n1 -- n2 )
2253: \ logarithmus dualis of n1>0, rounded down to the next integer
2254: assert( dup 0> )
2255: 2/ 0 begin
2256: over 0> while
2257: 1+ swap 2/ swap
2258: repeat
2259: nip ;
2260: 7 log2 .
2261: 8 log2 .
2262: @end example
2263:
2264: At run-time @code{while} consumes a flag; if it is 0, execution
2265: continues behind the @code{repeat}; if the flag is non-zero, execution
2266: continues behind the @code{while}. @code{Repeat} jumps back to
2267: @code{begin}, just like @code{again}.
2268:
2269: In Forth there are many combinations/abbreviations, like @code{1+}.
2270: However, @code{2/} is not one of them; it shifts it's argument right by
2271: one bit (arithmetic shift right):
2272:
2273: @example
2274: -5 2 / .
2275: -5 2/ .
2276: @end example
2277:
2278: @code{assert(} is no standard word, but you can get it on systems other
2279: then Gforth by including @file{compat/assert.fs}. You can see what it
2280: does by trying
2281:
2282: @example
2283: 0 log2 .
2284: @end example
2285:
2286: Here's a loop with an exit at the end:
2287:
2288: @example
2289: : log2 ( +n1 -- n2 )
2290: \ logarithmus dualis of n1>0, rounded down to the next integer
2291: assert( dup 0 > )
2292: -1 begin
2293: 1+ swap 2/ swap
2294: over 0 <=
2295: until
2296: nip ;
2297: @end example
2298:
2299: @code{Until} consumes a flag; if it is non-zero, execution continues at
2300: the @code{begin}, otherwise after the @code{until}.
2301:
2302: @assignment
2303: Write a definition for computing the greatest common divisor.
2304: @endassignment
2305:
2306: Reference: @ref{Simple Loops}.
2307:
2308:
2309: @node Counted loops Tutorial, Recursion Tutorial, General Loops Tutorial, Tutorial
2310: @section Counted loops
2311: @cindex loops, counted, tutorial
2312:
2313: @example
2314: : ^ ( n1 u -- n )
2315: \ n = the uth power of u1
2316: 1 swap 0 u+do
2317: over *
2318: loop
2319: nip ;
2320: 3 2 ^ .
2321: 4 3 ^ .
2322: @end example
2323:
2324: @code{U+do} (from @file{compat/loops.fs}, if your Forth system doesn't
2325: have it) takes two numbers of the stack @code{( u3 u4 -- )}, and then
2326: performs the code between @code{u+do} and @code{loop} for @code{u3-u4}
2327: times (or not at all, if @code{u3-u4<0}).
2328:
2329: You can see the stack effect design rules at work in the stack effect of
2330: the loop start words: Since the start value of the loop is more
2331: frequently constant than the end value, the start value is passed on
2332: the top-of-stack.
2333:
2334: You can access the counter of a counted loop with @code{i}:
2335:
2336: @example
2337: : fac ( u -- u! )
2338: 1 swap 1+ 1 u+do
2339: i *
2340: loop ;
2341: 5 fac .
2342: 7 fac .
2343: @end example
2344:
2345: There is also @code{+do}, which expects signed numbers (important for
2346: deciding whether to enter the loop).
2347:
2348: @assignment
2349: Write a definition for computing the nth Fibonacci number.
2350: @endassignment
2351:
2352: You can also use increments other than 1:
2353:
2354: @example
2355: : up2 ( n1 n2 -- )
2356: +do
2357: i .
2358: 2 +loop ;
2359: 10 0 up2
2360:
2361: : down2 ( n1 n2 -- )
2362: -do
2363: i .
2364: 2 -loop ;
2365: 0 10 down2
2366: @end example
2367:
2368: Reference: @ref{Counted Loops}.
2369:
2370:
2371: @node Recursion Tutorial, Leaving definitions or loops Tutorial, Counted loops Tutorial, Tutorial
2372: @section Recursion
2373: @cindex recursion tutorial
2374:
2375: Usually the name of a definition is not visible in the definition; but
2376: earlier definitions are usually visible:
2377:
2378: @example
2379: 1 0 / . \ "Floating-point unidentified fault" in Gforth on most platforms
2380: : / ( n1 n2 -- n )
2381: dup 0= if
2382: -10 throw \ report division by zero
2383: endif
2384: / \ old version
2385: ;
2386: 1 0 /
2387: @end example
2388:
2389: For recursive definitions you can use @code{recursive} (non-standard) or
2390: @code{recurse}:
2391:
2392: @example
2393: : fac1 ( n -- n! ) recursive
2394: dup 0> if
2395: dup 1- fac1 *
2396: else
2397: drop 1
2398: endif ;
2399: 7 fac1 .
2400:
2401: : fac2 ( n -- n! )
2402: dup 0> if
2403: dup 1- recurse *
2404: else
2405: drop 1
2406: endif ;
2407: 8 fac2 .
2408: @end example
2409:
2410: @assignment
2411: Write a recursive definition for computing the nth Fibonacci number.
2412: @endassignment
2413:
2414: Reference (including indirect recursion): @xref{Calls and returns}.
2415:
2416:
2417: @node Leaving definitions or loops Tutorial, Return Stack Tutorial, Recursion Tutorial, Tutorial
2418: @section Leaving definitions or loops
2419: @cindex leaving definitions, tutorial
2420: @cindex leaving loops, tutorial
2421:
2422: @code{EXIT} exits the current definition right away. For every counted
2423: loop that is left in this way, an @code{UNLOOP} has to be performed
2424: before the @code{EXIT}:
2425:
2426: @c !! real examples
2427: @example
2428: : ...
2429: ... u+do
2430: ... if
2431: ... unloop exit
2432: endif
2433: ...
2434: loop
2435: ... ;
2436: @end example
2437:
2438: @code{LEAVE} leaves the innermost counted loop right away:
2439:
2440: @example
2441: : ...
2442: ... u+do
2443: ... if
2444: ... leave
2445: endif
2446: ...
2447: loop
2448: ... ;
2449: @end example
2450:
2451: @c !! example
2452:
2453: Reference: @ref{Calls and returns}, @ref{Counted Loops}.
2454:
2455:
2456: @node Return Stack Tutorial, Memory Tutorial, Leaving definitions or loops Tutorial, Tutorial
2457: @section Return Stack
2458: @cindex return stack tutorial
2459:
2460: In addition to the data stack Forth also has a second stack, the return
2461: stack; most Forth systems store the return addresses of procedure calls
2462: there (thus its name). Programmers can also use this stack:
2463:
2464: @example
2465: : foo ( n1 n2 -- )
2466: .s
2467: >r .s
2468: r@@ .
2469: >r .s
2470: r@@ .
2471: r> .
2472: r@@ .
2473: r> . ;
2474: 1 2 foo
2475: @end example
2476:
2477: @code{>r} takes an element from the data stack and pushes it onto the
2478: return stack; conversely, @code{r>} moves an elementm from the return to
2479: the data stack; @code{r@@} pushes a copy of the top of the return stack
2480: on the return stack.
2481:
2482: Forth programmers usually use the return stack for storing data
2483: temporarily, if using the data stack alone would be too complex, and
2484: factoring and locals are not an option:
2485:
2486: @example
2487: : 2swap ( x1 x2 x3 x4 -- x3 x4 x1 x2 )
2488: rot >r rot r> ;
2489: @end example
2490:
2491: The return address of the definition and the loop control parameters of
2492: counted loops usually reside on the return stack, so you have to take
2493: all items, that you have pushed on the return stack in a colon
2494: definition or counted loop, from the return stack before the definition
2495: or loop ends. You cannot access items that you pushed on the return
2496: stack outside some definition or loop within the definition of loop.
2497:
2498: If you miscount the return stack items, this usually ends in a crash:
2499:
2500: @example
2501: : crash ( n -- )
2502: >r ;
2503: 5 crash
2504: @end example
2505:
2506: You cannot mix using locals and using the return stack (according to the
2507: standard; Gforth has no problem). However, they solve the same
2508: problems, so this shouldn't be an issue.
2509:
2510: @assignment
2511: Can you rewrite any of the definitions you wrote until now in a better
2512: way using the return stack?
2513: @endassignment
2514:
2515: Reference: @ref{Return stack}.
2516:
2517:
2518: @node Memory Tutorial, Characters and Strings Tutorial, Return Stack Tutorial, Tutorial
2519: @section Memory
2520: @cindex memory access/allocation tutorial
2521:
2522: You can create a global variable @code{v} with
2523:
2524: @example
2525: variable v ( -- addr )
2526: @end example
2527:
2528: @code{v} pushes the address of a cell in memory on the stack. This cell
2529: was reserved by @code{variable}. You can use @code{!} (store) to store
2530: values into this cell and @code{@@} (fetch) to load the value from the
2531: stack into memory:
2532:
2533: @example
2534: v .
2535: 5 v ! .s
2536: v @@ .
2537: @end example
2538:
2539: You can see a raw dump of memory with @code{dump}:
2540:
2541: @example
2542: v 1 cells .s dump
2543: @end example
2544:
2545: @code{Cells ( n1 -- n2 )} gives you the number of bytes (or, more
2546: generally, address units (aus)) that @code{n1 cells} occupy. You can
2547: also reserve more memory:
2548:
2549: @example
2550: create v2 20 cells allot
2551: v2 20 cells dump
2552: @end example
2553:
2554: creates a word @code{v2} and reserves 20 uninitialized cells; the
2555: address pushed by @code{v2} points to the start of these 20 cells. You
2556: can use address arithmetic to access these cells:
2557:
2558: @example
2559: 3 v2 5 cells + !
2560: v2 20 cells dump
2561: @end example
2562:
2563: You can reserve and initialize memory with @code{,}:
2564:
2565: @example
2566: create v3
2567: 5 , 4 , 3 , 2 , 1 ,
2568: v3 @@ .
2569: v3 cell+ @@ .
2570: v3 2 cells + @@ .
2571: v3 5 cells dump
2572: @end example
2573:
2574: @assignment
2575: Write a definition @code{vsum ( addr u -- n )} that computes the sum of
2576: @code{u} cells, with the first of these cells at @code{addr}, the next
2577: one at @code{addr cell+} etc.
2578: @endassignment
2579:
2580: You can also reserve memory without creating a new word:
2581:
2582: @example
2583: here 10 cells allot .
2584: here .
2585: @end example
2586:
2587: @code{Here} pushes the start address of the memory area. You should
2588: store it somewhere, or you will have a hard time finding the memory area
2589: again.
2590:
2591: @code{Allot} manages dictionary memory. The dictionary memory contains
2592: the system's data structures for words etc. on Gforth and most other
2593: Forth systems. It is managed like a stack: You can free the memory that
2594: you have just @code{allot}ed with
2595:
2596: @example
2597: -10 cells allot
2598: here .
2599: @end example
2600:
2601: Note that you cannot do this if you have created a new word in the
2602: meantime (because then your @code{allot}ed memory is no longer on the
2603: top of the dictionary ``stack'').
2604:
2605: Alternatively, you can use @code{allocate} and @code{free} which allow
2606: freeing memory in any order:
2607:
2608: @example
2609: 10 cells allocate throw .s
2610: 20 cells allocate throw .s
2611: swap
2612: free throw
2613: free throw
2614: @end example
2615:
2616: The @code{throw}s deal with errors (e.g., out of memory).
2617:
2618: And there is also a
2619: @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
2620: garbage collector}, which eliminates the need to @code{free} memory
2621: explicitly.
2622:
2623: Reference: @ref{Memory}.
2624:
2625:
2626: @node Characters and Strings Tutorial, Alignment Tutorial, Memory Tutorial, Tutorial
2627: @section Characters and Strings
2628: @cindex strings tutorial
2629: @cindex characters tutorial
2630:
2631: On the stack characters take up a cell, like numbers. In memory they
2632: have their own size (one 8-bit byte on most systems), and therefore
2633: require their own words for memory access:
2634:
2635: @example
2636: create v4
2637: 104 c, 97 c, 108 c, 108 c, 111 c,
2638: v4 4 chars + c@@ .
2639: v4 5 chars dump
2640: @end example
2641:
2642: The preferred representation of strings on the stack is @code{addr
2643: u-count}, where @code{addr} is the address of the first character and
2644: @code{u-count} is the number of characters in the string.
2645:
2646: @example
2647: v4 5 type
2648: @end example
2649:
2650: You get a string constant with
2651:
2652: @example
2653: s" hello, world" .s
2654: type
2655: @end example
2656:
2657: Make sure you have a space between @code{s"} and the string; @code{s"}
2658: is a normal Forth word and must be delimited with white space (try what
2659: happens when you remove the space).
2660:
2661: However, this interpretive use of @code{s"} is quite restricted: the
2662: string exists only until the next call of @code{s"} (some Forth systems
2663: keep more than one of these strings, but usually they still have a
2664: limited lifetime).
2665:
2666: @example
2667: s" hello," s" world" .s
2668: type
2669: type
2670: @end example
2671:
2672: You can also use @code{s"} in a definition, and the resulting
2673: strings then live forever (well, for as long as the definition):
2674:
2675: @example
2676: : foo s" hello," s" world" ;
2677: foo .s
2678: type
2679: type
2680: @end example
2681:
2682: @assignment
2683: @code{Emit ( c -- )} types @code{c} as character (not a number).
2684: Implement @code{type ( addr u -- )}.
2685: @endassignment
2686:
2687: Reference: @ref{Memory Blocks}.
2688:
2689:
2690: @node Alignment Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Characters and Strings Tutorial, Tutorial
2691: @section Alignment
2692: @cindex alignment tutorial
2693: @cindex memory alignment tutorial
2694:
2695: On many processors cells have to be aligned in memory, if you want to
2696: access them with @code{@@} and @code{!} (and even if the processor does
2697: not require alignment, access to aligned cells is faster).
2698:
2699: @code{Create} aligns @code{here} (i.e., the place where the next
2700: allocation will occur, and that the @code{create}d word points to).
2701: Likewise, the memory produced by @code{allocate} starts at an aligned
2702: address. Adding a number of @code{cells} to an aligned address produces
2703: another aligned address.
2704:
2705: However, address arithmetic involving @code{char+} and @code{chars} can
2706: create an address that is not cell-aligned. @code{Aligned ( addr --
2707: a-addr )} produces the next aligned address:
2708:
2709: @example
2710: v3 char+ aligned .s @@ .
2711: v3 char+ .s @@ .
2712: @end example
2713:
2714: Similarly, @code{align} advances @code{here} to the next aligned
2715: address:
2716:
2717: @example
2718: create v5 97 c,
2719: here .
2720: align here .
2721: 1000 ,
2722: @end example
2723:
2724: Note that you should use aligned addresses even if your processor does
2725: not require them, if you want your program to be portable.
2726:
2727: Reference: @ref{Address arithmetic}.
2728:
2729:
2730: @node Interpretation and Compilation Semantics and Immediacy Tutorial, Execution Tokens Tutorial, Alignment Tutorial, Tutorial
2731: @section Interpretation and Compilation Semantics and Immediacy
2732: @cindex semantics tutorial
2733: @cindex interpretation semantics tutorial
2734: @cindex compilation semantics tutorial
2735: @cindex immediate, tutorial
2736:
2737: When a word is compiled, it behaves differently from being interpreted.
2738: E.g., consider @code{+}:
2739:
2740: @example
2741: 1 2 + .
2742: : foo + ;
2743: @end example
2744:
2745: These two behaviours are known as compilation and interpretation
2746: semantics. For normal words (e.g., @code{+}), the compilation semantics
2747: is to append the interpretation semantics to the currently defined word
2748: (@code{foo} in the example above). I.e., when @code{foo} is executed
2749: later, the interpretation semantics of @code{+} (i.e., adding two
2750: numbers) will be performed.
2751:
2752: However, there are words with non-default compilation semantics, e.g.,
2753: the control-flow words like @code{if}. You can use @code{immediate} to
2754: change the compilation semantics of the last defined word to be equal to
2755: the interpretation semantics:
2756:
2757: @example
2758: : [FOO] ( -- )
2759: 5 . ; immediate
2760:
2761: [FOO]
2762: : bar ( -- )
2763: [FOO] ;
2764: bar
2765: see bar
2766: @end example
2767:
2768: Two conventions to mark words with non-default compilation semnatics are
2769: names with brackets (more frequently used) and to write them all in
2770: upper case (less frequently used).
2771:
2772: In Gforth (and many other systems) you can also remove the
2773: interpretation semantics with @code{compile-only} (the compilation
2774: semantics is derived from the original interpretation semantics):
2775:
2776: @example
2777: : flip ( -- )
2778: 6 . ; compile-only \ but not immediate
2779: flip
2780:
2781: : flop ( -- )
2782: flip ;
2783: flop
2784: @end example
2785:
2786: In this example the interpretation semantics of @code{flop} is equal to
2787: the original interpretation semantics of @code{flip}.
2788:
2789: The text interpreter has two states: in interpret state, it performs the
2790: interpretation semantics of words it encounters; in compile state, it
2791: performs the compilation semantics of these words.
2792:
2793: Among other things, @code{:} switches into compile state, and @code{;}
2794: switches back to interpret state. They contain the factors @code{]}
2795: (switch to compile state) and @code{[} (switch to interpret state), that
2796: do nothing but switch the state.
2797:
2798: @example
2799: : xxx ( -- )
2800: [ 5 . ]
2801: ;
2802:
2803: xxx
2804: see xxx
2805: @end example
2806:
2807: These brackets are also the source of the naming convention mentioned
2808: above.
2809:
2810: Reference: @ref{Interpretation and Compilation Semantics}.
2811:
2812:
2813: @node Execution Tokens Tutorial, Exceptions Tutorial, Interpretation and Compilation Semantics and Immediacy Tutorial, Tutorial
2814: @section Execution Tokens
2815: @cindex execution tokens tutorial
2816: @cindex XT tutorial
2817:
2818: @code{' word} gives you the execution token (XT) of a word. The XT is a
2819: cell representing the interpretation semantics of a word. You can
2820: execute this semantics with @code{execute}:
2821:
2822: @example
2823: ' + .s
2824: 1 2 rot execute .
2825: @end example
2826:
2827: The XT is similar to a function pointer in C. However, parameter
2828: passing through the stack makes it a little more flexible:
2829:
2830: @example
2831: : map-array ( ... addr u xt -- ... )
2832: \ executes xt ( ... x -- ... ) for every element of the array starting
2833: \ at addr and containing u elements
2834: @{ xt @}
2835: cells over + swap ?do
2836: i @@ xt execute
2837: 1 cells +loop ;
2838:
2839: create a 3 , 4 , 2 , -1 , 4 ,
2840: a 5 ' . map-array .s
2841: 0 a 5 ' + map-array .
2842: s" max-n" environment? drop .s
2843: a 5 ' min map-array .
2844: @end example
2845:
2846: You can use map-array with the XTs of words that consume one element
2847: more than they produce. In theory you can also use it with other XTs,
2848: but the stack effect then depends on the size of the array, which is
2849: hard to understand.
2850:
2851: Since XTs are cell-sized, you can store them in memory and manipulate
2852: them on the stack like other cells. You can also compile the XT into a
2853: word with @code{compile,}:
2854:
2855: @example
2856: : foo1 ( n1 n2 -- n )
2857: [ ' + compile, ] ;
2858: see foo
2859: @end example
2860:
2861: This is non-standard, because @code{compile,} has no compilation
2862: semantics in the standard, but it works in good Forth systems. For the
2863: broken ones, use
2864:
2865: @example
2866: : [compile,] compile, ; immediate
2867:
2868: : foo1 ( n1 n2 -- n )
2869: [ ' + ] [compile,] ;
2870: see foo
2871: @end example
2872:
2873: @code{'} is a word with default compilation semantics; it parses the
2874: next word when its interpretation semantics are executed, not during
2875: compilation:
2876:
2877: @example
2878: : foo ( -- xt )
2879: ' ;
2880: see foo
2881: : bar ( ... "word" -- ... )
2882: ' execute ;
2883: see bar
2884: 1 2 bar + .
2885: @end example
2886:
2887: You often want to parse a word during compilation and compile its XT so
2888: it will be pushed on the stack at run-time. @code{[']} does this:
2889:
2890: @example
2891: : xt-+ ( -- xt )
2892: ['] + ;
2893: see xt-+
2894: 1 2 xt-+ execute .
2895: @end example
2896:
2897: Many programmers tend to see @code{'} and the word it parses as one
2898: unit, and expect it to behave like @code{[']} when compiled, and are
2899: confused by the actual behaviour. If you are, just remember that the
2900: Forth system just takes @code{'} as one unit and has no idea that it is
2901: a parsing word (attempts to convenience programmers in this issue have
2902: usually resulted in even worse pitfalls, see
2903: @uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,
2904: @code{State}-smartness---Why it is evil and How to Exorcise it}).
2905:
2906: Note that the state of the interpreter does not come into play when
2907: creating and executing XTs. I.e., even when you execute @code{'} in
2908: compile state, it still gives you the interpretation semantics. And
2909: whatever that state is, @code{execute} performs the semantics
2910: represented by the XT (i.e., for XTs produced with @code{'} the
2911: interpretation semantics).
2912:
2913: Reference: @ref{Tokens for Words}.
2914:
2915:
2916: @node Exceptions Tutorial, Defining Words Tutorial, Execution Tokens Tutorial, Tutorial
2917: @section Exceptions
2918: @cindex exceptions tutorial
2919:
2920: @code{throw ( n -- )} causes an exception unless n is zero.
2921:
2922: @example
2923: 100 throw .s
2924: 0 throw .s
2925: @end example
2926:
2927: @code{catch ( ... xt -- ... n )} behaves similar to @code{execute}, but
2928: it catches exceptions and pushes the number of the exception on the
2929: stack (or 0, if the xt executed without exception). If there was an
2930: exception, the stacks have the same depth as when entering @code{catch}:
2931:
2932: @example
2933: .s
2934: 3 0 ' / catch .s
2935: 3 2 ' / catch .s
2936: @end example
2937:
2938: @assignment
2939: Try the same with @code{execute} instead of @code{catch}.
2940: @endassignment
2941:
2942: @code{Throw} always jumps to the dynamically next enclosing
2943: @code{catch}, even if it has to leave several call levels to achieve
2944: this:
2945:
2946: @example
2947: : foo 100 throw ;
2948: : foo1 foo ." after foo" ;
2949: : bar ['] foo1 catch ;
2950: bar .
2951: @end example
2952:
2953: It is often important to restore a value upon leaving a definition, even
2954: if the definition is left through an exception. You can ensure this
2955: like this:
2956:
2957: @example
2958: : ...
2959: save-x
2960: ['] word-changing-x catch ( ... n )
2961: restore-x
2962: ( ... n ) throw ;
2963: @end example
2964:
2965: Gforth provides an alternative syntax in addition to @code{catch}:
2966: @code{try ... recover ... endtry}. If the code between @code{try} and
2967: @code{recover} has an exception, the stack depths are restored, the
2968: exception number is pushed on the stack, and the code between
2969: @code{recover} and @code{endtry} is performed. E.g., the definition for
2970: @code{catch} is
2971:
2972: @example
2973: : catch ( x1 .. xn xt -- y1 .. ym 0 / z1 .. zn error ) \ exception
2974: try
2975: execute 0
2976: recover
2977: nip
2978: endtry ;
2979: @end example
2980:
2981: The equivalent to the restoration code above is
2982:
2983: @example
2984: : ...
2985: save-x
2986: try
2987: word-changing-x
2988: end-try
2989: restore-x
2990: throw ;
2991: @end example
2992:
2993: As you can see, the @code{recover} part is optional.
2994:
2995: Reference: @ref{Exception Handling}.
2996:
2997:
2998: @node Defining Words Tutorial, Arrays and Records Tutorial, Exceptions Tutorial, Tutorial
2999: @section Defining Words
3000: @cindex defining words tutorial
3001: @cindex does> tutorial
3002: @cindex create...does> tutorial
3003:
3004: @c before semantics?
3005:
3006: @code{:}, @code{create}, and @code{variable} are definition words: They
3007: define other words. @code{Constant} is another definition word:
3008:
3009: @example
3010: 5 constant foo
3011: foo .
3012: @end example
3013:
3014: You can also use the prefixes @code{2} (double-cell) and @code{f}
3015: (floating point) with @code{variable} and @code{constant}.
3016:
3017: You can also define your own defining words. E.g.:
3018:
3019: @example
3020: : variable ( "name" -- )
3021: create 0 , ;
3022: @end example
3023:
3024: You can also define defining words that create words that do something
3025: other than just producing their address:
3026:
3027: @example
3028: : constant ( n "name" -- )
3029: create ,
3030: does> ( -- n )
3031: ( addr ) @@ ;
3032:
3033: 5 constant foo
3034: foo .
3035: @end example
3036:
3037: The definition of @code{constant} above ends at the @code{does>}; i.e.,
3038: @code{does>} replaces @code{;}, but it also does something else: It
3039: changes the last defined word such that it pushes the address of the
3040: body of the word and then performs the code after the @code{does>}
3041: whenever it is called.
3042:
3043: In the example above, @code{constant} uses @code{,} to store 5 into the
3044: body of @code{foo}. When @code{foo} executes, it pushes the address of
3045: the body onto the stack, then (in the code after the @code{does>})
3046: fetches the 5 from there.
3047:
3048: The stack comment near the @code{does>} reflects the stack effect of the
3049: defined word, not the stack effect of the code after the @code{does>}
3050: (the difference is that the code expects the address of the body that
3051: the stack comment does not show).
3052:
3053: You can use these definition words to do factoring in cases that involve
3054: (other) definition words. E.g., a field offset is always added to an
3055: address. Instead of defining
3056:
3057: @example
3058: 2 cells constant offset-field1
3059: @end example
3060:
3061: and using this like
3062:
3063: @example
3064: ( addr ) offset-field1 +
3065: @end example
3066:
3067: you can define a definition word
3068:
3069: @example
3070: : simple-field ( n "name" -- )
3071: create ,
3072: does> ( n1 -- n1+n )
3073: ( addr ) @@ + ;
3074: @end example
3075:
3076: Definition and use of field offsets now look like this:
3077:
3078: @example
3079: 2 cells simple-field field1
3080: create mystruct 4 cells allot
3081: mystruct .s field1 .s drop
3082: @end example
3083:
3084: If you want to do something with the word without performing the code
3085: after the @code{does>}, you can access the body of a @code{create}d word
3086: with @code{>body ( xt -- addr )}:
3087:
3088: @example
3089: : value ( n "name" -- )
3090: create ,
3091: does> ( -- n1 )
3092: @@ ;
3093: : to ( n "name" -- )
3094: ' >body ! ;
3095:
3096: 5 value foo
3097: foo .
3098: 7 to foo
3099: foo .
3100: @end example
3101:
3102: @assignment
3103: Define @code{defer ( "name" -- )}, which creates a word that stores an
3104: XT (at the start the XT of @code{abort}), and upon execution
3105: @code{execute}s the XT. Define @code{is ( xt "name" -- )} that stores
3106: @code{xt} into @code{name}, a word defined with @code{defer}. Indirect
3107: recursion is one application of @code{defer}.
3108: @endassignment
3109:
3110: Reference: @ref{User-defined Defining Words}.
3111:
3112:
3113: @node Arrays and Records Tutorial, POSTPONE Tutorial, Defining Words Tutorial, Tutorial
3114: @section Arrays and Records
3115: @cindex arrays tutorial
3116: @cindex records tutorial
3117: @cindex structs tutorial
3118:
3119: Forth has no standard words for defining data structures such as arrays
3120: and records (structs in C terminology), but you can build them yourself
3121: based on address arithmetic. You can also define words for defining
3122: arrays and records (@pxref{Defining Words Tutorial,, Defining Words}).
3123:
3124: One of the first projects a Forth newcomer sets out upon when learning
3125: about defining words is an array defining word (possibly for
3126: n-dimensional arrays). Go ahead and do it, I did it, too; you will
3127: learn something from it. However, don't be disappointed when you later
3128: learn that you have little use for these words (inappropriate use would
3129: be even worse). I have not yet found a set of useful array words yet;
3130: the needs are just too diverse, and named, global arrays (the result of
3131: naive use of defining words) are often not flexible enough (e.g.,
3132: consider how to pass them as parameters). Another such project is a set
3133: of words to help dealing with strings.
3134:
3135: On the other hand, there is a useful set of record words, and it has
3136: been defined in @file{compat/struct.fs}; these words are predefined in
3137: Gforth. They are explained in depth elsewhere in this manual (see
3138: @pxref{Structures}). The @code{simple-field} example above is
3139: simplified variant of fields in this package.
3140:
3141:
3142: @node POSTPONE Tutorial, Literal Tutorial, Arrays and Records Tutorial, Tutorial
3143: @section @code{POSTPONE}
3144: @cindex postpone tutorial
3145:
3146: You can compile the compilation semantics (instead of compiling the
3147: interpretation semantics) of a word with @code{POSTPONE}:
3148:
3149: @example
3150: : MY-+ ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
3151: POSTPONE + ; immediate
3152: : foo ( n1 n2 -- n )
3153: MY-+ ;
3154: 1 2 foo .
3155: see foo
3156: @end example
3157:
3158: During the definition of @code{foo} the text interpreter performs the
3159: compilation semantics of @code{MY-+}, which performs the compilation
3160: semantics of @code{+}, i.e., it compiles @code{+} into @code{foo}.
3161:
3162: This example also displays separate stack comments for the compilation
3163: semantics and for the stack effect of the compiled code. For words with
3164: default compilation semantics these stack effects are usually not
3165: displayed; the stack effect of the compilation semantics is always
3166: @code{( -- )} for these words, the stack effect for the compiled code is
3167: the stack effect of the interpretation semantics.
3168:
3169: Note that the state of the interpreter does not come into play when
3170: performing the compilation semantics in this way. You can also perform
3171: it interpretively, e.g.:
3172:
3173: @example
3174: : foo2 ( n1 n2 -- n )
3175: [ MY-+ ] ;
3176: 1 2 foo .
3177: see foo
3178: @end example
3179:
3180: However, there are some broken Forth systems where this does not always
3181: work, and therefore this practice was been declared non-standard in
3182: 1999.
3183: @c !! repair.fs
3184:
3185: Here is another example for using @code{POSTPONE}:
3186:
3187: @example
3188: : MY-- ( Compilation: -- ; Run-time of compiled code: n1 n2 -- n )
3189: POSTPONE negate POSTPONE + ; immediate compile-only
3190: : bar ( n1 n2 -- n )
3191: MY-- ;
3192: 2 1 bar .
3193: see bar
3194: @end example
3195:
3196: You can define @code{ENDIF} in this way:
3197:
3198: @example
3199: : ENDIF ( Compilation: orig -- )
3200: POSTPONE then ; immediate
3201: @end example
3202:
3203: @assignment
3204: Write @code{MY-2DUP} that has compilation semantics equivalent to
3205: @code{2dup}, but compiles @code{over over}.
3206: @endassignment
3207:
3208: @c !! @xref{Macros} for reference
3209:
3210:
3211: @node Literal Tutorial, Advanced macros Tutorial, POSTPONE Tutorial, Tutorial
3212: @section @code{Literal}
3213: @cindex literal tutorial
3214:
3215: You cannot @code{POSTPONE} numbers:
3216:
3217: @example
3218: : [FOO] POSTPONE 500 ; immediate
3219: @end example
3220:
3221: Instead, you can use @code{LITERAL (compilation: n --; run-time: -- n )}:
3222:
3223: @example
3224: : [FOO] ( compilation: --; run-time: -- n )
3225: 500 POSTPONE literal ; immediate
3226:
3227: : flip [FOO] ;
3228: flip .
3229: see flip
3230: @end example
3231:
3232: @code{LITERAL} consumes a number at compile-time (when it's compilation
3233: semantics are executed) and pushes it at run-time (when the code it
3234: compiled is executed). A frequent use of @code{LITERAL} is to compile a
3235: number computed at compile time into the current word:
3236:
3237: @example
3238: : bar ( -- n )
3239: [ 2 2 + ] literal ;
3240: see bar
3241: @end example
3242:
3243: @assignment
3244: Write @code{]L} which allows writing the example above as @code{: bar (
3245: -- n ) [ 2 2 + ]L ;}
3246: @endassignment
3247:
3248: @c !! @xref{Macros} for reference
3249:
3250:
3251: @node Advanced macros Tutorial, Compilation Tokens Tutorial, Literal Tutorial, Tutorial
3252: @section Advanced macros
3253: @cindex macros, advanced tutorial
3254: @cindex run-time code generation, tutorial
3255:
3256: Reconsider @code{map-array} from @ref{Execution Tokens Tutorial,,
3257: Execution Tokens}. It frequently performs @code{execute}, a relatively
3258: expensive operation in some Forth implementations. You can use
3259: @code{compile,} and @code{POSTPONE} to eliminate these @code{execute}s
3260: and produce a word that contains the word to be performed directly:
3261:
3262: @c use ]] ... [[
3263: @example
3264: : compile-map-array ( compilation: xt -- ; run-time: ... addr u -- ... )
3265: \ at run-time, execute xt ( ... x -- ... ) for each element of the
3266: \ array beginning at addr and containing u elements
3267: @{ xt @}
3268: POSTPONE cells POSTPONE over POSTPONE + POSTPONE swap POSTPONE ?do
3269: POSTPONE i POSTPONE @@ xt compile,
3270: 1 cells POSTPONE literal POSTPONE +loop ;
3271:
3272: : sum-array ( addr u -- n )
3273: 0 rot rot [ ' + compile-map-array ] ;
3274: see sum-array
3275: a 5 sum-array .
3276: @end example
3277:
3278: You can use the full power of Forth for generating the code; here's an
3279: example where the code is generated in a loop:
3280:
3281: @example
3282: : compile-vmul-step ( compilation: n --; run-time: n1 addr1 -- n2 addr2 )
3283: \ n2=n1+(addr1)*n, addr2=addr1+cell
3284: POSTPONE tuck POSTPONE @@
3285: POSTPONE literal POSTPONE * POSTPONE +
3286: POSTPONE swap POSTPONE cell+ ;
3287:
3288: : compile-vmul ( compilation: addr1 u -- ; run-time: addr2 -- n )
3289: \ n=v1*v2 (inner product), where the v_i are represented as addr_i u
3290: 0 postpone literal postpone swap
3291: [ ' compile-vmul-step compile-map-array ]
3292: postpone drop ;
3293: see compile-vmul
3294:
3295: : a-vmul ( addr -- n )
3296: \ n=a*v, where v is a vector that's as long as a and starts at addr
3297: [ a 5 compile-vmul ] ;
3298: see a-vmul
3299: a a-vmul .
3300: @end example
3301:
3302: This example uses @code{compile-map-array} to show off, but you could
3303: also use @code{map-array} instead (try it now!).
3304:
3305: You can use this technique for efficient multiplication of large
3306: matrices. In matrix multiplication, you multiply every line of one
3307: matrix with every column of the other matrix. You can generate the code
3308: for one line once, and use it for every column. The only downside of
3309: this technique is that it is cumbersome to recover the memory consumed
3310: by the generated code when you are done (and in more complicated cases
3311: it is not possible portably).
3312:
3313: @c !! @xref{Macros} for reference
3314:
3315:
3316: @node Compilation Tokens Tutorial, Wordlists and Search Order Tutorial, Advanced macros Tutorial, Tutorial
3317: @section Compilation Tokens
3318: @cindex compilation tokens, tutorial
3319: @cindex CT, tutorial
3320:
3321: This section is Gforth-specific. You can skip it.
3322:
3323: @code{' word compile,} compiles the interpretation semantics. For words
3324: with default compilation semantics this is the same as performing the
3325: compilation semantics. To represent the compilation semantics of other
3326: words (e.g., words like @code{if} that have no interpretation
3327: semantics), Gforth has the concept of a compilation token (CT,
3328: consisting of two cells), and words @code{comp'} and @code{[comp']}.
3329: You can perform the compilation semantics represented by a CT with
3330: @code{execute}:
3331:
3332: @example
3333: : foo2 ( n1 n2 -- n )
3334: [ comp' + execute ] ;
3335: see foo
3336: @end example
3337:
3338: You can compile the compilation semantics represented by a CT with
3339: @code{postpone,}:
3340:
3341: @example
3342: : foo3 ( -- )
3343: [ comp' + postpone, ] ;
3344: see foo3
3345: @end example
3346:
3347: @code{[ comp' word postpone, ]} is equivalent to @code{POSTPONE word}.
3348: @code{comp'} is particularly useful for words that have no
3349: interpretation semantics:
3350:
3351: @example
3352: ' if
3353: comp' if .s 2drop
3354: @end example
3355:
3356: Reference: @ref{Tokens for Words}.
3357:
3358:
3359: @node Wordlists and Search Order Tutorial, , Compilation Tokens Tutorial, Tutorial
3360: @section Wordlists and Search Order
3361: @cindex wordlists tutorial
3362: @cindex search order, tutorial
3363:
3364: The dictionary is not just a memory area that allows you to allocate
3365: memory with @code{allot}, it also contains the Forth words, arranged in
3366: several wordlists. When searching for a word in a wordlist,
3367: conceptually you start searching at the youngest and proceed towards
3368: older words (in reality most systems nowadays use hash-tables); i.e., if
3369: you define a word with the same name as an older word, the new word
3370: shadows the older word.
3371:
3372: Which wordlists are searched in which order is determined by the search
3373: order. You can display the search order with @code{order}. It displays
3374: first the search order, starting with the wordlist searched first, then
3375: it displays the wordlist that will contain newly defined words.
3376:
3377: You can create a new, empty wordlist with @code{wordlist ( -- wid )}:
3378:
3379: @example
3380: wordlist constant mywords
3381: @end example
3382:
3383: @code{Set-current ( wid -- )} sets the wordlist that will contain newly
3384: defined words (the @emph{current} wordlist):
3385:
3386: @example
3387: mywords set-current
3388: order
3389: @end example
3390:
3391: Gforth does not display a name for the wordlist in @code{mywords}
3392: because this wordlist was created anonymously with @code{wordlist}.
3393:
3394: You can get the current wordlist with @code{get-current ( -- wid)}. If
3395: you want to put something into a specific wordlist without overall
3396: effect on the current wordlist, this typically looks like this:
3397:
3398: @example
3399: get-current mywords set-current ( wid )
3400: create someword
3401: ( wid ) set-current
3402: @end example
3403:
3404: You can write the search order with @code{set-order ( wid1 .. widn n --
3405: )} and read it with @code{get-order ( -- wid1 .. widn n )}. The first
3406: searched wordlist is topmost.
3407:
3408: @example
3409: get-order mywords swap 1+ set-order
3410: order
3411: @end example
3412:
3413: Yes, the order of wordlists in the output of @code{order} is reversed
3414: from stack comments and the output of @code{.s} and thus unintuitive.
3415:
3416: @assignment
3417: Define @code{>order ( wid -- )} with adds @code{wid} as first searched
3418: wordlist to the search order. Define @code{previous ( -- )}, which
3419: removes the first searched wordlist from the search order. Experiment
3420: with boundary conditions (you will see some crashes or situations that
3421: are hard or impossible to leave).
3422: @endassignment
3423:
3424: The search order is a powerful foundation for providing features similar
3425: to Modula-2 modules and C++ namespaces. However, trying to modularize
3426: programs in this way has disadvantages for debugging and reuse/factoring
3427: that overcome the advantages in my experience (I don't do huge projects,
3428: though). These disadvantages are not so clear in other
3429: languages/programming environments, because these langauges are not so
3430: strong in debugging and reuse.
3431:
3432: @c !! example
3433:
3434: Reference: @ref{Word Lists}.
3435:
3436: @c ******************************************************************
3437: @node Introduction, Words, Tutorial, Top
3438: @comment node-name, next, previous, up
3439: @chapter An Introduction to ANS Forth
3440: @cindex Forth - an introduction
3441:
3442: The primary purpose of this manual is to document Gforth. However, since
3443: Forth is not a widely-known language and there is a lack of up-to-date
3444: teaching material, it seems worthwhile to provide some introductory
3445: material. For other sources of Forth-related
3446: information, see @ref{Forth-related information}.
3447:
3448: The examples in this section should work on any ANS Forth; the
3449: output shown was produced using Gforth. Each example attempts to
3450: reproduce the exact output that Gforth produces. If you try out the
3451: examples (and you should), what you should type is shown @kbd{like this}
3452: and Gforth's response is shown @code{like this}. The single exception is
3453: that, where the example shows @key{RET} it means that you should
3454: press the ``carriage return'' key. Unfortunately, some output formats for
3455: this manual cannot show the difference between @kbd{this} and
3456: @code{this} which will make trying out the examples harder (but not
3457: impossible).
3458:
3459: Forth is an unusual language. It provides an interactive development
3460: environment which includes both an interpreter and compiler. Forth
3461: programming style encourages you to break a problem down into many
3462: @cindex factoring
3463: small fragments (@dfn{factoring}), and then to develop and test each
3464: fragment interactively. Forth advocates assert that breaking the
3465: edit-compile-test cycle used by conventional programming languages can
3466: lead to great productivity improvements.
3467:
3468: @menu
3469: * Introducing the Text Interpreter::
3470: * Stacks and Postfix notation::
3471: * Your first definition::
3472: * How does that work?::
3473: * Forth is written in Forth::
3474: * Review - elements of a Forth system::
3475: * Where to go next::
3476: * Exercises::
3477: @end menu
3478:
3479: @comment ----------------------------------------------
3480: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
3481: @section Introducing the Text Interpreter
3482: @cindex text interpreter
3483: @cindex outer interpreter
3484:
3485: @c IMO this is too detailed and the pace is too slow for
3486: @c an introduction. If you know German, take a look at
3487: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
3488: @c to see how I do it - anton
3489:
3490: @c nac-> Where I have accepted your comments 100% and modified the text
3491: @c accordingly, I have deleted your comments. Elsewhere I have added a
3492: @c response like this to attempt to rationalise what I have done. Of
3493: @c course, this is a very clumsy mechanism for something that would be
3494: @c done far more efficiently over a beer. Please delete any dialogue
3495: @c you consider closed.
3496:
3497: When you invoke the Forth image, you will see a startup banner printed
3498: and nothing else (if you have Gforth installed on your system, try
3499: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
3500: its command line interpreter, which is called the @dfn{Text Interpreter}
3501: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
3502: about the text interpreter as you read through this chapter, for more
3503: detail @pxref{The Text Interpreter}).
3504:
3505: Although it's not obvious, Forth is actually waiting for your
3506: input. Type a number and press the @key{RET} key:
3507:
3508: @example
3509: @kbd{45@key{RET}} ok
3510: @end example
3511:
3512: Rather than give you a prompt to invite you to input something, the text
3513: interpreter prints a status message @i{after} it has processed a line
3514: of input. The status message in this case (``@code{ ok}'' followed by
3515: carriage-return) indicates that the text interpreter was able to process
3516: all of your input successfully. Now type something illegal:
3517:
3518: @example
3519: @kbd{qwer341@key{RET}}
3520: :1: Undefined word
3521: qwer341
3522: ^^^^^^^
3523: $400D2BA8 Bounce
3524: $400DBDA8 no.extensions
3525: @end example
3526:
3527: The exact text, other than the ``Undefined word'' may differ slightly on
3528: your system, but the effect is the same; when the text interpreter
3529: detects an error, it discards any remaining text on a line, resets
3530: certain internal state and prints an error message. For a detailed description of error messages see @ref{Error
3531: messages}.
3532:
3533: The text interpreter waits for you to press carriage-return, and then
3534: processes your input line. Starting at the beginning of the line, it
3535: breaks the line into groups of characters separated by spaces. For each
3536: group of characters in turn, it makes two attempts to do something:
3537:
3538: @itemize @bullet
3539: @item
3540: @cindex name dictionary
3541: It tries to treat it as a command. It does this by searching a @dfn{name
3542: dictionary}. If the group of characters matches an entry in the name
3543: dictionary, the name dictionary provides the text interpreter with
3544: information that allows the text interpreter perform some actions. In
3545: Forth jargon, we say that the group
3546: @cindex word
3547: @cindex definition
3548: @cindex execution token
3549: @cindex xt
3550: of characters names a @dfn{word}, that the dictionary search returns an
3551: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
3552: word, and that the text interpreter executes the xt. Often, the terms
3553: @dfn{word} and @dfn{definition} are used interchangeably.
3554: @item
3555: If the text interpreter fails to find a match in the name dictionary, it
3556: tries to treat the group of characters as a number in the current number
3557: base (when you start up Forth, the current number base is base 10). If
3558: the group of characters legitimately represents a number, the text
3559: interpreter pushes the number onto a stack (we'll learn more about that
3560: in the next section).
3561: @end itemize
3562:
3563: If the text interpreter is unable to do either of these things with any
3564: group of characters, it discards the group of characters and the rest of
3565: the line, then prints an error message. If the text interpreter reaches
3566: the end of the line without error, it prints the status message ``@code{ ok}''
3567: followed by carriage-return.
3568:
3569: This is the simplest command we can give to the text interpreter:
3570:
3571: @example
3572: @key{RET} ok
3573: @end example
3574:
3575: The text interpreter did everything we asked it to do (nothing) without
3576: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
3577: command:
3578:
3579: @example
3580: @kbd{12 dup fred dup@key{RET}}
3581: :1: Undefined word
3582: 12 dup fred dup
3583: ^^^^
3584: $400D2BA8 Bounce
3585: $400DBDA8 no.extensions
3586: @end example
3587:
3588: When you press the carriage-return key, the text interpreter starts to
3589: work its way along the line:
3590:
3591: @itemize @bullet
3592: @item
3593: When it gets to the space after the @code{2}, it takes the group of
3594: characters @code{12} and looks them up in the name
3595: dictionary@footnote{We can't tell if it found them or not, but assume
3596: for now that it did not}. There is no match for this group of characters
3597: in the name dictionary, so it tries to treat them as a number. It is
3598: able to do this successfully, so it puts the number, 12, ``on the stack''
3599: (whatever that means).
3600: @item
3601: The text interpreter resumes scanning the line and gets the next group
3602: of characters, @code{dup}. It looks it up in the name dictionary and
3603: (you'll have to take my word for this) finds it, and executes the word
3604: @code{dup} (whatever that means).
3605: @item
3606: Once again, the text interpreter resumes scanning the line and gets the
3607: group of characters @code{fred}. It looks them up in the name
3608: dictionary, but can't find them. It tries to treat them as a number, but
3609: they don't represent any legal number.
3610: @end itemize
3611:
3612: At this point, the text interpreter gives up and prints an error
3613: message. The error message shows exactly how far the text interpreter
3614: got in processing the line. In particular, it shows that the text
3615: interpreter made no attempt to do anything with the final character
3616: group, @code{dup}, even though we have good reason to believe that the
3617: text interpreter would have no problem looking that word up and
3618: executing it a second time.
3619:
3620:
3621: @comment ----------------------------------------------
3622: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
3623: @section Stacks, postfix notation and parameter passing
3624: @cindex text interpreter
3625: @cindex outer interpreter
3626:
3627: In procedural programming languages (like C and Pascal), the
3628: building-block of programs is the @dfn{function} or @dfn{procedure}. These
3629: functions or procedures are called with @dfn{explicit parameters}. For
3630: example, in C we might write:
3631:
3632: @example
3633: total = total + new_volume(length,height,depth);
3634: @end example
3635:
3636: @noindent
3637: where new_volume is a function-call to another piece of code, and total,
3638: length, height and depth are all variables. length, height and depth are
3639: parameters to the function-call.
3640:
3641: In Forth, the equivalent of the function or procedure is the
3642: @dfn{definition} and parameters are implicitly passed between
3643: definitions using a shared stack that is visible to the
3644: programmer. Although Forth does support variables, the existence of the
3645: stack means that they are used far less often than in most other
3646: programming languages. When the text interpreter encounters a number, it
3647: will place (@dfn{push}) it on the stack. There are several stacks (the
3648: actual number is implementation-dependent ...) and the particular stack
3649: used for any operation is implied unambiguously by the operation being
3650: performed. The stack used for all integer operations is called the @dfn{data
3651: stack} and, since this is the stack used most commonly, references to
3652: ``the data stack'' are often abbreviated to ``the stack''.
3653:
3654: The stacks have a last-in, first-out (LIFO) organisation. If you type:
3655:
3656: @example
3657: @kbd{1 2 3@key{RET}} ok
3658: @end example
3659:
3660: Then this instructs the text interpreter to placed three numbers on the
3661: (data) stack. An analogy for the behaviour of the stack is to take a
3662: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
3663: the table. The 3 was the last card onto the pile (``last-in'') and if
3664: you take a card off the pile then, unless you're prepared to fiddle a
3665: bit, the card that you take off will be the 3 (``first-out''). The
3666: number that will be first-out of the stack is called the @dfn{top of
3667: stack}, which
3668: @cindex TOS definition
3669: is often abbreviated to @dfn{TOS}.
3670:
3671: To understand how parameters are passed in Forth, consider the
3672: behaviour of the definition @code{+} (pronounced ``plus''). You will not
3673: be surprised to learn that this definition performs addition. More
3674: precisely, it adds two number together and produces a result. Where does
3675: it get the two numbers from? It takes the top two numbers off the
3676: stack. Where does it place the result? On the stack. You can act-out the
3677: behaviour of @code{+} with your playing cards like this:
3678:
3679: @itemize @bullet
3680: @item
3681: Pick up two cards from the stack on the table
3682: @item
3683: Stare at them intently and ask yourself ``what @i{is} the sum of these two
3684: numbers''
3685: @item
3686: Decide that the answer is 5
3687: @item
3688: Shuffle the two cards back into the pack and find a 5
3689: @item
3690: Put a 5 on the remaining ace that's on the table.
3691: @end itemize
3692:
3693: If you don't have a pack of cards handy but you do have Forth running,
3694: you can use the definition @code{.s} to show the current state of the stack,
3695: without affecting the stack. Type:
3696:
3697: @example
3698: @kbd{clearstack 1 2 3@key{RET}} ok
3699: @kbd{.s@key{RET}} <3> 1 2 3 ok
3700: @end example
3701:
3702: The text interpreter looks up the word @code{clearstack} and executes
3703: it; it tidies up the stack and removes any entries that may have been
3704: left on it by earlier examples. The text interpreter pushes each of the
3705: three numbers in turn onto the stack. Finally, the text interpreter
3706: looks up the word @code{.s} and executes it. The effect of executing
3707: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
3708: followed by a list of all the items on the stack; the item on the far
3709: right-hand side is the TOS.
3710:
3711: You can now type:
3712:
3713: @example
3714: @kbd{+ .s@key{RET}} <2> 1 5 ok
3715: @end example
3716:
3717: @noindent
3718: which is correct; there are now 2 items on the stack and the result of
3719: the addition is 5.
3720:
3721: If you're playing with cards, try doing a second addition: pick up the
3722: two cards, work out that their sum is 6, shuffle them into the pack,
3723: look for a 6 and place that on the table. You now have just one item on
3724: the stack. What happens if you try to do a third addition? Pick up the
3725: first card, pick up the second card -- ah! There is no second card. This
3726: is called a @dfn{stack underflow} and consitutes an error. If you try to
3727: do the same thing with Forth it will report an error (probably a Stack
3728: Underflow or an Invalid Memory Address error).
3729:
3730: The opposite situation to a stack underflow is a @dfn{stack overflow},
3731: which simply accepts that there is a finite amount of storage space
3732: reserved for the stack. To stretch the playing card analogy, if you had
3733: enough packs of cards and you piled the cards up on the table, you would
3734: eventually be unable to add another card; you'd hit the ceiling. Gforth
3735: allows you to set the maximum size of the stacks. In general, the only
3736: time that you will get a stack overflow is because a definition has a
3737: bug in it and is generating data on the stack uncontrollably.
3738:
3739: There's one final use for the playing card analogy. If you model your
3740: stack using a pack of playing cards, the maximum number of items on
3741: your stack will be 52 (I assume you didn't use the Joker). The maximum
3742: @i{value} of any item on the stack is 13 (the King). In fact, the only
3743: possible numbers are positive integer numbers 1 through 13; you can't
3744: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
3745: think about some of the cards, you can accommodate different
3746: numbers. For example, you could think of the Jack as representing 0,
3747: the Queen as representing -1 and the King as representing -2. Your
3748: @i{range} remains unchanged (you can still only represent a total of 13
3749: numbers) but the numbers that you can represent are -2 through 10.
3750:
3751: In that analogy, the limit was the amount of information that a single
3752: stack entry could hold, and Forth has a similar limit. In Forth, the
3753: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
3754: implementation dependent and affects the maximum value that a stack
3755: entry can hold. A Standard Forth provides a cell size of at least
3756: 16-bits, and most desktop systems use a cell size of 32-bits.
3757:
3758: Forth does not do any type checking for you, so you are free to
3759: manipulate and combine stack items in any way you wish. A convenient way
3760: of treating stack items is as 2's complement signed integers, and that
3761: is what Standard words like @code{+} do. Therefore you can type:
3762:
3763: @example
3764: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
3765: @end example
3766:
3767: If you use numbers and definitions like @code{+} in order to turn Forth
3768: into a great big pocket calculator, you will realise that it's rather
3769: different from a normal calculator. Rather than typing 2 + 3 = you had
3770: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
3771: result). The terminology used to describe this difference is to say that
3772: your calculator uses @dfn{Infix Notation} (parameters and operators are
3773: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
3774: operators are separate), also called @dfn{Reverse Polish Notation}.
3775:
3776: Whilst postfix notation might look confusing to begin with, it has
3777: several important advantages:
3778:
3779: @itemize @bullet
3780: @item
3781: it is unambiguous
3782: @item
3783: it is more concise
3784: @item
3785: it fits naturally with a stack-based system
3786: @end itemize
3787:
3788: To examine these claims in more detail, consider these sums:
3789:
3790: @example
3791: 6 + 5 * 4 =
3792: 4 * 5 + 6 =
3793: @end example
3794:
3795: If you're just learning maths or your maths is very rusty, you will
3796: probably come up with the answer 44 for the first and 26 for the
3797: second. If you are a bit of a whizz at maths you will remember the
3798: @i{convention} that multiplication takes precendence over addition, and
3799: you'd come up with the answer 26 both times. To explain the answer 26
3800: to someone who got the answer 44, you'd probably rewrite the first sum
3801: like this:
3802:
3803: @example
3804: 6 + (5 * 4) =
3805: @end example
3806:
3807: If what you really wanted was to perform the addition before the
3808: multiplication, you would have to use parentheses to force it.
3809:
3810: If you did the first two sums on a pocket calculator you would probably
3811: get the right answers, unless you were very cautious and entered them using
3812: these keystroke sequences:
3813:
3814: 6 + 5 = * 4 =
3815: 4 * 5 = + 6 =
3816:
3817: Postfix notation is unambiguous because the order that the operators
3818: are applied is always explicit; that also means that parentheses are
3819: never required. The operators are @i{active} (the act of quoting the
3820: operator makes the operation occur) which removes the need for ``=''.
3821:
3822: The sum 6 + 5 * 4 can be written (in postfix notation) in two
3823: equivalent ways:
3824:
3825: @example
3826: 6 5 4 * + or:
3827: 5 4 * 6 +
3828: @end example
3829:
3830: An important thing that you should notice about this notation is that
3831: the @i{order} of the numbers does not change; if you want to subtract
3832: 2 from 10 you type @code{10 2 -}.
3833:
3834: The reason that Forth uses postfix notation is very simple to explain: it
3835: makes the implementation extremely simple, and it follows naturally from
3836: using the stack as a mechanism for passing parameters. Another way of
3837: thinking about this is to realise that all Forth definitions are
3838: @i{active}; they execute as they are encountered by the text
3839: interpreter. The result of this is that the syntax of Forth is trivially
3840: simple.
3841:
3842:
3843:
3844: @comment ----------------------------------------------
3845: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
3846: @section Your first Forth definition
3847: @cindex first definition
3848:
3849: Until now, the examples we've seen have been trivial; we've just been
3850: using Forth as a bigger-than-pocket calculator. Also, each calculation
3851: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
3852: again@footnote{That's not quite true. If you press the up-arrow key on
3853: your keyboard you should be able to scroll back to any earlier command,
3854: edit it and re-enter it.} In this section we'll see how to add new
3855: words to Forth's vocabulary.
3856:
3857: The easiest way to create a new word is to use a @dfn{colon
3858: definition}. We'll define a few and try them out before worrying too
3859: much about how they work. Try typing in these examples; be careful to
3860: copy the spaces accurately:
3861:
3862: @example
3863: : add-two 2 + . ;
3864: : greet ." Hello and welcome" ;
3865: : demo 5 add-two ;
3866: @end example
3867:
3868: @noindent
3869: Now try them out:
3870:
3871: @example
3872: @kbd{greet@key{RET}} Hello and welcome ok
3873: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
3874: @kbd{4 add-two@key{RET}} 6 ok
3875: @kbd{demo@key{RET}} 7 ok
3876: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
3877: @end example
3878:
3879: The first new thing that we've introduced here is the pair of words
3880: @code{:} and @code{;}. These are used to start and terminate a new
3881: definition, respectively. The first word after the @code{:} is the name
3882: for the new definition.
3883:
3884: As you can see from the examples, a definition is built up of words that
3885: have already been defined; Forth makes no distinction between
3886: definitions that existed when you started the system up, and those that
3887: you define yourself.
3888:
3889: The examples also introduce the words @code{.} (dot), @code{."}
3890: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
3891: the stack and displays it. It's like @code{.s} except that it only
3892: displays the top item of the stack and it is destructive; after it has
3893: executed, the number is no longer on the stack. There is always one
3894: space printed after the number, and no spaces before it. Dot-quote
3895: defines a string (a sequence of characters) that will be printed when
3896: the word is executed. The string can contain any printable characters
3897: except @code{"}. A @code{"} has a special function; it is not a Forth
3898: word but it acts as a delimiter (the way that delimiters work is
3899: described in the next section). Finally, @code{dup} duplicates the value
3900: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
3901:
3902: We already know that the text interpreter searches through the
3903: dictionary to locate names. If you've followed the examples earlier, you
3904: will already have a definition called @code{add-two}. Lets try modifying
3905: it by typing in a new definition:
3906:
3907: @example
3908: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
3909: @end example
3910:
3911: Forth recognised that we were defining a word that already exists, and
3912: printed a message to warn us of that fact. Let's try out the new
3913: definition:
3914:
3915: @example
3916: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
3917: @end example
3918:
3919: @noindent
3920: All that we've actually done here, though, is to create a new
3921: definition, with a particular name. The fact that there was already a
3922: definition with the same name did not make any difference to the way
3923: that the new definition was created (except that Forth printed a warning
3924: message). The old definition of add-two still exists (try @code{demo}
3925: again to see that this is true). Any new definition will use the new
3926: definition of @code{add-two}, but old definitions continue to use the
3927: version that already existed at the time that they were @code{compiled}.
3928:
3929: Before you go on to the next section, try defining and redefining some
3930: words of your own.
3931:
3932: @comment ----------------------------------------------
3933: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
3934: @section How does that work?
3935: @cindex parsing words
3936:
3937: @c That's pretty deep (IMO way too deep) for an introduction. - anton
3938:
3939: @c Is it a good idea to talk about the interpretation semantics of a
3940: @c number? We don't have an xt to go along with it. - anton
3941:
3942: @c Now that I have eliminated execution semantics, I wonder if it would not
3943: @c be better to keep them (or add run-time semantics), to make it easier to
3944: @c explain what compilation semantics usually does. - anton
3945:
3946: @c nac-> I removed the term ``default compilation sematics'' from the
3947: @c introductory chapter. Removing ``execution semantics'' was making
3948: @c everything simpler to explain, then I think the use of this term made
3949: @c everything more complex again. I replaced it with ``default
3950: @c semantics'' (which is used elsewhere in the manual) by which I mean
3951: @c ``a definition that has neither the immediate nor the compile-only
3952: @c flag set''. I reworded big chunks of the ``how does that work''
3953: @c section (and, unusually for me, I think I even made it shorter!). See
3954: @c what you think -- I know I have not addressed your primary concern
3955: @c that it is too heavy-going for an introduction. From what I understood
3956: @c of your course notes it looks as though they might be a good framework.
3957: @c Things that I've tried to capture here are some things that came as a
3958: @c great revelation here when I first understood them. Also, I like the
3959: @c fact that a very simple code example shows up almost all of the issues
3960: @c that you need to understand to see how Forth works. That's unique and
3961: @c worthwhile to emphasise.
3962:
3963: Now we're going to take another look at the definition of @code{add-two}
3964: from the previous section. From our knowledge of the way that the text
3965: interpreter works, we would have expected this result when we tried to
3966: define @code{add-two}:
3967:
3968: @example
3969: @kbd{: add-two 2 + . ;@key{RET}}
3970: ^^^^^^^
3971: Error: Undefined word
3972: @end example
3973:
3974: The reason that this didn't happen is bound up in the way that @code{:}
3975: works. The word @code{:} does two special things. The first special
3976: thing that it does prevents the text interpreter from ever seeing the
3977: characters @code{add-two}. The text interpreter uses a variable called
3978: @cindex modifying >IN
3979: @code{>IN} (pronounced ``to-in'') to keep track of where it is in the
3980: input line. When it encounters the word @code{:} it behaves in exactly
3981: the same way as it does for any other word; it looks it up in the name
3982: dictionary, finds its xt and executes it. When @code{:} executes, it
3983: looks at the input buffer, finds the word @code{add-two} and advances the
3984: value of @code{>IN} to point past it. It then does some other stuff
3985: associated with creating the new definition (including creating an entry
3986: for @code{add-two} in the name dictionary). When the execution of @code{:}
3987: completes, control returns to the text interpreter, which is oblivious
3988: to the fact that it has been tricked into ignoring part of the input
3989: line.
3990:
3991: @cindex parsing words
3992: Words like @code{:} -- words that advance the value of @code{>IN} and so
3993: prevent the text interpreter from acting on the whole of the input line
3994: -- are called @dfn{parsing words}.
3995:
3996: @cindex @code{state} - effect on the text interpreter
3997: @cindex text interpreter - effect of state
3998: The second special thing that @code{:} does is change the value of a
3999: variable called @code{state}, which affects the way that the text
4000: interpreter behaves. When Gforth starts up, @code{state} has the value
4001: 0, and the text interpreter is said to be @dfn{interpreting}. During a
4002: colon definition (started with @code{:}), @code{state} is set to -1 and
4003: the text interpreter is said to be @dfn{compiling}.
4004:
4005: In this example, the text interpreter is compiling when it processes the
4006: string ``@code{2 + . ;}''. It still breaks the string down into
4007: character sequences in the same way. However, instead of pushing the
4008: number @code{2} onto the stack, it lays down (@dfn{compiles}) some magic
4009: into the definition of @code{add-two} that will make the number @code{2} get
4010: pushed onto the stack when @code{add-two} is @dfn{executed}. Similarly,
4011: the behaviours of @code{+} and @code{.} are also compiled into the
4012: definition.
4013:
4014: One category of words don't get compiled. These so-called @dfn{immediate
4015: words} get executed (performed @i{now}) regardless of whether the text
4016: interpreter is interpreting or compiling. The word @code{;} is an
4017: immediate word. Rather than being compiled into the definition, it
4018: executes. Its effect is to terminate the current definition, which
4019: includes changing the value of @code{state} back to 0.
4020:
4021: When you execute @code{add-two}, it has a @dfn{run-time effect} that is
4022: exactly the same as if you had typed @code{2 + . @key{RET}} outside of a
4023: definition.
4024:
4025: In Forth, every word or number can be described in terms of two
4026: properties:
4027:
4028: @itemize @bullet
4029: @item
4030: @cindex interpretation semantics
4031: Its @dfn{interpretation semantics} describe how it will behave when the
4032: text interpreter encounters it in @dfn{interpret} state. The
4033: interpretation semantics of a word are represented by an @dfn{execution
4034: token}.
4035: @item
4036: @cindex compilation semantics
4037: Its @dfn{compilation semantics} describe how it will behave when the
4038: text interpreter encounters it in @dfn{compile} state. The compilation
4039: semantics of a word are represented in an implementation-dependent way;
4040: Gforth uses a @dfn{compilation token}.
4041: @end itemize
4042:
4043: @noindent
4044: Numbers are always treated in a fixed way:
4045:
4046: @itemize @bullet
4047: @item
4048: When the number is @dfn{interpreted}, its behaviour is to push the
4049: number onto the stack.
4050: @item
4051: When the number is @dfn{compiled}, a piece of code is appended to the
4052: current definition that pushes the number when it runs. (In other words,
4053: the compilation semantics of a number are to postpone its interpretation
4054: semantics until the run-time of the definition that it is being compiled
4055: into.)
4056: @end itemize
4057:
4058: Words don't behave in such a regular way, but most have @i{default
4059: semantics} which means that they behave like this:
4060:
4061: @itemize @bullet
4062: @item
4063: The @dfn{interpretation semantics} of the word are to do something useful.
4064: @item
4065: The @dfn{compilation semantics} of the word are to append its
4066: @dfn{interpretation semantics} to the current definition (so that its
4067: run-time behaviour is to do something useful).
4068: @end itemize
4069:
4070: @cindex immediate words
4071: The actual behaviour of any particular word can be controlled by using
4072: the words @code{immediate} and @code{compile-only} when the word is
4073: defined. These words set flags in the name dictionary entry of the most
4074: recently defined word, and these flags are retrieved by the text
4075: interpreter when it finds the word in the name dictionary.
4076:
4077: A word that is marked as @dfn{immediate} has compilation semantics that
4078: are identical to its interpretation semantics. In other words, it
4079: behaves like this:
4080:
4081: @itemize @bullet
4082: @item
4083: The @dfn{interpretation semantics} of the word are to do something useful.
4084: @item
4085: The @dfn{compilation semantics} of the word are to do something useful
4086: (and actually the same thing); i.e., it is executed during compilation.
4087: @end itemize
4088:
4089: Marking a word as @dfn{compile-only} prohibits the text interpreter from
4090: performing the interpretation semantics of the word directly; an attempt
4091: to do so will generate an error. It is never necessary to use
4092: @code{compile-only} (and it is not even part of ANS Forth, though it is
4093: provided by many implementations) but it is good etiquette to apply it
4094: to a word that will not behave correctly (and might have unexpected
4095: side-effects) in interpret state. For example, it is only legal to use
4096: the conditional word @code{IF} within a definition. If you forget this
4097: and try to use it elsewhere, the fact that (in Gforth) it is marked as
4098: @code{compile-only} allows the text interpreter to generate a helpful
4099: error message rather than subjecting you to the consequences of your
4100: folly.
4101:
4102: This example shows the difference between an immediate and a
4103: non-immediate word:
4104:
4105: @example
4106: : show-state state @@ . ;
4107: : show-state-now show-state ; immediate
4108: : word1 show-state ;
4109: : word2 show-state-now ;
4110: @end example
4111:
4112: The word @code{immediate} after the definition of @code{show-state-now}
4113: makes that word an immediate word. These definitions introduce a new
4114: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
4115: variable, and leaves it on the stack. Therefore, the behaviour of
4116: @code{show-state} is to print a number that represents the current value
4117: of @code{state}.
4118:
4119: When you execute @code{word1}, it prints the number 0, indicating that
4120: the system is interpreting. When the text interpreter compiled the
4121: definition of @code{word1}, it encountered @code{show-state} whose
4122: compilation semantics are to append its interpretation semantics to the
4123: current definition. When you execute @code{word1}, it performs the
4124: interpretation semantics of @code{show-state}. At the time that @code{word1}
4125: (and therefore @code{show-state}) are executed, the system is
4126: interpreting.
4127:
4128: When you pressed @key{RET} after entering the definition of @code{word2},
4129: you should have seen the number -1 printed, followed by ``@code{
4130: ok}''. When the text interpreter compiled the definition of
4131: @code{word2}, it encountered @code{show-state-now}, an immediate word,
4132: whose compilation semantics are therefore to perform its interpretation
4133: semantics. It is executed straight away (even before the text
4134: interpreter has moved on to process another group of characters; the
4135: @code{;} in this example). The effect of executing it are to display the
4136: value of @code{state} @i{at the time that the definition of}
4137: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
4138: system is compiling at this time. If you execute @code{word2} it does
4139: nothing at all.
4140:
4141: @cindex @code{."}, how it works
4142: Before leaving the subject of immediate words, consider the behaviour of
4143: @code{."} in the definition of @code{greet}, in the previous
4144: section. This word is both a parsing word and an immediate word. Notice
4145: that there is a space between @code{."} and the start of the text
4146: @code{Hello and welcome}, but that there is no space between the last
4147: letter of @code{welcome} and the @code{"} character. The reason for this
4148: is that @code{."} is a Forth word; it must have a space after it so that
4149: the text interpreter can identify it. The @code{"} is not a Forth word;
4150: it is a @dfn{delimiter}. The examples earlier show that, when the string
4151: is displayed, there is neither a space before the @code{H} nor after the
4152: @code{e}. Since @code{."} is an immediate word, it executes at the time
4153: that @code{greet} is defined. When it executes, its behaviour is to
4154: search forward in the input line looking for the delimiter. When it
4155: finds the delimiter, it updates @code{>IN} to point past the
4156: delimiter. It also compiles some magic code into the definition of
4157: @code{greet}; the xt of a run-time routine that prints a text string. It
4158: compiles the string @code{Hello and welcome} into memory so that it is
4159: available to be printed later. When the text interpreter gains control,
4160: the next word it finds in the input stream is @code{;} and so it
4161: terminates the definition of @code{greet}.
4162:
4163:
4164: @comment ----------------------------------------------
4165: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
4166: @section Forth is written in Forth
4167: @cindex structure of Forth programs
4168:
4169: When you start up a Forth compiler, a large number of definitions
4170: already exist. In Forth, you develop a new application using bottom-up
4171: programming techniques to create new definitions that are defined in
4172: terms of existing definitions. As you create each definition you can
4173: test and debug it interactively.
4174:
4175: If you have tried out the examples in this section, you will probably
4176: have typed them in by hand; when you leave Gforth, your definitions will
4177: be lost. You can avoid this by using a text editor to enter Forth source
4178: code into a file, and then loading code from the file using
4179: @code{include} (@pxref{Forth source files}). A Forth source file is
4180: processed by the text interpreter, just as though you had typed it in by
4181: hand@footnote{Actually, there are some subtle differences -- see
4182: @ref{The Text Interpreter}.}.
4183:
4184: Gforth also supports the traditional Forth alternative to using text
4185: files for program entry (@pxref{Blocks}).
4186:
4187: In common with many, if not most, Forth compilers, most of Gforth is
4188: actually written in Forth. All of the @file{.fs} files in the
4189: installation directory@footnote{For example,
4190: @file{/usr/local/share/gforth...}} are Forth source files, which you can
4191: study to see examples of Forth programming.
4192:
4193: Gforth maintains a history file that records every line that you type to
4194: the text interpreter. This file is preserved between sessions, and is
4195: used to provide a command-line recall facility. If you enter long
4196: definitions by hand, you can use a text editor to paste them out of the
4197: history file into a Forth source file for reuse at a later time
4198: (for more information @pxref{Command-line editing}).
4199:
4200:
4201: @comment ----------------------------------------------
4202: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
4203: @section Review - elements of a Forth system
4204: @cindex elements of a Forth system
4205:
4206: To summarise this chapter:
4207:
4208: @itemize @bullet
4209: @item
4210: Forth programs use @dfn{factoring} to break a problem down into small
4211: fragments called @dfn{words} or @dfn{definitions}.
4212: @item
4213: Forth program development is an interactive process.
4214: @item
4215: The main command loop that accepts input, and controls both
4216: interpretation and compilation, is called the @dfn{text interpreter}
4217: (also known as the @dfn{outer interpreter}).
4218: @item
4219: Forth has a very simple syntax, consisting of words and numbers
4220: separated by spaces or carriage-return characters. Any additional syntax
4221: is imposed by @dfn{parsing words}.
4222: @item
4223: Forth uses a stack to pass parameters between words. As a result, it
4224: uses postfix notation.
4225: @item
4226: To use a word that has previously been defined, the text interpreter
4227: searches for the word in the @dfn{name dictionary}.
4228: @item
4229: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
4230: @item
4231: The text interpreter uses the value of @code{state} to select between
4232: the use of the @dfn{interpretation semantics} and the @dfn{compilation
4233: semantics} of a word that it encounters.
4234: @item
4235: The relationship between the @dfn{interpretation semantics} and
4236: @dfn{compilation semantics} for a word
4237: depend upon the way in which the word was defined (for example, whether
4238: it is an @dfn{immediate} word).
4239: @item
4240: Forth definitions can be implemented in Forth (called @dfn{high-level
4241: definitions}) or in some other way (usually a lower-level language and
4242: as a result often called @dfn{low-level definitions}, @dfn{code
4243: definitions} or @dfn{primitives}).
4244: @item
4245: Many Forth systems are implemented mainly in Forth.
4246: @end itemize
4247:
4248:
4249: @comment ----------------------------------------------
4250: @node Where to go next, Exercises, Review - elements of a Forth system, Introduction
4251: @section Where To Go Next
4252: @cindex where to go next
4253:
4254: Amazing as it may seem, if you have read (and understood) this far, you
4255: know almost all the fundamentals about the inner workings of a Forth
4256: system. You certainly know enough to be able to read and understand the
4257: rest of this manual and the ANS Forth document, to learn more about the
4258: facilities that Forth in general and Gforth in particular provide. Even
4259: scarier, you know almost enough to implement your own Forth system.
4260: However, that's not a good idea just yet... better to try writing some
4261: programs in Gforth.
4262:
4263: Forth has such a rich vocabulary that it can be hard to know where to
4264: start in learning it. This section suggests a few sets of words that are
4265: enough to write small but useful programs. Use the word index in this
4266: document to learn more about each word, then try it out and try to write
4267: small definitions using it. Start by experimenting with these words:
4268:
4269: @itemize @bullet
4270: @item
4271: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
4272: @item
4273: Comparison: @code{MIN MAX =}
4274: @item
4275: Logic: @code{AND OR XOR NOT}
4276: @item
4277: Stack manipulation: @code{DUP DROP SWAP OVER}
4278: @item
4279: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
4280: @item
4281: Input/Output: @code{. ." EMIT CR KEY}
4282: @item
4283: Defining words: @code{: ; CREATE}
4284: @item
4285: Memory allocation words: @code{ALLOT ,}
4286: @item
4287: Tools: @code{SEE WORDS .S MARKER}
4288: @end itemize
4289:
4290: When you have mastered those, go on to:
4291:
4292: @itemize @bullet
4293: @item
4294: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
4295: @item
4296: Memory access: @code{@@ !}
4297: @end itemize
4298:
4299: When you have mastered these, there's nothing for it but to read through
4300: the whole of this manual and find out what you've missed.
4301:
4302: @comment ----------------------------------------------
4303: @node Exercises, , Where to go next, Introduction
4304: @section Exercises
4305: @cindex exercises
4306:
4307: TODO: provide a set of programming excercises linked into the stuff done
4308: already and into other sections of the manual. Provide solutions to all
4309: the exercises in a .fs file in the distribution.
4310:
4311: @c Get some inspiration from Starting Forth and Kelly&Spies.
4312:
4313: @c excercises:
4314: @c 1. take inches and convert to feet and inches.
4315: @c 2. take temperature and convert from fahrenheight to celcius;
4316: @c may need to care about symmetric vs floored??
4317: @c 3. take input line and do character substitution
4318: @c to encipher or decipher
4319: @c 4. as above but work on a file for in and out
4320: @c 5. take input line and convert to pig-latin
4321: @c
4322: @c thing of sets of things to exercise then come up with
4323: @c problems that need those things.
4324:
4325:
4326: @c ******************************************************************
4327: @node Words, Error messages, Introduction, Top
4328: @chapter Forth Words
4329: @cindex words
4330:
4331: @menu
4332: * Notation::
4333: * Case insensitivity::
4334: * Comments::
4335: * Boolean Flags::
4336: * Arithmetic::
4337: * Stack Manipulation::
4338: * Memory::
4339: * Control Structures::
4340: * Defining Words::
4341: * Interpretation and Compilation Semantics::
4342: * Tokens for Words::
4343: * The Text Interpreter::
4344: * Word Lists::
4345: * Environmental Queries::
4346: * Files::
4347: * Blocks::
4348: * Other I/O::
4349: * Programming Tools::
4350: * Assembler and Code Words::
4351: * Threading Words::
4352: * Locals::
4353: * Structures::
4354: * Object-oriented Forth::
4355: * Passing Commands to the OS::
4356: * Keeping track of Time::
4357: * Miscellaneous Words::
4358: @end menu
4359:
4360: @node Notation, Case insensitivity, Words, Words
4361: @section Notation
4362: @cindex notation of glossary entries
4363: @cindex format of glossary entries
4364: @cindex glossary notation format
4365: @cindex word glossary entry format
4366:
4367: The Forth words are described in this section in the glossary notation
4368: that has become a de-facto standard for Forth texts:
4369:
4370: @format
4371: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
4372: @end format
4373: @i{Description}
4374:
4375: @table @var
4376: @item word
4377: The name of the word.
4378:
4379: @item Stack effect
4380: @cindex stack effect
4381: The stack effect is written in the notation @code{@i{before} --
4382: @i{after}}, where @i{before} and @i{after} describe the top of
4383: stack entries before and after the execution of the word. The rest of
4384: the stack is not touched by the word. The top of stack is rightmost,
4385: i.e., a stack sequence is written as it is typed in. Note that Gforth
4386: uses a separate floating point stack, but a unified stack
4387: notation. Also, return stack effects are not shown in @i{stack
4388: effect}, but in @i{Description}. The name of a stack item describes
4389: the type and/or the function of the item. See below for a discussion of
4390: the types.
4391:
4392: All words have two stack effects: A compile-time stack effect and a
4393: run-time stack effect. The compile-time stack-effect of most words is
4394: @i{ -- }. If the compile-time stack-effect of a word deviates from
4395: this standard behaviour, or the word does other unusual things at
4396: compile time, both stack effects are shown; otherwise only the run-time
4397: stack effect is shown.
4398:
4399: @cindex pronounciation of words
4400: @item pronunciation
4401: How the word is pronounced.
4402:
4403: @cindex wordset
4404: @cindex environment wordset
4405: @item wordset
4406: The ANS Forth standard is divided into several word sets. A standard
4407: system need not support all of them. Therefore, in theory, the fewer
4408: word sets your program uses the more portable it will be. However, we
4409: suspect that most ANS Forth systems on personal machines will feature
4410: all word sets. Words that are not defined in ANS Forth have
4411: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
4412: describes words that will work in future releases of Gforth;
4413: @code{gforth-internal} words are more volatile. Environmental query
4414: strings are also displayed like words; you can recognize them by the
4415: @code{environment} in the word set field.
4416:
4417: @item Description
4418: A description of the behaviour of the word.
4419: @end table
4420:
4421: @cindex types of stack items
4422: @cindex stack item types
4423: The type of a stack item is specified by the character(s) the name
4424: starts with:
4425:
4426: @table @code
4427: @item f
4428: @cindex @code{f}, stack item type
4429: Boolean flags, i.e. @code{false} or @code{true}.
4430: @item c
4431: @cindex @code{c}, stack item type
4432: Char
4433: @item w
4434: @cindex @code{w}, stack item type
4435: Cell, can contain an integer or an address
4436: @item n
4437: @cindex @code{n}, stack item type
4438: signed integer
4439: @item u
4440: @cindex @code{u}, stack item type
4441: unsigned integer
4442: @item d
4443: @cindex @code{d}, stack item type
4444: double sized signed integer
4445: @item ud
4446: @cindex @code{ud}, stack item type
4447: double sized unsigned integer
4448: @item r
4449: @cindex @code{r}, stack item type
4450: Float (on the FP stack)
4451: @item a-
4452: @cindex @code{a_}, stack item type
4453: Cell-aligned address
4454: @item c-
4455: @cindex @code{c_}, stack item type
4456: Char-aligned address (note that a Char may have two bytes in Windows NT)
4457: @item f-
4458: @cindex @code{f_}, stack item type
4459: Float-aligned address
4460: @item df-
4461: @cindex @code{df_}, stack item type
4462: Address aligned for IEEE double precision float
4463: @item sf-
4464: @cindex @code{sf_}, stack item type
4465: Address aligned for IEEE single precision float
4466: @item xt
4467: @cindex @code{xt}, stack item type
4468: Execution token, same size as Cell
4469: @item wid
4470: @cindex @code{wid}, stack item type
4471: Word list ID, same size as Cell
4472: @item ior, wior
4473: @cindex ior type description
4474: @cindex wior type description
4475: I/O result code, cell-sized. In Gforth, you can @code{throw} iors.
4476: @item f83name
4477: @cindex @code{f83name}, stack item type
4478: Pointer to a name structure
4479: @item "
4480: @cindex @code{"}, stack item type
4481: string in the input stream (not on the stack). The terminating character
4482: is a blank by default. If it is not a blank, it is shown in @code{<>}
4483: quotes.
4484: @end table
4485:
4486: @comment ----------------------------------------------
4487: @node Case insensitivity, Comments, Notation, Words
4488: @section Case insensitivity
4489: @cindex case sensitivity
4490: @cindex upper and lower case
4491:
4492: Gforth is case-insensitive; you can enter definitions and invoke
4493: Standard words using upper, lower or mixed case (however,
4494: @pxref{core-idef, Implementation-defined options, Implementation-defined
4495: options}).
4496:
4497: ANS Forth only @i{requires} implementations to recognise Standard words
4498: when they are typed entirely in upper case. Therefore, a Standard
4499: program must use upper case for all Standard words. You can use whatever
4500: case you like for words that you define, but in a Standard program you
4501: have to use the words in the same case that you defined them.
4502:
4503: Gforth supports case sensitivity through @code{table}s (case-sensitive
4504: wordlists, @pxref{Word Lists}).
4505:
4506: Two people have asked how to convert Gforth to be case-sensitive; while
4507: we think this is a bad idea, you can change all wordlists into tables
4508: like this:
4509:
4510: @example
4511: ' table-find forth-wordlist wordlist-map @ !
4512: @end example
4513:
4514: Note that you now have to type the predefined words in the same case
4515: that we defined them, which are varying. You may want to convert them
4516: to your favourite case before doing this operation (I won't explain how,
4517: because if you are even contemplating doing this, you'd better have
4518: enough knowledge of Forth systems to know this already).
4519:
4520: @node Comments, Boolean Flags, Case insensitivity, Words
4521: @section Comments
4522: @cindex comments
4523:
4524: Forth supports two styles of comment; the traditional @i{in-line} comment,
4525: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
4526:
4527:
4528: doc-(
4529: doc-\
4530: doc-\G
4531:
4532:
4533: @node Boolean Flags, Arithmetic, Comments, Words
4534: @section Boolean Flags
4535: @cindex Boolean flags
4536:
4537: A Boolean flag is cell-sized. A cell with all bits clear represents the
4538: flag @code{false} and a flag with all bits set represents the flag
4539: @code{true}. Words that check a flag (for example, @code{IF}) will treat
4540: a cell that has @i{any} bit set as @code{true}.
4541: @c on and off to Memory?
4542: @c true and false to "Bitwise operations" or "Numeric comparison"?
4543:
4544: doc-true
4545: doc-false
4546: doc-on
4547: doc-off
4548:
4549:
4550: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
4551: @section Arithmetic
4552: @cindex arithmetic words
4553:
4554: @cindex division with potentially negative operands
4555: Forth arithmetic is not checked, i.e., you will not hear about integer
4556: overflow on addition or multiplication, you may hear about division by
4557: zero if you are lucky. The operator is written after the operands, but
4558: the operands are still in the original order. I.e., the infix @code{2-1}
4559: corresponds to @code{2 1 -}. Forth offers a variety of division
4560: operators. If you perform division with potentially negative operands,
4561: you do not want to use @code{/} or @code{/mod} with its undefined
4562: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
4563: former, @pxref{Mixed precision}).
4564: @comment TODO discuss the different division forms and the std approach
4565:
4566: @menu
4567: * Single precision::
4568: * Double precision:: Double-cell integer arithmetic
4569: * Bitwise operations::
4570: * Numeric comparison::
4571: * Mixed precision:: Operations with single and double-cell integers
4572: * Floating Point::
4573: @end menu
4574:
4575: @node Single precision, Double precision, Arithmetic, Arithmetic
4576: @subsection Single precision
4577: @cindex single precision arithmetic words
4578:
4579: @c !! cell undefined
4580:
4581: By default, numbers in Forth are single-precision integers that are one
4582: cell in size. They can be signed or unsigned, depending upon how you
4583: treat them. For the rules used by the text interpreter for recognising
4584: single-precision integers see @ref{Number Conversion}.
4585:
4586: These words are all defined for signed operands, but some of them also
4587: work for unsigned numbers: @code{+}, @code{1+}, @code{-}, @code{1-},
4588: @code{*}.
4589:
4590: doc-+
4591: doc-1+
4592: doc--
4593: doc-1-
4594: doc-*
4595: doc-/
4596: doc-mod
4597: doc-/mod
4598: doc-negate
4599: doc-abs
4600: doc-min
4601: doc-max
4602: doc-floored
4603:
4604:
4605: @node Double precision, Bitwise operations, Single precision, Arithmetic
4606: @subsection Double precision
4607: @cindex double precision arithmetic words
4608:
4609: For the rules used by the text interpreter for
4610: recognising double-precision integers, see @ref{Number Conversion}.
4611:
4612: A double precision number is represented by a cell pair, with the most
4613: significant cell at the TOS. It is trivial to convert an unsigned single
4614: to a double: simply push a @code{0} onto the TOS. Since numbers are
4615: represented by Gforth using 2's complement arithmetic, converting a
4616: signed single to a (signed) double requires sign-extension across the
4617: most significant cell. This can be achieved using @code{s>d}. The moral
4618: of the story is that you cannot convert a number without knowing whether
4619: it represents an unsigned or a signed number.
4620:
4621: These words are all defined for signed operands, but some of them also
4622: work for unsigned numbers: @code{d+}, @code{d-}.
4623:
4624: doc-s>d
4625: doc-d>s
4626: doc-d+
4627: doc-d-
4628: doc-dnegate
4629: doc-dabs
4630: doc-dmin
4631: doc-dmax
4632:
4633:
4634: @node Bitwise operations, Numeric comparison, Double precision, Arithmetic
4635: @subsection Bitwise operations
4636: @cindex bitwise operation words
4637:
4638:
4639: doc-and
4640: doc-or
4641: doc-xor
4642: doc-invert
4643: doc-lshift
4644: doc-rshift
4645: doc-2*
4646: doc-d2*
4647: doc-2/
4648: doc-d2/
4649:
4650:
4651: @node Numeric comparison, Mixed precision, Bitwise operations, Arithmetic
4652: @subsection Numeric comparison
4653: @cindex numeric comparison words
4654:
4655: Note that the words that compare for equality (@code{= <> 0= 0<> d= d<>
4656: d0= d0<>}) work for for both signed and unsigned numbers.
4657:
4658: doc-<
4659: doc-<=
4660: doc-<>
4661: doc-=
4662: doc->
4663: doc->=
4664:
4665: doc-0<
4666: doc-0<=
4667: doc-0<>
4668: doc-0=
4669: doc-0>
4670: doc-0>=
4671:
4672: doc-u<
4673: doc-u<=
4674: @c u<> and u= exist but are the same as <> and =
4675: @c doc-u<>
4676: @c doc-u=
4677: doc-u>
4678: doc-u>=
4679:
4680: doc-within
4681:
4682: doc-d<
4683: doc-d<=
4684: doc-d<>
4685: doc-d=
4686: doc-d>
4687: doc-d>=
4688:
4689: doc-d0<
4690: doc-d0<=
4691: doc-d0<>
4692: doc-d0=
4693: doc-d0>
4694: doc-d0>=
4695:
4696: doc-du<
4697: doc-du<=
4698: @c du<> and du= exist but are the same as d<> and d=
4699: @c doc-du<>
4700: @c doc-du=
4701: doc-du>
4702: doc-du>=
4703:
4704:
4705: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
4706: @subsection Mixed precision
4707: @cindex mixed precision arithmetic words
4708:
4709:
4710: doc-m+
4711: doc-*/
4712: doc-*/mod
4713: doc-m*
4714: doc-um*
4715: doc-m*/
4716: doc-um/mod
4717: doc-fm/mod
4718: doc-sm/rem
4719:
4720:
4721: @node Floating Point, , Mixed precision, Arithmetic
4722: @subsection Floating Point
4723: @cindex floating point arithmetic words
4724:
4725: For the rules used by the text interpreter for
4726: recognising floating-point numbers see @ref{Number Conversion}.
4727:
4728: Gforth has a separate floating point stack, but the documentation uses
4729: the unified notation.@footnote{It's easy to generate the separate
4730: notation from that by just separating the floating-point numbers out:
4731: e.g. @code{( n r1 u r2 -- r3 )} becomes @code{( n u -- ) ( F: r1 r2 --
4732: r3 )}.}
4733:
4734: @cindex floating-point arithmetic, pitfalls
4735: Floating point numbers have a number of unpleasant surprises for the
4736: unwary (e.g., floating point addition is not associative) and even a few
4737: for the wary. You should not use them unless you know what you are doing
4738: or you don't care that the results you get are totally bogus. If you
4739: want to learn about the problems of floating point numbers (and how to
4740: avoid them), you might start with @cite{David Goldberg,
4741: @uref{http://www.validgh.com/goldberg/paper.ps,What Every Computer
4742: Scientist Should Know About Floating-Point Arithmetic}, ACM Computing
4743: Surveys 23(1):5@minus{}48, March 1991}.
4744:
4745:
4746: doc-d>f
4747: doc-f>d
4748: doc-f+
4749: doc-f-
4750: doc-f*
4751: doc-f/
4752: doc-fnegate
4753: doc-fabs
4754: doc-fmax
4755: doc-fmin
4756: doc-floor
4757: doc-fround
4758: doc-f**
4759: doc-fsqrt
4760: doc-fexp
4761: doc-fexpm1
4762: doc-fln
4763: doc-flnp1
4764: doc-flog
4765: doc-falog
4766: doc-f2*
4767: doc-f2/
4768: doc-1/f
4769: doc-precision
4770: doc-set-precision
4771:
4772: @cindex angles in trigonometric operations
4773: @cindex trigonometric operations
4774: Angles in floating point operations are given in radians (a full circle
4775: has 2 pi radians).
4776:
4777: doc-fsin
4778: doc-fcos
4779: doc-fsincos
4780: doc-ftan
4781: doc-fasin
4782: doc-facos
4783: doc-fatan
4784: doc-fatan2
4785: doc-fsinh
4786: doc-fcosh
4787: doc-ftanh
4788: doc-fasinh
4789: doc-facosh
4790: doc-fatanh
4791: doc-pi
4792:
4793: @cindex equality of floats
4794: @cindex floating-point comparisons
4795: One particular problem with floating-point arithmetic is that comparison
4796: for equality often fails when you would expect it to succeed. For this
4797: reason approximate equality is often preferred (but you still have to
4798: know what you are doing). Also note that IEEE NaNs may compare
4799: differently from what you might expect. The comparison words are:
4800:
4801: doc-f~rel
4802: doc-f~abs
4803: doc-f~
4804: doc-f=
4805: doc-f<>
4806:
4807: doc-f<
4808: doc-f<=
4809: doc-f>
4810: doc-f>=
4811:
4812: doc-f0<
4813: doc-f0<=
4814: doc-f0<>
4815: doc-f0=
4816: doc-f0>
4817: doc-f0>=
4818:
4819:
4820: @node Stack Manipulation, Memory, Arithmetic, Words
4821: @section Stack Manipulation
4822: @cindex stack manipulation words
4823:
4824: @cindex floating-point stack in the standard
4825: Gforth maintains a number of separate stacks:
4826:
4827: @cindex data stack
4828: @cindex parameter stack
4829: @itemize @bullet
4830: @item
4831: A data stack (also known as the @dfn{parameter stack}) -- for
4832: characters, cells, addresses, and double cells.
4833:
4834: @cindex floating-point stack
4835: @item
4836: A floating point stack -- for holding floating point (FP) numbers.
4837:
4838: @cindex return stack
4839: @item
4840: A return stack -- for holding the return addresses of colon
4841: definitions and other (non-FP) data.
4842:
4843: @cindex locals stack
4844: @item
4845: A locals stack -- for holding local variables.
4846: @end itemize
4847:
4848: @menu
4849: * Data stack::
4850: * Floating point stack::
4851: * Return stack::
4852: * Locals stack::
4853: * Stack pointer manipulation::
4854: @end menu
4855:
4856: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
4857: @subsection Data stack
4858: @cindex data stack manipulation words
4859: @cindex stack manipulations words, data stack
4860:
4861:
4862: doc-drop
4863: doc-nip
4864: doc-dup
4865: doc-over
4866: doc-tuck
4867: doc-swap
4868: doc-pick
4869: doc-rot
4870: doc--rot
4871: doc-?dup
4872: doc-roll
4873: doc-2drop
4874: doc-2nip
4875: doc-2dup
4876: doc-2over
4877: doc-2tuck
4878: doc-2swap
4879: doc-2rot
4880:
4881:
4882: @node Floating point stack, Return stack, Data stack, Stack Manipulation
4883: @subsection Floating point stack
4884: @cindex floating-point stack manipulation words
4885: @cindex stack manipulation words, floating-point stack
4886:
4887: Whilst every sane Forth has a separate floating-point stack, it is not
4888: strictly required; an ANS Forth system could theoretically keep
4889: floating-point numbers on the data stack. As an additional difficulty,
4890: you don't know how many cells a floating-point number takes. It is
4891: reportedly possible to write words in a way that they work also for a
4892: unified stack model, but we do not recommend trying it. Instead, just
4893: say that your program has an environmental dependency on a separate
4894: floating-point stack.
4895:
4896: doc-floating-stack
4897:
4898: doc-fdrop
4899: doc-fnip
4900: doc-fdup
4901: doc-fover
4902: doc-ftuck
4903: doc-fswap
4904: doc-fpick
4905: doc-frot
4906:
4907:
4908: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
4909: @subsection Return stack
4910: @cindex return stack manipulation words
4911: @cindex stack manipulation words, return stack
4912:
4913: @cindex return stack and locals
4914: @cindex locals and return stack
4915: A Forth system is allowed to keep local variables on the
4916: return stack. This is reasonable, as local variables usually eliminate
4917: the need to use the return stack explicitly. So, if you want to produce
4918: a standard compliant program and you are using local variables in a
4919: word, forget about return stack manipulations in that word (refer to the
4920: standard document for the exact rules).
4921:
4922: doc->r
4923: doc-r>
4924: doc-r@
4925: doc-rdrop
4926: doc-2>r
4927: doc-2r>
4928: doc-2r@
4929: doc-2rdrop
4930:
4931:
4932: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
4933: @subsection Locals stack
4934:
4935: Gforth uses an extra locals stack. It is described, along with the
4936: reasons for its existence, in @ref{Implementation,Implementation of locals}.
4937:
4938: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
4939: @subsection Stack pointer manipulation
4940: @cindex stack pointer manipulation words
4941:
4942: @c removed s0 r0 l0 -- they are obsolete aliases for sp0 rp0 lp0
4943: doc-sp0
4944: doc-sp@
4945: doc-sp!
4946: doc-fp0
4947: doc-fp@
4948: doc-fp!
4949: doc-rp0
4950: doc-rp@
4951: doc-rp!
4952: doc-lp0
4953: doc-lp@
4954: doc-lp!
4955:
4956:
4957: @node Memory, Control Structures, Stack Manipulation, Words
4958: @section Memory
4959: @cindex memory words
4960:
4961: @menu
4962: * Memory model::
4963: * Dictionary allocation::
4964: * Heap Allocation::
4965: * Memory Access::
4966: * Address arithmetic::
4967: * Memory Blocks::
4968: @end menu
4969:
4970: In addition to the standard Forth memory allocation words, there is also
4971: a @uref{http://www.complang.tuwien.ac.at/forth/garbage-collection.zip,
4972: garbage collector}.
4973:
4974: @node Memory model, Dictionary allocation, Memory, Memory
4975: @subsection ANS Forth and Gforth memory models
4976:
4977: @c The ANS Forth description is a mess (e.g., is the heap part of
4978: @c the dictionary?), so let's not stick to closely with it.
4979:
4980: ANS Forth considers a Forth system as consisting of several address
4981: spaces, of which only @dfn{data space} is managed and accessible with
4982: the memory words. Memory not necessarily in data space includes the
4983: stacks, the code (called code space) and the headers (called name
4984: space). In Gforth everything is in data space, but the code for the
4985: primitives is usually read-only.
4986:
4987: Data space is divided into a number of areas: The (data space portion of
4988: the) dictionary@footnote{Sometimes, the term @dfn{dictionary} is used to
4989: refer to the search data structure embodied in word lists and headers,
4990: because it is used for looking up names, just as you would in a
4991: conventional dictionary.}, the heap, and a number of system-allocated
4992: buffers.
4993:
4994: @cindex address arithmetic restrictions, ANS vs. Gforth
4995: @cindex contiguous regions, ANS vs. Gforth
4996: In ANS Forth data space is also divided into contiguous regions. You
4997: can only use address arithmetic within a contiguous region, not between
4998: them. Usually each allocation gives you one contiguous region, but the
4999: dictionary allocation words have additional rules (@pxref{Dictionary
5000: allocation}).
5001:
5002: Gforth provides one big address space, and address arithmetic can be
5003: performed between any addresses. However, in the dictionary headers or
5004: code are interleaved with data, so almost the only contiguous data space
5005: regions there are those described by ANS Forth as contiguous; but you
5006: can be sure that the dictionary is allocated towards increasing
5007: addresses even between contiguous regions. The memory order of
5008: allocations in the heap is platform-dependent (and possibly different
5009: from one run to the next).
5010:
5011:
5012: @node Dictionary allocation, Heap Allocation, Memory model, Memory
5013: @subsection Dictionary allocation
5014: @cindex reserving data space
5015: @cindex data space - reserving some
5016:
5017: Dictionary allocation is a stack-oriented allocation scheme, i.e., if
5018: you want to deallocate X, you also deallocate everything
5019: allocated after X.
5020:
5021: @cindex contiguous regions in dictionary allocation
5022: The allocations using the words below are contiguous and grow the region
5023: towards increasing addresses. Other words that allocate dictionary
5024: memory of any kind (i.e., defining words including @code{:noname}) end
5025: the contiguous region and start a new one.
5026:
5027: In ANS Forth only @code{create}d words are guaranteed to produce an
5028: address that is the start of the following contiguous region. In
5029: particular, the cell allocated by @code{variable} is not guaranteed to
5030: be contiguous with following @code{allot}ed memory.
5031:
5032: You can deallocate memory by using @code{allot} with a negative argument
5033: (with some restrictions, see @code{allot}). For larger deallocations use
5034: @code{marker}.
5035:
5036:
5037: doc-here
5038: doc-unused
5039: doc-allot
5040: doc-c,
5041: doc-f,
5042: doc-,
5043: doc-2,
5044:
5045: Memory accesses have to be aligned (@pxref{Address arithmetic}). So of
5046: course you should allocate memory in an aligned way, too. I.e., before
5047: allocating allocating a cell, @code{here} must be cell-aligned, etc.
5048: The words below align @code{here} if it is not already. Basically it is
5049: only already aligned for a type, if the last allocation was a multiple
5050: of the size of this type and if @code{here} was aligned for this type
5051: before.
5052:
5053: After freshly @code{create}ing a word, @code{here} is @code{align}ed in
5054: ANS Forth (@code{maxalign}ed in Gforth).
5055:
5056: doc-align
5057: doc-falign
5058: doc-sfalign
5059: doc-dfalign
5060: doc-maxalign
5061: doc-cfalign
5062:
5063:
5064: @node Heap Allocation, Memory Access, Dictionary allocation, Memory
5065: @subsection Heap allocation
5066: @cindex heap allocation
5067: @cindex dynamic allocation of memory
5068: @cindex memory-allocation word set
5069:
5070: @cindex contiguous regions and heap allocation
5071: Heap allocation supports deallocation of allocated memory in any
5072: order. Dictionary allocation is not affected by it (i.e., it does not
5073: end a contiguous region). In Gforth, these words are implemented using
5074: the standard C library calls malloc(), free() and resize().
5075:
5076: The memory region produced by one invocation of @code{allocate} or
5077: @code{resize} is internally contiguous. There is no contiguity between
5078: such a region and any other region (including others allocated from the
5079: heap).
5080:
5081: doc-allocate
5082: doc-free
5083: doc-resize
5084:
5085:
5086: @node Memory Access, Address arithmetic, Heap Allocation, Memory
5087: @subsection Memory Access
5088: @cindex memory access words
5089:
5090: doc-@
5091: doc-!
5092: doc-+!
5093: doc-c@
5094: doc-c!
5095: doc-2@
5096: doc-2!
5097: doc-f@
5098: doc-f!
5099: doc-sf@
5100: doc-sf!
5101: doc-df@
5102: doc-df!
5103:
5104:
5105: @node Address arithmetic, Memory Blocks, Memory Access, Memory
5106: @subsection Address arithmetic
5107: @cindex address arithmetic words
5108:
5109: Address arithmetic is the foundation on which you can build data
5110: structures like arrays, records (@pxref{Structures}) and objects
5111: (@pxref{Object-oriented Forth}).
5112:
5113: @cindex address unit
5114: @cindex au (address unit)
5115: ANS Forth does not specify the sizes of the data types. Instead, it
5116: offers a number of words for computing sizes and doing address
5117: arithmetic. Address arithmetic is performed in terms of address units
5118: (aus); on most systems the address unit is one byte. Note that a
5119: character may have more than one au, so @code{chars} is no noop (on
5120: platforms where it is a noop, it compiles to nothing).
5121:
5122: The basic address arithmetic words are @code{+} and @code{-}. E.g., if
5123: you have the address of a cell, perform @code{1 cells +}, and you will
5124: have the address of the next cell.
5125:
5126: @cindex contiguous regions and address arithmetic
5127: In ANS Forth you can perform address arithmetic only within a contiguous
5128: region, i.e., if you have an address into one region, you can only add
5129: and subtract such that the result is still within the region; you can
5130: only subtract or compare addresses from within the same contiguous
5131: region. Reasons: several contiguous regions can be arranged in memory
5132: in any way; on segmented systems addresses may have unusual
5133: representations, such that address arithmetic only works within a
5134: region. Gforth provides a few more guarantees (linear address space,
5135: dictionary grows upwards), but in general I have found it easy to stay
5136: within contiguous regions (exception: computing and comparing to the
5137: address just beyond the end of an array).
5138:
5139: @cindex alignment of addresses for types
5140: ANS Forth also defines words for aligning addresses for specific
5141: types. Many computers require that accesses to specific data types
5142: must only occur at specific addresses; e.g., that cells may only be
5143: accessed at addresses divisible by 4. Even if a machine allows unaligned
5144: accesses, it can usually perform aligned accesses faster.
5145:
5146: For the performance-conscious: alignment operations are usually only
5147: necessary during the definition of a data structure, not during the
5148: (more frequent) accesses to it.
5149:
5150: ANS Forth defines no words for character-aligning addresses. This is not
5151: an oversight, but reflects the fact that addresses that are not
5152: char-aligned have no use in the standard and therefore will not be
5153: created.
5154:
5155: @cindex @code{CREATE} and alignment
5156: ANS Forth guarantees that addresses returned by @code{CREATE}d words
5157: are cell-aligned; in addition, Gforth guarantees that these addresses
5158: are aligned for all purposes.
5159:
5160: Note that the ANS Forth word @code{char} has nothing to do with address
5161: arithmetic.
5162:
5163:
5164: doc-chars
5165: doc-char+
5166: doc-cells
5167: doc-cell+
5168: doc-cell
5169: doc-aligned
5170: doc-floats
5171: doc-float+
5172: doc-float
5173: doc-faligned
5174: doc-sfloats
5175: doc-sfloat+
5176: doc-sfaligned
5177: doc-dfloats
5178: doc-dfloat+
5179: doc-dfaligned
5180: doc-maxaligned
5181: doc-cfaligned
5182: doc-address-unit-bits
5183:
5184:
5185: @node Memory Blocks, , Address arithmetic, Memory
5186: @subsection Memory Blocks
5187: @cindex memory block words
5188: @cindex character strings - moving and copying
5189:
5190: Memory blocks often represent character strings; For ways of storing
5191: character strings in memory see @ref{String Formats}. For other
5192: string-processing words see @ref{Displaying characters and strings}.
5193:
5194: A few of these words work on address unit blocks. In that case, you
5195: usually have to insert @code{CHARS} before the word when working on
5196: character strings. Most words work on character blocks, and expect a
5197: char-aligned address.
5198:
5199: When copying characters between overlapping memory regions, use
5200: @code{chars move} or choose carefully between @code{cmove} and
5201: @code{cmove>}.
5202:
5203: doc-move
5204: doc-erase
5205: doc-cmove
5206: doc-cmove>
5207: doc-fill
5208: doc-blank
5209: doc-compare
5210: doc-search
5211: doc--trailing
5212: doc-/string
5213:
5214:
5215: @comment TODO examples
5216:
5217:
5218: @node Control Structures, Defining Words, Memory, Words
5219: @section Control Structures
5220: @cindex control structures
5221:
5222: Control structures in Forth cannot be used interpretively, only in a
5223: colon definition@footnote{To be precise, they have no interpretation
5224: semantics (@pxref{Interpretation and Compilation Semantics}).}. We do
5225: not like this limitation, but have not seen a satisfying way around it
5226: yet, although many schemes have been proposed.
5227:
5228: @menu
5229: * Selection:: IF ... ELSE ... ENDIF
5230: * Simple Loops:: BEGIN ...
5231: * Counted Loops:: DO
5232: * Arbitrary control structures::
5233: * Calls and returns::
5234: * Exception Handling::
5235: @end menu
5236:
5237: @node Selection, Simple Loops, Control Structures, Control Structures
5238: @subsection Selection
5239: @cindex selection control structures
5240: @cindex control structures for selection
5241:
5242: @cindex @code{IF} control structure
5243: @example
5244: @i{flag}
5245: IF
5246: @i{code}
5247: ENDIF
5248: @end example
5249: @noindent
5250:
5251: If @i{flag} is non-zero (as far as @code{IF} etc. are concerned, a cell
5252: with any bit set represents truth) @i{code} is executed.
5253:
5254: @example
5255: @i{flag}
5256: IF
5257: @i{code1}
5258: ELSE
5259: @i{code2}
5260: ENDIF
5261: @end example
5262:
5263: If @var{flag} is true, @i{code1} is executed, otherwise @i{code2} is
5264: executed.
5265:
5266: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
5267: standard, and @code{ENDIF} is not, although it is quite popular. We
5268: recommend using @code{ENDIF}, because it is less confusing for people
5269: who also know other languages (and is not prone to reinforcing negative
5270: prejudices against Forth in these people). Adding @code{ENDIF} to a
5271: system that only supplies @code{THEN} is simple:
5272: @example
5273: : ENDIF POSTPONE THEN ; immediate
5274: @end example
5275:
5276: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
5277: (adv.)} has the following meanings:
5278: @quotation
5279: ... 2b: following next after in order ... 3d: as a necessary consequence
5280: (if you were there, then you saw them).
5281: @end quotation
5282: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
5283: and many other programming languages has the meaning 3d.]
5284:
5285: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
5286: you can avoid using @code{?dup}. Using these alternatives is also more
5287: efficient than using @code{?dup}. Definitions in ANS Forth
5288: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
5289: @file{compat/control.fs}.
5290:
5291: @cindex @code{CASE} control structure
5292: @example
5293: @i{n}
5294: CASE
5295: @i{n1} OF @i{code1} ENDOF
5296: @i{n2} OF @i{code2} ENDOF
5297: @dots{}
5298: ( n ) @i{default-code} ( n )
5299: ENDCASE
5300: @end example
5301:
5302: Executes the first @i{codei}, where the @i{ni} is equal to @i{n}. If no
5303: @i{ni} matches, the optional @i{default-code} is executed. The optional
5304: default case can be added by simply writing the code after the last
5305: @code{ENDOF}. It may use @i{n}, which is on top of the stack, but must
5306: not consume it.
5307:
5308: @progstyle
5309: To keep the code understandable, you should ensure that on all paths
5310: through a selection construct the stack is changed in the same way
5311: (wrt. number and types of stack items consumed and pushed).
5312:
5313: @node Simple Loops, Counted Loops, Selection, Control Structures
5314: @subsection Simple Loops
5315: @cindex simple loops
5316: @cindex loops without count
5317:
5318: @cindex @code{WHILE} loop
5319: @example
5320: BEGIN
5321: @i{code1}
5322: @i{flag}
5323: WHILE
5324: @i{code2}
5325: REPEAT
5326: @end example
5327:
5328: @i{code1} is executed and @i{flag} is computed. If it is true,
5329: @i{code2} is executed and the loop is restarted; If @i{flag} is
5330: false, execution continues after the @code{REPEAT}.
5331:
5332: @cindex @code{UNTIL} loop
5333: @example
5334: BEGIN
5335: @i{code}
5336: @i{flag}
5337: UNTIL
5338: @end example
5339:
5340: @i{code} is executed. The loop is restarted if @code{flag} is false.
5341:
5342: @progstyle
5343: To keep the code understandable, a complete iteration of the loop should
5344: not change the number and types of the items on the stacks.
5345:
5346: @cindex endless loop
5347: @cindex loops, endless
5348: @example
5349: BEGIN
5350: @i{code}
5351: AGAIN
5352: @end example
5353:
5354: This is an endless loop.
5355:
5356: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
5357: @subsection Counted Loops
5358: @cindex counted loops
5359: @cindex loops, counted
5360: @cindex @code{DO} loops
5361:
5362: The basic counted loop is:
5363: @example
5364: @i{limit} @i{start}
5365: ?DO
5366: @i{body}
5367: LOOP
5368: @end example
5369:
5370: This performs one iteration for every integer, starting from @i{start}
5371: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
5372: accessed with @code{i}. For example, the loop:
5373: @example
5374: 10 0 ?DO
5375: i .
5376: LOOP
5377: @end example
5378: @noindent
5379: prints @code{0 1 2 3 4 5 6 7 8 9}
5380:
5381: The index of the innermost loop can be accessed with @code{i}, the index
5382: of the next loop with @code{j}, and the index of the third loop with
5383: @code{k}.
5384:
5385:
5386: doc-i
5387: doc-j
5388: doc-k
5389:
5390:
5391: The loop control data are kept on the return stack, so there are some
5392: restrictions on mixing return stack accesses and counted loop words. In
5393: particuler, if you put values on the return stack outside the loop, you
5394: cannot read them inside the loop@footnote{well, not in a way that is
5395: portable.}. If you put values on the return stack within a loop, you
5396: have to remove them before the end of the loop and before accessing the
5397: index of the loop.
5398:
5399: There are several variations on the counted loop:
5400:
5401: @itemize @bullet
5402: @item
5403: @code{LEAVE} leaves the innermost counted loop immediately; execution
5404: continues after the associated @code{LOOP} or @code{NEXT}. For example:
5405:
5406: @example
5407: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
5408: @end example
5409: prints @code{0 1 2 3}
5410:
5411:
5412: @item
5413: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
5414: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
5415: return stack so @code{EXIT} can get to its return address. For example:
5416:
5417: @example
5418: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
5419: @end example
5420: prints @code{0 1 2 3}
5421:
5422:
5423: @item
5424: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
5425: (and @code{LOOP} iterates until they become equal by wrap-around
5426: arithmetic). This behaviour is usually not what you want. Therefore,
5427: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
5428: @code{?DO}), which do not enter the loop if @i{start} is greater than
5429: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
5430: unsigned loop parameters.
5431:
5432: @item
5433: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
5434: the loop, independent of the loop parameters. Do not use @code{DO}, even
5435: if you know that the loop is entered in any case. Such knowledge tends
5436: to become invalid during maintenance of a program, and then the
5437: @code{DO} will make trouble.
5438:
5439: @item
5440: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
5441: index by @i{n} instead of by 1. The loop is terminated when the border
5442: between @i{limit-1} and @i{limit} is crossed. E.g.:
5443:
5444: @example
5445: 4 0 +DO i . 2 +LOOP
5446: @end example
5447: @noindent
5448: prints @code{0 2}
5449:
5450: @example
5451: 4 1 +DO i . 2 +LOOP
5452: @end example
5453: @noindent
5454: prints @code{1 3}
5455:
5456: @item
5457: @cindex negative increment for counted loops
5458: @cindex counted loops with negative increment
5459: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
5460:
5461: @example
5462: -1 0 ?DO i . -1 +LOOP
5463: @end example
5464: @noindent
5465: prints @code{0 -1}
5466:
5467: @example
5468: 0 0 ?DO i . -1 +LOOP
5469: @end example
5470: prints nothing.
5471:
5472: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
5473: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
5474: index by @i{u} each iteration. The loop is terminated when the border
5475: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
5476: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
5477:
5478: @example
5479: -2 0 -DO i . 1 -LOOP
5480: @end example
5481: @noindent
5482: prints @code{0 -1}
5483:
5484: @example
5485: -1 0 -DO i . 1 -LOOP
5486: @end example
5487: @noindent
5488: prints @code{0}
5489:
5490: @example
5491: 0 0 -DO i . 1 -LOOP
5492: @end example
5493: @noindent
5494: prints nothing.
5495:
5496: @end itemize
5497:
5498: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
5499: @code{-LOOP} are not defined in ANS Forth. However, an implementation
5500: for these words that uses only standard words is provided in
5501: @file{compat/loops.fs}.
5502:
5503:
5504: @cindex @code{FOR} loops
5505: Another counted loop is:
5506: @example
5507: @i{n}
5508: FOR
5509: @i{body}
5510: NEXT
5511: @end example
5512: This is the preferred loop of native code compiler writers who are too
5513: lazy to optimize @code{?DO} loops properly. This loop structure is not
5514: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
5515: @code{i} produces values starting with @i{n} and ending with 0. Other
5516: Forth systems may behave differently, even if they support @code{FOR}
5517: loops. To avoid problems, don't use @code{FOR} loops.
5518:
5519: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
5520: @subsection Arbitrary control structures
5521: @cindex control structures, user-defined
5522:
5523: @cindex control-flow stack
5524: ANS Forth permits and supports using control structures in a non-nested
5525: way. Information about incomplete control structures is stored on the
5526: control-flow stack. This stack may be implemented on the Forth data
5527: stack, and this is what we have done in Gforth.
5528:
5529: @cindex @code{orig}, control-flow stack item
5530: @cindex @code{dest}, control-flow stack item
5531: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
5532: entry represents a backward branch target. A few words are the basis for
5533: building any control structure possible (except control structures that
5534: need storage, like calls, coroutines, and backtracking).
5535:
5536:
5537: doc-if
5538: doc-ahead
5539: doc-then
5540: doc-begin
5541: doc-until
5542: doc-again
5543: doc-cs-pick
5544: doc-cs-roll
5545:
5546:
5547: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
5548: manipulate the control-flow stack in a portable way. Without them, you
5549: would need to know how many stack items are occupied by a control-flow
5550: entry (many systems use one cell. In Gforth they currently take three,
5551: but this may change in the future).
5552:
5553: Some standard control structure words are built from these words:
5554:
5555:
5556: doc-else
5557: doc-while
5558: doc-repeat
5559:
5560:
5561: @noindent
5562: Gforth adds some more control-structure words:
5563:
5564:
5565: doc-endif
5566: doc-?dup-if
5567: doc-?dup-0=-if
5568:
5569:
5570: @noindent
5571: Counted loop words constitute a separate group of words:
5572:
5573:
5574: doc-?do
5575: doc-+do
5576: doc-u+do
5577: doc--do
5578: doc-u-do
5579: doc-do
5580: doc-for
5581: doc-loop
5582: doc-+loop
5583: doc--loop
5584: doc-next
5585: doc-leave
5586: doc-?leave
5587: doc-unloop
5588: doc-done
5589:
5590:
5591: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
5592: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
5593: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
5594: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
5595: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
5596: resolved (by using one of the loop-ending words or @code{DONE}).
5597:
5598: @noindent
5599: Another group of control structure words are:
5600:
5601:
5602: doc-case
5603: doc-endcase
5604: doc-of
5605: doc-endof
5606:
5607:
5608: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
5609: @code{CS-ROLL}.
5610:
5611: @subsubsection Programming Style
5612: @cindex control structures programming style
5613: @cindex programming style, arbitrary control structures
5614:
5615: In order to ensure readability we recommend that you do not create
5616: arbitrary control structures directly, but define new control structure
5617: words for the control structure you want and use these words in your
5618: program. For example, instead of writing:
5619:
5620: @example
5621: BEGIN
5622: ...
5623: IF [ 1 CS-ROLL ]
5624: ...
5625: AGAIN THEN
5626: @end example
5627:
5628: @noindent
5629: we recommend defining control structure words, e.g.,
5630:
5631: @example
5632: : WHILE ( DEST -- ORIG DEST )
5633: POSTPONE IF
5634: 1 CS-ROLL ; immediate
5635:
5636: : REPEAT ( orig dest -- )
5637: POSTPONE AGAIN
5638: POSTPONE THEN ; immediate
5639: @end example
5640:
5641: @noindent
5642: and then using these to create the control structure:
5643:
5644: @example
5645: BEGIN
5646: ...
5647: WHILE
5648: ...
5649: REPEAT
5650: @end example
5651:
5652: That's much easier to read, isn't it? Of course, @code{REPEAT} and
5653: @code{WHILE} are predefined, so in this example it would not be
5654: necessary to define them.
5655:
5656: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
5657: @subsection Calls and returns
5658: @cindex calling a definition
5659: @cindex returning from a definition
5660:
5661: @cindex recursive definitions
5662: A definition can be called simply be writing the name of the definition
5663: to be called. Normally a definition is invisible during its own
5664: definition. If you want to write a directly recursive definition, you
5665: can use @code{recursive} to make the current definition visible, or
5666: @code{recurse} to call the current definition directly.
5667:
5668:
5669: doc-recursive
5670: doc-recurse
5671:
5672:
5673: @comment TODO add example of the two recursion methods
5674: @quotation
5675: @progstyle
5676: I prefer using @code{recursive} to @code{recurse}, because calling the
5677: definition by name is more descriptive (if the name is well-chosen) than
5678: the somewhat cryptic @code{recurse}. E.g., in a quicksort
5679: implementation, it is much better to read (and think) ``now sort the
5680: partitions'' than to read ``now do a recursive call''.
5681: @end quotation
5682:
5683: For mutual recursion, use @code{Defer}red words, like this:
5684:
5685: @example
5686: Defer foo
5687:
5688: : bar ( ... -- ... )
5689: ... foo ... ;
5690:
5691: :noname ( ... -- ... )
5692: ... bar ... ;
5693: IS foo
5694: @end example
5695:
5696: Deferred words are discussed in more detail in @ref{Deferred words}.
5697:
5698: The current definition returns control to the calling definition when
5699: the end of the definition is reached or @code{EXIT} is encountered.
5700:
5701: doc-exit
5702: doc-;s
5703:
5704:
5705: @node Exception Handling, , Calls and returns, Control Structures
5706: @subsection Exception Handling
5707: @cindex exceptions
5708:
5709: @c quit is a very bad idea for error handling,
5710: @c because it does not translate into a THROW
5711: @c it also does not belong into this chapter
5712:
5713: If a word detects an error condition that it cannot handle, it can
5714: @code{throw} an exception. In the simplest case, this will terminate
5715: your program, and report an appropriate error.
5716:
5717: doc-throw
5718:
5719: @code{Throw} consumes a cell-sized error number on the stack. There are
5720: some predefined error numbers in ANS Forth (see @file{errors.fs}). In
5721: Gforth (and most other systems) you can use the iors produced by various
5722: words as error numbers (e.g., a typical use of @code{allocate} is
5723: @code{allocate throw}). Gforth also provides the word @code{exception}
5724: to define your own error numbers (with decent error reporting); an ANS
5725: Forth version of this word (but without the error messages) is available
5726: in @code{compat/except.fs}. And finally, you can use your own error
5727: numbers (anything outside the range -4095..0), but won't get nice error
5728: messages, only numbers. For example, try:
5729:
5730: @example
5731: -10 throw \ ANS defined
5732: -267 throw \ system defined
5733: s" my error" exception throw \ user defined
5734: 7 throw \ arbitrary number
5735: @end example
5736:
5737: doc---exception-exception
5738:
5739: A common idiom to @code{THROW} a specific error if a flag is true is
5740: this:
5741:
5742: @example
5743: @code{( flag ) 0<> @i{errno} and throw}
5744: @end example
5745:
5746: Your program can provide exception handlers to catch exceptions. An
5747: exception handler can be used to correct the problem, or to clean up
5748: some data structures and just throw the exception to the next exception
5749: handler. Note that @code{throw} jumps to the dynamically innermost
5750: exception handler. The system's exception handler is outermost, and just
5751: prints an error and restarts command-line interpretation (or, in batch
5752: mode (i.e., while processing the shell command line), leaves Gforth).
5753:
5754: The ANS Forth way to catch exceptions is @code{catch}:
5755:
5756: doc-catch
5757:
5758: The most common use of exception handlers is to clean up the state when
5759: an error happens. E.g.,
5760:
5761: @example
5762: base @ >r hex \ actually the hex should be inside foo, or we h
5763: ['] foo catch ( nerror|0 )
5764: r> base !
5765: ( nerror|0 ) throw \ pass it on
5766: @end example
5767:
5768: A use of @code{catch} for handling the error @code{myerror} might look
5769: like this:
5770:
5771: @example
5772: ['] foo catch
5773: CASE
5774: myerror OF ... ( do something about it ) ENDOF
5775: dup throw \ default: pass other errors on, do nothing on non-errors
5776: ENDCASE
5777: @end example
5778:
5779: Having to wrap the code into a separate word is often cumbersome,
5780: therefore Gforth provides an alternative syntax:
5781:
5782: @example
5783: TRY
5784: @i{code1}
5785: RECOVER \ optional
5786: @i{code2} \ optional
5787: ENDTRY
5788: @end example
5789:
5790: This performs @i{Code1}. If @i{code1} completes normally, execution
5791: continues after the @code{endtry}. If @i{Code1} throws, the stacks are
5792: reset to the state during @code{try}, the throw value is pushed on the
5793: data stack, and execution constinues at @i{code2}, and finally falls
5794: through the @code{endtry} into the following code. If there is no
5795: @code{recover} clause, this works like an empty recover clause.
5796:
5797: doc-try
5798: doc-recover
5799: doc-endtry
5800:
5801: The cleanup example from above in this syntax:
5802:
5803: @example
5804: base @ >r TRY
5805: hex foo \ now the hex is placed correctly
5806: 0 \ value for throw
5807: ENDTRY
5808: r> base ! throw
5809: @end example
5810:
5811: And here's the error handling example:
5812:
5813: @example
5814: TRY
5815: foo
5816: RECOVER
5817: CASE
5818: myerror OF ... ( do something about it ) ENDOF
5819: throw \ pass other errors on
5820: ENDCASE
5821: ENDTRY
5822: @end example
5823:
5824: @progstyle
5825: As usual, you should ensure that the stack depth is statically known at
5826: the end: either after the @code{throw} for passing on errors, or after
5827: the @code{ENDTRY} (or, if you use @code{catch}, after the end of the
5828: selection construct for handling the error).
5829:
5830: There are two alternatives to @code{throw}: @code{Abort"} is conditional
5831: and you can provide an error message. @code{Abort} just produces an
5832: ``Aborted'' error.
5833:
5834: The problem with these words is that exception handlers cannot
5835: differentiate between different @code{abort"}s; they just look like
5836: @code{-2 throw} to them (the error message cannot be accessed by
5837: standard programs). Similar @code{abort} looks like @code{-1 throw} to
5838: exception handlers.
5839:
5840: doc-abort"
5841: doc-abort
5842:
5843:
5844:
5845: @c -------------------------------------------------------------
5846: @node Defining Words, Interpretation and Compilation Semantics, Control Structures, Words
5847: @section Defining Words
5848: @cindex defining words
5849:
5850: Defining words are used to extend Forth by creating new entries in the dictionary.
5851:
5852: @menu
5853: * CREATE::
5854: * Variables:: Variables and user variables
5855: * Constants::
5856: * Values:: Initialised variables
5857: * Colon Definitions::
5858: * Anonymous Definitions:: Definitions without names
5859: * Supplying names:: Passing definition names as strings
5860: * User-defined Defining Words::
5861: * Deferred words:: Allow forward references
5862: * Aliases::
5863: @end menu
5864:
5865: @node CREATE, Variables, Defining Words, Defining Words
5866: @subsection @code{CREATE}
5867: @cindex simple defining words
5868: @cindex defining words, simple
5869:
5870: Defining words are used to create new entries in the dictionary. The
5871: simplest defining word is @code{CREATE}. @code{CREATE} is used like
5872: this:
5873:
5874: @example
5875: CREATE new-word1
5876: @end example
5877:
5878: @code{CREATE} is a parsing word, i.e., it takes an argument from the
5879: input stream (@code{new-word1} in our example). It generates a
5880: dictionary entry for @code{new-word1}. When @code{new-word1} is
5881: executed, all that it does is leave an address on the stack. The address
5882: represents the value of the data space pointer (@code{HERE}) at the time
5883: that @code{new-word1} was defined. Therefore, @code{CREATE} is a way of
5884: associating a name with the address of a region of memory.
5885:
5886: doc-create
5887:
5888: Note that in ANS Forth guarantees only for @code{create} that its body
5889: is in dictionary data space (i.e., where @code{here}, @code{allot}
5890: etc. work, @pxref{Dictionary allocation}). Also, in ANS Forth only
5891: @code{create}d words can be modified with @code{does>}
5892: (@pxref{User-defined Defining Words}). And in ANS Forth @code{>body}
5893: can only be applied to @code{create}d words.
5894:
5895: By extending this example to reserve some memory in data space, we end
5896: up with something like a @i{variable}. Here are two different ways to do
5897: it:
5898:
5899: @example
5900: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
5901: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
5902: @end example
5903:
5904: The variable can be examined and modified using @code{@@} (``fetch'') and
5905: @code{!} (``store'') like this:
5906:
5907: @example
5908: new-word2 @@ . \ get address, fetch from it and display
5909: 1234 new-word2 ! \ new value, get address, store to it
5910: @end example
5911:
5912: @cindex arrays
5913: A similar mechanism can be used to create arrays. For example, an
5914: 80-character text input buffer:
5915:
5916: @example
5917: CREATE text-buf 80 chars allot
5918:
5919: text-buf 0 chars c@@ \ the 1st character (offset 0)
5920: text-buf 3 chars c@@ \ the 4th character (offset 3)
5921: @end example
5922:
5923: You can build arbitrarily complex data structures by allocating
5924: appropriate areas of memory. For further discussions of this, and to
5925: learn about some Gforth tools that make it easier,
5926: @xref{Structures}.
5927:
5928:
5929: @node Variables, Constants, CREATE, Defining Words
5930: @subsection Variables
5931: @cindex variables
5932:
5933: The previous section showed how a sequence of commands could be used to
5934: generate a variable. As a final refinement, the whole code sequence can
5935: be wrapped up in a defining word (pre-empting the subject of the next
5936: section), making it easier to create new variables:
5937:
5938: @example
5939: : myvariableX ( "name" -- a-addr ) CREATE 1 cells allot ;
5940: : myvariable0 ( "name" -- a-addr ) CREATE 0 , ;
5941:
5942: myvariableX foo \ variable foo starts off with an unknown value
5943: myvariable0 joe \ whilst joe is initialised to 0
5944:
5945: 45 3 * foo ! \ set foo to 135
5946: 1234 joe ! \ set joe to 1234
5947: 3 joe +! \ increment joe by 3.. to 1237
5948: @end example
5949:
5950: Not surprisingly, there is no need to define @code{myvariable}, since
5951: Forth already has a definition @code{Variable}. ANS Forth does not
5952: guarantee that a @code{Variable} is initialised when it is created
5953: (i.e., it may behave like @code{myvariableX}). In contrast, Gforth's
5954: @code{Variable} initialises the variable to 0 (i.e., it behaves exactly
5955: like @code{myvariable0}). Forth also provides @code{2Variable} and
5956: @code{fvariable} for double and floating-point variables, respectively
5957: -- they are initialised to 0. and 0e in Gforth. If you use a @code{Variable} to
5958: store a boolean, you can use @code{on} and @code{off} to toggle its
5959: state.
5960:
5961: doc-variable
5962: doc-2variable
5963: doc-fvariable
5964:
5965: @cindex user variables
5966: @cindex user space
5967: The defining word @code{User} behaves in the same way as @code{Variable}.
5968: The difference is that it reserves space in @i{user (data) space} rather
5969: than normal data space. In a Forth system that has a multi-tasker, each
5970: task has its own set of user variables.
5971:
5972: doc-user
5973: @c doc-udp
5974: @c doc-uallot
5975:
5976: @comment TODO is that stuff about user variables strictly correct? Is it
5977: @comment just terminal tasks that have user variables?
5978: @comment should document tasker.fs (with some examples) elsewhere
5979: @comment in this manual, then expand on user space and user variables.
5980:
5981: @node Constants, Values, Variables, Defining Words
5982: @subsection Constants
5983: @cindex constants
5984:
5985: @code{Constant} allows you to declare a fixed value and refer to it by
5986: name. For example:
5987:
5988: @example
5989: 12 Constant INCHES-PER-FOOT
5990: 3E+08 fconstant SPEED-O-LIGHT
5991: @end example
5992:
5993: A @code{Variable} can be both read and written, so its run-time
5994: behaviour is to supply an address through which its current value can be
5995: manipulated. In contrast, the value of a @code{Constant} cannot be
5996: changed once it has been declared@footnote{Well, often it can be -- but
5997: not in a Standard, portable way. It's safer to use a @code{Value} (read
5998: on).} so it's not necessary to supply the address -- it is more
5999: efficient to return the value of the constant directly. That's exactly
6000: what happens; the run-time effect of a constant is to put its value on
6001: the top of the stack (You can find one
6002: way of implementing @code{Constant} in @ref{User-defined Defining Words}).
6003:
6004: Forth also provides @code{2Constant} and @code{fconstant} for defining
6005: double and floating-point constants, respectively.
6006:
6007: doc-constant
6008: doc-2constant
6009: doc-fconstant
6010:
6011: @c that's too deep, and it's not necessarily true for all ANS Forths. - anton
6012: @c nac-> How could that not be true in an ANS Forth? You can't define a
6013: @c constant, use it and then delete the definition of the constant..
6014:
6015: @c anton->An ANS Forth system can compile a constant to a literal; On
6016: @c decompilation you would see only the number, just as if it had been used
6017: @c in the first place. The word will stay, of course, but it will only be
6018: @c used by the text interpreter (no run-time duties, except when it is
6019: @c POSTPONEd or somesuch).
6020:
6021: @c nac:
6022: @c I agree that it's rather deep, but IMO it is an important difference
6023: @c relative to other programming languages.. often it's annoying: it
6024: @c certainly changes my programming style relative to C.
6025:
6026: @c anton: In what way?
6027:
6028: Constants in Forth behave differently from their equivalents in other
6029: programming languages. In other languages, a constant (such as an EQU in
6030: assembler or a #define in C) only exists at compile-time; in the
6031: executable program the constant has been translated into an absolute
6032: number and, unless you are using a symbolic debugger, it's impossible to
6033: know what abstract thing that number represents. In Forth a constant has
6034: an entry in the header space and remains there after the code that uses
6035: it has been defined. In fact, it must remain in the dictionary since it
6036: has run-time duties to perform. For example:
6037:
6038: @example
6039: 12 Constant INCHES-PER-FOOT
6040: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
6041: @end example
6042:
6043: @cindex in-lining of constants
6044: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
6045: associated with the constant @code{INCHES-PER-FOOT}. If you use
6046: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
6047: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
6048: attempt to optimise constants by in-lining them where they are used. You
6049: can force Gforth to in-line a constant like this:
6050:
6051: @example
6052: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
6053: @end example
6054:
6055: If you use @code{see} to decompile @i{this} version of
6056: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
6057: longer present. To understand how this works, read
6058: @ref{Interpret/Compile states}, and @ref{Literals}.
6059:
6060: In-lining constants in this way might improve execution time
6061: fractionally, and can ensure that a constant is now only referenced at
6062: compile-time. However, the definition of the constant still remains in
6063: the dictionary. Some Forth compilers provide a mechanism for controlling
6064: a second dictionary for holding transient words such that this second
6065: dictionary can be deleted later in order to recover memory
6066: space. However, there is no standard way of doing this.
6067:
6068:
6069: @node Values, Colon Definitions, Constants, Defining Words
6070: @subsection Values
6071: @cindex values
6072:
6073: A @code{Value} behaves like a @code{Constant}, but it can be changed.
6074: @code{TO} is a parsing word that changes a @code{Values}. In Gforth
6075: (not in ANS Forth) you can access (and change) a @code{value} also with
6076: @code{>body}.
6077:
6078: Here are some
6079: examples:
6080:
6081: @example
6082: 12 Value APPLES \ Define APPLES with an initial value of 12
6083: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
6084: 1 ' APPLES >body +! \ Increment APPLES. Non-standard usage.
6085: APPLES \ puts 35 on the top of the stack.
6086: @end example
6087:
6088: doc-value
6089: doc-to
6090:
6091:
6092:
6093: @node Colon Definitions, Anonymous Definitions, Values, Defining Words
6094: @subsection Colon Definitions
6095: @cindex colon definitions
6096:
6097: @example
6098: : name ( ... -- ... )
6099: word1 word2 word3 ;
6100: @end example
6101:
6102: @noindent
6103: Creates a word called @code{name} that, upon execution, executes
6104: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
6105:
6106: The explanation above is somewhat superficial. For simple examples of
6107: colon definitions see @ref{Your first definition}. For an in-depth
6108: discussion of some of the issues involved, @xref{Interpretation and
6109: Compilation Semantics}.
6110:
6111: doc-:
6112: doc-;
6113:
6114:
6115: @node Anonymous Definitions, Supplying names, Colon Definitions, Defining Words
6116: @subsection Anonymous Definitions
6117: @cindex colon definitions
6118: @cindex defining words without name
6119:
6120: Sometimes you want to define an @dfn{anonymous word}; a word without a
6121: name. You can do this with:
6122:
6123: doc-:noname
6124:
6125: This leaves the execution token for the word on the stack after the
6126: closing @code{;}. Here's an example in which a deferred word is
6127: initialised with an @code{xt} from an anonymous colon definition:
6128:
6129: @example
6130: Defer deferred
6131: :noname ( ... -- ... )
6132: ... ;
6133: IS deferred
6134: @end example
6135:
6136: @noindent
6137: Gforth provides an alternative way of doing this, using two separate
6138: words:
6139:
6140: doc-noname
6141: @cindex execution token of last defined word
6142: doc-lastxt
6143:
6144: @noindent
6145: The previous example can be rewritten using @code{noname} and
6146: @code{lastxt}:
6147:
6148: @example
6149: Defer deferred
6150: noname : ( ... -- ... )
6151: ... ;
6152: lastxt IS deferred
6153: @end example
6154:
6155: @noindent
6156: @code{noname} works with any defining word, not just @code{:}.
6157:
6158: @code{lastxt} also works when the last word was not defined as
6159: @code{noname}. It does not work for combined words, though. It also has
6160: the useful property that is is valid as soon as the header for a
6161: definition has been built. Thus:
6162:
6163: @example
6164: lastxt . : foo [ lastxt . ] ; ' foo .
6165: @end example
6166:
6167: @noindent
6168: prints 3 numbers; the last two are the same.
6169:
6170: @node Supplying names, User-defined Defining Words, Anonymous Definitions, Defining Words
6171: @subsection Supplying the name of a defined word
6172: @cindex names for defined words
6173: @cindex defining words, name given in a string
6174:
6175: By default, a defining word takes the name for the defined word from the
6176: input stream. Sometimes you want to supply the name from a string. You
6177: can do this with:
6178:
6179: doc-nextname
6180:
6181: For example:
6182:
6183: @example
6184: s" foo" nextname create
6185: @end example
6186:
6187: @noindent
6188: is equivalent to:
6189:
6190: @example
6191: create foo
6192: @end example
6193:
6194: @noindent
6195: @code{nextname} works with any defining word.
6196:
6197:
6198: @node User-defined Defining Words, Deferred words, Supplying names, Defining Words
6199: @subsection User-defined Defining Words
6200: @cindex user-defined defining words
6201: @cindex defining words, user-defined
6202:
6203: You can create a new defining word by wrapping defining-time code around
6204: an existing defining word and putting the sequence in a colon
6205: definition.
6206:
6207: @c anton: This example is very complex and leads in a quite different
6208: @c direction from the CREATE-DOES> stuff that follows. It should probably
6209: @c be done elsewhere, or as a subsubsection of this subsection (or as a
6210: @c subsection of Defining Words)
6211:
6212: For example, suppose that you have a word @code{stats} that
6213: gathers statistics about colon definitions given the @i{xt} of the
6214: definition, and you want every colon definition in your application to
6215: make a call to @code{stats}. You can define and use a new version of
6216: @code{:} like this:
6217:
6218: @example
6219: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
6220: ... ; \ other code
6221:
6222: : my: : lastxt postpone literal ['] stats compile, ;
6223:
6224: my: foo + - ;
6225: @end example
6226:
6227: When @code{foo} is defined using @code{my:} these steps occur:
6228:
6229: @itemize @bullet
6230: @item
6231: @code{my:} is executed.
6232: @item
6233: The @code{:} within the definition (the one between @code{my:} and
6234: @code{lastxt}) is executed, and does just what it always does; it parses
6235: the input stream for a name, builds a dictionary header for the name
6236: @code{foo} and switches @code{state} from interpret to compile.
6237: @item
6238: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
6239: being defined -- @code{foo} -- onto the stack.
6240: @item
6241: The code that was produced by @code{postpone literal} is executed; this
6242: causes the value on the stack to be compiled as a literal in the code
6243: area of @code{foo}.
6244: @item
6245: The code @code{['] stats} compiles a literal into the definition of
6246: @code{my:}. When @code{compile,} is executed, that literal -- the
6247: execution token for @code{stats} -- is layed down in the code area of
6248: @code{foo} , following the literal@footnote{Strictly speaking, the
6249: mechanism that @code{compile,} uses to convert an @i{xt} into something
6250: in the code area is implementation-dependent. A threaded implementation
6251: might spit out the execution token directly whilst another
6252: implementation might spit out a native code sequence.}.
6253: @item
6254: At this point, the execution of @code{my:} is complete, and control
6255: returns to the text interpreter. The text interpreter is in compile
6256: state, so subsequent text @code{+ -} is compiled into the definition of
6257: @code{foo} and the @code{;} terminates the definition as always.
6258: @end itemize
6259:
6260: You can use @code{see} to decompile a word that was defined using
6261: @code{my:} and see how it is different from a normal @code{:}
6262: definition. For example:
6263:
6264: @example
6265: : bar + - ; \ like foo but using : rather than my:
6266: see bar
6267: : bar
6268: + - ;
6269: see foo
6270: : foo
6271: 107645672 stats + - ;
6272:
6273: \ use ' stats . to show that 107645672 is the xt for stats
6274: @end example
6275:
6276: You can use techniques like this to make new defining words in terms of
6277: @i{any} existing defining word.
6278:
6279:
6280: @cindex defining defining words
6281: @cindex @code{CREATE} ... @code{DOES>}
6282: If you want the words defined with your defining words to behave
6283: differently from words defined with standard defining words, you can
6284: write your defining word like this:
6285:
6286: @example
6287: : def-word ( "name" -- )
6288: CREATE @i{code1}
6289: DOES> ( ... -- ... )
6290: @i{code2} ;
6291:
6292: def-word name
6293: @end example
6294:
6295: @cindex child words
6296: This fragment defines a @dfn{defining word} @code{def-word} and then
6297: executes it. When @code{def-word} executes, it @code{CREATE}s a new
6298: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
6299: is not executed at this time. The word @code{name} is sometimes called a
6300: @dfn{child} of @code{def-word}.
6301:
6302: When you execute @code{name}, the address of the body of @code{name} is
6303: put on the data stack and @i{code2} is executed (the address of the body
6304: of @code{name} is the address @code{HERE} returns immediately after the
6305: @code{CREATE}, i.e., the address a @code{create}d word returns by
6306: default).
6307:
6308: @c anton:
6309: @c www.dictionary.com says:
6310: @c at·a·vism: 1.The reappearance of a characteristic in an organism after
6311: @c several generations of absence, usually caused by the chance
6312: @c recombination of genes. 2.An individual or a part that exhibits
6313: @c atavism. Also called throwback. 3.The return of a trait or recurrence
6314: @c of previous behavior after a period of absence.
6315: @c
6316: @c Doesn't seem to fit.
6317:
6318: @c @cindex atavism in child words
6319: You can use @code{def-word} to define a set of child words that behave
6320: similarly; they all have a common run-time behaviour determined by
6321: @i{code2}. Typically, the @i{code1} sequence builds a data area in the
6322: body of the child word. The structure of the data is common to all
6323: children of @code{def-word}, but the data values are specific -- and
6324: private -- to each child word. When a child word is executed, the
6325: address of its private data area is passed as a parameter on TOS to be
6326: used and manipulated@footnote{It is legitimate both to read and write to
6327: this data area.} by @i{code2}.
6328:
6329: The two fragments of code that make up the defining words act (are
6330: executed) at two completely separate times:
6331:
6332: @itemize @bullet
6333: @item
6334: At @i{define time}, the defining word executes @i{code1} to generate a
6335: child word
6336: @item
6337: At @i{child execution time}, when a child word is invoked, @i{code2}
6338: is executed, using parameters (data) that are private and specific to
6339: the child word.
6340: @end itemize
6341:
6342: Another way of understanding the behaviour of @code{def-word} and
6343: @code{name} is to say that, if you make the following definitions:
6344: @example
6345: : def-word1 ( "name" -- )
6346: CREATE @i{code1} ;
6347:
6348: : action1 ( ... -- ... )
6349: @i{code2} ;
6350:
6351: def-word1 name1
6352: @end example
6353:
6354: @noindent
6355: Then using @code{name1 action1} is equivalent to using @code{name}.
6356:
6357: The classic example is that you can define @code{CONSTANT} in this way:
6358:
6359: @example
6360: : CONSTANT ( w "name" -- )
6361: CREATE ,
6362: DOES> ( -- w )
6363: @@ ;
6364: @end example
6365:
6366: @comment There is a beautiful description of how this works and what
6367: @comment it does in the Forthwrite 100th edition.. as well as an elegant
6368: @comment commentary on the Counting Fruits problem.
6369:
6370: When you create a constant with @code{5 CONSTANT five}, a set of
6371: define-time actions take place; first a new word @code{five} is created,
6372: then the value 5 is laid down in the body of @code{five} with
6373: @code{,}. When @code{five} is executed, the address of the body is put on
6374: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
6375: no code of its own; it simply contains a data field and a pointer to the
6376: code that follows @code{DOES>} in its defining word. That makes words
6377: created in this way very compact.
6378:
6379: The final example in this section is intended to remind you that space
6380: reserved in @code{CREATE}d words is @i{data} space and therefore can be
6381: both read and written by a Standard program@footnote{Exercise: use this
6382: example as a starting point for your own implementation of @code{Value}
6383: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
6384: @code{[']}.}:
6385:
6386: @example
6387: : foo ( "name" -- )
6388: CREATE -1 ,
6389: DOES> ( -- )
6390: @@ . ;
6391:
6392: foo first-word
6393: foo second-word
6394:
6395: 123 ' first-word >BODY !
6396: @end example
6397:
6398: If @code{first-word} had been a @code{CREATE}d word, we could simply
6399: have executed it to get the address of its data field. However, since it
6400: was defined to have @code{DOES>} actions, its execution semantics are to
6401: perform those @code{DOES>} actions. To get the address of its data field
6402: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
6403: translate the xt into the address of the data field. When you execute
6404: @code{first-word}, it will display @code{123}. When you execute
6405: @code{second-word} it will display @code{-1}.
6406:
6407: @cindex stack effect of @code{DOES>}-parts
6408: @cindex @code{DOES>}-parts, stack effect
6409: In the examples above the stack comment after the @code{DOES>} specifies
6410: the stack effect of the defined words, not the stack effect of the
6411: following code (the following code expects the address of the body on
6412: the top of stack, which is not reflected in the stack comment). This is
6413: the convention that I use and recommend (it clashes a bit with using
6414: locals declarations for stack effect specification, though).
6415:
6416: @menu
6417: * CREATE..DOES> applications::
6418: * CREATE..DOES> details::
6419: * Advanced does> usage example::
6420: @end menu
6421:
6422: @node CREATE..DOES> applications, CREATE..DOES> details, User-defined Defining Words, User-defined Defining Words
6423: @subsubsection Applications of @code{CREATE..DOES>}
6424: @cindex @code{CREATE} ... @code{DOES>}, applications
6425:
6426: You may wonder how to use this feature. Here are some usage patterns:
6427:
6428: @cindex factoring similar colon definitions
6429: When you see a sequence of code occurring several times, and you can
6430: identify a meaning, you will factor it out as a colon definition. When
6431: you see similar colon definitions, you can factor them using
6432: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
6433: that look very similar:
6434: @example
6435: : ori, ( reg-target reg-source n -- )
6436: 0 asm-reg-reg-imm ;
6437: : andi, ( reg-target reg-source n -- )
6438: 1 asm-reg-reg-imm ;
6439: @end example
6440:
6441: @noindent
6442: This could be factored with:
6443: @example
6444: : reg-reg-imm ( op-code -- )
6445: CREATE ,
6446: DOES> ( reg-target reg-source n -- )
6447: @@ asm-reg-reg-imm ;
6448:
6449: 0 reg-reg-imm ori,
6450: 1 reg-reg-imm andi,
6451: @end example
6452:
6453: @cindex currying
6454: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
6455: supply a part of the parameters for a word (known as @dfn{currying} in
6456: the functional language community). E.g., @code{+} needs two
6457: parameters. Creating versions of @code{+} with one parameter fixed can
6458: be done like this:
6459: @example
6460: : curry+ ( n1 -- )
6461: CREATE ,
6462: DOES> ( n2 -- n1+n2 )
6463: @@ + ;
6464:
6465: 3 curry+ 3+
6466: -2 curry+ 2-
6467: @end example
6468:
6469: @node CREATE..DOES> details, Advanced does> usage example, CREATE..DOES> applications, User-defined Defining Words
6470: @subsubsection The gory details of @code{CREATE..DOES>}
6471: @cindex @code{CREATE} ... @code{DOES>}, details
6472:
6473: doc-does>
6474:
6475: @cindex @code{DOES>} in a separate definition
6476: This means that you need not use @code{CREATE} and @code{DOES>} in the
6477: same definition; you can put the @code{DOES>}-part in a separate
6478: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
6479: @example
6480: : does1
6481: DOES> ( ... -- ... )
6482: ... ;
6483:
6484: : does2
6485: DOES> ( ... -- ... )
6486: ... ;
6487:
6488: : def-word ( ... -- ... )
6489: create ...
6490: IF
6491: does1
6492: ELSE
6493: does2
6494: ENDIF ;
6495: @end example
6496:
6497: In this example, the selection of whether to use @code{does1} or
6498: @code{does2} is made at definition-time; at the time that the child word is
6499: @code{CREATE}d.
6500:
6501: @cindex @code{DOES>} in interpretation state
6502: In a standard program you can apply a @code{DOES>}-part only if the last
6503: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
6504: will override the behaviour of the last word defined in any case. In a
6505: standard program, you can use @code{DOES>} only in a colon
6506: definition. In Gforth, you can also use it in interpretation state, in a
6507: kind of one-shot mode; for example:
6508: @example
6509: CREATE name ( ... -- ... )
6510: @i{initialization}
6511: DOES>
6512: @i{code} ;
6513: @end example
6514:
6515: @noindent
6516: is equivalent to the standard:
6517: @example
6518: :noname
6519: DOES>
6520: @i{code} ;
6521: CREATE name EXECUTE ( ... -- ... )
6522: @i{initialization}
6523: @end example
6524:
6525: doc->body
6526:
6527: @node Advanced does> usage example, , CREATE..DOES> details, User-defined Defining Words
6528: @subsubsection Advanced does> usage example
6529:
6530: The MIPS disassembler (@file{arch/mips/disasm.fs}) contains many words
6531: for disassembling instructions, that follow a very repetetive scheme:
6532:
6533: @example
6534: :noname @var{disasm-operands} s" @var{inst-name}" type ;
6535: @var{entry-num} cells @var{table} + !
6536: @end example
6537:
6538: Of course, this inspires the idea to factor out the commonalities to
6539: allow a definition like
6540:
6541: @example
6542: @var{disasm-operands} @var{entry-num} @var{table} define-inst @var{inst-name}
6543: @end example
6544:
6545: The parameters @var{disasm-operands} and @var{table} are usually
6546: correlated. Moreover, before I wrote the disassembler, there already
6547: existed code that defines instructions like this:
6548:
6549: @example
6550: @var{entry-num} @var{inst-format} @var{inst-name}
6551: @end example
6552:
6553: This code comes from the assembler and resides in
6554: @file{arch/mips/insts.fs}.
6555:
6556: So I had to define the @var{inst-format} words that performed the scheme
6557: above when executed. At first I chose to use run-time code-generation:
6558:
6559: @example
6560: : @var{inst-format} ( entry-num "name" -- ; compiled code: addr w -- )
6561: :noname Postpone @var{disasm-operands}
6562: name Postpone sliteral Postpone type Postpone ;
6563: swap cells @var{table} + ! ;
6564: @end example
6565:
6566: Note that this supplies the other two parameters of the scheme above.
6567:
6568: An alternative would have been to write this using
6569: @code{create}/@code{does>}:
6570:
6571: @example
6572: : @var{inst-format} ( entry-num "name" -- )
6573: here name string, ( entry-num c-addr ) \ parse and save "name"
6574: noname create , ( entry-num )
6575: lastxt swap cells @var{table} + !
6576: does> ( addr w -- )
6577: \ disassemble instruction w at addr
6578: @@ >r
6579: @var{disasm-operands}
6580: r> count type ;
6581: @end example
6582:
6583: Somehow the first solution is simpler, mainly because it's simpler to
6584: shift a string from definition-time to use-time with @code{sliteral}
6585: than with @code{string,} and friends.
6586:
6587: I wrote a lot of words following this scheme and soon thought about
6588: factoring out the commonalities among them. Note that this uses a
6589: two-level defining word, i.e., a word that defines ordinary defining
6590: words.
6591:
6592: This time a solution involving @code{postpone} and friends seemed more
6593: difficult (try it as an exercise), so I decided to use a
6594: @code{create}/@code{does>} word; since I was already at it, I also used
6595: @code{create}/@code{does>} for the lower level (try using
6596: @code{postpone} etc. as an exercise), resulting in the following
6597: definition:
6598:
6599: @example
6600: : define-format ( disasm-xt table-xt -- )
6601: \ define an instruction format that uses disasm-xt for
6602: \ disassembling and enters the defined instructions into table
6603: \ table-xt
6604: create 2,
6605: does> ( u "inst" -- )
6606: \ defines an anonymous word for disassembling instruction inst,
6607: \ and enters it as u-th entry into table-xt
6608: 2@@ swap here name string, ( u table-xt disasm-xt c-addr ) \ remember string
6609: noname create 2, \ define anonymous word
6610: execute lastxt swap ! \ enter xt of defined word into table-xt
6611: does> ( addr w -- )
6612: \ disassemble instruction w at addr
6613: 2@@ >r ( addr w disasm-xt R: c-addr )
6614: execute ( R: c-addr ) \ disassemble operands
6615: r> count type ; \ print name
6616: @end example
6617:
6618: Note that the tables here (in contrast to above) do the @code{cells +}
6619: by themselves (that's why you have to pass an xt). This word is used in
6620: the following way:
6621:
6622: @example
6623: ' @var{disasm-operands} ' @var{table} define-format @var{inst-format}
6624: @end example
6625:
6626: As shown above, the defined instruction format is then used like this:
6627:
6628: @example
6629: @var{entry-num} @var{inst-format} @var{inst-name}
6630: @end example
6631:
6632: In terms of currying, this kind of two-level defining word provides the
6633: parameters in three stages: first @var{disasm-operands} and @var{table},
6634: then @var{entry-num} and @var{inst-name}, finally @code{addr w}, i.e.,
6635: the instruction to be disassembled.
6636:
6637: Of course this did not quite fit all the instruction format names used
6638: in @file{insts.fs}, so I had to define a few wrappers that conditioned
6639: the parameters into the right form.
6640:
6641: If you have trouble following this section, don't worry. First, this is
6642: involved and takes time (and probably some playing around) to
6643: understand; second, this is the first two-level
6644: @code{create}/@code{does>} word I have written in seventeen years of
6645: Forth; and if I did not have @file{insts.fs} to start with, I may well
6646: have elected to use just a one-level defining word (with some repeating
6647: of parameters when using the defining word). So it is not necessary to
6648: understand this, but it may improve your understanding of Forth.
6649:
6650:
6651: @node Deferred words, Aliases, User-defined Defining Words, Defining Words
6652: @subsection Deferred words
6653: @cindex deferred words
6654:
6655: The defining word @code{Defer} allows you to define a word by name
6656: without defining its behaviour; the definition of its behaviour is
6657: deferred. Here are two situation where this can be useful:
6658:
6659: @itemize @bullet
6660: @item
6661: Where you want to allow the behaviour of a word to be altered later, and
6662: for all precompiled references to the word to change when its behaviour
6663: is changed.
6664: @item
6665: For mutual recursion; @xref{Calls and returns}.
6666: @end itemize
6667:
6668: In the following example, @code{foo} always invokes the version of
6669: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
6670: always invokes the version that prints ``@code{Hello}''. There is no way
6671: of getting @code{foo} to use the later version without re-ordering the
6672: source code and recompiling it.
6673:
6674: @example
6675: : greet ." Good morning" ;
6676: : foo ... greet ... ;
6677: : greet ." Hello" ;
6678: : bar ... greet ... ;
6679: @end example
6680:
6681: This problem can be solved by defining @code{greet} as a @code{Defer}red
6682: word. The behaviour of a @code{Defer}red word can be defined and
6683: redefined at any time by using @code{IS} to associate the xt of a
6684: previously-defined word with it. The previous example becomes:
6685:
6686: @example
6687: Defer greet ( -- )
6688: : foo ... greet ... ;
6689: : bar ... greet ... ;
6690: : greet1 ( -- ) ." Good morning" ;
6691: : greet2 ( -- ) ." Hello" ;
6692: ' greet2 <IS> greet \ make greet behave like greet2
6693: @end example
6694:
6695: @progstyle
6696: You should write a stack comment for every deferred word, and put only
6697: XTs into deferred words that conform to this stack effect. Otherwise
6698: it's too difficult to use the deferred word.
6699:
6700: A deferred word can be used to improve the statistics-gathering example
6701: from @ref{User-defined Defining Words}; rather than edit the
6702: application's source code to change every @code{:} to a @code{my:}, do
6703: this:
6704:
6705: @example
6706: : real: : ; \ retain access to the original
6707: defer : \ redefine as a deferred word
6708: ' my: <IS> : \ use special version of :
6709: \
6710: \ load application here
6711: \
6712: ' real: <IS> : \ go back to the original
6713: @end example
6714:
6715:
6716: One thing to note is that @code{<IS>} consumes its name when it is
6717: executed. If you want to specify the name at compile time, use
6718: @code{[IS]}:
6719:
6720: @example
6721: : set-greet ( xt -- )
6722: [IS] greet ;
6723:
6724: ' greet1 set-greet
6725: @end example
6726:
6727: A deferred word can only inherit execution semantics from the xt
6728: (because that is all that an xt can represent -- for more discussion of
6729: this @pxref{Tokens for Words}); by default it will have default
6730: interpretation and compilation semantics deriving from this execution
6731: semantics. However, you can change the interpretation and compilation
6732: semantics of the deferred word in the usual ways:
6733:
6734: @example
6735: : bar .... ; compile-only
6736: Defer fred immediate
6737: Defer jim
6738:
6739: ' bar <IS> jim \ jim has default semantics
6740: ' bar <IS> fred \ fred is immediate
6741: @end example
6742:
6743: doc-defer
6744: doc-<is>
6745: doc-[is]
6746: doc-is
6747: @comment TODO document these: what's defers [is]
6748: doc-what's
6749: doc-defers
6750:
6751: @c Use @code{words-deferred} to see a list of deferred words.
6752:
6753: Definitions in ANS Forth for @code{defer}, @code{<is>} and @code{[is]}
6754: are provided in @file{compat/defer.fs}.
6755:
6756:
6757: @node Aliases, , Deferred words, Defining Words
6758: @subsection Aliases
6759: @cindex aliases
6760:
6761: The defining word @code{Alias} allows you to define a word by name that
6762: has the same behaviour as some other word. Here are two situation where
6763: this can be useful:
6764:
6765: @itemize @bullet
6766: @item
6767: When you want access to a word's definition from a different word list
6768: (for an example of this, see the definition of the @code{Root} word list
6769: in the Gforth source).
6770: @item
6771: When you want to create a synonym; a definition that can be known by
6772: either of two names (for example, @code{THEN} and @code{ENDIF} are
6773: aliases).
6774: @end itemize
6775:
6776: Like deferred words, an alias has default compilation and interpretation
6777: semantics at the beginning (not the modifications of the other word),
6778: but you can change them in the usual ways (@code{immediate},
6779: @code{compile-only}). For example:
6780:
6781: @example
6782: : foo ... ; immediate
6783:
6784: ' foo Alias bar \ bar is not an immediate word
6785: ' foo Alias fooby immediate \ fooby is an immediate word
6786: @end example
6787:
6788: Words that are aliases have the same xt, different headers in the
6789: dictionary, and consequently different name tokens (@pxref{Tokens for
6790: Words}) and possibly different immediate flags. An alias can only have
6791: default or immediate compilation semantics; you can define aliases for
6792: combined words with @code{interpret/compile:} -- see @ref{Combined words}.
6793:
6794: doc-alias
6795:
6796:
6797: @node Interpretation and Compilation Semantics, Tokens for Words, Defining Words, Words
6798: @section Interpretation and Compilation Semantics
6799: @cindex semantics, interpretation and compilation
6800:
6801: @c !! state and ' are used without explanation
6802: @c example for immediate/compile-only? or is the tutorial enough
6803:
6804: @cindex interpretation semantics
6805: The @dfn{interpretation semantics} of a (named) word are what the text
6806: interpreter does when it encounters the word in interpret state. It also
6807: appears in some other contexts, e.g., the execution token returned by
6808: @code{' @i{word}} identifies the interpretation semantics of @i{word}
6809: (in other words, @code{' @i{word} execute} is equivalent to
6810: interpret-state text interpretation of @code{@i{word}}).
6811:
6812: @cindex compilation semantics
6813: The @dfn{compilation semantics} of a (named) word are what the text
6814: interpreter does when it encounters the word in compile state. It also
6815: appears in other contexts, e.g, @code{POSTPONE @i{word}}
6816: compiles@footnote{In standard terminology, ``appends to the current
6817: definition''.} the compilation semantics of @i{word}.
6818:
6819: @cindex execution semantics
6820: The standard also talks about @dfn{execution semantics}. They are used
6821: only for defining the interpretation and compilation semantics of many
6822: words. By default, the interpretation semantics of a word are to
6823: @code{execute} its execution semantics, and the compilation semantics of
6824: a word are to @code{compile,} its execution semantics.@footnote{In
6825: standard terminology: The default interpretation semantics are its
6826: execution semantics; the default compilation semantics are to append its
6827: execution semantics to the execution semantics of the current
6828: definition.}
6829:
6830: Unnamed words (@pxref{Anonymous Definitions}) cannot be encountered by
6831: the text interpreter, ticked, or @code{postpone}d, so they have no
6832: interpretation or compilation semantics. Their behaviour is represented
6833: by their XT (@pxref{Tokens for Words}), and we call it execution
6834: semantics, too.
6835:
6836: @comment TODO expand, make it co-operate with new sections on text interpreter.
6837:
6838: @cindex immediate words
6839: @cindex compile-only words
6840: You can change the semantics of the most-recently defined word:
6841:
6842:
6843: doc-immediate
6844: doc-compile-only
6845: doc-restrict
6846:
6847:
6848: Note that ticking (@code{'}) a compile-only word gives an error
6849: (``Interpreting a compile-only word'').
6850:
6851: @menu
6852: * Combined words::
6853: @end menu
6854:
6855:
6856: @node Combined words, , Interpretation and Compilation Semantics, Interpretation and Compilation Semantics
6857: @subsection Combined Words
6858: @cindex combined words
6859:
6860: Gforth allows you to define @dfn{combined words} -- words that have an
6861: arbitrary combination of interpretation and compilation semantics.
6862:
6863: doc-interpret/compile:
6864:
6865: This feature was introduced for implementing @code{TO} and @code{S"}. I
6866: recommend that you do not define such words, as cute as they may be:
6867: they make it hard to get at both parts of the word in some contexts.
6868: E.g., assume you want to get an execution token for the compilation
6869: part. Instead, define two words, one that embodies the interpretation
6870: part, and one that embodies the compilation part. Once you have done
6871: that, you can define a combined word with @code{interpret/compile:} for
6872: the convenience of your users.
6873:
6874: You might try to use this feature to provide an optimizing
6875: implementation of the default compilation semantics of a word. For
6876: example, by defining:
6877: @example
6878: :noname
6879: foo bar ;
6880: :noname
6881: POSTPONE foo POSTPONE bar ;
6882: interpret/compile: opti-foobar
6883: @end example
6884:
6885: @noindent
6886: as an optimizing version of:
6887:
6888: @example
6889: : foobar
6890: foo bar ;
6891: @end example
6892:
6893: Unfortunately, this does not work correctly with @code{[compile]},
6894: because @code{[compile]} assumes that the compilation semantics of all
6895: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
6896: opti-foobar} would compile compilation semantics, whereas
6897: @code{[compile] foobar} would compile interpretation semantics.
6898:
6899: @cindex state-smart words (are a bad idea)
6900: Some people try to use @dfn{state-smart} words to emulate the feature provided
6901: by @code{interpret/compile:} (words are state-smart if they check
6902: @code{STATE} during execution). E.g., they would try to code
6903: @code{foobar} like this:
6904:
6905: @example
6906: : foobar
6907: STATE @@
6908: IF ( compilation state )
6909: POSTPONE foo POSTPONE bar
6910: ELSE
6911: foo bar
6912: ENDIF ; immediate
6913: @end example
6914:
6915: Although this works if @code{foobar} is only processed by the text
6916: interpreter, it does not work in other contexts (like @code{'} or
6917: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
6918: for a state-smart word, not for the interpretation semantics of the
6919: original @code{foobar}; when you execute this execution token (directly
6920: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
6921: state, the result will not be what you expected (i.e., it will not
6922: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
6923: write them@footnote{For a more detailed discussion of this topic, see
6924: M. Anton Ertl,
6925: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl98.ps.gz,@code{State}-smartness---Why
6926: it is Evil and How to Exorcise it}}, EuroForth '98.}!
6927:
6928: @cindex defining words with arbitrary semantics combinations
6929: It is also possible to write defining words that define words with
6930: arbitrary combinations of interpretation and compilation semantics. In
6931: general, they look like this:
6932:
6933: @example
6934: : def-word
6935: create-interpret/compile
6936: @i{code1}
6937: interpretation>
6938: @i{code2}
6939: <interpretation
6940: compilation>
6941: @i{code3}
6942: <compilation ;
6943: @end example
6944:
6945: For a @i{word} defined with @code{def-word}, the interpretation
6946: semantics are to push the address of the body of @i{word} and perform
6947: @i{code2}, and the compilation semantics are to push the address of
6948: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
6949: can also be defined like this (except that the defined constants don't
6950: behave correctly when @code{[compile]}d):
6951:
6952: @example
6953: : constant ( n "name" -- )
6954: create-interpret/compile
6955: ,
6956: interpretation> ( -- n )
6957: @@
6958: <interpretation
6959: compilation> ( compilation. -- ; run-time. -- n )
6960: @@ postpone literal
6961: <compilation ;
6962: @end example
6963:
6964:
6965: doc-create-interpret/compile
6966: doc-interpretation>
6967: doc-<interpretation
6968: doc-compilation>
6969: doc-<compilation
6970:
6971:
6972: Words defined with @code{interpret/compile:} and
6973: @code{create-interpret/compile} have an extended header structure that
6974: differs from other words; however, unless you try to access them with
6975: plain address arithmetic, you should not notice this. Words for
6976: accessing the header structure usually know how to deal with this; e.g.,
6977: @code{'} @i{word} @code{>body} also gives you the body of a word created
6978: with @code{create-interpret/compile}.
6979:
6980:
6981: doc-postpone
6982:
6983: @comment TODO -- expand glossary text for POSTPONE
6984:
6985:
6986: @c -------------------------------------------------------------
6987: @node Tokens for Words, The Text Interpreter, Interpretation and Compilation Semantics, Words
6988: @section Tokens for Words
6989: @cindex tokens for words
6990:
6991: This section describes the creation and use of tokens that represent
6992: words.
6993:
6994: @menu
6995: * Execution token:: represents execution/interpretation semantics
6996: * Compilation token:: represents compilation semantics
6997: * Name token:: represents named words
6998: @end menu
6999:
7000: @node Execution token, Compilation token, Tokens for Words, Tokens for Words
7001: @subsection Execution token
7002:
7003: @cindex xt
7004: @cindex execution token
7005: An @dfn{execution token} (@i{XT}) represents some behaviour of a word.
7006: You can use @code{execute} to invoke this behaviour.
7007:
7008: @cindex tick (')
7009: You can use @code{'} to get an execution token that represents the
7010: interpretation semantics of a named word:
7011:
7012: @example
7013: 5 ' .
7014: execute
7015: @end example
7016:
7017: doc-'
7018:
7019: @code{'} parses at run-time; there is also a word @code{[']} that parses
7020: when it is compiled, and compiles the resulting XT:
7021:
7022: @example
7023: : foo ['] . execute ;
7024: 5 foo
7025: : bar ' execute ; \ by contrast,
7026: 5 bar . \ ' parses "." when bar executes
7027: @end example
7028:
7029: doc-[']
7030:
7031: If you want the execution token of @i{word}, write @code{['] @i{word}}
7032: in compiled code and @code{' @i{word}} in interpreted code. Gforth's
7033: @code{'} and @code{[']} behave somewhat unusually by complaining about
7034: compile-only words (because these words have no interpretation
7035: semantics). You might get what you want by using @code{COMP' @i{word}
7036: DROP} or @code{[COMP'] @i{word} DROP} (for details @pxref{Compilation
7037: token}).
7038:
7039: Another way to get an XT is @code{:noname} or @code{lastxt}
7040: (@pxref{Anonymous Definitions}). For anonymous words this gives an xt
7041: for the only behaviour the word has (the execution semantics). For
7042: named words, @code{lastxt} produces an XT for the same behaviour it
7043: would produce if the word was defined anonymously.
7044:
7045: @example
7046: :noname ." hello" ;
7047: execute
7048: @end example
7049:
7050: An XT occupies one cell and can be manipulated like any other cell.
7051:
7052: @cindex code field address
7053: @cindex CFA
7054: In ANS Forth the XT is just an abstract data type (i.e., defined by the
7055: operations that produce or consume it). For old hands: In Gforth, the
7056: XT is implemented as a code field address (CFA).
7057:
7058: @c !! discuss "compile," some more (or in Macros).
7059:
7060: doc-execute
7061: doc-perform
7062: doc-compile,
7063:
7064: @node Compilation token, Name token, Execution token, Tokens for Words
7065: @subsection Compilation token
7066:
7067: @cindex compilation token
7068: @cindex CT (compilation token)
7069: Gforth represents the compilation semantics of a named word by a
7070: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
7071: @i{xt} is an execution token. The compilation semantics represented by
7072: the compilation token can be performed with @code{execute}, which
7073: consumes the whole compilation token, with an additional stack effect
7074: determined by the represented compilation semantics.
7075:
7076: At present, the @i{w} part of a compilation token is an execution token,
7077: and the @i{xt} part represents either @code{execute} or
7078: @code{compile,}@footnote{Depending upon the compilation semantics of the
7079: word. If the word has default compilation semantics, the @i{xt} will
7080: represent @code{compile,}. Otherwise (e.g., for immediate words), the
7081: @i{xt} will represent @code{execute}.}. However, don't rely on that
7082: knowledge, unless necessary; future versions of Gforth may introduce
7083: unusual compilation tokens (e.g., a compilation token that represents
7084: the compilation semantics of a literal).
7085:
7086: You can perform the compilation semantics represented by the compilation
7087: token with @code{execute}. You can compile the compilation semantics
7088: with @code{postpone,}. I.e., @code{COMP' @i{word} postpone,} is
7089: equivalent to @code{postpone @i{word}}.
7090:
7091: doc-[comp']
7092: doc-comp'
7093: doc-postpone,
7094:
7095: @node Name token, , Compilation token, Tokens for Words
7096: @subsection Name token
7097:
7098: @cindex name token
7099: @cindex name field address
7100: @cindex NFA
7101: Gforth represents named words by the @dfn{name token}, (@i{nt}). In
7102: Gforth, the abstract data type @emph{name token} is implemented as a
7103: name field address (NFA).
7104:
7105: doc-find-name
7106: doc-name>int
7107: doc-name?int
7108: doc-name>comp
7109: doc-name>string
7110:
7111:
7112: @c ----------------------------------------------------------
7113: @node The Text Interpreter, Word Lists, Tokens for Words, Words
7114: @section The Text Interpreter
7115: @cindex interpreter - outer
7116: @cindex text interpreter
7117: @cindex outer interpreter
7118:
7119: @c Should we really describe all these ugly details? IMO the text
7120: @c interpreter should be much cleaner, but that may not be possible within
7121: @c ANS Forth. - anton
7122: @c nac-> I wanted to explain how it works to show how you can exploit
7123: @c it in your own programs. When I was writing a cross-compiler, figuring out
7124: @c some of these gory details was very helpful to me. None of the textbooks
7125: @c I've seen cover it, and the most modern Forth textbook -- Forth Inc's,
7126: @c seems to positively avoid going into too much detail for some of
7127: @c the internals.
7128:
7129: @c anton: ok. I wonder, though, if this is the right place; for some stuff
7130: @c it is; for the ugly details, I would prefer another place. I wonder
7131: @c whether we should have a chapter before "Words" that describes some
7132: @c basic concepts referred to in words, and a chapter after "Words" that
7133: @c describes implementation details.
7134:
7135: The text interpreter@footnote{This is an expanded version of the
7136: material in @ref{Introducing the Text Interpreter}.} is an endless loop
7137: that processes input from the current input device. It is also called
7138: the outer interpreter, in contrast to the inner interpreter
7139: (@pxref{Engine}) which executes the compiled Forth code on interpretive
7140: implementations.
7141:
7142: @cindex interpret state
7143: @cindex compile state
7144: The text interpreter operates in one of two states: @dfn{interpret
7145: state} and @dfn{compile state}. The current state is defined by the
7146: aptly-named variable @code{state}.
7147:
7148: This section starts by describing how the text interpreter behaves when
7149: it is in interpret state, processing input from the user input device --
7150: the keyboard. This is the mode that a Forth system is in after it starts
7151: up.
7152:
7153: @cindex input buffer
7154: @cindex terminal input buffer
7155: The text interpreter works from an area of memory called the @dfn{input
7156: buffer}@footnote{When the text interpreter is processing input from the
7157: keyboard, this area of memory is called the @dfn{terminal input buffer}
7158: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
7159: @code{#TIB}.}, which stores your keyboard input when you press the
7160: @key{RET} key. Starting at the beginning of the input buffer, it skips
7161: leading spaces (called @dfn{delimiters}) then parses a string (a
7162: sequence of non-space characters) until it reaches either a space
7163: character or the end of the buffer. Having parsed a string, it makes two
7164: attempts to process it:
7165:
7166: @cindex dictionary
7167: @itemize @bullet
7168: @item
7169: It looks for the string in a @dfn{dictionary} of definitions. If the
7170: string is found, the string names a @dfn{definition} (also known as a
7171: @dfn{word}) and the dictionary search returns information that allows
7172: the text interpreter to perform the word's @dfn{interpretation
7173: semantics}. In most cases, this simply means that the word will be
7174: executed.
7175: @item
7176: If the string is not found in the dictionary, the text interpreter
7177: attempts to treat it as a number, using the rules described in
7178: @ref{Number Conversion}. If the string represents a legal number in the
7179: current radix, the number is pushed onto a parameter stack (the data
7180: stack for integers, the floating-point stack for floating-point
7181: numbers).
7182: @end itemize
7183:
7184: If both attempts fail, or if the word is found in the dictionary but has
7185: no interpretation semantics@footnote{This happens if the word was
7186: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
7187: remainder of the input buffer, issues an error message and waits for
7188: more input. If one of the attempts succeeds, the text interpreter
7189: repeats the parsing process until the whole of the input buffer has been
7190: processed, at which point it prints the status message ``@code{ ok}''
7191: and waits for more input.
7192:
7193: @c anton: this should be in the input stream subsection (or below it)
7194:
7195: @cindex parse area
7196: The text interpreter keeps track of its position in the input buffer by
7197: updating a variable called @code{>IN} (pronounced ``to-in''). The value
7198: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
7199: of the input buffer. The region from offset @code{>IN @@} to the end of
7200: the input buffer is called the @dfn{parse area}@footnote{In other words,
7201: the text interpreter processes the contents of the input buffer by
7202: parsing strings from the parse area until the parse area is empty.}.
7203: This example shows how @code{>IN} changes as the text interpreter parses
7204: the input buffer:
7205:
7206: @example
7207: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
7208: CR ." ->" TYPE ." <-" ; IMMEDIATE
7209:
7210: 1 2 3 remaining + remaining .
7211:
7212: : foo 1 2 3 remaining SWAP remaining ;
7213: @end example
7214:
7215: @noindent
7216: The result is:
7217:
7218: @example
7219: ->+ remaining .<-
7220: ->.<-5 ok
7221:
7222: ->SWAP remaining ;-<
7223: ->;<- ok
7224: @end example
7225:
7226: @cindex parsing words
7227: The value of @code{>IN} can also be modified by a word in the input
7228: buffer that is executed by the text interpreter. This means that a word
7229: can ``trick'' the text interpreter into either skipping a section of the
7230: input buffer@footnote{This is how parsing words work.} or into parsing a
7231: section twice. For example:
7232:
7233: @example
7234: : lat ." <<foo>>" ;
7235: : flat ." <<bar>>" >IN DUP @@ 3 - SWAP ! ;
7236: @end example
7237:
7238: @noindent
7239: When @code{flat} is executed, this output is produced@footnote{Exercise
7240: for the reader: what would happen if the @code{3} were replaced with
7241: @code{4}?}:
7242:
7243: @example
7244: <<bar>><<foo>>
7245: @end example
7246:
7247: This technique can be used to work around some of the interoperability
7248: problems of parsing words. Of course, it's better to avoid parsing
7249: words where possible.
7250:
7251: @noindent
7252: Two important notes about the behaviour of the text interpreter:
7253:
7254: @itemize @bullet
7255: @item
7256: It processes each input string to completion before parsing additional
7257: characters from the input buffer.
7258: @item
7259: It treats the input buffer as a read-only region (and so must your code).
7260: @end itemize
7261:
7262: @noindent
7263: When the text interpreter is in compile state, its behaviour changes in
7264: these ways:
7265:
7266: @itemize @bullet
7267: @item
7268: If a parsed string is found in the dictionary, the text interpreter will
7269: perform the word's @dfn{compilation semantics}. In most cases, this
7270: simply means that the execution semantics of the word will be appended
7271: to the current definition.
7272: @item
7273: When a number is encountered, it is compiled into the current definition
7274: (as a literal) rather than being pushed onto a parameter stack.
7275: @item
7276: If an error occurs, @code{state} is modified to put the text interpreter
7277: back into interpret state.
7278: @item
7279: Each time a line is entered from the keyboard, Gforth prints
7280: ``@code{ compiled}'' rather than `` @code{ok}''.
7281: @end itemize
7282:
7283: @cindex text interpreter - input sources
7284: When the text interpreter is using an input device other than the
7285: keyboard, its behaviour changes in these ways:
7286:
7287: @itemize @bullet
7288: @item
7289: When the parse area is empty, the text interpreter attempts to refill
7290: the input buffer from the input source. When the input source is
7291: exhausted, the input source is set back to the previous input source.
7292: @item
7293: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
7294: time the parse area is emptied.
7295: @item
7296: If an error occurs, the input source is set back to the user input
7297: device.
7298: @end itemize
7299:
7300: You can read about this in more detail in @ref{Input Sources}.
7301:
7302: doc->in
7303: doc-source
7304:
7305: doc-tib
7306: doc-#tib
7307:
7308:
7309: @menu
7310: * Input Sources::
7311: * Number Conversion::
7312: * Interpret/Compile states::
7313: * Literals::
7314: * Interpreter Directives::
7315: @end menu
7316:
7317: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
7318: @subsection Input Sources
7319: @cindex input sources
7320: @cindex text interpreter - input sources
7321:
7322: By default, the text interpreter processes input from the user input
7323: device (the keyboard) when Forth starts up. The text interpreter can
7324: process input from any of these sources:
7325:
7326: @itemize @bullet
7327: @item
7328: The user input device -- the keyboard.
7329: @item
7330: A file, using the words described in @ref{Forth source files}.
7331: @item
7332: A block, using the words described in @ref{Blocks}.
7333: @item
7334: A text string, using @code{evaluate}.
7335: @end itemize
7336:
7337: A program can identify the current input device from the values of
7338: @code{source-id} and @code{blk}.
7339:
7340:
7341: doc-source-id
7342: doc-blk
7343:
7344: doc-save-input
7345: doc-restore-input
7346:
7347: doc-evaluate
7348:
7349:
7350:
7351: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
7352: @subsection Number Conversion
7353: @cindex number conversion
7354: @cindex double-cell numbers, input format
7355: @cindex input format for double-cell numbers
7356: @cindex single-cell numbers, input format
7357: @cindex input format for single-cell numbers
7358: @cindex floating-point numbers, input format
7359: @cindex input format for floating-point numbers
7360:
7361: This section describes the rules that the text interpreter uses when it
7362: tries to convert a string into a number.
7363:
7364: Let <digit> represent any character that is a legal digit in the current
7365: number base@footnote{For example, 0-9 when the number base is decimal or
7366: 0-9, A-F when the number base is hexadecimal.}.
7367:
7368: Let <decimal digit> represent any character in the range 0-9.
7369:
7370: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
7371: in the braces (@i{a} or @i{b} or neither).
7372:
7373: Let * represent any number of instances of the previous character
7374: (including none).
7375:
7376: Let any other character represent itself.
7377:
7378: @noindent
7379: Now, the conversion rules are:
7380:
7381: @itemize @bullet
7382: @item
7383: A string of the form <digit><digit>* is treated as a single-precision
7384: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
7385: @item
7386: A string of the form -<digit><digit>* is treated as a single-precision
7387: (cell-sized) negative integer, and is represented using 2's-complement
7388: arithmetic. Examples are -45 -5681 -0
7389: @item
7390: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
7391: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
7392: (all three of these represent the same number).
7393: @item
7394: A string of the form -<digit><digit>*.<digit>* is treated as a
7395: double-precision (double-cell-sized) negative integer, and is
7396: represented using 2's-complement arithmetic. Examples are -3465. -3.465
7397: -34.65 (all three of these represent the same number).
7398: @item
7399: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
7400: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
7401: number. Examples are 1e 1e0 1.e 1.e0 +1e+0 (which all represent the same
7402: number) +12.E-4
7403: @end itemize
7404:
7405: By default, the number base used for integer number conversion is given
7406: by the contents of the variable @code{base}. Note that a lot of
7407: confusion can result from unexpected values of @code{base}. If you
7408: change @code{base} anywhere, make sure to save the old value and restore
7409: it afterwards. In general I recommend keeping @code{base} decimal, and
7410: using the prefixes described below for the popular non-decimal bases.
7411:
7412: doc-dpl
7413: doc-base
7414: doc-hex
7415: doc-decimal
7416:
7417:
7418: @cindex '-prefix for character strings
7419: @cindex &-prefix for decimal numbers
7420: @cindex %-prefix for binary numbers
7421: @cindex $-prefix for hexadecimal numbers
7422: Gforth allows you to override the value of @code{base} by using a
7423: prefix@footnote{Some Forth implementations provide a similar scheme by
7424: implementing @code{$} etc. as parsing words that process the subsequent
7425: number in the input stream and push it onto the stack. For example, see
7426: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
7427: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
7428: is required between the prefix and the number.} before the first digit
7429: of an (integer) number. Four prefixes are supported:
7430:
7431: @itemize @bullet
7432: @item
7433: @code{&} -- decimal
7434: @item
7435: @code{%} -- binary
7436: @item
7437: @code{$} -- hexadecimal
7438: @item
7439: @code{'} -- base @code{max-char+1}
7440: @end itemize
7441:
7442: Here are some examples, with the equivalent decimal number shown after
7443: in braces:
7444:
7445: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
7446: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
7447: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
7448: &905 (905), $abc (2478), $ABC (2478).
7449:
7450: @cindex number conversion - traps for the unwary
7451: @noindent
7452: Number conversion has a number of traps for the unwary:
7453:
7454: @itemize @bullet
7455: @item
7456: You cannot determine the current number base using the code sequence
7457: @code{base @@ .} -- the number base is always 10 in the current number
7458: base. Instead, use something like @code{base @@ dec.}
7459: @item
7460: If the number base is set to a value greater than 14 (for example,
7461: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
7462: it to be intepreted as either a single-precision integer or a
7463: floating-point number (Gforth treats it as an integer). The ambiguity
7464: can be resolved by explicitly stating the sign of the mantissa and/or
7465: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
7466: ambiguity arises; either representation will be treated as a
7467: floating-point number.
7468: @item
7469: There is a word @code{bin} but it does @i{not} set the number base!
7470: It is used to specify file types.
7471: @item
7472: ANS Forth requires the @code{.} of a double-precision number to be the
7473: final character in the string. Gforth allows the @code{.} to be
7474: anywhere after the first digit.
7475: @item
7476: The number conversion process does not check for overflow.
7477: @item
7478: In an ANS Forth program @code{base} is required to be decimal when
7479: converting floating-point numbers. In Gforth, number conversion to
7480: floating-point numbers always uses base &10, irrespective of the value
7481: of @code{base}.
7482: @end itemize
7483:
7484: You can read numbers into your programs with the words described in
7485: @ref{Input}.
7486:
7487: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
7488: @subsection Interpret/Compile states
7489: @cindex Interpret/Compile states
7490:
7491: A standard program is not permitted to change @code{state}
7492: explicitly. However, it can change @code{state} implicitly, using the
7493: words @code{[} and @code{]}. When @code{[} is executed it switches
7494: @code{state} to interpret state, and therefore the text interpreter
7495: starts interpreting. When @code{]} is executed it switches @code{state}
7496: to compile state and therefore the text interpreter starts
7497: compiling. The most common usage for these words is for switching into
7498: interpret state and back from within a colon definition; this technique
7499: can be used to compile a literal (for an example, @pxref{Literals}) or
7500: for conditional compilation (for an example, @pxref{Interpreter
7501: Directives}).
7502:
7503:
7504: @c This is a bad example: It's non-standard, and it's not necessary.
7505: @c However, I can't think of a good example for switching into compile
7506: @c state when there is no current word (@code{state}-smart words are not a
7507: @c good reason). So maybe we should use an example for switching into
7508: @c interpret @code{state} in a colon def. - anton
7509: @c nac-> I agree. I started out by putting in the example, then realised
7510: @c that it was non-ANS, so wrote more words around it. I hope this
7511: @c re-written version is acceptable to you. I do want to keep the example
7512: @c as it is helpful for showing what is and what is not portable, particularly
7513: @c where it outlaws a style in common use.
7514:
7515: @c anton: it's more important to show what's portable. After we have done
7516: @c that, we can also show what's not. In any case, I intend to write a
7517: @c section Macros (or so) which will also deal with [ ].
7518:
7519: @code{[} and @code{]} also give you the ability to switch into compile
7520: state and back, but we cannot think of any useful Standard application
7521: for this ability. Pre-ANS Forth textbooks have examples like this:
7522:
7523: @example
7524: : AA ." this is A" ;
7525: : BB ." this is B" ;
7526: : CC ." this is C" ;
7527:
7528: create table ] aa bb cc [
7529:
7530: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
7531: cells table + @ execute ;
7532: @end example
7533:
7534: This example builds a jump table; @code{0 go} will display ``@code{this
7535: is A}''. Using @code{[} and @code{]} in this example is equivalent to
7536: defining @code{table} like this:
7537:
7538: @example
7539: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
7540: @end example
7541:
7542: The problem with this code is that the definition of @code{table} is not
7543: portable -- it @i{compile}s execution tokens into code space. Whilst it
7544: @i{may} work on systems where code space and data space co-incide, the
7545: Standard only allows data space to be assigned for a @code{CREATE}d
7546: word. In addition, the Standard only allows @code{@@} to access data
7547: space, whilst this example is using it to access code space. The only
7548: portable, Standard way to build this table is to build it in data space,
7549: like this:
7550:
7551: @example
7552: create table ' aa , ' bb , ' cc ,
7553: @end example
7554:
7555: doc-state
7556: doc-[
7557: doc-]
7558:
7559:
7560: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
7561: @subsection Literals
7562: @cindex Literals
7563:
7564: Often, you want to use a number within a colon definition. When you do
7565: this, the text interpreter automatically compiles the number as a
7566: @i{literal}. A literal is a number whose run-time effect is to be pushed
7567: onto the stack. If you had to do some maths to generate the number, you
7568: might write it like this:
7569:
7570: @example
7571: : HOUR-TO-SEC ( n1 -- n2 )
7572: 60 * \ to minutes
7573: 60 * ; \ to seconds
7574: @end example
7575:
7576: It is very clear what this definition is doing, but it's inefficient
7577: since it is performing 2 multiples at run-time. An alternative would be
7578: to write:
7579:
7580: @example
7581: : HOUR-TO-SEC ( n1 -- n2 )
7582: 3600 * ; \ to seconds
7583: @end example
7584:
7585: Which does the same thing, and has the advantage of using a single
7586: multiply. Ideally, we'd like the efficiency of the second with the
7587: readability of the first.
7588:
7589: @code{Literal} allows us to achieve that. It takes a number from the
7590: stack and lays it down in the current definition just as though the
7591: number had been typed directly into the definition. Our first attempt
7592: might look like this:
7593:
7594: @example
7595: 60 \ mins per hour
7596: 60 * \ seconds per minute
7597: : HOUR-TO-SEC ( n1 -- n2 )
7598: Literal * ; \ to seconds
7599: @end example
7600:
7601: But this produces the error message @code{unstructured}. What happened?
7602: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
7603: @i{colon-sys} is implementation-defined. In other words, once we start a
7604: colon definition we can't portably access anything that was on the stack
7605: before the definition began@footnote{@cite{Two Problems in ANS Forth},
7606: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
7607: some situations where you might want to access stack items above
7608: colon-sys, and provides a solution to the problem.}. The correct way of
7609: solving this problem in this instance is to use @code{[ ]} like this:
7610:
7611: @example
7612: : HOUR-TO-SEC ( n1 -- n2 )
7613: [ 60 \ minutes per hour
7614: 60 * ] \ seconds per minute
7615: LITERAL * ; \ to seconds
7616: @end example
7617:
7618:
7619: doc-literal
7620: doc-]L
7621: doc-2literal
7622: doc-fliteral
7623:
7624:
7625: @node Interpreter Directives, , Literals, The Text Interpreter
7626: @subsection Interpreter Directives
7627: @cindex interpreter directives
7628: @cindex conditional compilation
7629:
7630: These words are usually used in interpret state; typically to control
7631: which parts of a source file are processed by the text
7632: interpreter. There are only a few ANS Forth Standard words, but Gforth
7633: supplements these with a rich set of immediate control structure words
7634: to compensate for the fact that the non-immediate versions can only be
7635: used in compile state (@pxref{Control Structures}). Typical usages:
7636:
7637: @example
7638: FALSE Constant HAVE-ASSEMBLER
7639: .
7640: .
7641: HAVE-ASSEMBLER [IF]
7642: : ASSEMBLER-FEATURE
7643: ...
7644: ;
7645: [ENDIF]
7646: .
7647: .
7648: : SEE
7649: ... \ general-purpose SEE code
7650: [ HAVE-ASSEMBLER [IF] ]
7651: ... \ assembler-specific SEE code
7652: [ [ENDIF] ]
7653: ;
7654: @end example
7655:
7656:
7657: doc-[IF]
7658: doc-[ELSE]
7659: doc-[THEN]
7660: doc-[ENDIF]
7661:
7662: doc-[IFDEF]
7663: doc-[IFUNDEF]
7664:
7665: doc-[?DO]
7666: doc-[DO]
7667: doc-[FOR]
7668: doc-[LOOP]
7669: doc-[+LOOP]
7670: doc-[NEXT]
7671:
7672: doc-[BEGIN]
7673: doc-[UNTIL]
7674: doc-[AGAIN]
7675: doc-[WHILE]
7676: doc-[REPEAT]
7677:
7678:
7679: @c -------------------------------------------------------------
7680: @node Word Lists, Environmental Queries, The Text Interpreter, Words
7681: @section Word Lists
7682: @cindex word lists
7683: @cindex header space
7684:
7685: A wordlist is a list of named words; you can add new words and look up
7686: words by name (and you can remove words in a restricted way with
7687: markers). Every named (and @code{reveal}ed) word is in one wordlist.
7688:
7689: @cindex search order stack
7690: The text interpreter searches the wordlists present in the search order
7691: (a stack of wordlists), from the top to the bottom. Within each
7692: wordlist, the search starts conceptually at the newest word; i.e., if
7693: two words in a wordlist have the same name, the newer word is found.
7694:
7695: @cindex compilation word list
7696: New words are added to the @dfn{compilation wordlist} (aka current
7697: wordlist).
7698:
7699: @cindex wid
7700: A word list is identified by a cell-sized word list identifier (@i{wid})
7701: in much the same way as a file is identified by a file handle. The
7702: numerical value of the wid has no (portable) meaning, and might change
7703: from session to session.
7704:
7705: The ANS Forth ``Search order'' word set is intended to provide a set of
7706: low-level tools that allow various different schemes to be
7707: implemented. Gforth provides @code{vocabulary}, a traditional Forth
7708: word. @file{compat/vocabulary.fs} provides an implementation in ANS
7709: Forth.
7710:
7711: @comment TODO: locals section refers to here, saying that every word list (aka
7712: @comment vocabulary) has its own methods for searching etc. Need to document that.
7713:
7714: @comment TODO: document markers, reveal, tables, mappedwordlist
7715:
7716: @comment the gforthman- prefix is used to pick out the true definition of a
7717: @comment word from the source files, rather than some alias.
7718:
7719: doc-forth-wordlist
7720: doc-definitions
7721: doc-get-current
7722: doc-set-current
7723: doc-get-order
7724: doc---gforthman-set-order
7725: doc-wordlist
7726: doc-table
7727: doc-push-order
7728: doc-previous
7729: doc-also
7730: doc---gforthman-forth
7731: doc-only
7732: doc---gforthman-order
7733:
7734: doc-find
7735: doc-search-wordlist
7736:
7737: doc-words
7738: doc-vlist
7739: @c doc-words-deferred
7740:
7741: doc-mappedwordlist
7742: doc-root
7743: doc-vocabulary
7744: doc-seal
7745: doc-vocs
7746: doc-current
7747: doc-context
7748:
7749:
7750: @menu
7751: * Why use word lists?::
7752: * Word list examples::
7753: @end menu
7754:
7755: @node Why use word lists?, Word list examples, Word Lists, Word Lists
7756: @subsection Why use word lists?
7757: @cindex word lists - why use them?
7758:
7759: Here are some reasons for using multiple word lists:
7760:
7761: @itemize @bullet
7762: @item
7763: To improve compilation speed by reducing the number of header space
7764: entries that must be searched. This is achieved by creating a new
7765: word list that contains all of the definitions that are used in the
7766: definition of a Forth system but which would not usually be used by
7767: programs running on that system. That word list would be on the search
7768: list when the Forth system was compiled but would be removed from the
7769: search list for normal operation. This can be a useful technique for
7770: low-performance systems (for example, 8-bit processors in embedded
7771: systems) but is unlikely to be necessary in high-performance desktop
7772: systems.
7773: @item
7774: To prevent a set of words from being used outside the context in which
7775: they are valid. Two classic examples of this are an integrated editor
7776: (all of the edit commands are defined in a separate word list; the
7777: search order is set to the editor word list when the editor is invoked;
7778: the old search order is restored when the editor is terminated) and an
7779: integrated assembler (the op-codes for the machine are defined in a
7780: separate word list which is used when a @code{CODE} word is defined).
7781: @item
7782: To prevent a name-space clash between multiple definitions with the same
7783: name. For example, when building a cross-compiler you might have a word
7784: @code{IF} that generates conditional code for your target system. By
7785: placing this definition in a different word list you can control whether
7786: the host system's @code{IF} or the target system's @code{IF} get used in
7787: any particular context by controlling the order of the word lists on the
7788: search order stack.
7789: @end itemize
7790:
7791: @node Word list examples, , Why use word lists?, Word Lists
7792: @subsection Word list examples
7793: @cindex word lists - examples
7794:
7795: Here is an example of creating and using a new wordlist using ANS
7796: Forth Standard words:
7797:
7798: @example
7799: wordlist constant my-new-words-wordlist
7800: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
7801:
7802: \ add it to the search order
7803: also my-new-words
7804:
7805: \ alternatively, add it to the search order and make it
7806: \ the compilation word list
7807: also my-new-words definitions
7808: \ type "order" to see the problem
7809: @end example
7810:
7811: The problem with this example is that @code{order} has no way to
7812: associate the name @code{my-new-words} with the wid of the word list (in
7813: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
7814: that has no associated name). There is no Standard way of associating a
7815: name with a wid.
7816:
7817: In Gforth, this example can be re-coded using @code{vocabulary}, which
7818: associates a name with a wid:
7819:
7820: @example
7821: vocabulary my-new-words
7822:
7823: \ add it to the search order
7824: also my-new-words
7825:
7826: \ alternatively, add it to the search order and make it
7827: \ the compilation word list
7828: my-new-words definitions
7829: \ type "order" to see that the problem is solved
7830: @end example
7831:
7832: @c -------------------------------------------------------------
7833: @node Environmental Queries, Files, Word Lists, Words
7834: @section Environmental Queries
7835: @cindex environmental queries
7836:
7837: ANS Forth introduced the idea of ``environmental queries'' as a way
7838: for a program running on a system to determine certain characteristics of the system.
7839: The Standard specifies a number of strings that might be recognised by a system.
7840:
7841: The Standard requires that the header space used for environmental queries
7842: be distinct from the header space used for definitions.
7843:
7844: Typically, environmental queries are supported by creating a set of
7845: definitions in a word list that is @i{only} used during environmental
7846: queries; that is what Gforth does. There is no Standard way of adding
7847: definitions to the set of recognised environmental queries, but any
7848: implementation that supports the loading of optional word sets must have
7849: some mechanism for doing this (after loading the word set, the
7850: associated environmental query string must return @code{true}). In
7851: Gforth, the word list used to honour environmental queries can be
7852: manipulated just like any other word list.
7853:
7854:
7855: doc-environment?
7856: doc-environment-wordlist
7857:
7858: doc-gforth
7859: doc-os-class
7860:
7861:
7862: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
7863: returning two items on the stack, querying it using @code{environment?}
7864: will return an additional item; the @code{true} flag that shows that the
7865: string was recognised.
7866:
7867: @comment TODO Document the standard strings or note where they are documented herein
7868:
7869: Here are some examples of using environmental queries:
7870:
7871: @example
7872: s" address-unit-bits" environment? 0=
7873: [IF]
7874: cr .( environmental attribute address-units-bits unknown... ) cr
7875: [THEN]
7876:
7877: s" block" environment? [IF] DROP include block.fs [THEN]
7878:
7879: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
7880:
7881: s" gforth" environment? [IF] .( Gforth version ) TYPE
7882: [ELSE] .( Not Gforth..) [THEN]
7883: @end example
7884:
7885:
7886: Here is an example of adding a definition to the environment word list:
7887:
7888: @example
7889: get-current environment-wordlist set-current
7890: true constant block
7891: true constant block-ext
7892: set-current
7893: @end example
7894:
7895: You can see what definitions are in the environment word list like this:
7896:
7897: @example
7898: get-order 1+ environment-wordlist swap set-order words previous
7899: @end example
7900:
7901:
7902: @c -------------------------------------------------------------
7903: @node Files, Blocks, Environmental Queries, Words
7904: @section Files
7905: @cindex files
7906: @cindex I/O - file-handling
7907:
7908: Gforth provides facilities for accessing files that are stored in the
7909: host operating system's file-system. Files that are processed by Gforth
7910: can be divided into two categories:
7911:
7912: @itemize @bullet
7913: @item
7914: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
7915: @item
7916: Files that are processed by some other program (@dfn{general files}).
7917: @end itemize
7918:
7919: doc-loadfilename
7920: doc-sourcefilename
7921: doc-sourceline#
7922:
7923: @menu
7924: * Forth source files::
7925: * General files::
7926: * Search Paths::
7927: @end menu
7928:
7929:
7930: @c -------------------------------------------------------------
7931: @node Forth source files, General files, Files, Files
7932: @subsection Forth source files
7933: @cindex including files
7934: @cindex Forth source files
7935:
7936: The simplest way to interpret the contents of a file is to use one of
7937: these two formats:
7938:
7939: @example
7940: include mysource.fs
7941: s" mysource.fs" included
7942: @end example
7943:
7944: Sometimes you want to include a file only if it is not included already
7945: (by, say, another source file). In that case, you can use one of these
7946: three formats:
7947:
7948: @example
7949: require mysource.fs
7950: needs mysource.fs
7951: s" mysource.fs" required
7952: @end example
7953:
7954: @cindex stack effect of included files
7955: @cindex including files, stack effect
7956: It is good practice to write your source files such that interpreting them
7957: does not change the stack. Source files designed in this way can be used with
7958: @code{required} and friends without complications. For example:
7959:
7960: @example
7961: 1 require foo.fs drop
7962: @end example
7963:
7964:
7965: doc-include-file
7966: doc-included
7967: doc-included?
7968: doc-include
7969: doc-required
7970: doc-require
7971: doc-needs
7972: doc-init-included-files
7973:
7974:
7975: A definition in ANS Forth for @code{required} is provided in
7976: @file{compat/required.fs}.
7977:
7978: @c -------------------------------------------------------------
7979: @node General files, Search Paths, Forth source files, Files
7980: @subsection General files
7981: @cindex general files
7982: @cindex file-handling
7983:
7984: Files are opened/created by name and type. The following types are
7985: recognised:
7986:
7987:
7988: doc-r/o
7989: doc-r/w
7990: doc-w/o
7991: doc-bin
7992:
7993:
7994: When a file is opened/created, it returns a file identifier,
7995: @i{wfileid} that is used for all other file commands. All file
7996: commands also return a status value, @i{wior}, that is 0 for a
7997: successful operation and an implementation-defined non-zero value in the
7998: case of an error.
7999:
8000:
8001: doc-open-file
8002: doc-create-file
8003:
8004: doc-close-file
8005: doc-delete-file
8006: doc-rename-file
8007: doc-read-file
8008: doc-read-line
8009: doc-write-file
8010: doc-write-line
8011: doc-emit-file
8012: doc-flush-file
8013:
8014: doc-file-status
8015: doc-file-position
8016: doc-reposition-file
8017: doc-file-size
8018: doc-resize-file
8019:
8020:
8021: @c ---------------------------------------------------------
8022: @node Search Paths, , General files, Files
8023: @subsection Search Paths
8024: @cindex path for @code{included}
8025: @cindex file search path
8026: @cindex @code{include} search path
8027: @cindex search path for files
8028:
8029: If you specify an absolute filename (i.e., a filename starting with
8030: @file{/} or @file{~}, or with @file{:} in the second position (as in
8031: @samp{C:...})) for @code{included} and friends, that file is included
8032: just as you would expect.
8033:
8034: For relative filenames, Gforth uses a search path similar to Forth's
8035: search order (@pxref{Word Lists}). It tries to find the given filename
8036: in the directories present in the path, and includes the first one it
8037: finds. There are separate search paths for Forth source files and
8038: general files.
8039:
8040: If the search path contains the directory @file{.} (as it should), this
8041: refers to the directory that the present file was @code{included}
8042: from. This allows files to include other files relative to their own
8043: position (irrespective of the current working directory or the absolute
8044: position). This feature is essential for libraries consisting of
8045: several files, where a file may include other files from the library.
8046: It corresponds to @code{#include "..."} in C. If the current input
8047: source is not a file, @file{.} refers to the directory of the innermost
8048: file being included, or, if there is no file being included, to the
8049: current working directory.
8050:
8051: Use @file{~+} to refer to the current working directory (as in the
8052: @code{bash}).
8053:
8054: If the filename starts with @file{./}, the search path is not searched
8055: (just as with absolute filenames), and the @file{.} has the same meaning
8056: as described above.
8057:
8058: @menu
8059: * Forth Search Paths::
8060: * General Search Paths::
8061: @end menu
8062:
8063: @c ---------------------------------------------------------
8064: @node Forth Search Paths, General Search Paths, Search Paths, Search Paths
8065: @subsubsection Forth Search Paths
8066: @cindex search path control - Forth
8067:
8068: The search path is initialized when you start Gforth (@pxref{Invoking
8069: Gforth}). You can display it and change it using these words:
8070:
8071:
8072: doc-.fpath
8073: doc-fpath+
8074: doc-fpath=
8075: doc-open-fpath-file
8076:
8077:
8078: @noindent
8079: Here is an example of using @code{fpath} and @code{require}:
8080:
8081: @example
8082: fpath= /usr/lib/forth/|./
8083: require timer.fs
8084: @end example
8085:
8086: @c ---------------------------------------------------------
8087: @node General Search Paths, , Forth Search Paths, Search Paths
8088: @subsubsection General Search Paths
8089: @cindex search path control - for user applications
8090:
8091: Your application may need to search files in several directories, like
8092: @code{included} does. To facilitate this, Gforth allows you to define
8093: and use your own search paths, by providing generic equivalents of the
8094: Forth search path words:
8095:
8096:
8097: doc-.path
8098: doc-path+
8099: doc-path=
8100: doc-open-path-file
8101:
8102:
8103: Here's an example of creating a search path:
8104:
8105: @example
8106: \ Make a buffer for the path:
8107: create mypath 100 chars , \ maximum length (is checked)
8108: 0 , \ real len
8109: 100 chars allot \ space for path
8110: @end example
8111:
8112: @c -------------------------------------------------------------
8113: @node Blocks, Other I/O, Files, Words
8114: @section Blocks
8115: @cindex I/O - blocks
8116: @cindex blocks
8117:
8118: When you run Gforth on a modern desk-top computer, it runs under the
8119: control of an operating system which provides certain services. One of
8120: these services is @var{file services}, which allows Forth source code
8121: and data to be stored in files and read into Gforth (@pxref{Files}).
8122:
8123: Traditionally, Forth has been an important programming language on
8124: systems where it has interfaced directly to the underlying hardware with
8125: no intervening operating system. Forth provides a mechanism, called
8126: @dfn{blocks}, for accessing mass storage on such systems.
8127:
8128: A block is a 1024-byte data area, which can be used to hold data or
8129: Forth source code. No structure is imposed on the contents of the
8130: block. A block is identified by its number; blocks are numbered
8131: contiguously from 1 to an implementation-defined maximum.
8132:
8133: A typical system that used blocks but no operating system might use a
8134: single floppy-disk drive for mass storage, with the disks formatted to
8135: provide 256-byte sectors. Blocks would be implemented by assigning the
8136: first four sectors of the disk to block 1, the second four sectors to
8137: block 2 and so on, up to the limit of the capacity of the disk. The disk
8138: would not contain any file system information, just the set of blocks.
8139:
8140: @cindex blocks file
8141: On systems that do provide file services, blocks are typically
8142: implemented by storing a sequence of blocks within a single @dfn{blocks
8143: file}. The size of the blocks file will be an exact multiple of 1024
8144: bytes, corresponding to the number of blocks it contains. This is the
8145: mechanism that Gforth uses.
8146:
8147: @cindex @file{blocks.fb}
8148: Only 1 blocks file can be open at a time. If you use block words without
8149: having specified a blocks file, Gforth defaults to the blocks file
8150: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
8151: locate a blocks file (@pxref{Forth Search Paths}).
8152:
8153: @cindex block buffers
8154: When you read and write blocks under program control, Gforth uses a
8155: number of @dfn{block buffers} as intermediate storage. These buffers are
8156: not used when you use @code{load} to interpret the contents of a block.
8157:
8158: The behaviour of the block buffers is directly analagous to that of a
8159: cache. Each block buffer has three states:
8160:
8161: @itemize @bullet
8162: @item
8163: Unassigned
8164: @item
8165: Assigned-clean
8166: @item
8167: Assigned-dirty
8168: @end itemize
8169:
8170: Initially, all block buffers are @i{unassigned}. In order to access a
8171: block, the block (specified by its block number) must be assigned to a
8172: block buffer.
8173:
8174: The assignment of a block to a block buffer is performed by @code{block}
8175: or @code{buffer}. Use @code{block} when you wish to modify the existing
8176: contents of a block. Use @code{buffer} when you don't care about the
8177: existing contents of the block@footnote{The ANS Forth definition of
8178: @code{buffer} is intended not to cause disk I/O; if the data associated
8179: with the particular block is already stored in a block buffer due to an
8180: earlier @code{block} command, @code{buffer} will return that block
8181: buffer and the existing contents of the block will be
8182: available. Otherwise, @code{buffer} will simply assign a new, empty
8183: block buffer for the block.}.
8184:
8185: Once a block has been assigned to a block buffer using @code{block} or
8186: @code{buffer}, that block buffer becomes the @i{current block buffer}
8187: and its state changes to @i{assigned-clean}. Data may only be
8188: manipulated (read or written) within the current block buffer.
8189:
8190: When the contents of the current block buffer has been modified it is
8191: necessary, @emph{before calling @code{block} or @code{buffer} again}, to
8192: either abandon the changes (by doing nothing) or commit the changes,
8193: using @code{update}. Using @code{update} does not change the blocks
8194: file; it simply changes a block buffer's state to @i{assigned-dirty}.
8195:
8196: The word @code{flush} causes all @i{assigned-dirty} blocks to be
8197: written back to the blocks file on disk. Leaving Gforth using @code{bye}
8198: also causes a @code{flush} to be performed.
8199:
8200: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
8201: algorithm to assign a block buffer to a block. That means that any
8202: particular block can only be assigned to one specific block buffer,
8203: called (for the particular operation) the @i{victim buffer}. If the
8204: victim buffer is @i{unassigned} or @i{assigned-clean} it is allocated to
8205: the new block immediately. If it is @i{assigned-dirty} its current
8206: contents are written back to the blocks file on disk before it is
8207: allocated to the new block.
8208:
8209: Although no structure is imposed on the contents of a block, it is
8210: traditional to display the contents as 16 lines each of 64 characters. A
8211: block provides a single, continuous stream of input (for example, it
8212: acts as a single parse area) -- there are no end-of-line characters
8213: within a block, and no end-of-file character at the end of a
8214: block. There are two consequences of this:
8215:
8216: @itemize @bullet
8217: @item
8218: The last character of one line wraps straight into the first character
8219: of the following line
8220: @item
8221: The word @code{\} -- comment to end of line -- requires special
8222: treatment; in the context of a block it causes all characters until the
8223: end of the current 64-character ``line'' to be ignored.
8224: @end itemize
8225:
8226: In Gforth, when you use @code{block} with a non-existent block number,
8227: the current blocks file will be extended to the appropriate size and the
8228: block buffer will be initialised with spaces.
8229:
8230: Gforth includes a simple block editor (type @code{use blocked.fb 0 list}
8231: for details) but doesn't encourage the use of blocks; the mechanism is
8232: only provided for backward compatibility -- ANS Forth requires blocks to
8233: be available when files are.
8234:
8235: Common techniques that are used when working with blocks include:
8236:
8237: @itemize @bullet
8238: @item
8239: A screen editor that allows you to edit blocks without leaving the Forth
8240: environment.
8241: @item
8242: Shadow screens; where every code block has an associated block
8243: containing comments (for example: code in odd block numbers, comments in
8244: even block numbers). Typically, the block editor provides a convenient
8245: mechanism to toggle between code and comments.
8246: @item
8247: Load blocks; a single block (typically block 1) contains a number of
8248: @code{thru} commands which @code{load} the whole of the application.
8249: @end itemize
8250:
8251: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
8252: integrated into a Forth programming environment.
8253:
8254: @comment TODO what about errors on open-blocks?
8255:
8256: doc-open-blocks
8257: doc-use
8258: doc-get-block-fid
8259: doc-block-position
8260:
8261: doc-scr
8262: doc-list
8263:
8264: doc---gforthman-block
8265: doc-buffer
8266:
8267: doc-update
8268: doc-updated?
8269: doc-save-buffers
8270: doc-empty-buffers
8271: doc-empty-buffer
8272: doc-flush
8273:
8274: doc-load
8275: doc-thru
8276: doc-+load
8277: doc-+thru
8278: doc---gforthman--->
8279: doc-block-included
8280:
8281:
8282: @c -------------------------------------------------------------
8283: @node Other I/O, Programming Tools, Blocks, Words
8284: @section Other I/O
8285: @cindex I/O - keyboard and display
8286:
8287: @menu
8288: * Simple numeric output:: Predefined formats
8289: * Formatted numeric output:: Formatted (pictured) output
8290: * String Formats:: How Forth stores strings in memory
8291: * Displaying characters and strings:: Other stuff
8292: * Input:: Input
8293: @end menu
8294:
8295: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
8296: @subsection Simple numeric output
8297: @cindex numeric output - simple/free-format
8298:
8299: The simplest output functions are those that display numbers from the
8300: data or floating-point stacks. Floating-point output is always displayed
8301: using base 10. Numbers displayed from the data stack use the value stored
8302: in @code{base}.
8303:
8304:
8305: doc-.
8306: doc-dec.
8307: doc-hex.
8308: doc-u.
8309: doc-.r
8310: doc-u.r
8311: doc-d.
8312: doc-ud.
8313: doc-d.r
8314: doc-ud.r
8315: doc-f.
8316: doc-fe.
8317: doc-fs.
8318:
8319:
8320: Examples of printing the number 1234.5678E23 in the different floating-point output
8321: formats are shown below:
8322:
8323: @example
8324: f. 123456779999999000000000000.
8325: fe. 123.456779999999E24
8326: fs. 1.23456779999999E26
8327: @end example
8328:
8329:
8330: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
8331: @subsection Formatted numeric output
8332: @cindex formatted numeric output
8333: @cindex pictured numeric output
8334: @cindex numeric output - formatted
8335:
8336: Forth traditionally uses a technique called @dfn{pictured numeric
8337: output} for formatted printing of integers. In this technique, digits
8338: are extracted from the number (using the current output radix defined by
8339: @code{base}), converted to ASCII codes and appended to a string that is
8340: built in a scratch-pad area of memory (@pxref{core-idef,
8341: Implementation-defined options, Implementation-defined
8342: options}). Arbitrary characters can be appended to the string during the
8343: extraction process. The completed string is specified by an address
8344: and length and can be manipulated (@code{TYPE}ed, copied, modified)
8345: under program control.
8346:
8347: All of the words described in the previous section for simple numeric
8348: output are implemented in Gforth using pictured numeric output.
8349:
8350: Three important things to remember about pictured numeric output:
8351:
8352: @itemize @bullet
8353: @item
8354: It always operates on double-precision numbers; to display a
8355: single-precision number, convert it first (for ways of doing this
8356: @pxref{Double precision}).
8357: @item
8358: It always treats the double-precision number as though it were
8359: unsigned. The examples below show ways of printing signed numbers.
8360: @item
8361: The string is built up from right to left; least significant digit first.
8362: @end itemize
8363:
8364:
8365: doc-<#
8366: doc-<<#
8367: doc-#
8368: doc-#s
8369: doc-hold
8370: doc-sign
8371: doc-#>
8372: doc-#>>
8373:
8374: doc-represent
8375:
8376:
8377: @noindent
8378: Here are some examples of using pictured numeric output:
8379:
8380: @example
8381: : my-u. ( u -- )
8382: \ Simplest use of pns.. behaves like Standard u.
8383: 0 \ convert to unsigned double
8384: <# \ start conversion
8385: #s \ convert all digits
8386: #> \ complete conversion
8387: TYPE SPACE ; \ display, with trailing space
8388:
8389: : cents-only ( u -- )
8390: 0 \ convert to unsigned double
8391: <# \ start conversion
8392: # # \ convert two least-significant digits
8393: #> \ complete conversion, discard other digits
8394: TYPE SPACE ; \ display, with trailing space
8395:
8396: : dollars-and-cents ( u -- )
8397: 0 \ convert to unsigned double
8398: <# \ start conversion
8399: # # \ convert two least-significant digits
8400: [char] . hold \ insert decimal point
8401: #s \ convert remaining digits
8402: [char] $ hold \ append currency symbol
8403: #> \ complete conversion
8404: TYPE SPACE ; \ display, with trailing space
8405:
8406: : my-. ( n -- )
8407: \ handling negatives.. behaves like Standard .
8408: s>d \ convert to signed double
8409: swap over dabs \ leave sign byte followed by unsigned double
8410: <# \ start conversion
8411: #s \ convert all digits
8412: rot sign \ get at sign byte, append "-" if needed
8413: #> \ complete conversion
8414: TYPE SPACE ; \ display, with trailing space
8415:
8416: : account. ( n -- )
8417: \ accountants don't like minus signs, they use braces
8418: \ for negative numbers
8419: s>d \ convert to signed double
8420: swap over dabs \ leave sign byte followed by unsigned double
8421: <# \ start conversion
8422: 2 pick \ get copy of sign byte
8423: 0< IF [char] ) hold THEN \ right-most character of output
8424: #s \ convert all digits
8425: rot \ get at sign byte
8426: 0< IF [char] ( hold THEN
8427: #> \ complete conversion
8428: TYPE SPACE ; \ display, with trailing space
8429: @end example
8430:
8431: Here are some examples of using these words:
8432:
8433: @example
8434: 1 my-u. 1
8435: hex -1 my-u. decimal FFFFFFFF
8436: 1 cents-only 01
8437: 1234 cents-only 34
8438: 2 dollars-and-cents $0.02
8439: 1234 dollars-and-cents $12.34
8440: 123 my-. 123
8441: -123 my. -123
8442: 123 account. 123
8443: -456 account. (456)
8444: @end example
8445:
8446:
8447: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
8448: @subsection String Formats
8449: @cindex strings - see character strings
8450: @cindex character strings - formats
8451: @cindex I/O - see character strings
8452:
8453: Forth commonly uses two different methods for representing character
8454: strings:
8455:
8456: @itemize @bullet
8457: @item
8458: @cindex address of counted string
8459: @cindex counted string
8460: As a @dfn{counted string}, represented by a @i{c-addr}. The char
8461: addressed by @i{c-addr} contains a character-count, @i{n}, of the
8462: string and the string occupies the subsequent @i{n} char addresses in
8463: memory.
8464: @item
8465: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
8466: of the string in characters, and @i{c-addr} is the address of the
8467: first byte of the string.
8468: @end itemize
8469:
8470: ANS Forth encourages the use of the second format when representing
8471: strings on the stack, whilst conceeding that the counted string format
8472: remains useful as a way of storing strings in memory.
8473:
8474:
8475: doc-count
8476:
8477:
8478: For words that move, copy and search for strings see @ref{Memory
8479: Blocks}. For words that display characters and strings see
8480: @ref{Displaying characters and strings}.
8481:
8482: @node Displaying characters and strings, Input, String Formats, Other I/O
8483: @subsection Displaying characters and strings
8484: @cindex characters - compiling and displaying
8485: @cindex character strings - compiling and displaying
8486:
8487: This section starts with a glossary of Forth words and ends with a set
8488: of examples.
8489:
8490:
8491: doc-bl
8492: doc-space
8493: doc-spaces
8494: doc-emit
8495: doc-toupper
8496: doc-."
8497: doc-.(
8498: doc-type
8499: doc-typewhite
8500: doc-cr
8501: @cindex cursor control
8502: doc-at-xy
8503: doc-page
8504: doc-s"
8505: doc-c"
8506: doc-char
8507: doc-[char]
8508: doc-sliteral
8509:
8510:
8511: @noindent
8512: As an example, consider the following text, stored in a file @file{test.fs}:
8513:
8514: @example
8515: .( text-1)
8516: : my-word
8517: ." text-2" cr
8518: .( text-3)
8519: ;
8520:
8521: ." text-4"
8522:
8523: : my-char
8524: [char] ALPHABET emit
8525: char emit
8526: ;
8527: @end example
8528:
8529: When you load this code into Gforth, the following output is generated:
8530:
8531: @example
8532: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
8533: @end example
8534:
8535: @itemize @bullet
8536: @item
8537: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
8538: is an immediate word; it behaves in the same way whether it is used inside
8539: or outside a colon definition.
8540: @item
8541: Message @code{text-4} is displayed because of Gforth's added interpretation
8542: semantics for @code{."}.
8543: @item
8544: Message @code{text-2} is @i{not} displayed, because the text interpreter
8545: performs the compilation semantics for @code{."} within the definition of
8546: @code{my-word}.
8547: @end itemize
8548:
8549: Here are some examples of executing @code{my-word} and @code{my-char}:
8550:
8551: @example
8552: @kbd{my-word @key{RET}} text-2
8553: ok
8554: @kbd{my-char fred @key{RET}} Af ok
8555: @kbd{my-char jim @key{RET}} Aj ok
8556: @end example
8557:
8558: @itemize @bullet
8559: @item
8560: Message @code{text-2} is displayed because of the run-time behaviour of
8561: @code{."}.
8562: @item
8563: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
8564: on the stack at run-time. @code{emit} always displays the character
8565: when @code{my-char} is executed.
8566: @item
8567: @code{char} parses a string at run-time and the second @code{emit} displays
8568: the first character of the string.
8569: @item
8570: If you type @code{see my-char} you can see that @code{[char]} discarded
8571: the text ``LPHABET'' and only compiled the display code for ``A'' into the
8572: definition of @code{my-char}.
8573: @end itemize
8574:
8575:
8576:
8577: @node Input, , Displaying characters and strings, Other I/O
8578: @subsection Input
8579: @cindex input
8580: @cindex I/O - see input
8581: @cindex parsing a string
8582:
8583: For ways of storing character strings in memory see @ref{String Formats}.
8584:
8585: @comment TODO examples for >number >float accept key key? pad parse word refill
8586: @comment then index them
8587:
8588:
8589: doc-key
8590: doc-key?
8591: doc-ekey
8592: doc-ekey?
8593: doc-ekey>char
8594: doc->number
8595: doc->float
8596: doc-accept
8597: doc-pad
8598: doc-parse
8599: doc-word
8600: doc-sword
8601: doc-(name)
8602: doc-refill
8603: @comment obsolescent words..
8604: doc-convert
8605: doc-query
8606: doc-expect
8607: doc-span
8608:
8609:
8610:
8611: @c -------------------------------------------------------------
8612: @node Programming Tools, Assembler and Code Words, Other I/O, Words
8613: @section Programming Tools
8614: @cindex programming tools
8615:
8616: @menu
8617: * Debugging:: Simple and quick.
8618: * Assertions:: Making your programs self-checking.
8619: * Singlestep Debugger:: Executing your program word by word.
8620: @end menu
8621:
8622: @node Debugging, Assertions, Programming Tools, Programming Tools
8623: @subsection Debugging
8624: @cindex debugging
8625:
8626: Languages with a slow edit/compile/link/test development loop tend to
8627: require sophisticated tracing/stepping debuggers to facilate
8628: productive debugging.
8629:
8630: A much better (faster) way in fast-compiling languages is to add
8631: printing code at well-selected places, let the program run, look at
8632: the output, see where things went wrong, add more printing code, etc.,
8633: until the bug is found.
8634:
8635: The simple debugging aids provided in @file{debugs.fs}
8636: are meant to support this style of debugging. In addition, there are
8637: words for non-destructively inspecting the stack and memory:
8638:
8639:
8640: doc-.s
8641: doc-f.s
8642:
8643:
8644: There is a word @code{.r} but it does @i{not} display the return
8645: stack! It is used for formatted numeric output.
8646:
8647:
8648: doc-depth
8649: doc-fdepth
8650: doc-clearstack
8651: doc-?
8652: doc-dump
8653:
8654:
8655: The word @code{~~} prints debugging information (by default the source
8656: location and the stack contents). It is easy to insert. If you use Emacs
8657: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
8658: query-replace them with nothing). The deferred words
8659: @code{printdebugdata} and @code{printdebugline} control the output of
8660: @code{~~}. The default source location output format works well with
8661: Emacs' compilation mode, so you can step through the program at the
8662: source level using @kbd{C-x `} (the advantage over a stepping debugger
8663: is that you can step in any direction and you know where the crash has
8664: happened or where the strange data has occurred).
8665:
8666: The default actions of @code{~~} clobber the contents of the pictured
8667: numeric output string, so you should not use @code{~~}, e.g., between
8668: @code{<#} and @code{#>}.
8669:
8670:
8671: doc-~~
8672: doc-printdebugdata
8673: doc-printdebugline
8674:
8675: doc-see
8676: doc-marker
8677:
8678:
8679: Here's an example of using @code{marker} at the start of a source file
8680: that you are debugging; it ensures that you only ever have one copy of
8681: the file's definitions compiled at any time:
8682:
8683: @example
8684: [IFDEF] my-code
8685: my-code
8686: [ENDIF]
8687:
8688: marker my-code
8689: init-included-files
8690:
8691: \ .. definitions start here
8692: \ .
8693: \ .
8694: \ end
8695: @end example
8696:
8697:
8698:
8699: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
8700: @subsection Assertions
8701: @cindex assertions
8702:
8703: It is a good idea to make your programs self-checking, especially if you
8704: make an assumption that may become invalid during maintenance (for
8705: example, that a certain field of a data structure is never zero). Gforth
8706: supports @dfn{assertions} for this purpose. They are used like this:
8707:
8708: @example
8709: assert( @i{flag} )
8710: @end example
8711:
8712: The code between @code{assert(} and @code{)} should compute a flag, that
8713: should be true if everything is alright and false otherwise. It should
8714: not change anything else on the stack. The overall stack effect of the
8715: assertion is @code{( -- )}. E.g.
8716:
8717: @example
8718: assert( 1 1 + 2 = ) \ what we learn in school
8719: assert( dup 0<> ) \ assert that the top of stack is not zero
8720: assert( false ) \ this code should not be reached
8721: @end example
8722:
8723: The need for assertions is different at different times. During
8724: debugging, we want more checking, in production we sometimes care more
8725: for speed. Therefore, assertions can be turned off, i.e., the assertion
8726: becomes a comment. Depending on the importance of an assertion and the
8727: time it takes to check it, you may want to turn off some assertions and
8728: keep others turned on. Gforth provides several levels of assertions for
8729: this purpose:
8730:
8731:
8732: doc-assert0(
8733: doc-assert1(
8734: doc-assert2(
8735: doc-assert3(
8736: doc-assert(
8737: doc-)
8738:
8739:
8740: The variable @code{assert-level} specifies the highest assertions that
8741: are turned on. I.e., at the default @code{assert-level} of one,
8742: @code{assert0(} and @code{assert1(} assertions perform checking, while
8743: @code{assert2(} and @code{assert3(} assertions are treated as comments.
8744:
8745: The value of @code{assert-level} is evaluated at compile-time, not at
8746: run-time. Therefore you cannot turn assertions on or off at run-time;
8747: you have to set the @code{assert-level} appropriately before compiling a
8748: piece of code. You can compile different pieces of code at different
8749: @code{assert-level}s (e.g., a trusted library at level 1 and
8750: newly-written code at level 3).
8751:
8752:
8753: doc-assert-level
8754:
8755:
8756: If an assertion fails, a message compatible with Emacs' compilation mode
8757: is produced and the execution is aborted (currently with @code{ABORT"}.
8758: If there is interest, we will introduce a special throw code. But if you
8759: intend to @code{catch} a specific condition, using @code{throw} is
8760: probably more appropriate than an assertion).
8761:
8762: Definitions in ANS Forth for these assertion words are provided
8763: in @file{compat/assert.fs}.
8764:
8765:
8766: @node Singlestep Debugger, , Assertions, Programming Tools
8767: @subsection Singlestep Debugger
8768: @cindex singlestep Debugger
8769: @cindex debugging Singlestep
8770:
8771: When you create a new word there's often the need to check whether it
8772: behaves correctly or not. You can do this by typing @code{dbg
8773: badword}. A debug session might look like this:
8774:
8775: @example
8776: : badword 0 DO i . LOOP ; ok
8777: 2 dbg badword
8778: : badword
8779: Scanning code...
8780:
8781: Nesting debugger ready!
8782:
8783: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
8784: 400D4740 8049F68 DO -> [ 0 ]
8785: 400D4744 804A0C8 i -> [ 1 ] 00000
8786: 400D4748 400C5E60 . -> 0 [ 0 ]
8787: 400D474C 8049D0C LOOP -> [ 0 ]
8788: 400D4744 804A0C8 i -> [ 1 ] 00001
8789: 400D4748 400C5E60 . -> 1 [ 0 ]
8790: 400D474C 8049D0C LOOP -> [ 0 ]
8791: 400D4758 804B384 ; -> ok
8792: @end example
8793:
8794: Each line displayed is one step. You always have to hit return to
8795: execute the next word that is displayed. If you don't want to execute
8796: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
8797: an overview what keys are available:
8798:
8799: @table @i
8800:
8801: @item @key{RET}
8802: Next; Execute the next word.
8803:
8804: @item n
8805: Nest; Single step through next word.
8806:
8807: @item u
8808: Unnest; Stop debugging and execute rest of word. If we got to this word
8809: with nest, continue debugging with the calling word.
8810:
8811: @item d
8812: Done; Stop debugging and execute rest.
8813:
8814: @item s
8815: Stop; Abort immediately.
8816:
8817: @end table
8818:
8819: Debugging large application with this mechanism is very difficult, because
8820: you have to nest very deeply into the program before the interesting part
8821: begins. This takes a lot of time.
8822:
8823: To do it more directly put a @code{BREAK:} command into your source code.
8824: When program execution reaches @code{BREAK:} the single step debugger is
8825: invoked and you have all the features described above.
8826:
8827: If you have more than one part to debug it is useful to know where the
8828: program has stopped at the moment. You can do this by the
8829: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
8830: string is typed out when the ``breakpoint'' is reached.
8831:
8832:
8833: doc-dbg
8834: doc-break:
8835: doc-break"
8836:
8837:
8838:
8839: @c -------------------------------------------------------------
8840: @node Assembler and Code Words, Threading Words, Programming Tools, Words
8841: @section Assembler and Code Words
8842: @cindex assembler
8843: @cindex code words
8844:
8845: @menu
8846: * Code and ;code::
8847: * Common Assembler:: Assembler Syntax
8848: * Common Disassembler::
8849: * 386 Assembler:: Deviations and special cases
8850: * Alpha Assembler:: Deviations and special cases
8851: * MIPS assembler:: Deviations and special cases
8852: * Other assemblers:: How to write them
8853: @end menu
8854:
8855: @node Code and ;code, Common Assembler, Assembler and Code Words, Assembler and Code Words
8856: @subsection @code{Code} and @code{;code}
8857:
8858: Gforth provides some words for defining primitives (words written in
8859: machine code), and for defining the machine-code equivalent of
8860: @code{DOES>}-based defining words. However, the machine-independent
8861: nature of Gforth poses a few problems: First of all, Gforth runs on
8862: several architectures, so it can provide no standard assembler. What's
8863: worse is that the register allocation not only depends on the processor,
8864: but also on the @code{gcc} version and options used.
8865:
8866: The words that Gforth offers encapsulate some system dependences (e.g.,
8867: the header structure), so a system-independent assembler may be used in
8868: Gforth. If you do not have an assembler, you can compile machine code
8869: directly with @code{,} and @code{c,}@footnote{This isn't portable,
8870: because these words emit stuff in @i{data} space; it works because
8871: Gforth has unified code/data spaces. Assembler isn't likely to be
8872: portable anyway.}.
8873:
8874:
8875: doc-assembler
8876: doc-init-asm
8877: doc-code
8878: doc-end-code
8879: doc-;code
8880: doc-flush-icache
8881:
8882:
8883: If @code{flush-icache} does not work correctly, @code{code} words
8884: etc. will not work (reliably), either.
8885:
8886: The typical usage of these @code{code} words can be shown most easily by
8887: analogy to the equivalent high-level defining words:
8888:
8889: @example
8890: : foo code foo
8891: <high-level Forth words> <assembler>
8892: ; end-code
8893:
8894: : bar : bar
8895: <high-level Forth words> <high-level Forth words>
8896: CREATE CREATE
8897: <high-level Forth words> <high-level Forth words>
8898: DOES> ;code
8899: <high-level Forth words> <assembler>
8900: ; end-code
8901: @end example
8902:
8903: @code{flush-icache} is always present. The other words are rarely used
8904: and reside in @code{code.fs}, which is usually not loaded. You can load
8905: it with @code{require code.fs}.
8906:
8907: @cindex registers of the inner interpreter
8908: In the assembly code you will want to refer to the inner interpreter's
8909: registers (e.g., the data stack pointer) and you may want to use other
8910: registers for temporary storage. Unfortunately, the register allocation
8911: is installation-dependent.
8912:
8913: The easiest solution is to use explicit register declarations
8914: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
8915: GNU C Manual}) for all of the inner interpreter's registers: You have to
8916: compile Gforth with @code{-DFORCE_REG} (configure option
8917: @code{--enable-force-reg}) and the appropriate declarations must be
8918: present in the @code{machine.h} file (see @code{mips.h} for an example;
8919: you can find a full list of all declarable register symbols with
8920: @code{grep register engine.c}). If you give explicit registers to all
8921: variables that are declared at the beginning of @code{engine()}, you
8922: should be able to use the other caller-saved registers for temporary
8923: storage. Alternatively, you can use the @code{gcc} option
8924: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
8925: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
8926: (however, this restriction on register allocation may slow Gforth
8927: significantly).
8928:
8929: If this solution is not viable (e.g., because @code{gcc} does not allow
8930: you to explicitly declare all the registers you need), you have to find
8931: out by looking at the code where the inner interpreter's registers
8932: reside and which registers can be used for temporary storage. You can
8933: get an assembly listing of the engine's code with @code{make engine.s}.
8934:
8935: In any case, it is good practice to abstract your assembly code from the
8936: actual register allocation. E.g., if the data stack pointer resides in
8937: register @code{$17}, create an alias for this register called @code{sp},
8938: and use that in your assembly code.
8939:
8940: @cindex code words, portable
8941: Another option for implementing normal and defining words efficiently
8942: is to add the desired functionality to the source of Gforth. For normal
8943: words you just have to edit @file{primitives} (@pxref{Automatic
8944: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
8945: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
8946: @file{prims2x.fs}, and possibly @file{cross.fs}.
8947:
8948: @node Common Assembler, Common Disassembler, Code and ;code, Assembler and Code Words
8949: @subsection Common Assembler
8950:
8951: The assemblers in Gforth generally use a postfix syntax, i.e., the
8952: instruction name follows the operands.
8953:
8954: The operands are passed in the usual order (the same that is used in the
8955: manual of the architecture). Since they all are Forth words, they have
8956: to be separated by spaces; you can also use Forth words to compute the
8957: operands.
8958:
8959: The instruction names usually end with a @code{,}. This makes it easier
8960: to visually separate instructions if you put several of them on one
8961: line; it also avoids shadowing other Forth words (e.g., @code{and}).
8962:
8963: Registers are usually specified by number; e.g., (decimal) @code{11}
8964: specifies registers R11 and F11 on the Alpha architecture (which one,
8965: depends on the instruction). The usual names are also available, e.g.,
8966: @code{s2} for R11 on Alpha.
8967:
8968: Control flow is specified similar to normal Forth code (@pxref{Arbitrary
8969: control structures}), with @code{if,}, @code{ahead,}, @code{then,},
8970: @code{begin,}, @code{until,}, @code{again,}, @code{cs-roll},
8971: @code{cs-pick}, @code{else,}, @code{while,}, and @code{repeat,}. The
8972: conditions are specified in a way specific to each assembler.
8973:
8974: Note that the register assignments of the Gforth engine can change
8975: between Gforth versions, or even between different compilations of the
8976: same Gforth version (e.g., if you use a different GCC version). So if
8977: you want to refer to Gforth's registers (e.g., the stack pointer or
8978: TOS), I recommend defining your own words for refering to these
8979: registers, and using them later on; then you can easily adapt to a
8980: changed register assignment. The stability of the register assignment
8981: is usually better if you build Gforth with @code{--enable-force-reg}.
8982:
8983: In particular, the resturn stack pointer and the instruction pointer are
8984: in memory in @code{gforth}, and usually in registers in
8985: @code{gforth-fast}. The most common use of these registers is to
8986: dispatch to the next word (the @code{next} routine). A portable way to
8987: do this is to jump to @code{' noop >code-address} (of course, this is
8988: less efficient than integrating the @code{next} code and scheduling it
8989: well).
8990:
8991: @node Common Disassembler, 386 Assembler, Common Assembler, Assembler and Code Words
8992: @subsection Common Disassembler
8993:
8994: You can disassemble a @code{code} word with @code{see}
8995: (@pxref{Debugging}). You can disassemble a section of memory with
8996:
8997: doc-disasm
8998:
8999: The disassembler generally produces output that can be fed into the
9000: assembler (i.e., same syntax, etc.). It also includes additional
9001: information in comments. In particular, the address of the instruction
9002: is given in a comment before the instruction.
9003:
9004: @code{See} may display more or less than the actual code of the word,
9005: because the recognition of the end of the code is unreliable. You can
9006: use @code{disasm} if it did not display enough. It may display more, if
9007: the code word is not immediately followed by a named word. If you have
9008: something else there, you can follow the word with @code{align last @ ,}
9009: to ensure that the end is recognized.
9010:
9011: @node 386 Assembler, Alpha Assembler, Common Disassembler, Assembler and Code Words
9012: @subsection 386 Assembler
9013:
9014: The 386 assembler included in Gforth was written by Bernd Paysan, it's
9015: available under GPL, and originally part of bigFORTH.
9016:
9017: The 386 disassembler included in Gforth was written by Andrew McKewan
9018: and is in the public domain.
9019:
9020: The disassembler displays code in prefix Intel syntax.
9021:
9022: The assembler uses a postfix syntax with reversed parameters.
9023:
9024: The assembler includes all instruction of the Athlon, i.e. 486 core
9025: instructions, Pentium and PPro extensions, floating point, MMX, 3Dnow!,
9026: but not ISSE. It's an integrated 16- and 32-bit assembler. Default is 32
9027: bit, you can switch to 16 bit with .86 and back to 32 bit with .386.
9028:
9029: There are several prefixes to switch between different operation sizes,
9030: @code{.b} for byte accesses, @code{.w} for word accesses, @code{.d} for
9031: double-word accesses. Addressing modes can be switched with @code{.wa}
9032: for 16 bit addresses, and @code{.da} for 32 bit addresses. You don't
9033: need a prefix for byte register names (@code{AL} et al).
9034:
9035: For floating point operations, the prefixes are @code{.fs} (IEEE
9036: single), @code{.fl} (IEEE double), @code{.fx} (extended), @code{.fw}
9037: (word), @code{.fd} (double-word), and @code{.fq} (quad-word).
9038:
9039: The MMX opcodes don't have size prefixes, they are spelled out like in
9040: the Intel assembler. Instead of move from and to memory, there are
9041: PLDQ/PLDD and PSTQ/PSTD.
9042:
9043: The registers lack the 'e' prefix; even in 32 bit mode, eax is called
9044: ax. Immediate values are indicated by postfixing them with @code{#},
9045: e.g., @code{3 #}. Here are some examples of addressing modes:
9046:
9047: @example
9048: 3 # \ immediate
9049: ax \ register
9050: 100 di d) \ 100[edi]
9051: 4 bx cx di) \ 4[ebx][ecx]
9052: di ax *4 i) \ [edi][eax*4]
9053: 20 ax *4 i#) \ 20[eax*4]
9054: @end example
9055:
9056: Some example of instructions are:
9057:
9058: @example
9059: ax bx mov \ move ebx,eax
9060: 3 # ax mov \ mov eax,3
9061: 100 di ) ax mov \ mov eax,100[edi]
9062: 4 bx cx di) ax mov \ mov eax,4[ebx][ecx]
9063: .w ax bx mov \ mov bx,ax
9064: @end example
9065:
9066: The following forms are supported for binary instructions:
9067:
9068: @example
9069: <reg> <reg> <inst>
9070: <n> # <reg> <inst>
9071: <mem> <reg> <inst>
9072: <reg> <mem> <inst>
9073: @end example
9074:
9075: Immediate to memory is not supported. The shift/rotate syntax is:
9076:
9077: @example
9078: <reg/mem> 1 # shl \ shortens to shift without immediate
9079: <reg/mem> 4 # shl
9080: <reg/mem> cl shl
9081: @end example
9082:
9083: Precede string instructions (@code{movs} etc.) with @code{.b} to get
9084: the byte version.
9085:
9086: The control structure words @code{IF} @code{UNTIL} etc. must be preceded
9087: by one of these conditions: @code{vs vc u< u>= 0= 0<> u<= u> 0< 0>= ps
9088: pc < >= <= >}. (Note that most of these words shadow some Forth words
9089: when @code{assembler} is in front of @code{forth} in the search path,
9090: e.g., in @code{code} words). Currently the control structure words use
9091: one stack item, so you have to use @code{roll} instead of @code{cs-roll}
9092: to shuffle them (you can also use @code{swap} etc.).
9093:
9094: Here is an example of a @code{code} word (assumes that the stack pointer
9095: is in esi and the TOS is in ebx):
9096:
9097: @example
9098: code my+ ( n1 n2 -- n )
9099: 4 si D) bx add
9100: 4 # si add
9101: Next
9102: end-code
9103: @end example
9104:
9105: @node Alpha Assembler, MIPS assembler, 386 Assembler, Assembler and Code Words
9106: @subsection Alpha Assembler
9107:
9108: The Alpha assembler and disassembler were originally written by Bernd
9109: Thallner.
9110:
9111: The register names @code{a0}--@code{a5} are not available to avoid
9112: shadowing hex numbers.
9113:
9114: Immediate forms of arithmetic instructions are distinguished by a
9115: @code{#} just before the @code{,}, e.g., @code{and#,} (note: @code{lda,}
9116: does not count as arithmetic instruction).
9117:
9118: You have to specify all operands to an instruction, even those that
9119: other assemblers consider optional, e.g., the destination register for
9120: @code{br,}, or the destination register and hint for @code{jmp,}.
9121:
9122: You can specify conditions for @code{if,} by removing the first @code{b}
9123: and the trailing @code{,} from a branch with a corresponding name; e.g.,
9124:
9125: @example
9126: 11 fgt if, \ if F11>0e
9127: ...
9128: endif,
9129: @end example
9130:
9131: @code{fbgt,} gives @code{fgt}.
9132:
9133: @node MIPS assembler, Other assemblers, Alpha Assembler, Assembler and Code Words
9134: @subsection MIPS assembler
9135:
9136: The MIPS assembler was originally written by Christian Pirker.
9137:
9138: Currently the assembler and disassembler only cover the MIPS-I
9139: architecture (R3000), and don't support FP instructions.
9140:
9141: The register names @code{$a0}--@code{$a3} are not available to avoid
9142: shadowing hex numbers.
9143:
9144: Because there is no way to distinguish registers from immediate values,
9145: you have to explicitly use the immediate forms of instructions, i.e.,
9146: @code{addiu,}, not just @code{addu,} (@command{as} does this
9147: implicitly).
9148:
9149: If the architecture manual specifies several formats for the instruction
9150: (e.g., for @code{jalr,}), you usually have to use the one with more
9151: arguments (i.e., two for @code{jalr,}). When in doubt, see
9152: @code{arch/mips/testasm.fs} for an example of correct use.
9153:
9154: Branches and jumps in the MIPS architecture have a delay slot. You have
9155: to fill it yourself (the simplest way is to use @code{nop,}), the
9156: assembler does not do it for you (unlike @command{as}). Even
9157: @code{if,}, @code{ahead,}, @code{until,}, @code{again,}, @code{while,},
9158: @code{else,} and @code{repeat,} need a delay slot. Since @code{begin,}
9159: and @code{then,} just specify branch targets, they are not affected.
9160:
9161: Note that you must not put branches, jumps, or @code{li,} into the delay
9162: slot: @code{li,} may expand to several instructions, and control flow
9163: instructions may not be put into the branch delay slot in any case.
9164:
9165: For branches the argument specifying the target is a relative address;
9166: You have to add the address of the delay slot to get the absolute
9167: address.
9168:
9169: The MIPS architecture also has load delay slots and restrictions on
9170: using @code{mfhi,} and @code{mflo,}; you have to order the instructions
9171: yourself to satisfy these restrictions, the assembler does not do it for
9172: you.
9173:
9174: You can specify the conditions for @code{if,} etc. by taking a
9175: conditional branch and leaving away the @code{b} at the start and the
9176: @code{,} at the end. E.g.,
9177:
9178: @example
9179: 4 5 eq if,
9180: ... \ do something if $4 equals $5
9181: then,
9182: @end example
9183:
9184: @node Other assemblers, , MIPS assembler, Assembler and Code Words
9185: @subsection Other assemblers
9186:
9187: If you want to contribute another assembler/disassembler, please contact
9188: us (@email{bug-gforth@@gnu.org}) to check if we have such an assembler
9189: already. If you are writing them from scratch, please use a similar
9190: syntax style as the one we use (i.e., postfix, commas at the end of the
9191: instruction names, @pxref{Common Assembler}); make the output of the
9192: disassembler be valid input for the assembler, and keep the style
9193: similar to the style we used.
9194:
9195: Hints on implementation: The most important part is to have a good test
9196: suite that contains all instructions. Once you have that, the rest is
9197: easy. For actual coding you can take a look at
9198: @file{arch/mips/disasm.fs} to get some ideas on how to use data for both
9199: the assembler and disassembler, avoiding redundancy and some potential
9200: bugs. You can also look at that file (and @pxref{Advanced does> usage
9201: example}) to get ideas how to factor a disassembler.
9202:
9203: Start with the disassembler, because it's easier to reuse data from the
9204: disassembler for the assembler than the other way round.
9205:
9206: For the assembler, take a look at @file{arch/alpha/asm.fs}, which shows
9207: how simple it can be.
9208:
9209: @c -------------------------------------------------------------
9210: @node Threading Words, Locals, Assembler and Code Words, Words
9211: @section Threading Words
9212: @cindex threading words
9213:
9214: @cindex code address
9215: These words provide access to code addresses and other threading stuff
9216: in Gforth (and, possibly, other interpretive Forths). It more or less
9217: abstracts away the differences between direct and indirect threading
9218: (and, for direct threading, the machine dependences). However, at
9219: present this wordset is still incomplete. It is also pretty low-level;
9220: some day it will hopefully be made unnecessary by an internals wordset
9221: that abstracts implementation details away completely.
9222:
9223:
9224: doc-threading-method
9225: doc->code-address
9226: doc->does-code
9227: doc-code-address!
9228: doc-does-code!
9229: doc-does-handler!
9230: doc-/does-handler
9231:
9232:
9233: The code addresses produced by various defining words are produced by
9234: the following words:
9235:
9236:
9237: doc-docol:
9238: doc-docon:
9239: doc-dovar:
9240: doc-douser:
9241: doc-dodefer:
9242: doc-dofield:
9243:
9244:
9245: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
9246: with @code{>does-code}. If the word was defined in that way, the value
9247: returned is non-zero and identifies the @code{DOES>} used by the
9248: defining word.
9249: @comment TODO should that be ``identifies the xt of the DOES> ??''
9250:
9251: @c -------------------------------------------------------------
9252: @node Locals, Structures, Threading Words, Words
9253: @section Locals
9254: @cindex locals
9255:
9256: Local variables can make Forth programming more enjoyable and Forth
9257: programs easier to read. Unfortunately, the locals of ANS Forth are
9258: laden with restrictions. Therefore, we provide not only the ANS Forth
9259: locals wordset, but also our own, more powerful locals wordset (we
9260: implemented the ANS Forth locals wordset through our locals wordset).
9261:
9262: The ideas in this section have also been published in M. Anton Ertl,
9263: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz,
9264: Automatic Scoping of Local Variables}}, EuroForth '94.
9265:
9266: @menu
9267: * Gforth locals::
9268: * ANS Forth locals::
9269: @end menu
9270:
9271: @node Gforth locals, ANS Forth locals, Locals, Locals
9272: @subsection Gforth locals
9273: @cindex Gforth locals
9274: @cindex locals, Gforth style
9275:
9276: Locals can be defined with
9277:
9278: @example
9279: @{ local1 local2 ... -- comment @}
9280: @end example
9281: or
9282: @example
9283: @{ local1 local2 ... @}
9284: @end example
9285:
9286: E.g.,
9287: @example
9288: : max @{ n1 n2 -- n3 @}
9289: n1 n2 > if
9290: n1
9291: else
9292: n2
9293: endif ;
9294: @end example
9295:
9296: The similarity of locals definitions with stack comments is intended. A
9297: locals definition often replaces the stack comment of a word. The order
9298: of the locals corresponds to the order in a stack comment and everything
9299: after the @code{--} is really a comment.
9300:
9301: This similarity has one disadvantage: It is too easy to confuse locals
9302: declarations with stack comments, causing bugs and making them hard to
9303: find. However, this problem can be avoided by appropriate coding
9304: conventions: Do not use both notations in the same program. If you do,
9305: they should be distinguished using additional means, e.g. by position.
9306:
9307: @cindex types of locals
9308: @cindex locals types
9309: The name of the local may be preceded by a type specifier, e.g.,
9310: @code{F:} for a floating point value:
9311:
9312: @example
9313: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
9314: \ complex multiplication
9315: Ar Br f* Ai Bi f* f-
9316: Ar Bi f* Ai Br f* f+ ;
9317: @end example
9318:
9319: @cindex flavours of locals
9320: @cindex locals flavours
9321: @cindex value-flavoured locals
9322: @cindex variable-flavoured locals
9323: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
9324: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
9325: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
9326: with @code{W:}, @code{D:} etc.) produces its value and can be changed
9327: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
9328: produces its address (which becomes invalid when the variable's scope is
9329: left). E.g., the standard word @code{emit} can be defined in terms of
9330: @code{type} like this:
9331:
9332: @example
9333: : emit @{ C^ char* -- @}
9334: char* 1 type ;
9335: @end example
9336:
9337: @cindex default type of locals
9338: @cindex locals, default type
9339: A local without type specifier is a @code{W:} local. Both flavours of
9340: locals are initialized with values from the data or FP stack.
9341:
9342: Currently there is no way to define locals with user-defined data
9343: structures, but we are working on it.
9344:
9345: Gforth allows defining locals everywhere in a colon definition. This
9346: poses the following questions:
9347:
9348: @menu
9349: * Where are locals visible by name?::
9350: * How long do locals live?::
9351: * Programming Style::
9352: * Implementation::
9353: @end menu
9354:
9355: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
9356: @subsubsection Where are locals visible by name?
9357: @cindex locals visibility
9358: @cindex visibility of locals
9359: @cindex scope of locals
9360:
9361: Basically, the answer is that locals are visible where you would expect
9362: it in block-structured languages, and sometimes a little longer. If you
9363: want to restrict the scope of a local, enclose its definition in
9364: @code{SCOPE}...@code{ENDSCOPE}.
9365:
9366:
9367: doc-scope
9368: doc-endscope
9369:
9370:
9371: These words behave like control structure words, so you can use them
9372: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
9373: arbitrary ways.
9374:
9375: If you want a more exact answer to the visibility question, here's the
9376: basic principle: A local is visible in all places that can only be
9377: reached through the definition of the local@footnote{In compiler
9378: construction terminology, all places dominated by the definition of the
9379: local.}. In other words, it is not visible in places that can be reached
9380: without going through the definition of the local. E.g., locals defined
9381: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
9382: defined in @code{BEGIN}...@code{UNTIL} are visible after the
9383: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
9384:
9385: The reasoning behind this solution is: We want to have the locals
9386: visible as long as it is meaningful. The user can always make the
9387: visibility shorter by using explicit scoping. In a place that can
9388: only be reached through the definition of a local, the meaning of a
9389: local name is clear. In other places it is not: How is the local
9390: initialized at the control flow path that does not contain the
9391: definition? Which local is meant, if the same name is defined twice in
9392: two independent control flow paths?
9393:
9394: This should be enough detail for nearly all users, so you can skip the
9395: rest of this section. If you really must know all the gory details and
9396: options, read on.
9397:
9398: In order to implement this rule, the compiler has to know which places
9399: are unreachable. It knows this automatically after @code{AHEAD},
9400: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
9401: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
9402: compiler that the control flow never reaches that place. If
9403: @code{UNREACHABLE} is not used where it could, the only consequence is
9404: that the visibility of some locals is more limited than the rule above
9405: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
9406: lie to the compiler), buggy code will be produced.
9407:
9408:
9409: doc-unreachable
9410:
9411:
9412: Another problem with this rule is that at @code{BEGIN}, the compiler
9413: does not know which locals will be visible on the incoming
9414: back-edge. All problems discussed in the following are due to this
9415: ignorance of the compiler (we discuss the problems using @code{BEGIN}
9416: loops as examples; the discussion also applies to @code{?DO} and other
9417: loops). Perhaps the most insidious example is:
9418: @example
9419: AHEAD
9420: BEGIN
9421: x
9422: [ 1 CS-ROLL ] THEN
9423: @{ x @}
9424: ...
9425: UNTIL
9426: @end example
9427:
9428: This should be legal according to the visibility rule. The use of
9429: @code{x} can only be reached through the definition; but that appears
9430: textually below the use.
9431:
9432: From this example it is clear that the visibility rules cannot be fully
9433: implemented without major headaches. Our implementation treats common
9434: cases as advertised and the exceptions are treated in a safe way: The
9435: compiler makes a reasonable guess about the locals visible after a
9436: @code{BEGIN}; if it is too pessimistic, the
9437: user will get a spurious error about the local not being defined; if the
9438: compiler is too optimistic, it will notice this later and issue a
9439: warning. In the case above the compiler would complain about @code{x}
9440: being undefined at its use. You can see from the obscure examples in
9441: this section that it takes quite unusual control structures to get the
9442: compiler into trouble, and even then it will often do fine.
9443:
9444: If the @code{BEGIN} is reachable from above, the most optimistic guess
9445: is that all locals visible before the @code{BEGIN} will also be
9446: visible after the @code{BEGIN}. This guess is valid for all loops that
9447: are entered only through the @code{BEGIN}, in particular, for normal
9448: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
9449: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
9450: compiler. When the branch to the @code{BEGIN} is finally generated by
9451: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
9452: warns the user if it was too optimistic:
9453: @example
9454: IF
9455: @{ x @}
9456: BEGIN
9457: \ x ?
9458: [ 1 cs-roll ] THEN
9459: ...
9460: UNTIL
9461: @end example
9462:
9463: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
9464: optimistically assumes that it lives until the @code{THEN}. It notices
9465: this difference when it compiles the @code{UNTIL} and issues a
9466: warning. The user can avoid the warning, and make sure that @code{x}
9467: is not used in the wrong area by using explicit scoping:
9468: @example
9469: IF
9470: SCOPE
9471: @{ x @}
9472: ENDSCOPE
9473: BEGIN
9474: [ 1 cs-roll ] THEN
9475: ...
9476: UNTIL
9477: @end example
9478:
9479: Since the guess is optimistic, there will be no spurious error messages
9480: about undefined locals.
9481:
9482: If the @code{BEGIN} is not reachable from above (e.g., after
9483: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
9484: optimistic guess, as the locals visible after the @code{BEGIN} may be
9485: defined later. Therefore, the compiler assumes that no locals are
9486: visible after the @code{BEGIN}. However, the user can use
9487: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
9488: visible at the BEGIN as at the point where the top control-flow stack
9489: item was created.
9490:
9491:
9492: doc-assume-live
9493:
9494:
9495: @noindent
9496: E.g.,
9497: @example
9498: @{ x @}
9499: AHEAD
9500: ASSUME-LIVE
9501: BEGIN
9502: x
9503: [ 1 CS-ROLL ] THEN
9504: ...
9505: UNTIL
9506: @end example
9507:
9508: Other cases where the locals are defined before the @code{BEGIN} can be
9509: handled by inserting an appropriate @code{CS-ROLL} before the
9510: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
9511: behind the @code{ASSUME-LIVE}).
9512:
9513: Cases where locals are defined after the @code{BEGIN} (but should be
9514: visible immediately after the @code{BEGIN}) can only be handled by
9515: rearranging the loop. E.g., the ``most insidious'' example above can be
9516: arranged into:
9517: @example
9518: BEGIN
9519: @{ x @}
9520: ... 0=
9521: WHILE
9522: x
9523: REPEAT
9524: @end example
9525:
9526: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
9527: @subsubsection How long do locals live?
9528: @cindex locals lifetime
9529: @cindex lifetime of locals
9530:
9531: The right answer for the lifetime question would be: A local lives at
9532: least as long as it can be accessed. For a value-flavoured local this
9533: means: until the end of its visibility. However, a variable-flavoured
9534: local could be accessed through its address far beyond its visibility
9535: scope. Ultimately, this would mean that such locals would have to be
9536: garbage collected. Since this entails un-Forth-like implementation
9537: complexities, I adopted the same cowardly solution as some other
9538: languages (e.g., C): The local lives only as long as it is visible;
9539: afterwards its address is invalid (and programs that access it
9540: afterwards are erroneous).
9541:
9542: @node Programming Style, Implementation, How long do locals live?, Gforth locals
9543: @subsubsection Programming Style
9544: @cindex locals programming style
9545: @cindex programming style, locals
9546:
9547: The freedom to define locals anywhere has the potential to change
9548: programming styles dramatically. In particular, the need to use the
9549: return stack for intermediate storage vanishes. Moreover, all stack
9550: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
9551: determined arguments) can be eliminated: If the stack items are in the
9552: wrong order, just write a locals definition for all of them; then
9553: write the items in the order you want.
9554:
9555: This seems a little far-fetched and eliminating stack manipulations is
9556: unlikely to become a conscious programming objective. Still, the number
9557: of stack manipulations will be reduced dramatically if local variables
9558: are used liberally (e.g., compare @code{max} (@pxref{Gforth locals}) with
9559: a traditional implementation of @code{max}).
9560:
9561: This shows one potential benefit of locals: making Forth programs more
9562: readable. Of course, this benefit will only be realized if the
9563: programmers continue to honour the principle of factoring instead of
9564: using the added latitude to make the words longer.
9565:
9566: @cindex single-assignment style for locals
9567: Using @code{TO} can and should be avoided. Without @code{TO},
9568: every value-flavoured local has only a single assignment and many
9569: advantages of functional languages apply to Forth. I.e., programs are
9570: easier to analyse, to optimize and to read: It is clear from the
9571: definition what the local stands for, it does not turn into something
9572: different later.
9573:
9574: E.g., a definition using @code{TO} might look like this:
9575: @example
9576: : strcmp @{ addr1 u1 addr2 u2 -- n @}
9577: u1 u2 min 0
9578: ?do
9579: addr1 c@@ addr2 c@@ -
9580: ?dup-if
9581: unloop exit
9582: then
9583: addr1 char+ TO addr1
9584: addr2 char+ TO addr2
9585: loop
9586: u1 u2 - ;
9587: @end example
9588: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
9589: every loop iteration. @code{strcmp} is a typical example of the
9590: readability problems of using @code{TO}. When you start reading
9591: @code{strcmp}, you think that @code{addr1} refers to the start of the
9592: string. Only near the end of the loop you realize that it is something
9593: else.
9594:
9595: This can be avoided by defining two locals at the start of the loop that
9596: are initialized with the right value for the current iteration.
9597: @example
9598: : strcmp @{ addr1 u1 addr2 u2 -- n @}
9599: addr1 addr2
9600: u1 u2 min 0
9601: ?do @{ s1 s2 @}
9602: s1 c@@ s2 c@@ -
9603: ?dup-if
9604: unloop exit
9605: then
9606: s1 char+ s2 char+
9607: loop
9608: 2drop
9609: u1 u2 - ;
9610: @end example
9611: Here it is clear from the start that @code{s1} has a different value
9612: in every loop iteration.
9613:
9614: @node Implementation, , Programming Style, Gforth locals
9615: @subsubsection Implementation
9616: @cindex locals implementation
9617: @cindex implementation of locals
9618:
9619: @cindex locals stack
9620: Gforth uses an extra locals stack. The most compelling reason for
9621: this is that the return stack is not float-aligned; using an extra stack
9622: also eliminates the problems and restrictions of using the return stack
9623: as locals stack. Like the other stacks, the locals stack grows toward
9624: lower addresses. A few primitives allow an efficient implementation:
9625:
9626:
9627: doc-@local#
9628: doc-f@local#
9629: doc-laddr#
9630: doc-lp+!#
9631: doc-lp!
9632: doc->l
9633: doc-f>l
9634:
9635:
9636: In addition to these primitives, some specializations of these
9637: primitives for commonly occurring inline arguments are provided for
9638: efficiency reasons, e.g., @code{@@local0} as specialization of
9639: @code{@@local#} for the inline argument 0. The following compiling words
9640: compile the right specialized version, or the general version, as
9641: appropriate:
9642:
9643:
9644: doc-compile-@local
9645: doc-compile-f@local
9646: doc-compile-lp+!
9647:
9648:
9649: Combinations of conditional branches and @code{lp+!#} like
9650: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
9651: is taken) are provided for efficiency and correctness in loops.
9652:
9653: A special area in the dictionary space is reserved for keeping the
9654: local variable names. @code{@{} switches the dictionary pointer to this
9655: area and @code{@}} switches it back and generates the locals
9656: initializing code. @code{W:} etc.@ are normal defining words. This
9657: special area is cleared at the start of every colon definition.
9658:
9659: @cindex word list for defining locals
9660: A special feature of Gforth's dictionary is used to implement the
9661: definition of locals without type specifiers: every word list (aka
9662: vocabulary) has its own methods for searching
9663: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
9664: with a special search method: When it is searched for a word, it
9665: actually creates that word using @code{W:}. @code{@{} changes the search
9666: order to first search the word list containing @code{@}}, @code{W:} etc.,
9667: and then the word list for defining locals without type specifiers.
9668:
9669: The lifetime rules support a stack discipline within a colon
9670: definition: The lifetime of a local is either nested with other locals
9671: lifetimes or it does not overlap them.
9672:
9673: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
9674: pointer manipulation is generated. Between control structure words
9675: locals definitions can push locals onto the locals stack. @code{AGAIN}
9676: is the simplest of the other three control flow words. It has to
9677: restore the locals stack depth of the corresponding @code{BEGIN}
9678: before branching. The code looks like this:
9679: @format
9680: @code{lp+!#} current-locals-size @minus{} dest-locals-size
9681: @code{branch} <begin>
9682: @end format
9683:
9684: @code{UNTIL} is a little more complicated: If it branches back, it
9685: must adjust the stack just like @code{AGAIN}. But if it falls through,
9686: the locals stack must not be changed. The compiler generates the
9687: following code:
9688: @format
9689: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
9690: @end format
9691: The locals stack pointer is only adjusted if the branch is taken.
9692:
9693: @code{THEN} can produce somewhat inefficient code:
9694: @format
9695: @code{lp+!#} current-locals-size @minus{} orig-locals-size
9696: <orig target>:
9697: @code{lp+!#} orig-locals-size @minus{} new-locals-size
9698: @end format
9699: The second @code{lp+!#} adjusts the locals stack pointer from the
9700: level at the @i{orig} point to the level after the @code{THEN}. The
9701: first @code{lp+!#} adjusts the locals stack pointer from the current
9702: level to the level at the orig point, so the complete effect is an
9703: adjustment from the current level to the right level after the
9704: @code{THEN}.
9705:
9706: @cindex locals information on the control-flow stack
9707: @cindex control-flow stack items, locals information
9708: In a conventional Forth implementation a dest control-flow stack entry
9709: is just the target address and an orig entry is just the address to be
9710: patched. Our locals implementation adds a word list to every orig or dest
9711: item. It is the list of locals visible (or assumed visible) at the point
9712: described by the entry. Our implementation also adds a tag to identify
9713: the kind of entry, in particular to differentiate between live and dead
9714: (reachable and unreachable) orig entries.
9715:
9716: A few unusual operations have to be performed on locals word lists:
9717:
9718:
9719: doc-common-list
9720: doc-sub-list?
9721: doc-list-size
9722:
9723:
9724: Several features of our locals word list implementation make these
9725: operations easy to implement: The locals word lists are organised as
9726: linked lists; the tails of these lists are shared, if the lists
9727: contain some of the same locals; and the address of a name is greater
9728: than the address of the names behind it in the list.
9729:
9730: Another important implementation detail is the variable
9731: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
9732: determine if they can be reached directly or only through the branch
9733: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
9734: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
9735: definition, by @code{BEGIN} and usually by @code{THEN}.
9736:
9737: Counted loops are similar to other loops in most respects, but
9738: @code{LEAVE} requires special attention: It performs basically the same
9739: service as @code{AHEAD}, but it does not create a control-flow stack
9740: entry. Therefore the information has to be stored elsewhere;
9741: traditionally, the information was stored in the target fields of the
9742: branches created by the @code{LEAVE}s, by organizing these fields into a
9743: linked list. Unfortunately, this clever trick does not provide enough
9744: space for storing our extended control flow information. Therefore, we
9745: introduce another stack, the leave stack. It contains the control-flow
9746: stack entries for all unresolved @code{LEAVE}s.
9747:
9748: Local names are kept until the end of the colon definition, even if
9749: they are no longer visible in any control-flow path. In a few cases
9750: this may lead to increased space needs for the locals name area, but
9751: usually less than reclaiming this space would cost in code size.
9752:
9753:
9754: @node ANS Forth locals, , Gforth locals, Locals
9755: @subsection ANS Forth locals
9756: @cindex locals, ANS Forth style
9757:
9758: The ANS Forth locals wordset does not define a syntax for locals, but
9759: words that make it possible to define various syntaxes. One of the
9760: possible syntaxes is a subset of the syntax we used in the Gforth locals
9761: wordset, i.e.:
9762:
9763: @example
9764: @{ local1 local2 ... -- comment @}
9765: @end example
9766: @noindent
9767: or
9768: @example
9769: @{ local1 local2 ... @}
9770: @end example
9771:
9772: The order of the locals corresponds to the order in a stack comment. The
9773: restrictions are:
9774:
9775: @itemize @bullet
9776: @item
9777: Locals can only be cell-sized values (no type specifiers are allowed).
9778: @item
9779: Locals can be defined only outside control structures.
9780: @item
9781: Locals can interfere with explicit usage of the return stack. For the
9782: exact (and long) rules, see the standard. If you don't use return stack
9783: accessing words in a definition using locals, you will be all right. The
9784: purpose of this rule is to make locals implementation on the return
9785: stack easier.
9786: @item
9787: The whole definition must be in one line.
9788: @end itemize
9789:
9790: Locals defined in this way behave like @code{VALUE}s
9791: (@pxref{Values}). I.e., they are initialized from the stack. Using their
9792: name produces their value. Their value can be changed using @code{TO}.
9793:
9794: Since this syntax is supported by Gforth directly, you need not do
9795: anything to use it. If you want to port a program using this syntax to
9796: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
9797: syntax on the other system.
9798:
9799: Note that a syntax shown in the standard, section A.13 looks
9800: similar, but is quite different in having the order of locals
9801: reversed. Beware!
9802:
9803: The ANS Forth locals wordset itself consists of a word:
9804:
9805:
9806: doc-(local)
9807:
9808:
9809: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
9810: awful that we strongly recommend not to use it. We have implemented this
9811: syntax to make porting to Gforth easy, but do not document it here. The
9812: problem with this syntax is that the locals are defined in an order
9813: reversed with respect to the standard stack comment notation, making
9814: programs harder to read, and easier to misread and miswrite. The only
9815: merit of this syntax is that it is easy to implement using the ANS Forth
9816: locals wordset.
9817:
9818:
9819: @c ----------------------------------------------------------
9820: @node Structures, Object-oriented Forth, Locals, Words
9821: @section Structures
9822: @cindex structures
9823: @cindex records
9824:
9825: This section presents the structure package that comes with Gforth. A
9826: version of the package implemented in ANS Forth is available in
9827: @file{compat/struct.fs}. This package was inspired by a posting on
9828: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
9829: possibly John Hayes). A version of this section has been published in
9830: ???. Marcel Hendrix provided helpful comments.
9831:
9832: @menu
9833: * Why explicit structure support?::
9834: * Structure Usage::
9835: * Structure Naming Convention::
9836: * Structure Implementation::
9837: * Structure Glossary::
9838: @end menu
9839:
9840: @node Why explicit structure support?, Structure Usage, Structures, Structures
9841: @subsection Why explicit structure support?
9842:
9843: @cindex address arithmetic for structures
9844: @cindex structures using address arithmetic
9845: If we want to use a structure containing several fields, we could simply
9846: reserve memory for it, and access the fields using address arithmetic
9847: (@pxref{Address arithmetic}). As an example, consider a structure with
9848: the following fields
9849:
9850: @table @code
9851: @item a
9852: is a float
9853: @item b
9854: is a cell
9855: @item c
9856: is a float
9857: @end table
9858:
9859: Given the (float-aligned) base address of the structure we get the
9860: address of the field
9861:
9862: @table @code
9863: @item a
9864: without doing anything further.
9865: @item b
9866: with @code{float+}
9867: @item c
9868: with @code{float+ cell+ faligned}
9869: @end table
9870:
9871: It is easy to see that this can become quite tiring.
9872:
9873: Moreover, it is not very readable, because seeing a
9874: @code{cell+} tells us neither which kind of structure is
9875: accessed nor what field is accessed; we have to somehow infer the kind
9876: of structure, and then look up in the documentation, which field of
9877: that structure corresponds to that offset.
9878:
9879: Finally, this kind of address arithmetic also causes maintenance
9880: troubles: If you add or delete a field somewhere in the middle of the
9881: structure, you have to find and change all computations for the fields
9882: afterwards.
9883:
9884: So, instead of using @code{cell+} and friends directly, how
9885: about storing the offsets in constants:
9886:
9887: @example
9888: 0 constant a-offset
9889: 0 float+ constant b-offset
9890: 0 float+ cell+ faligned c-offset
9891: @end example
9892:
9893: Now we can get the address of field @code{x} with @code{x-offset
9894: +}. This is much better in all respects. Of course, you still
9895: have to change all later offset definitions if you add a field. You can
9896: fix this by declaring the offsets in the following way:
9897:
9898: @example
9899: 0 constant a-offset
9900: a-offset float+ constant b-offset
9901: b-offset cell+ faligned constant c-offset
9902: @end example
9903:
9904: Since we always use the offsets with @code{+}, we could use a defining
9905: word @code{cfield} that includes the @code{+} in the action of the
9906: defined word:
9907:
9908: @example
9909: : cfield ( n "name" -- )
9910: create ,
9911: does> ( name execution: addr1 -- addr2 )
9912: @@ + ;
9913:
9914: 0 cfield a
9915: 0 a float+ cfield b
9916: 0 b cell+ faligned cfield c
9917: @end example
9918:
9919: Instead of @code{x-offset +}, we now simply write @code{x}.
9920:
9921: The structure field words now can be used quite nicely. However,
9922: their definition is still a bit cumbersome: We have to repeat the
9923: name, the information about size and alignment is distributed before
9924: and after the field definitions etc. The structure package presented
9925: here addresses these problems.
9926:
9927: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
9928: @subsection Structure Usage
9929: @cindex structure usage
9930:
9931: @cindex @code{field} usage
9932: @cindex @code{struct} usage
9933: @cindex @code{end-struct} usage
9934: You can define a structure for a (data-less) linked list with:
9935: @example
9936: struct
9937: cell% field list-next
9938: end-struct list%
9939: @end example
9940:
9941: With the address of the list node on the stack, you can compute the
9942: address of the field that contains the address of the next node with
9943: @code{list-next}. E.g., you can determine the length of a list
9944: with:
9945:
9946: @example
9947: : list-length ( list -- n )
9948: \ "list" is a pointer to the first element of a linked list
9949: \ "n" is the length of the list
9950: 0 BEGIN ( list1 n1 )
9951: over
9952: WHILE ( list1 n1 )
9953: 1+ swap list-next @@ swap
9954: REPEAT
9955: nip ;
9956: @end example
9957:
9958: You can reserve memory for a list node in the dictionary with
9959: @code{list% %allot}, which leaves the address of the list node on the
9960: stack. For the equivalent allocation on the heap you can use @code{list%
9961: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
9962: use @code{list% %allocate}). You can get the the size of a list
9963: node with @code{list% %size} and its alignment with @code{list%
9964: %alignment}.
9965:
9966: Note that in ANS Forth the body of a @code{create}d word is
9967: @code{aligned} but not necessarily @code{faligned};
9968: therefore, if you do a:
9969: @example
9970: create @emph{name} foo% %allot
9971: @end example
9972:
9973: @noindent
9974: then the memory alloted for @code{foo%} is
9975: guaranteed to start at the body of @code{@emph{name}} only if
9976: @code{foo%} contains only character, cell and double fields.
9977:
9978: @cindex structures containing structures
9979: You can include a structure @code{foo%} as a field of
9980: another structure, like this:
9981: @example
9982: struct
9983: ...
9984: foo% field ...
9985: ...
9986: end-struct ...
9987: @end example
9988:
9989: @cindex structure extension
9990: @cindex extended records
9991: Instead of starting with an empty structure, you can extend an
9992: existing structure. E.g., a plain linked list without data, as defined
9993: above, is hardly useful; You can extend it to a linked list of integers,
9994: like this:@footnote{This feature is also known as @emph{extended
9995: records}. It is the main innovation in the Oberon language; in other
9996: words, adding this feature to Modula-2 led Wirth to create a new
9997: language, write a new compiler etc. Adding this feature to Forth just
9998: required a few lines of code.}
9999:
10000: @example
10001: list%
10002: cell% field intlist-int
10003: end-struct intlist%
10004: @end example
10005:
10006: @code{intlist%} is a structure with two fields:
10007: @code{list-next} and @code{intlist-int}.
10008:
10009: @cindex structures containing arrays
10010: You can specify an array type containing @emph{n} elements of
10011: type @code{foo%} like this:
10012:
10013: @example
10014: foo% @emph{n} *
10015: @end example
10016:
10017: You can use this array type in any place where you can use a normal
10018: type, e.g., when defining a @code{field}, or with
10019: @code{%allot}.
10020:
10021: @cindex first field optimization
10022: The first field is at the base address of a structure and the word
10023: for this field (e.g., @code{list-next}) actually does not change
10024: the address on the stack. You may be tempted to leave it away in the
10025: interest of run-time and space efficiency. This is not necessary,
10026: because the structure package optimizes this case and compiling such
10027: words does not generate any code. So, in the interest of readability
10028: and maintainability you should include the word for the field when
10029: accessing the field.
10030:
10031: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
10032: @subsection Structure Naming Convention
10033: @cindex structure naming convention
10034:
10035: The field names that come to (my) mind are often quite generic, and,
10036: if used, would cause frequent name clashes. E.g., many structures
10037: probably contain a @code{counter} field. The structure names
10038: that come to (my) mind are often also the logical choice for the names
10039: of words that create such a structure.
10040:
10041: Therefore, I have adopted the following naming conventions:
10042:
10043: @itemize @bullet
10044: @cindex field naming convention
10045: @item
10046: The names of fields are of the form
10047: @code{@emph{struct}-@emph{field}}, where
10048: @code{@emph{struct}} is the basic name of the structure, and
10049: @code{@emph{field}} is the basic name of the field. You can
10050: think of field words as converting the (address of the)
10051: structure into the (address of the) field.
10052:
10053: @cindex structure naming convention
10054: @item
10055: The names of structures are of the form
10056: @code{@emph{struct}%}, where
10057: @code{@emph{struct}} is the basic name of the structure.
10058: @end itemize
10059:
10060: This naming convention does not work that well for fields of extended
10061: structures; e.g., the integer list structure has a field
10062: @code{intlist-int}, but has @code{list-next}, not
10063: @code{intlist-next}.
10064:
10065: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
10066: @subsection Structure Implementation
10067: @cindex structure implementation
10068: @cindex implementation of structures
10069:
10070: The central idea in the implementation is to pass the data about the
10071: structure being built on the stack, not in some global
10072: variable. Everything else falls into place naturally once this design
10073: decision is made.
10074:
10075: The type description on the stack is of the form @emph{align
10076: size}. Keeping the size on the top-of-stack makes dealing with arrays
10077: very simple.
10078:
10079: @code{field} is a defining word that uses @code{Create}
10080: and @code{DOES>}. The body of the field contains the offset
10081: of the field, and the normal @code{DOES>} action is simply:
10082:
10083: @example
10084: @@ +
10085: @end example
10086:
10087: @noindent
10088: i.e., add the offset to the address, giving the stack effect
10089: @i{addr1 -- addr2} for a field.
10090:
10091: @cindex first field optimization, implementation
10092: This simple structure is slightly complicated by the optimization
10093: for fields with offset 0, which requires a different
10094: @code{DOES>}-part (because we cannot rely on there being
10095: something on the stack if such a field is invoked during
10096: compilation). Therefore, we put the different @code{DOES>}-parts
10097: in separate words, and decide which one to invoke based on the
10098: offset. For a zero offset, the field is basically a noop; it is
10099: immediate, and therefore no code is generated when it is compiled.
10100:
10101: @node Structure Glossary, , Structure Implementation, Structures
10102: @subsection Structure Glossary
10103: @cindex structure glossary
10104:
10105:
10106: doc-%align
10107: doc-%alignment
10108: doc-%alloc
10109: doc-%allocate
10110: doc-%allot
10111: doc-cell%
10112: doc-char%
10113: doc-dfloat%
10114: doc-double%
10115: doc-end-struct
10116: doc-field
10117: doc-float%
10118: doc-naligned
10119: doc-sfloat%
10120: doc-%size
10121: doc-struct
10122:
10123:
10124: @c -------------------------------------------------------------
10125: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
10126: @section Object-oriented Forth
10127:
10128: Gforth comes with three packages for object-oriented programming:
10129: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
10130: is preloaded, so you have to @code{include} them before use. The most
10131: important differences between these packages (and others) are discussed
10132: in @ref{Comparison with other object models}. All packages are written
10133: in ANS Forth and can be used with any other ANS Forth.
10134:
10135: @menu
10136: * Why object-oriented programming?::
10137: * Object-Oriented Terminology::
10138: * Objects::
10139: * OOF::
10140: * Mini-OOF::
10141: * Comparison with other object models::
10142: @end menu
10143:
10144: @c ----------------------------------------------------------------
10145: @node Why object-oriented programming?, Object-Oriented Terminology, Object-oriented Forth, Object-oriented Forth
10146: @subsection Why object-oriented programming?
10147: @cindex object-oriented programming motivation
10148: @cindex motivation for object-oriented programming
10149:
10150: Often we have to deal with several data structures (@emph{objects}),
10151: that have to be treated similarly in some respects, but differently in
10152: others. Graphical objects are the textbook example: circles, triangles,
10153: dinosaurs, icons, and others, and we may want to add more during program
10154: development. We want to apply some operations to any graphical object,
10155: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
10156: has to do something different for every kind of object.
10157: @comment TODO add some other operations eg perimeter, area
10158: @comment and tie in to concrete examples later..
10159:
10160: We could implement @code{draw} as a big @code{CASE}
10161: control structure that executes the appropriate code depending on the
10162: kind of object to be drawn. This would be not be very elegant, and,
10163: moreover, we would have to change @code{draw} every time we add
10164: a new kind of graphical object (say, a spaceship).
10165:
10166: What we would rather do is: When defining spaceships, we would tell
10167: the system: ``Here's how you @code{draw} a spaceship; you figure
10168: out the rest''.
10169:
10170: This is the problem that all systems solve that (rightfully) call
10171: themselves object-oriented; the object-oriented packages presented here
10172: solve this problem (and not much else).
10173: @comment TODO ?list properties of oo systems.. oo vs o-based?
10174:
10175: @c ------------------------------------------------------------------------
10176: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
10177: @subsection Object-Oriented Terminology
10178: @cindex object-oriented terminology
10179: @cindex terminology for object-oriented programming
10180:
10181: This section is mainly for reference, so you don't have to understand
10182: all of it right away. The terminology is mainly Smalltalk-inspired. In
10183: short:
10184:
10185: @table @emph
10186: @cindex class
10187: @item class
10188: a data structure definition with some extras.
10189:
10190: @cindex object
10191: @item object
10192: an instance of the data structure described by the class definition.
10193:
10194: @cindex instance variables
10195: @item instance variables
10196: fields of the data structure.
10197:
10198: @cindex selector
10199: @cindex method selector
10200: @cindex virtual function
10201: @item selector
10202: (or @emph{method selector}) a word (e.g.,
10203: @code{draw}) that performs an operation on a variety of data
10204: structures (classes). A selector describes @emph{what} operation to
10205: perform. In C++ terminology: a (pure) virtual function.
10206:
10207: @cindex method
10208: @item method
10209: the concrete definition that performs the operation
10210: described by the selector for a specific class. A method specifies
10211: @emph{how} the operation is performed for a specific class.
10212:
10213: @cindex selector invocation
10214: @cindex message send
10215: @cindex invoking a selector
10216: @item selector invocation
10217: a call of a selector. One argument of the call (the TOS (top-of-stack))
10218: is used for determining which method is used. In Smalltalk terminology:
10219: a message (consisting of the selector and the other arguments) is sent
10220: to the object.
10221:
10222: @cindex receiving object
10223: @item receiving object
10224: the object used for determining the method executed by a selector
10225: invocation. In the @file{objects.fs} model, it is the object that is on
10226: the TOS when the selector is invoked. (@emph{Receiving} comes from
10227: the Smalltalk @emph{message} terminology.)
10228:
10229: @cindex child class
10230: @cindex parent class
10231: @cindex inheritance
10232: @item child class
10233: a class that has (@emph{inherits}) all properties (instance variables,
10234: selectors, methods) from a @emph{parent class}. In Smalltalk
10235: terminology: The subclass inherits from the superclass. In C++
10236: terminology: The derived class inherits from the base class.
10237:
10238: @end table
10239:
10240: @c If you wonder about the message sending terminology, it comes from
10241: @c a time when each object had it's own task and objects communicated via
10242: @c message passing; eventually the Smalltalk developers realized that
10243: @c they can do most things through simple (indirect) calls. They kept the
10244: @c terminology.
10245:
10246: @c --------------------------------------------------------------
10247: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
10248: @subsection The @file{objects.fs} model
10249: @cindex objects
10250: @cindex object-oriented programming
10251:
10252: @cindex @file{objects.fs}
10253: @cindex @file{oof.fs}
10254:
10255: This section describes the @file{objects.fs} package. This material also
10256: has been published in M. Anton Ertl,
10257: @cite{@uref{http://www.complang.tuwien.ac.at/forth/objects/objects.html,
10258: Yet Another Forth Objects Package}}, Forth Dimensions 19(2), pages
10259: 37--43.
10260: @c McKewan's and Zsoter's packages
10261:
10262: This section assumes that you have read @ref{Structures}.
10263:
10264: The techniques on which this model is based have been used to implement
10265: the parser generator, Gray, and have also been used in Gforth for
10266: implementing the various flavours of word lists (hashed or not,
10267: case-sensitive or not, special-purpose word lists for locals etc.).
10268:
10269:
10270: @menu
10271: * Properties of the Objects model::
10272: * Basic Objects Usage::
10273: * The Objects base class::
10274: * Creating objects::
10275: * Object-Oriented Programming Style::
10276: * Class Binding::
10277: * Method conveniences::
10278: * Classes and Scoping::
10279: * Dividing classes::
10280: * Object Interfaces::
10281: * Objects Implementation::
10282: * Objects Glossary::
10283: @end menu
10284:
10285: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
10286: and Bernd Paysan helped me with the related works section.
10287:
10288: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
10289: @subsubsection Properties of the @file{objects.fs} model
10290: @cindex @file{objects.fs} properties
10291:
10292: @itemize @bullet
10293: @item
10294: It is straightforward to pass objects on the stack. Passing
10295: selectors on the stack is a little less convenient, but possible.
10296:
10297: @item
10298: Objects are just data structures in memory, and are referenced by their
10299: address. You can create words for objects with normal defining words
10300: like @code{constant}. Likewise, there is no difference between instance
10301: variables that contain objects and those that contain other data.
10302:
10303: @item
10304: Late binding is efficient and easy to use.
10305:
10306: @item
10307: It avoids parsing, and thus avoids problems with state-smartness
10308: and reduced extensibility; for convenience there are a few parsing
10309: words, but they have non-parsing counterparts. There are also a few
10310: defining words that parse. This is hard to avoid, because all standard
10311: defining words parse (except @code{:noname}); however, such
10312: words are not as bad as many other parsing words, because they are not
10313: state-smart.
10314:
10315: @item
10316: It does not try to incorporate everything. It does a few things and does
10317: them well (IMO). In particular, this model was not designed to support
10318: information hiding (although it has features that may help); you can use
10319: a separate package for achieving this.
10320:
10321: @item
10322: It is layered; you don't have to learn and use all features to use this
10323: model. Only a few features are necessary (@pxref{Basic Objects Usage},
10324: @pxref{The Objects base class}, @pxref{Creating objects}.), the others
10325: are optional and independent of each other.
10326:
10327: @item
10328: An implementation in ANS Forth is available.
10329:
10330: @end itemize
10331:
10332:
10333: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
10334: @subsubsection Basic @file{objects.fs} Usage
10335: @cindex basic objects usage
10336: @cindex objects, basic usage
10337:
10338: You can define a class for graphical objects like this:
10339:
10340: @cindex @code{class} usage
10341: @cindex @code{end-class} usage
10342: @cindex @code{selector} usage
10343: @example
10344: object class \ "object" is the parent class
10345: selector draw ( x y graphical -- )
10346: end-class graphical
10347: @end example
10348:
10349: This code defines a class @code{graphical} with an
10350: operation @code{draw}. We can perform the operation
10351: @code{draw} on any @code{graphical} object, e.g.:
10352:
10353: @example
10354: 100 100 t-rex draw
10355: @end example
10356:
10357: @noindent
10358: where @code{t-rex} is a word (say, a constant) that produces a
10359: graphical object.
10360:
10361: @comment TODO add a 2nd operation eg perimeter.. and use for
10362: @comment a concrete example
10363:
10364: @cindex abstract class
10365: How do we create a graphical object? With the present definitions,
10366: we cannot create a useful graphical object. The class
10367: @code{graphical} describes graphical objects in general, but not
10368: any concrete graphical object type (C++ users would call it an
10369: @emph{abstract class}); e.g., there is no method for the selector
10370: @code{draw} in the class @code{graphical}.
10371:
10372: For concrete graphical objects, we define child classes of the
10373: class @code{graphical}, e.g.:
10374:
10375: @cindex @code{overrides} usage
10376: @cindex @code{field} usage in class definition
10377: @example
10378: graphical class \ "graphical" is the parent class
10379: cell% field circle-radius
10380:
10381: :noname ( x y circle -- )
10382: circle-radius @@ draw-circle ;
10383: overrides draw
10384:
10385: :noname ( n-radius circle -- )
10386: circle-radius ! ;
10387: overrides construct
10388:
10389: end-class circle
10390: @end example
10391:
10392: Here we define a class @code{circle} as a child of @code{graphical},
10393: with field @code{circle-radius} (which behaves just like a field
10394: (@pxref{Structures}); it defines (using @code{overrides}) new methods
10395: for the selectors @code{draw} and @code{construct} (@code{construct} is
10396: defined in @code{object}, the parent class of @code{graphical}).
10397:
10398: Now we can create a circle on the heap (i.e.,
10399: @code{allocate}d memory) with:
10400:
10401: @cindex @code{heap-new} usage
10402: @example
10403: 50 circle heap-new constant my-circle
10404: @end example
10405:
10406: @noindent
10407: @code{heap-new} invokes @code{construct}, thus
10408: initializing the field @code{circle-radius} with 50. We can draw
10409: this new circle at (100,100) with:
10410:
10411: @example
10412: 100 100 my-circle draw
10413: @end example
10414:
10415: @cindex selector invocation, restrictions
10416: @cindex class definition, restrictions
10417: Note: You can only invoke a selector if the object on the TOS
10418: (the receiving object) belongs to the class where the selector was
10419: defined or one of its descendents; e.g., you can invoke
10420: @code{draw} only for objects belonging to @code{graphical}
10421: or its descendents (e.g., @code{circle}). Immediately before
10422: @code{end-class}, the search order has to be the same as
10423: immediately after @code{class}.
10424:
10425: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
10426: @subsubsection The @file{object.fs} base class
10427: @cindex @code{object} class
10428:
10429: When you define a class, you have to specify a parent class. So how do
10430: you start defining classes? There is one class available from the start:
10431: @code{object}. It is ancestor for all classes and so is the
10432: only class that has no parent. It has two selectors: @code{construct}
10433: and @code{print}.
10434:
10435: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
10436: @subsubsection Creating objects
10437: @cindex creating objects
10438: @cindex object creation
10439: @cindex object allocation options
10440:
10441: @cindex @code{heap-new} discussion
10442: @cindex @code{dict-new} discussion
10443: @cindex @code{construct} discussion
10444: You can create and initialize an object of a class on the heap with
10445: @code{heap-new} ( ... class -- object ) and in the dictionary
10446: (allocation with @code{allot}) with @code{dict-new} (
10447: ... class -- object ). Both words invoke @code{construct}, which
10448: consumes the stack items indicated by "..." above.
10449:
10450: @cindex @code{init-object} discussion
10451: @cindex @code{class-inst-size} discussion
10452: If you want to allocate memory for an object yourself, you can get its
10453: alignment and size with @code{class-inst-size 2@@} ( class --
10454: align size ). Once you have memory for an object, you can initialize
10455: it with @code{init-object} ( ... class object -- );
10456: @code{construct} does only a part of the necessary work.
10457:
10458: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
10459: @subsubsection Object-Oriented Programming Style
10460: @cindex object-oriented programming style
10461: @cindex programming style, object-oriented
10462:
10463: This section is not exhaustive.
10464:
10465: @cindex stack effects of selectors
10466: @cindex selectors and stack effects
10467: In general, it is a good idea to ensure that all methods for the
10468: same selector have the same stack effect: when you invoke a selector,
10469: you often have no idea which method will be invoked, so, unless all
10470: methods have the same stack effect, you will not know the stack effect
10471: of the selector invocation.
10472:
10473: One exception to this rule is methods for the selector
10474: @code{construct}. We know which method is invoked, because we
10475: specify the class to be constructed at the same place. Actually, I
10476: defined @code{construct} as a selector only to give the users a
10477: convenient way to specify initialization. The way it is used, a
10478: mechanism different from selector invocation would be more natural
10479: (but probably would take more code and more space to explain).
10480:
10481: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
10482: @subsubsection Class Binding
10483: @cindex class binding
10484: @cindex early binding
10485:
10486: @cindex late binding
10487: Normal selector invocations determine the method at run-time depending
10488: on the class of the receiving object. This run-time selection is called
10489: @i{late binding}.
10490:
10491: Sometimes it's preferable to invoke a different method. For example,
10492: you might want to use the simple method for @code{print}ing
10493: @code{object}s instead of the possibly long-winded @code{print} method
10494: of the receiver class. You can achieve this by replacing the invocation
10495: of @code{print} with:
10496:
10497: @cindex @code{[bind]} usage
10498: @example
10499: [bind] object print
10500: @end example
10501:
10502: @noindent
10503: in compiled code or:
10504:
10505: @cindex @code{bind} usage
10506: @example
10507: bind object print
10508: @end example
10509:
10510: @cindex class binding, alternative to
10511: @noindent
10512: in interpreted code. Alternatively, you can define the method with a
10513: name (e.g., @code{print-object}), and then invoke it through the
10514: name. Class binding is just a (often more convenient) way to achieve
10515: the same effect; it avoids name clutter and allows you to invoke
10516: methods directly without naming them first.
10517:
10518: @cindex superclass binding
10519: @cindex parent class binding
10520: A frequent use of class binding is this: When we define a method
10521: for a selector, we often want the method to do what the selector does
10522: in the parent class, and a little more. There is a special word for
10523: this purpose: @code{[parent]}; @code{[parent]
10524: @emph{selector}} is equivalent to @code{[bind] @emph{parent
10525: selector}}, where @code{@emph{parent}} is the parent
10526: class of the current class. E.g., a method definition might look like:
10527:
10528: @cindex @code{[parent]} usage
10529: @example
10530: :noname
10531: dup [parent] foo \ do parent's foo on the receiving object
10532: ... \ do some more
10533: ; overrides foo
10534: @end example
10535:
10536: @cindex class binding as optimization
10537: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
10538: March 1997), Andrew McKewan presents class binding as an optimization
10539: technique. I recommend not using it for this purpose unless you are in
10540: an emergency. Late binding is pretty fast with this model anyway, so the
10541: benefit of using class binding is small; the cost of using class binding
10542: where it is not appropriate is reduced maintainability.
10543:
10544: While we are at programming style questions: You should bind
10545: selectors only to ancestor classes of the receiving object. E.g., say,
10546: you know that the receiving object is of class @code{foo} or its
10547: descendents; then you should bind only to @code{foo} and its
10548: ancestors.
10549:
10550: @node Method conveniences, Classes and Scoping, Class Binding, Objects
10551: @subsubsection Method conveniences
10552: @cindex method conveniences
10553:
10554: In a method you usually access the receiving object pretty often. If
10555: you define the method as a plain colon definition (e.g., with
10556: @code{:noname}), you may have to do a lot of stack
10557: gymnastics. To avoid this, you can define the method with @code{m:
10558: ... ;m}. E.g., you could define the method for
10559: @code{draw}ing a @code{circle} with
10560:
10561: @cindex @code{this} usage
10562: @cindex @code{m:} usage
10563: @cindex @code{;m} usage
10564: @example
10565: m: ( x y circle -- )
10566: ( x y ) this circle-radius @@ draw-circle ;m
10567: @end example
10568:
10569: @cindex @code{exit} in @code{m: ... ;m}
10570: @cindex @code{exitm} discussion
10571: @cindex @code{catch} in @code{m: ... ;m}
10572: When this method is executed, the receiver object is removed from the
10573: stack; you can access it with @code{this} (admittedly, in this
10574: example the use of @code{m: ... ;m} offers no advantage). Note
10575: that I specify the stack effect for the whole method (i.e. including
10576: the receiver object), not just for the code between @code{m:}
10577: and @code{;m}. You cannot use @code{exit} in
10578: @code{m:...;m}; instead, use
10579: @code{exitm}.@footnote{Moreover, for any word that calls
10580: @code{catch} and was defined before loading
10581: @code{objects.fs}, you have to redefine it like I redefined
10582: @code{catch}: @code{: catch this >r catch r> to-this ;}}
10583:
10584: @cindex @code{inst-var} usage
10585: You will frequently use sequences of the form @code{this
10586: @emph{field}} (in the example above: @code{this
10587: circle-radius}). If you use the field only in this way, you can
10588: define it with @code{inst-var} and eliminate the
10589: @code{this} before the field name. E.g., the @code{circle}
10590: class above could also be defined with:
10591:
10592: @example
10593: graphical class
10594: cell% inst-var radius
10595:
10596: m: ( x y circle -- )
10597: radius @@ draw-circle ;m
10598: overrides draw
10599:
10600: m: ( n-radius circle -- )
10601: radius ! ;m
10602: overrides construct
10603:
10604: end-class circle
10605: @end example
10606:
10607: @code{radius} can only be used in @code{circle} and its
10608: descendent classes and inside @code{m:...;m}.
10609:
10610: @cindex @code{inst-value} usage
10611: You can also define fields with @code{inst-value}, which is
10612: to @code{inst-var} what @code{value} is to
10613: @code{variable}. You can change the value of such a field with
10614: @code{[to-inst]}. E.g., we could also define the class
10615: @code{circle} like this:
10616:
10617: @example
10618: graphical class
10619: inst-value radius
10620:
10621: m: ( x y circle -- )
10622: radius draw-circle ;m
10623: overrides draw
10624:
10625: m: ( n-radius circle -- )
10626: [to-inst] radius ;m
10627: overrides construct
10628:
10629: end-class circle
10630: @end example
10631:
10632: Finally, you can define named methods with @code{:m}. One use of this
10633: feature is the definition of words that occur only in one class and are
10634: not intended to be overridden, but which still need method context
10635: (e.g., for accessing @code{inst-var}s). Another use is for methods that
10636: would be bound frequently, if defined anonymously.
10637:
10638:
10639: @node Classes and Scoping, Dividing classes, Method conveniences, Objects
10640: @subsubsection Classes and Scoping
10641: @cindex classes and scoping
10642: @cindex scoping and classes
10643:
10644: Inheritance is frequent, unlike structure extension. This exacerbates
10645: the problem with the field name convention (@pxref{Structure Naming
10646: Convention}): One always has to remember in which class the field was
10647: originally defined; changing a part of the class structure would require
10648: changes for renaming in otherwise unaffected code.
10649:
10650: @cindex @code{inst-var} visibility
10651: @cindex @code{inst-value} visibility
10652: To solve this problem, I added a scoping mechanism (which was not in my
10653: original charter): A field defined with @code{inst-var} (or
10654: @code{inst-value}) is visible only in the class where it is defined and in
10655: the descendent classes of this class. Using such fields only makes
10656: sense in @code{m:}-defined methods in these classes anyway.
10657:
10658: This scoping mechanism allows us to use the unadorned field name,
10659: because name clashes with unrelated words become much less likely.
10660:
10661: @cindex @code{protected} discussion
10662: @cindex @code{private} discussion
10663: Once we have this mechanism, we can also use it for controlling the
10664: visibility of other words: All words defined after
10665: @code{protected} are visible only in the current class and its
10666: descendents. @code{public} restores the compilation
10667: (i.e. @code{current}) word list that was in effect before. If you
10668: have several @code{protected}s without an intervening
10669: @code{public} or @code{set-current}, @code{public}
10670: will restore the compilation word list in effect before the first of
10671: these @code{protected}s.
10672:
10673: @node Dividing classes, Object Interfaces, Classes and Scoping, Objects
10674: @subsubsection Dividing classes
10675: @cindex Dividing classes
10676: @cindex @code{methods}...@code{end-methods}
10677:
10678: You may want to do the definition of methods separate from the
10679: definition of the class, its selectors, fields, and instance variables,
10680: i.e., separate the implementation from the definition. You can do this
10681: in the following way:
10682:
10683: @example
10684: graphical class
10685: inst-value radius
10686: end-class circle
10687:
10688: ... \ do some other stuff
10689:
10690: circle methods \ now we are ready
10691:
10692: m: ( x y circle -- )
10693: radius draw-circle ;m
10694: overrides draw
10695:
10696: m: ( n-radius circle -- )
10697: [to-inst] radius ;m
10698: overrides construct
10699:
10700: end-methods
10701: @end example
10702:
10703: You can use several @code{methods}...@code{end-methods} sections. The
10704: only things you can do to the class in these sections are: defining
10705: methods, and overriding the class's selectors. You must not define new
10706: selectors or fields.
10707:
10708: Note that you often have to override a selector before using it. In
10709: particular, you usually have to override @code{construct} with a new
10710: method before you can invoke @code{heap-new} and friends. E.g., you
10711: must not create a circle before the @code{overrides construct} sequence
10712: in the example above.
10713:
10714: @node Object Interfaces, Objects Implementation, Dividing classes, Objects
10715: @subsubsection Object Interfaces
10716: @cindex object interfaces
10717: @cindex interfaces for objects
10718:
10719: In this model you can only call selectors defined in the class of the
10720: receiving objects or in one of its ancestors. If you call a selector
10721: with a receiving object that is not in one of these classes, the
10722: result is undefined; if you are lucky, the program crashes
10723: immediately.
10724:
10725: @cindex selectors common to hardly-related classes
10726: Now consider the case when you want to have a selector (or several)
10727: available in two classes: You would have to add the selector to a
10728: common ancestor class, in the worst case to @code{object}. You
10729: may not want to do this, e.g., because someone else is responsible for
10730: this ancestor class.
10731:
10732: The solution for this problem is interfaces. An interface is a
10733: collection of selectors. If a class implements an interface, the
10734: selectors become available to the class and its descendents. A class
10735: can implement an unlimited number of interfaces. For the problem
10736: discussed above, we would define an interface for the selector(s), and
10737: both classes would implement the interface.
10738:
10739: As an example, consider an interface @code{storage} for
10740: writing objects to disk and getting them back, and a class
10741: @code{foo} that implements it. The code would look like this:
10742:
10743: @cindex @code{interface} usage
10744: @cindex @code{end-interface} usage
10745: @cindex @code{implementation} usage
10746: @example
10747: interface
10748: selector write ( file object -- )
10749: selector read1 ( file object -- )
10750: end-interface storage
10751:
10752: bar class
10753: storage implementation
10754:
10755: ... overrides write
10756: ... overrides read1
10757: ...
10758: end-class foo
10759: @end example
10760:
10761: @noindent
10762: (I would add a word @code{read} @i{( file -- object )} that uses
10763: @code{read1} internally, but that's beyond the point illustrated
10764: here.)
10765:
10766: Note that you cannot use @code{protected} in an interface; and
10767: of course you cannot define fields.
10768:
10769: In the Neon model, all selectors are available for all classes;
10770: therefore it does not need interfaces. The price you pay in this model
10771: is slower late binding, and therefore, added complexity to avoid late
10772: binding.
10773:
10774: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
10775: @subsubsection @file{objects.fs} Implementation
10776: @cindex @file{objects.fs} implementation
10777:
10778: @cindex @code{object-map} discussion
10779: An object is a piece of memory, like one of the data structures
10780: described with @code{struct...end-struct}. It has a field
10781: @code{object-map} that points to the method map for the object's
10782: class.
10783:
10784: @cindex method map
10785: @cindex virtual function table
10786: The @emph{method map}@footnote{This is Self terminology; in C++
10787: terminology: virtual function table.} is an array that contains the
10788: execution tokens (@i{xt}s) of the methods for the object's class. Each
10789: selector contains an offset into a method map.
10790:
10791: @cindex @code{selector} implementation, class
10792: @code{selector} is a defining word that uses
10793: @code{CREATE} and @code{DOES>}. The body of the
10794: selector contains the offset; the @code{DOES>} action for a
10795: class selector is, basically:
10796:
10797: @example
10798: ( object addr ) @@ over object-map @@ + @@ execute
10799: @end example
10800:
10801: Since @code{object-map} is the first field of the object, it
10802: does not generate any code. As you can see, calling a selector has a
10803: small, constant cost.
10804:
10805: @cindex @code{current-interface} discussion
10806: @cindex class implementation and representation
10807: A class is basically a @code{struct} combined with a method
10808: map. During the class definition the alignment and size of the class
10809: are passed on the stack, just as with @code{struct}s, so
10810: @code{field} can also be used for defining class
10811: fields. However, passing more items on the stack would be
10812: inconvenient, so @code{class} builds a data structure in memory,
10813: which is accessed through the variable
10814: @code{current-interface}. After its definition is complete, the
10815: class is represented on the stack by a pointer (e.g., as parameter for
10816: a child class definition).
10817:
10818: A new class starts off with the alignment and size of its parent,
10819: and a copy of the parent's method map. Defining new fields extends the
10820: size and alignment; likewise, defining new selectors extends the
10821: method map. @code{overrides} just stores a new @i{xt} in the method
10822: map at the offset given by the selector.
10823:
10824: @cindex class binding, implementation
10825: Class binding just gets the @i{xt} at the offset given by the selector
10826: from the class's method map and @code{compile,}s (in the case of
10827: @code{[bind]}) it.
10828:
10829: @cindex @code{this} implementation
10830: @cindex @code{catch} and @code{this}
10831: @cindex @code{this} and @code{catch}
10832: I implemented @code{this} as a @code{value}. At the
10833: start of an @code{m:...;m} method the old @code{this} is
10834: stored to the return stack and restored at the end; and the object on
10835: the TOS is stored @code{TO this}. This technique has one
10836: disadvantage: If the user does not leave the method via
10837: @code{;m}, but via @code{throw} or @code{exit},
10838: @code{this} is not restored (and @code{exit} may
10839: crash). To deal with the @code{throw} problem, I have redefined
10840: @code{catch} to save and restore @code{this}; the same
10841: should be done with any word that can catch an exception. As for
10842: @code{exit}, I simply forbid it (as a replacement, there is
10843: @code{exitm}).
10844:
10845: @cindex @code{inst-var} implementation
10846: @code{inst-var} is just the same as @code{field}, with
10847: a different @code{DOES>} action:
10848: @example
10849: @@ this +
10850: @end example
10851: Similar for @code{inst-value}.
10852:
10853: @cindex class scoping implementation
10854: Each class also has a word list that contains the words defined with
10855: @code{inst-var} and @code{inst-value}, and its protected
10856: words. It also has a pointer to its parent. @code{class} pushes
10857: the word lists of the class and all its ancestors onto the search order stack,
10858: and @code{end-class} drops them.
10859:
10860: @cindex interface implementation
10861: An interface is like a class without fields, parent and protected
10862: words; i.e., it just has a method map. If a class implements an
10863: interface, its method map contains a pointer to the method map of the
10864: interface. The positive offsets in the map are reserved for class
10865: methods, therefore interface map pointers have negative
10866: offsets. Interfaces have offsets that are unique throughout the
10867: system, unlike class selectors, whose offsets are only unique for the
10868: classes where the selector is available (invokable).
10869:
10870: This structure means that interface selectors have to perform one
10871: indirection more than class selectors to find their method. Their body
10872: contains the interface map pointer offset in the class method map, and
10873: the method offset in the interface method map. The
10874: @code{does>} action for an interface selector is, basically:
10875:
10876: @example
10877: ( object selector-body )
10878: 2dup selector-interface @@ ( object selector-body object interface-offset )
10879: swap object-map @@ + @@ ( object selector-body map )
10880: swap selector-offset @@ + @@ execute
10881: @end example
10882:
10883: where @code{object-map} and @code{selector-offset} are
10884: first fields and generate no code.
10885:
10886: As a concrete example, consider the following code:
10887:
10888: @example
10889: interface
10890: selector if1sel1
10891: selector if1sel2
10892: end-interface if1
10893:
10894: object class
10895: if1 implementation
10896: selector cl1sel1
10897: cell% inst-var cl1iv1
10898:
10899: ' m1 overrides construct
10900: ' m2 overrides if1sel1
10901: ' m3 overrides if1sel2
10902: ' m4 overrides cl1sel2
10903: end-class cl1
10904:
10905: create obj1 object dict-new drop
10906: create obj2 cl1 dict-new drop
10907: @end example
10908:
10909: The data structure created by this code (including the data structure
10910: for @code{object}) is shown in the <a
10911: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
10912: @comment TODO add this diagram..
10913:
10914: @node Objects Glossary, , Objects Implementation, Objects
10915: @subsubsection @file{objects.fs} Glossary
10916: @cindex @file{objects.fs} Glossary
10917:
10918:
10919: doc---objects-bind
10920: doc---objects-<bind>
10921: doc---objects-bind'
10922: doc---objects-[bind]
10923: doc---objects-class
10924: doc---objects-class->map
10925: doc---objects-class-inst-size
10926: doc---objects-class-override!
10927: doc---objects-construct
10928: doc---objects-current'
10929: doc---objects-[current]
10930: doc---objects-current-interface
10931: doc---objects-dict-new
10932: doc---objects-drop-order
10933: doc---objects-end-class
10934: doc---objects-end-class-noname
10935: doc---objects-end-interface
10936: doc---objects-end-interface-noname
10937: doc---objects-end-methods
10938: doc---objects-exitm
10939: doc---objects-heap-new
10940: doc---objects-implementation
10941: doc---objects-init-object
10942: doc---objects-inst-value
10943: doc---objects-inst-var
10944: doc---objects-interface
10945: doc---objects-m:
10946: doc---objects-:m
10947: doc---objects-;m
10948: doc---objects-method
10949: doc---objects-methods
10950: doc---objects-object
10951: doc---objects-overrides
10952: doc---objects-[parent]
10953: doc---objects-print
10954: doc---objects-protected
10955: doc---objects-public
10956: doc---objects-push-order
10957: doc---objects-selector
10958: doc---objects-this
10959: doc---objects-<to-inst>
10960: doc---objects-[to-inst]
10961: doc---objects-to-this
10962: doc---objects-xt-new
10963:
10964:
10965: @c -------------------------------------------------------------
10966: @node OOF, Mini-OOF, Objects, Object-oriented Forth
10967: @subsection The @file{oof.fs} model
10968: @cindex oof
10969: @cindex object-oriented programming
10970:
10971: @cindex @file{objects.fs}
10972: @cindex @file{oof.fs}
10973:
10974: This section describes the @file{oof.fs} package.
10975:
10976: The package described in this section has been used in bigFORTH since 1991, and
10977: used for two large applications: a chromatographic system used to
10978: create new medicaments, and a graphic user interface library (MINOS).
10979:
10980: You can find a description (in German) of @file{oof.fs} in @cite{Object
10981: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
10982: 10(2), 1994.
10983:
10984: @menu
10985: * Properties of the OOF model::
10986: * Basic OOF Usage::
10987: * The OOF base class::
10988: * Class Declaration::
10989: * Class Implementation::
10990: @end menu
10991:
10992: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
10993: @subsubsection Properties of the @file{oof.fs} model
10994: @cindex @file{oof.fs} properties
10995:
10996: @itemize @bullet
10997: @item
10998: This model combines object oriented programming with information
10999: hiding. It helps you writing large application, where scoping is
11000: necessary, because it provides class-oriented scoping.
11001:
11002: @item
11003: Named objects, object pointers, and object arrays can be created,
11004: selector invocation uses the ``object selector'' syntax. Selector invocation
11005: to objects and/or selectors on the stack is a bit less convenient, but
11006: possible.
11007:
11008: @item
11009: Selector invocation and instance variable usage of the active object is
11010: straightforward, since both make use of the active object.
11011:
11012: @item
11013: Late binding is efficient and easy to use.
11014:
11015: @item
11016: State-smart objects parse selectors. However, extensibility is provided
11017: using a (parsing) selector @code{postpone} and a selector @code{'}.
11018:
11019: @item
11020: An implementation in ANS Forth is available.
11021:
11022: @end itemize
11023:
11024:
11025: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
11026: @subsubsection Basic @file{oof.fs} Usage
11027: @cindex @file{oof.fs} usage
11028:
11029: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
11030:
11031: You can define a class for graphical objects like this:
11032:
11033: @cindex @code{class} usage
11034: @cindex @code{class;} usage
11035: @cindex @code{method} usage
11036: @example
11037: object class graphical \ "object" is the parent class
11038: method draw ( x y graphical -- )
11039: class;
11040: @end example
11041:
11042: This code defines a class @code{graphical} with an
11043: operation @code{draw}. We can perform the operation
11044: @code{draw} on any @code{graphical} object, e.g.:
11045:
11046: @example
11047: 100 100 t-rex draw
11048: @end example
11049:
11050: @noindent
11051: where @code{t-rex} is an object or object pointer, created with e.g.
11052: @code{graphical : t-rex}.
11053:
11054: @cindex abstract class
11055: How do we create a graphical object? With the present definitions,
11056: we cannot create a useful graphical object. The class
11057: @code{graphical} describes graphical objects in general, but not
11058: any concrete graphical object type (C++ users would call it an
11059: @emph{abstract class}); e.g., there is no method for the selector
11060: @code{draw} in the class @code{graphical}.
11061:
11062: For concrete graphical objects, we define child classes of the
11063: class @code{graphical}, e.g.:
11064:
11065: @example
11066: graphical class circle \ "graphical" is the parent class
11067: cell var circle-radius
11068: how:
11069: : draw ( x y -- )
11070: circle-radius @@ draw-circle ;
11071:
11072: : init ( n-radius -- (
11073: circle-radius ! ;
11074: class;
11075: @end example
11076:
11077: Here we define a class @code{circle} as a child of @code{graphical},
11078: with a field @code{circle-radius}; it defines new methods for the
11079: selectors @code{draw} and @code{init} (@code{init} is defined in
11080: @code{object}, the parent class of @code{graphical}).
11081:
11082: Now we can create a circle in the dictionary with:
11083:
11084: @example
11085: 50 circle : my-circle
11086: @end example
11087:
11088: @noindent
11089: @code{:} invokes @code{init}, thus initializing the field
11090: @code{circle-radius} with 50. We can draw this new circle at (100,100)
11091: with:
11092:
11093: @example
11094: 100 100 my-circle draw
11095: @end example
11096:
11097: @cindex selector invocation, restrictions
11098: @cindex class definition, restrictions
11099: Note: You can only invoke a selector if the receiving object belongs to
11100: the class where the selector was defined or one of its descendents;
11101: e.g., you can invoke @code{draw} only for objects belonging to
11102: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
11103: mechanism will check if you try to invoke a selector that is not
11104: defined in this class hierarchy, so you'll get an error at compilation
11105: time.
11106:
11107:
11108: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
11109: @subsubsection The @file{oof.fs} base class
11110: @cindex @file{oof.fs} base class
11111:
11112: When you define a class, you have to specify a parent class. So how do
11113: you start defining classes? There is one class available from the start:
11114: @code{object}. You have to use it as ancestor for all classes. It is the
11115: only class that has no parent. Classes are also objects, except that
11116: they don't have instance variables; class manipulation such as
11117: inheritance or changing definitions of a class is handled through
11118: selectors of the class @code{object}.
11119:
11120: @code{object} provides a number of selectors:
11121:
11122: @itemize @bullet
11123: @item
11124: @code{class} for subclassing, @code{definitions} to add definitions
11125: later on, and @code{class?} to get type informations (is the class a
11126: subclass of the class passed on the stack?).
11127:
11128: doc---object-class
11129: doc---object-definitions
11130: doc---object-class?
11131:
11132:
11133: @item
11134: @code{init} and @code{dispose} as constructor and destructor of the
11135: object. @code{init} is invocated after the object's memory is allocated,
11136: while @code{dispose} also handles deallocation. Thus if you redefine
11137: @code{dispose}, you have to call the parent's dispose with @code{super
11138: dispose}, too.
11139:
11140: doc---object-init
11141: doc---object-dispose
11142:
11143:
11144: @item
11145: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
11146: @code{[]} to create named and unnamed objects and object arrays or
11147: object pointers.
11148:
11149: doc---object-new
11150: doc---object-new[]
11151: doc---object-:
11152: doc---object-ptr
11153: doc---object-asptr
11154: doc---object-[]
11155:
11156:
11157: @item
11158: @code{::} and @code{super} for explicit scoping. You should use explicit
11159: scoping only for super classes or classes with the same set of instance
11160: variables. Explicitly-scoped selectors use early binding.
11161:
11162: doc---object-::
11163: doc---object-super
11164:
11165:
11166: @item
11167: @code{self} to get the address of the object
11168:
11169: doc---object-self
11170:
11171:
11172: @item
11173: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
11174: pointers and instance defers.
11175:
11176: doc---object-bind
11177: doc---object-bound
11178: doc---object-link
11179: doc---object-is
11180:
11181:
11182: @item
11183: @code{'} to obtain selector tokens, @code{send} to invocate selectors
11184: form the stack, and @code{postpone} to generate selector invocation code.
11185:
11186: doc---object-'
11187: doc---object-postpone
11188:
11189:
11190: @item
11191: @code{with} and @code{endwith} to select the active object from the
11192: stack, and enable its scope. Using @code{with} and @code{endwith}
11193: also allows you to create code using selector @code{postpone} without being
11194: trapped by the state-smart objects.
11195:
11196: doc---object-with
11197: doc---object-endwith
11198:
11199:
11200: @end itemize
11201:
11202: @node Class Declaration, Class Implementation, The OOF base class, OOF
11203: @subsubsection Class Declaration
11204: @cindex class declaration
11205:
11206: @itemize @bullet
11207: @item
11208: Instance variables
11209:
11210: doc---oof-var
11211:
11212:
11213: @item
11214: Object pointers
11215:
11216: doc---oof-ptr
11217: doc---oof-asptr
11218:
11219:
11220: @item
11221: Instance defers
11222:
11223: doc---oof-defer
11224:
11225:
11226: @item
11227: Method selectors
11228:
11229: doc---oof-early
11230: doc---oof-method
11231:
11232:
11233: @item
11234: Class-wide variables
11235:
11236: doc---oof-static
11237:
11238:
11239: @item
11240: End declaration
11241:
11242: doc---oof-how:
11243: doc---oof-class;
11244:
11245:
11246: @end itemize
11247:
11248: @c -------------------------------------------------------------
11249: @node Class Implementation, , Class Declaration, OOF
11250: @subsubsection Class Implementation
11251: @cindex class implementation
11252:
11253: @c -------------------------------------------------------------
11254: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
11255: @subsection The @file{mini-oof.fs} model
11256: @cindex mini-oof
11257:
11258: Gforth's third object oriented Forth package is a 12-liner. It uses a
11259: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
11260: and reduces to the bare minimum of features. This is based on a posting
11261: of Bernd Paysan in comp.lang.forth.
11262:
11263: @menu
11264: * Basic Mini-OOF Usage::
11265: * Mini-OOF Example::
11266: * Mini-OOF Implementation::
11267: @end menu
11268:
11269: @c -------------------------------------------------------------
11270: @node Basic Mini-OOF Usage, Mini-OOF Example, Mini-OOF, Mini-OOF
11271: @subsubsection Basic @file{mini-oof.fs} Usage
11272: @cindex mini-oof usage
11273:
11274: There is a base class (@code{class}, which allocates one cell for the
11275: object pointer) plus seven other words: to define a method, a variable,
11276: a class; to end a class, to resolve binding, to allocate an object and
11277: to compile a class method.
11278: @comment TODO better description of the last one
11279:
11280:
11281: doc-object
11282: doc-method
11283: doc-var
11284: doc-class
11285: doc-end-class
11286: doc-defines
11287: doc-new
11288: doc-::
11289:
11290:
11291:
11292: @c -------------------------------------------------------------
11293: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
11294: @subsubsection Mini-OOF Example
11295: @cindex mini-oof example
11296:
11297: A short example shows how to use this package. This example, in slightly
11298: extended form, is supplied as @file{moof-exm.fs}
11299: @comment TODO could flesh this out with some comments from the Forthwrite article
11300:
11301: @example
11302: object class
11303: method init
11304: method draw
11305: end-class graphical
11306: @end example
11307:
11308: This code defines a class @code{graphical} with an
11309: operation @code{draw}. We can perform the operation
11310: @code{draw} on any @code{graphical} object, e.g.:
11311:
11312: @example
11313: 100 100 t-rex draw
11314: @end example
11315:
11316: where @code{t-rex} is an object or object pointer, created with e.g.
11317: @code{graphical new Constant t-rex}.
11318:
11319: For concrete graphical objects, we define child classes of the
11320: class @code{graphical}, e.g.:
11321:
11322: @example
11323: graphical class
11324: cell var circle-radius
11325: end-class circle \ "graphical" is the parent class
11326:
11327: :noname ( x y -- )
11328: circle-radius @@ draw-circle ; circle defines draw
11329: :noname ( r -- )
11330: circle-radius ! ; circle defines init
11331: @end example
11332:
11333: There is no implicit init method, so we have to define one. The creation
11334: code of the object now has to call init explicitely.
11335:
11336: @example
11337: circle new Constant my-circle
11338: 50 my-circle init
11339: @end example
11340:
11341: It is also possible to add a function to create named objects with
11342: automatic call of @code{init}, given that all objects have @code{init}
11343: on the same place:
11344:
11345: @example
11346: : new: ( .. o "name" -- )
11347: new dup Constant init ;
11348: 80 circle new: large-circle
11349: @end example
11350:
11351: We can draw this new circle at (100,100) with:
11352:
11353: @example
11354: 100 100 my-circle draw
11355: @end example
11356:
11357: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
11358: @subsubsection @file{mini-oof.fs} Implementation
11359:
11360: Object-oriented systems with late binding typically use a
11361: ``vtable''-approach: the first variable in each object is a pointer to a
11362: table, which contains the methods as function pointers. The vtable
11363: may also contain other information.
11364:
11365: So first, let's declare methods:
11366:
11367: @example
11368: : method ( m v -- m' v ) Create over , swap cell+ swap
11369: DOES> ( ... o -- ... ) @ over @ + @ execute ;
11370: @end example
11371:
11372: During method declaration, the number of methods and instance
11373: variables is on the stack (in address units). @code{method} creates
11374: one method and increments the method number. To execute a method, it
11375: takes the object, fetches the vtable pointer, adds the offset, and
11376: executes the @i{xt} stored there. Each method takes the object it is
11377: invoked from as top of stack parameter. The method itself should
11378: consume that object.
11379:
11380: Now, we also have to declare instance variables
11381:
11382: @example
11383: : var ( m v size -- m v' ) Create over , +
11384: DOES> ( o -- addr ) @ + ;
11385: @end example
11386:
11387: As before, a word is created with the current offset. Instance
11388: variables can have different sizes (cells, floats, doubles, chars), so
11389: all we do is take the size and add it to the offset. If your machine
11390: has alignment restrictions, put the proper @code{aligned} or
11391: @code{faligned} before the variable, to adjust the variable
11392: offset. That's why it is on the top of stack.
11393:
11394: We need a starting point (the base object) and some syntactic sugar:
11395:
11396: @example
11397: Create object 1 cells , 2 cells ,
11398: : class ( class -- class methods vars ) dup 2@ ;
11399: @end example
11400:
11401: For inheritance, the vtable of the parent object has to be
11402: copied when a new, derived class is declared. This gives all the
11403: methods of the parent class, which can be overridden, though.
11404:
11405: @example
11406: : end-class ( class methods vars -- )
11407: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
11408: cell+ dup cell+ r> rot @ 2 cells /string move ;
11409: @end example
11410:
11411: The first line creates the vtable, initialized with
11412: @code{noop}s. The second line is the inheritance mechanism, it
11413: copies the xts from the parent vtable.
11414:
11415: We still have no way to define new methods, let's do that now:
11416:
11417: @example
11418: : defines ( xt class -- ) ' >body @ + ! ;
11419: @end example
11420:
11421: To allocate a new object, we need a word, too:
11422:
11423: @example
11424: : new ( class -- o ) here over @ allot swap over ! ;
11425: @end example
11426:
11427: Sometimes derived classes want to access the method of the
11428: parent object. There are two ways to achieve this with Mini-OOF:
11429: first, you could use named words, and second, you could look up the
11430: vtable of the parent object.
11431:
11432: @example
11433: : :: ( class "name" -- ) ' >body @ + @ compile, ;
11434: @end example
11435:
11436:
11437: Nothing can be more confusing than a good example, so here is
11438: one. First let's declare a text object (called
11439: @code{button}), that stores text and position:
11440:
11441: @example
11442: object class
11443: cell var text
11444: cell var len
11445: cell var x
11446: cell var y
11447: method init
11448: method draw
11449: end-class button
11450: @end example
11451:
11452: @noindent
11453: Now, implement the two methods, @code{draw} and @code{init}:
11454:
11455: @example
11456: :noname ( o -- )
11457: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
11458: button defines draw
11459: :noname ( addr u o -- )
11460: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
11461: button defines init
11462: @end example
11463:
11464: @noindent
11465: To demonstrate inheritance, we define a class @code{bold-button}, with no
11466: new data and no new methods:
11467:
11468: @example
11469: button class
11470: end-class bold-button
11471:
11472: : bold 27 emit ." [1m" ;
11473: : normal 27 emit ." [0m" ;
11474: @end example
11475:
11476: @noindent
11477: The class @code{bold-button} has a different draw method to
11478: @code{button}, but the new method is defined in terms of the draw method
11479: for @code{button}:
11480:
11481: @example
11482: :noname bold [ button :: draw ] normal ; bold-button defines draw
11483: @end example
11484:
11485: @noindent
11486: Finally, create two objects and apply methods:
11487:
11488: @example
11489: button new Constant foo
11490: s" thin foo" foo init
11491: page
11492: foo draw
11493: bold-button new Constant bar
11494: s" fat bar" bar init
11495: 1 bar y !
11496: bar draw
11497: @end example
11498:
11499:
11500: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
11501: @subsection Comparison with other object models
11502: @cindex comparison of object models
11503: @cindex object models, comparison
11504:
11505: Many object-oriented Forth extensions have been proposed (@cite{A survey
11506: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
11507: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
11508: relation of the object models described here to two well-known and two
11509: closely-related (by the use of method maps) models.
11510:
11511: @cindex Neon model
11512: The most popular model currently seems to be the Neon model (see
11513: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
11514: 1997) by Andrew McKewan) but this model has a number of limitations
11515: @footnote{A longer version of this critique can be
11516: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
11517: Dimensions, May 1997) by Anton Ertl.}:
11518:
11519: @itemize @bullet
11520: @item
11521: It uses a @code{@emph{selector object}} syntax, which makes it unnatural
11522: to pass objects on the stack.
11523:
11524: @item
11525: It requires that the selector parses the input stream (at
11526: compile time); this leads to reduced extensibility and to bugs that are+
11527: hard to find.
11528:
11529: @item
11530: It allows using every selector to every object;
11531: this eliminates the need for classes, but makes it harder to create
11532: efficient implementations.
11533: @end itemize
11534:
11535: @cindex Pountain's object-oriented model
11536: Another well-known publication is @cite{Object-Oriented Forth} (Academic
11537: Press, London, 1987) by Dick Pountain. However, it is not really about
11538: object-oriented programming, because it hardly deals with late
11539: binding. Instead, it focuses on features like information hiding and
11540: overloading that are characteristic of modular languages like Ada (83).
11541:
11542: @cindex Zsoter's object-oriented model
11543: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1)
11544: 1996, pages 31-35) Andras Zsoter describes a model that makes heavy use
11545: of an active object (like @code{this} in @file{objects.fs}): The active
11546: object is not only used for accessing all fields, but also specifies the
11547: receiving object of every selector invocation; you have to change the
11548: active object explicitly with @code{@{ ... @}}, whereas in
11549: @file{objects.fs} it changes more or less implicitly at @code{m:
11550: ... ;m}. Such a change at the method entry point is unnecessary with the
11551: Zsoter's model, because the receiving object is the active object
11552: already. On the other hand, the explicit change is absolutely necessary
11553: in that model, because otherwise no one could ever change the active
11554: object. An ANS Forth implementation of this model is available at
11555: @uref{http://www.forth.org/fig/oopf.html}.
11556:
11557: @cindex @file{oof.fs}, differences to other models
11558: The @file{oof.fs} model combines information hiding and overloading
11559: resolution (by keeping names in various word lists) with object-oriented
11560: programming. It sets the active object implicitly on method entry, but
11561: also allows explicit changing (with @code{>o...o>} or with
11562: @code{with...endwith}). It uses parsing and state-smart objects and
11563: classes for resolving overloading and for early binding: the object or
11564: class parses the selector and determines the method from this. If the
11565: selector is not parsed by an object or class, it performs a call to the
11566: selector for the active object (late binding), like Zsoter's model.
11567: Fields are always accessed through the active object. The big
11568: disadvantage of this model is the parsing and the state-smartness, which
11569: reduces extensibility and increases the opportunities for subtle bugs;
11570: essentially, you are only safe if you never tick or @code{postpone} an
11571: object or class (Bernd disagrees, but I (Anton) am not convinced).
11572:
11573: @cindex @file{mini-oof.fs}, differences to other models
11574: The @file{mini-oof.fs} model is quite similar to a very stripped-down
11575: version of the @file{objects.fs} model, but syntactically it is a
11576: mixture of the @file{objects.fs} and @file{oof.fs} models.
11577:
11578: @c -------------------------------------------------------------
11579: @node Passing Commands to the OS, Keeping track of Time, Object-oriented Forth, Words
11580: @section Passing Commands to the Operating System
11581: @cindex operating system - passing commands
11582: @cindex shell commands
11583:
11584: Gforth allows you to pass an arbitrary string to the host operating
11585: system shell (if such a thing exists) for execution.
11586:
11587:
11588: doc-sh
11589: doc-system
11590: doc-$?
11591: doc-getenv
11592:
11593:
11594: @c -------------------------------------------------------------
11595: @node Keeping track of Time, Miscellaneous Words, Passing Commands to the OS, Words
11596: @section Keeping track of Time
11597: @cindex time-related words
11598:
11599: Gforth implements time-related operations by making calls to the C
11600: library function, @code{gettimeofday}.
11601:
11602: doc-ms
11603: doc-time&date
11604:
11605:
11606:
11607: @c -------------------------------------------------------------
11608: @node Miscellaneous Words, , Keeping track of Time, Words
11609: @section Miscellaneous Words
11610: @cindex miscellaneous words
11611:
11612: @comment TODO find homes for these
11613:
11614: These section lists the ANS Forth words that are not documented
11615: elsewhere in this manual. Ultimately, they all need proper homes.
11616:
11617: doc-[compile]
11618: doc-quit
11619:
11620: The following ANS Forth words are not currently supported by Gforth
11621: (@pxref{ANS conformance}):
11622:
11623: @code{EDITOR}
11624: @code{EMIT?}
11625: @code{FORGET}
11626:
11627: @c ******************************************************************
11628: @node Error messages, Tools, Words, Top
11629: @chapter Error messages
11630: @cindex error messages
11631: @cindex backtrace
11632:
11633: A typical Gforth error message looks like this:
11634:
11635: @example
11636: in file included from :-1
11637: in file included from ./yyy.fs:1
11638: ./xxx.fs:4: Invalid memory address
11639: bar
11640: ^^^
11641: $400E664C @@
11642: $400E6664 foo
11643: @end example
11644:
11645: The message identifying the error is @code{Invalid memory address}. The
11646: error happened when text-interpreting line 4 of the file
11647: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
11648: word on the line where the error happened, is pointed out (with
11649: @code{^^^}).
11650:
11651: The file containing the error was included in line 1 of @file{./yyy.fs},
11652: and @file{yyy.fs} was included from a non-file (in this case, by giving
11653: @file{yyy.fs} as command-line parameter to Gforth).
11654:
11655: At the end of the error message you find a return stack dump that can be
11656: interpreted as a backtrace (possibly empty). On top you find the top of
11657: the return stack when the @code{throw} happened, and at the bottom you
11658: find the return stack entry just above the return stack of the topmost
11659: text interpreter.
11660:
11661: To the right of most return stack entries you see a guess for the word
11662: that pushed that return stack entry as its return address. This gives a
11663: backtrace. In our case we see that @code{bar} called @code{foo}, and
11664: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
11665: address} exception).
11666:
11667: Note that the backtrace is not perfect: We don't know which return stack
11668: entries are return addresses (so we may get false positives); and in
11669: some cases (e.g., for @code{abort"}) we cannot determine from the return
11670: address the word that pushed the return address, so for some return
11671: addresses you see no names in the return stack dump.
11672:
11673: @cindex @code{catch} and backtraces
11674: The return stack dump represents the return stack at the time when a
11675: specific @code{throw} was executed. In programs that make use of
11676: @code{catch}, it is not necessarily clear which @code{throw} should be
11677: used for the return stack dump (e.g., consider one @code{throw} that
11678: indicates an error, which is caught, and during recovery another error
11679: happens; which @code{throw} should be used for the stack dump?). Gforth
11680: presents the return stack dump for the first @code{throw} after the last
11681: executed (not returned-to) @code{catch}; this works well in the usual
11682: case.
11683:
11684: @cindex @code{gforth-fast} and backtraces
11685: @cindex @code{gforth-fast}, difference from @code{gforth}
11686: @cindex backtraces with @code{gforth-fast}
11687: @cindex return stack dump with @code{gforth-fast}
11688: @code{gforth} is able to do a return stack dump for throws generated
11689: from primitives (e.g., invalid memory address, stack empty etc.);
11690: @code{gforth-fast} is only able to do a return stack dump from a
11691: directly called @code{throw} (including @code{abort} etc.). This is the
11692: only difference (apart from a speed factor of between 1.15 (K6-2) and
11693: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
11694: exception caused by a primitive in @code{gforth-fast}, you will
11695: typically see no return stack dump at all; however, if the exception is
11696: caught by @code{catch} (e.g., for restoring some state), and then
11697: @code{throw}n again, the return stack dump will be for the first such
11698: @code{throw}.
11699:
11700: @c ******************************************************************
11701: @node Tools, ANS conformance, Error messages, Top
11702: @chapter Tools
11703:
11704: @menu
11705: * ANS Report:: Report the words used, sorted by wordset.
11706: @end menu
11707:
11708: See also @ref{Emacs and Gforth}.
11709:
11710: @node ANS Report, , Tools, Tools
11711: @section @file{ans-report.fs}: Report the words used, sorted by wordset
11712: @cindex @file{ans-report.fs}
11713: @cindex report the words used in your program
11714: @cindex words used in your program
11715:
11716: If you want to label a Forth program as ANS Forth Program, you must
11717: document which wordsets the program uses; for extension wordsets, it is
11718: helpful to list the words the program requires from these wordsets
11719: (because Forth systems are allowed to provide only some words of them).
11720:
11721: The @file{ans-report.fs} tool makes it easy for you to determine which
11722: words from which wordset and which non-ANS words your application
11723: uses. You simply have to include @file{ans-report.fs} before loading the
11724: program you want to check. After loading your program, you can get the
11725: report with @code{print-ans-report}. A typical use is to run this as
11726: batch job like this:
11727: @example
11728: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
11729: @end example
11730:
11731: The output looks like this (for @file{compat/control.fs}):
11732: @example
11733: The program uses the following words
11734: from CORE :
11735: : POSTPONE THEN ; immediate ?dup IF 0=
11736: from BLOCK-EXT :
11737: \
11738: from FILE :
11739: (
11740: @end example
11741:
11742: @subsection Caveats
11743:
11744: Note that @file{ans-report.fs} just checks which words are used, not whether
11745: they are used in an ANS Forth conforming way!
11746:
11747: Some words are defined in several wordsets in the
11748: standard. @file{ans-report.fs} reports them for only one of the
11749: wordsets, and not necessarily the one you expect. It depends on usage
11750: which wordset is the right one to specify. E.g., if you only use the
11751: compilation semantics of @code{S"}, it is a Core word; if you also use
11752: its interpretation semantics, it is a File word.
11753:
11754: @c ******************************************************************
11755: @node ANS conformance, Standard vs Extensions, Tools, Top
11756: @chapter ANS conformance
11757: @cindex ANS conformance of Gforth
11758:
11759: To the best of our knowledge, Gforth is an
11760:
11761: ANS Forth System
11762: @itemize @bullet
11763: @item providing the Core Extensions word set
11764: @item providing the Block word set
11765: @item providing the Block Extensions word set
11766: @item providing the Double-Number word set
11767: @item providing the Double-Number Extensions word set
11768: @item providing the Exception word set
11769: @item providing the Exception Extensions word set
11770: @item providing the Facility word set
11771: @item providing @code{EKEY}, @code{EKEY>CHAR}, @code{EKEY?}, @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
11772: @item providing the File Access word set
11773: @item providing the File Access Extensions word set
11774: @item providing the Floating-Point word set
11775: @item providing the Floating-Point Extensions word set
11776: @item providing the Locals word set
11777: @item providing the Locals Extensions word set
11778: @item providing the Memory-Allocation word set
11779: @item providing the Memory-Allocation Extensions word set (that one's easy)
11780: @item providing the Programming-Tools word set
11781: @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
11782: @item providing the Search-Order word set
11783: @item providing the Search-Order Extensions word set
11784: @item providing the String word set
11785: @item providing the String Extensions word set (another easy one)
11786: @end itemize
11787:
11788: @cindex system documentation
11789: In addition, ANS Forth systems are required to document certain
11790: implementation choices. This chapter tries to meet these
11791: requirements. In many cases it gives a way to ask the system for the
11792: information instead of providing the information directly, in
11793: particular, if the information depends on the processor, the operating
11794: system or the installation options chosen, or if they are likely to
11795: change during the maintenance of Gforth.
11796:
11797: @comment The framework for the rest has been taken from pfe.
11798:
11799: @menu
11800: * The Core Words::
11801: * The optional Block word set::
11802: * The optional Double Number word set::
11803: * The optional Exception word set::
11804: * The optional Facility word set::
11805: * The optional File-Access word set::
11806: * The optional Floating-Point word set::
11807: * The optional Locals word set::
11808: * The optional Memory-Allocation word set::
11809: * The optional Programming-Tools word set::
11810: * The optional Search-Order word set::
11811: @end menu
11812:
11813:
11814: @c =====================================================================
11815: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
11816: @comment node-name, next, previous, up
11817: @section The Core Words
11818: @c =====================================================================
11819: @cindex core words, system documentation
11820: @cindex system documentation, core words
11821:
11822: @menu
11823: * core-idef:: Implementation Defined Options
11824: * core-ambcond:: Ambiguous Conditions
11825: * core-other:: Other System Documentation
11826: @end menu
11827:
11828: @c ---------------------------------------------------------------------
11829: @node core-idef, core-ambcond, The Core Words, The Core Words
11830: @subsection Implementation Defined Options
11831: @c ---------------------------------------------------------------------
11832: @cindex core words, implementation-defined options
11833: @cindex implementation-defined options, core words
11834:
11835:
11836: @table @i
11837: @item (Cell) aligned addresses:
11838: @cindex cell-aligned addresses
11839: @cindex aligned addresses
11840: processor-dependent. Gforth's alignment words perform natural alignment
11841: (e.g., an address aligned for a datum of size 8 is divisible by
11842: 8). Unaligned accesses usually result in a @code{-23 THROW}.
11843:
11844: @item @code{EMIT} and non-graphic characters:
11845: @cindex @code{EMIT} and non-graphic characters
11846: @cindex non-graphic characters and @code{EMIT}
11847: The character is output using the C library function (actually, macro)
11848: @code{putc}.
11849:
11850: @item character editing of @code{ACCEPT} and @code{EXPECT}:
11851: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
11852: @cindex editing in @code{ACCEPT} and @code{EXPECT}
11853: @cindex @code{ACCEPT}, editing
11854: @cindex @code{EXPECT}, editing
11855: This is modeled on the GNU readline library (@pxref{Readline
11856: Interaction, , Command Line Editing, readline, The GNU Readline
11857: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
11858: producing a full word completion every time you type it (instead of
11859: producing the common prefix of all completions). @xref{Command-line editing}.
11860:
11861: @item character set:
11862: @cindex character set
11863: The character set of your computer and display device. Gforth is
11864: 8-bit-clean (but some other component in your system may make trouble).
11865:
11866: @item Character-aligned address requirements:
11867: @cindex character-aligned address requirements
11868: installation-dependent. Currently a character is represented by a C
11869: @code{unsigned char}; in the future we might switch to @code{wchar_t}
11870: (Comments on that requested).
11871:
11872: @item character-set extensions and matching of names:
11873: @cindex character-set extensions and matching of names
11874: @cindex case-sensitivity for name lookup
11875: @cindex name lookup, case-sensitivity
11876: @cindex locale and case-sensitivity
11877: Any character except the ASCII NUL character can be used in a
11878: name. Matching is case-insensitive (except in @code{TABLE}s). The
11879: matching is performed using the C library function @code{strncasecmp}, whose
11880: function is probably influenced by the locale. E.g., the @code{C} locale
11881: does not know about accents and umlauts, so they are matched
11882: case-sensitively in that locale. For portability reasons it is best to
11883: write programs such that they work in the @code{C} locale. Then one can
11884: use libraries written by a Polish programmer (who might use words
11885: containing ISO Latin-2 encoded characters) and by a French programmer
11886: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
11887: funny results for some of the words (which ones, depends on the font you
11888: are using)). Also, the locale you prefer may not be available in other
11889: operating systems. Hopefully, Unicode will solve these problems one day.
11890:
11891: @item conditions under which control characters match a space delimiter:
11892: @cindex space delimiters
11893: @cindex control characters as delimiters
11894: If @code{WORD} is called with the space character as a delimiter, all
11895: white-space characters (as identified by the C macro @code{isspace()})
11896: are delimiters. @code{PARSE}, on the other hand, treats space like other
11897: delimiters. @code{SWORD} treats space like @code{WORD}, but behaves
11898: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
11899: interpreter (aka text interpreter) by default, treats all white-space
11900: characters as delimiters.
11901:
11902: @item format of the control-flow stack:
11903: @cindex control-flow stack, format
11904: The data stack is used as control-flow stack. The size of a control-flow
11905: stack item in cells is given by the constant @code{cs-item-size}. At the
11906: time of this writing, an item consists of a (pointer to a) locals list
11907: (third), an address in the code (second), and a tag for identifying the
11908: item (TOS). The following tags are used: @code{defstart},
11909: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
11910: @code{scopestart}.
11911:
11912: @item conversion of digits > 35
11913: @cindex digits > 35
11914: The characters @code{[\]^_'} are the digits with the decimal value
11915: 36@minus{}41. There is no way to input many of the larger digits.
11916:
11917: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
11918: @cindex @code{EXPECT}, display after end of input
11919: @cindex @code{ACCEPT}, display after end of input
11920: The cursor is moved to the end of the entered string. If the input is
11921: terminated using the @kbd{Return} key, a space is typed.
11922:
11923: @item exception abort sequence of @code{ABORT"}:
11924: @cindex exception abort sequence of @code{ABORT"}
11925: @cindex @code{ABORT"}, exception abort sequence
11926: The error string is stored into the variable @code{"error} and a
11927: @code{-2 throw} is performed.
11928:
11929: @item input line terminator:
11930: @cindex input line terminator
11931: @cindex line terminator on input
11932: @cindex newline character on input
11933: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
11934: lines. One of these characters is typically produced when you type the
11935: @kbd{Enter} or @kbd{Return} key.
11936:
11937: @item maximum size of a counted string:
11938: @cindex maximum size of a counted string
11939: @cindex counted string, maximum size
11940: @code{s" /counted-string" environment? drop .}. Currently 255 characters
11941: on all ports, but this may change.
11942:
11943: @item maximum size of a parsed string:
11944: @cindex maximum size of a parsed string
11945: @cindex parsed string, maximum size
11946: Given by the constant @code{/line}. Currently 255 characters.
11947:
11948: @item maximum size of a definition name, in characters:
11949: @cindex maximum size of a definition name, in characters
11950: @cindex name, maximum length
11951: 31
11952:
11953: @item maximum string length for @code{ENVIRONMENT?}, in characters:
11954: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
11955: @cindex @code{ENVIRONMENT?} string length, maximum
11956: 31
11957:
11958: @item method of selecting the user input device:
11959: @cindex user input device, method of selecting
11960: The user input device is the standard input. There is currently no way to
11961: change it from within Gforth. However, the input can typically be
11962: redirected in the command line that starts Gforth.
11963:
11964: @item method of selecting the user output device:
11965: @cindex user output device, method of selecting
11966: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
11967: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
11968: output when the user output device is a terminal, otherwise the output
11969: is buffered.
11970:
11971: @item methods of dictionary compilation:
11972: What are we expected to document here?
11973:
11974: @item number of bits in one address unit:
11975: @cindex number of bits in one address unit
11976: @cindex address unit, size in bits
11977: @code{s" address-units-bits" environment? drop .}. 8 in all current
11978: ports.
11979:
11980: @item number representation and arithmetic:
11981: @cindex number representation and arithmetic
11982: Processor-dependent. Binary two's complement on all current ports.
11983:
11984: @item ranges for integer types:
11985: @cindex ranges for integer types
11986: @cindex integer types, ranges
11987: Installation-dependent. Make environmental queries for @code{MAX-N},
11988: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
11989: unsigned (and positive) types is 0. The lower bound for signed types on
11990: two's complement and one's complement machines machines can be computed
11991: by adding 1 to the upper bound.
11992:
11993: @item read-only data space regions:
11994: @cindex read-only data space regions
11995: @cindex data-space, read-only regions
11996: The whole Forth data space is writable.
11997:
11998: @item size of buffer at @code{WORD}:
11999: @cindex size of buffer at @code{WORD}
12000: @cindex @code{WORD} buffer size
12001: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
12002: shared with the pictured numeric output string. If overwriting
12003: @code{PAD} is acceptable, it is as large as the remaining dictionary
12004: space, although only as much can be sensibly used as fits in a counted
12005: string.
12006:
12007: @item size of one cell in address units:
12008: @cindex cell size
12009: @code{1 cells .}.
12010:
12011: @item size of one character in address units:
12012: @cindex char size
12013: @code{1 chars .}. 1 on all current ports.
12014:
12015: @item size of the keyboard terminal buffer:
12016: @cindex size of the keyboard terminal buffer
12017: @cindex terminal buffer, size
12018: Varies. You can determine the size at a specific time using @code{lp@@
12019: tib - .}. It is shared with the locals stack and TIBs of files that
12020: include the current file. You can change the amount of space for TIBs
12021: and locals stack at Gforth startup with the command line option
12022: @code{-l}.
12023:
12024: @item size of the pictured numeric output buffer:
12025: @cindex size of the pictured numeric output buffer
12026: @cindex pictured numeric output buffer, size
12027: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
12028: shared with @code{WORD}.
12029:
12030: @item size of the scratch area returned by @code{PAD}:
12031: @cindex size of the scratch area returned by @code{PAD}
12032: @cindex @code{PAD} size
12033: The remainder of dictionary space. @code{unused pad here - - .}.
12034:
12035: @item system case-sensitivity characteristics:
12036: @cindex case-sensitivity characteristics
12037: Dictionary searches are case-insensitive (except in
12038: @code{TABLE}s). However, as explained above under @i{character-set
12039: extensions}, the matching for non-ASCII characters is determined by the
12040: locale you are using. In the default @code{C} locale all non-ASCII
12041: characters are matched case-sensitively.
12042:
12043: @item system prompt:
12044: @cindex system prompt
12045: @cindex prompt
12046: @code{ ok} in interpret state, @code{ compiled} in compile state.
12047:
12048: @item division rounding:
12049: @cindex division rounding
12050: installation dependent. @code{s" floored" environment? drop .}. We leave
12051: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
12052: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
12053:
12054: @item values of @code{STATE} when true:
12055: @cindex @code{STATE} values
12056: -1.
12057:
12058: @item values returned after arithmetic overflow:
12059: On two's complement machines, arithmetic is performed modulo
12060: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12061: arithmetic (with appropriate mapping for signed types). Division by zero
12062: typically results in a @code{-55 throw} (Floating-point unidentified
12063: fault), although a @code{-10 throw} (divide by zero) would be more
12064: appropriate.
12065:
12066: @item whether the current definition can be found after @t{DOES>}:
12067: @cindex @t{DOES>}, visibility of current definition
12068: No.
12069:
12070: @end table
12071:
12072: @c ---------------------------------------------------------------------
12073: @node core-ambcond, core-other, core-idef, The Core Words
12074: @subsection Ambiguous conditions
12075: @c ---------------------------------------------------------------------
12076: @cindex core words, ambiguous conditions
12077: @cindex ambiguous conditions, core words
12078:
12079: @table @i
12080:
12081: @item a name is neither a word nor a number:
12082: @cindex name not found
12083: @cindex undefined word
12084: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
12085: preserves the data and FP stack, so you don't lose more work than
12086: necessary.
12087:
12088: @item a definition name exceeds the maximum length allowed:
12089: @cindex word name too long
12090: @code{-19 throw} (Word name too long)
12091:
12092: @item addressing a region not inside the various data spaces of the forth system:
12093: @cindex Invalid memory address
12094: The stacks, code space and header space are accessible. Machine code space is
12095: typically readable. Accessing other addresses gives results dependent on
12096: the operating system. On decent systems: @code{-9 throw} (Invalid memory
12097: address).
12098:
12099: @item argument type incompatible with parameter:
12100: @cindex argument type mismatch
12101: This is usually not caught. Some words perform checks, e.g., the control
12102: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
12103: mismatch).
12104:
12105: @item attempting to obtain the execution token of a word with undefined execution semantics:
12106: @cindex Interpreting a compile-only word, for @code{'} etc.
12107: @cindex execution token of words with undefined execution semantics
12108: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
12109: get an execution token for @code{compile-only-error} (which performs a
12110: @code{-14 throw} when executed).
12111:
12112: @item dividing by zero:
12113: @cindex dividing by zero
12114: @cindex floating point unidentified fault, integer division
12115: On better platforms, this produces a @code{-10 throw} (Division by
12116: zero); on other systems, this typically results in a @code{-55 throw}
12117: (Floating-point unidentified fault).
12118:
12119: @item insufficient data stack or return stack space:
12120: @cindex insufficient data stack or return stack space
12121: @cindex stack overflow
12122: @cindex address alignment exception, stack overflow
12123: @cindex Invalid memory address, stack overflow
12124: Depending on the operating system, the installation, and the invocation
12125: of Gforth, this is either checked by the memory management hardware, or
12126: it is not checked. If it is checked, you typically get a @code{-3 throw}
12127: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
12128: throw} (Invalid memory address) (depending on the platform and how you
12129: achieved the overflow) as soon as the overflow happens. If it is not
12130: checked, overflows typically result in mysterious illegal memory
12131: accesses, producing @code{-9 throw} (Invalid memory address) or
12132: @code{-23 throw} (Address alignment exception); they might also destroy
12133: the internal data structure of @code{ALLOCATE} and friends, resulting in
12134: various errors in these words.
12135:
12136: @item insufficient space for loop control parameters:
12137: @cindex insufficient space for loop control parameters
12138: like other return stack overflows.
12139:
12140: @item insufficient space in the dictionary:
12141: @cindex insufficient space in the dictionary
12142: @cindex dictionary overflow
12143: If you try to allot (either directly with @code{allot}, or indirectly
12144: with @code{,}, @code{create} etc.) more memory than available in the
12145: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
12146: to access memory beyond the end of the dictionary, the results are
12147: similar to stack overflows.
12148:
12149: @item interpreting a word with undefined interpretation semantics:
12150: @cindex interpreting a word with undefined interpretation semantics
12151: @cindex Interpreting a compile-only word
12152: For some words, we have defined interpretation semantics. For the
12153: others: @code{-14 throw} (Interpreting a compile-only word).
12154:
12155: @item modifying the contents of the input buffer or a string literal:
12156: @cindex modifying the contents of the input buffer or a string literal
12157: These are located in writable memory and can be modified.
12158:
12159: @item overflow of the pictured numeric output string:
12160: @cindex overflow of the pictured numeric output string
12161: @cindex pictured numeric output string, overflow
12162: @code{-17 throw} (Pictured numeric ouput string overflow).
12163:
12164: @item parsed string overflow:
12165: @cindex parsed string overflow
12166: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
12167:
12168: @item producing a result out of range:
12169: @cindex result out of range
12170: On two's complement machines, arithmetic is performed modulo
12171: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
12172: arithmetic (with appropriate mapping for signed types). Division by zero
12173: typically results in a @code{-10 throw} (divide by zero) or @code{-55
12174: throw} (floating point unidentified fault). @code{convert} and
12175: @code{>number} currently overflow silently.
12176:
12177: @item reading from an empty data or return stack:
12178: @cindex stack empty
12179: @cindex stack underflow
12180: @cindex return stack underflow
12181: The data stack is checked by the outer (aka text) interpreter after
12182: every word executed. If it has underflowed, a @code{-4 throw} (Stack
12183: underflow) is performed. Apart from that, stacks may be checked or not,
12184: depending on operating system, installation, and invocation. If they are
12185: caught by a check, they typically result in @code{-4 throw} (Stack
12186: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
12187: (Invalid memory address), depending on the platform and which stack
12188: underflows and by how much. Note that even if the system uses checking
12189: (through the MMU), your program may have to underflow by a significant
12190: number of stack items to trigger the reaction (the reason for this is
12191: that the MMU, and therefore the checking, works with a page-size
12192: granularity). If there is no checking, the symptoms resulting from an
12193: underflow are similar to those from an overflow. Unbalanced return
12194: stack errors result in a variaty of symptoms, including @code{-9 throw}
12195: (Invalid memory address) and Illegal Instruction (typically @code{-260
12196: throw}).
12197:
12198: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
12199: @cindex unexpected end of the input buffer
12200: @cindex zero-length string as a name
12201: @cindex Attempt to use zero-length string as a name
12202: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
12203: use zero-length string as a name). Words like @code{'} probably will not
12204: find what they search. Note that it is possible to create zero-length
12205: names with @code{nextname} (should it not?).
12206:
12207: @item @code{>IN} greater than input buffer:
12208: @cindex @code{>IN} greater than input buffer
12209: The next invocation of a parsing word returns a string with length 0.
12210:
12211: @item @code{RECURSE} appears after @code{DOES>}:
12212: @cindex @code{RECURSE} appears after @code{DOES>}
12213: Compiles a recursive call to the defining word, not to the defined word.
12214:
12215: @item argument input source different than current input source for @code{RESTORE-INPUT}:
12216: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
12217: @cindex argument type mismatch, @code{RESTORE-INPUT}
12218: @cindex @code{RESTORE-INPUT}, Argument type mismatch
12219: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
12220: the end of the file was reached), its source-id may be
12221: reused. Therefore, restoring an input source specification referencing a
12222: closed file may lead to unpredictable results instead of a @code{-12
12223: THROW}.
12224:
12225: In the future, Gforth may be able to restore input source specifications
12226: from other than the current input source.
12227:
12228: @item data space containing definitions gets de-allocated:
12229: @cindex data space containing definitions gets de-allocated
12230: Deallocation with @code{allot} is not checked. This typically results in
12231: memory access faults or execution of illegal instructions.
12232:
12233: @item data space read/write with incorrect alignment:
12234: @cindex data space read/write with incorrect alignment
12235: @cindex alignment faults
12236: @cindex address alignment exception
12237: Processor-dependent. Typically results in a @code{-23 throw} (Address
12238: alignment exception). Under Linux-Intel on a 486 or later processor with
12239: alignment turned on, incorrect alignment results in a @code{-9 throw}
12240: (Invalid memory address). There are reportedly some processors with
12241: alignment restrictions that do not report violations.
12242:
12243: @item data space pointer not properly aligned, @code{,}, @code{C,}:
12244: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
12245: Like other alignment errors.
12246:
12247: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
12248: Like other stack underflows.
12249:
12250: @item loop control parameters not available:
12251: @cindex loop control parameters not available
12252: Not checked. The counted loop words simply assume that the top of return
12253: stack items are loop control parameters and behave accordingly.
12254:
12255: @item most recent definition does not have a name (@code{IMMEDIATE}):
12256: @cindex most recent definition does not have a name (@code{IMMEDIATE})
12257: @cindex last word was headerless
12258: @code{abort" last word was headerless"}.
12259:
12260: @item name not defined by @code{VALUE} used by @code{TO}:
12261: @cindex name not defined by @code{VALUE} used by @code{TO}
12262: @cindex @code{TO} on non-@code{VALUE}s
12263: @cindex Invalid name argument, @code{TO}
12264: @code{-32 throw} (Invalid name argument) (unless name is a local or was
12265: defined by @code{CONSTANT}; in the latter case it just changes the constant).
12266:
12267: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
12268: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
12269: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
12270: @code{-13 throw} (Undefined word)
12271:
12272: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
12273: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
12274: Gforth behaves as if they were of the same type. I.e., you can predict
12275: the behaviour by interpreting all parameters as, e.g., signed.
12276:
12277: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
12278: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
12279: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
12280: compilation semantics of @code{TO}.
12281:
12282: @item String longer than a counted string returned by @code{WORD}:
12283: @cindex string longer than a counted string returned by @code{WORD}
12284: @cindex @code{WORD}, string overflow
12285: Not checked. The string will be ok, but the count will, of course,
12286: contain only the least significant bits of the length.
12287:
12288: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
12289: @cindex @code{LSHIFT}, large shift counts
12290: @cindex @code{RSHIFT}, large shift counts
12291: Processor-dependent. Typical behaviours are returning 0 and using only
12292: the low bits of the shift count.
12293:
12294: @item word not defined via @code{CREATE}:
12295: @cindex @code{>BODY} of non-@code{CREATE}d words
12296: @code{>BODY} produces the PFA of the word no matter how it was defined.
12297:
12298: @cindex @code{DOES>} of non-@code{CREATE}d words
12299: @code{DOES>} changes the execution semantics of the last defined word no
12300: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
12301: @code{CREATE , DOES>}.
12302:
12303: @item words improperly used outside @code{<#} and @code{#>}:
12304: Not checked. As usual, you can expect memory faults.
12305:
12306: @end table
12307:
12308:
12309: @c ---------------------------------------------------------------------
12310: @node core-other, , core-ambcond, The Core Words
12311: @subsection Other system documentation
12312: @c ---------------------------------------------------------------------
12313: @cindex other system documentation, core words
12314: @cindex core words, other system documentation
12315:
12316: @table @i
12317: @item nonstandard words using @code{PAD}:
12318: @cindex @code{PAD} use by nonstandard words
12319: None.
12320:
12321: @item operator's terminal facilities available:
12322: @cindex operator's terminal facilities available
12323: After processing the command line, Gforth goes into interactive mode,
12324: and you can give commands to Gforth interactively. The actual facilities
12325: available depend on how you invoke Gforth.
12326:
12327: @item program data space available:
12328: @cindex program data space available
12329: @cindex data space available
12330: @code{UNUSED .} gives the remaining dictionary space. The total
12331: dictionary space can be specified with the @code{-m} switch
12332: (@pxref{Invoking Gforth}) when Gforth starts up.
12333:
12334: @item return stack space available:
12335: @cindex return stack space available
12336: You can compute the total return stack space in cells with
12337: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
12338: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
12339:
12340: @item stack space available:
12341: @cindex stack space available
12342: You can compute the total data stack space in cells with
12343: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
12344: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
12345:
12346: @item system dictionary space required, in address units:
12347: @cindex system dictionary space required, in address units
12348: Type @code{here forthstart - .} after startup. At the time of this
12349: writing, this gives 80080 (bytes) on a 32-bit system.
12350: @end table
12351:
12352:
12353: @c =====================================================================
12354: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
12355: @section The optional Block word set
12356: @c =====================================================================
12357: @cindex system documentation, block words
12358: @cindex block words, system documentation
12359:
12360: @menu
12361: * block-idef:: Implementation Defined Options
12362: * block-ambcond:: Ambiguous Conditions
12363: * block-other:: Other System Documentation
12364: @end menu
12365:
12366:
12367: @c ---------------------------------------------------------------------
12368: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
12369: @subsection Implementation Defined Options
12370: @c ---------------------------------------------------------------------
12371: @cindex implementation-defined options, block words
12372: @cindex block words, implementation-defined options
12373:
12374: @table @i
12375: @item the format for display by @code{LIST}:
12376: @cindex @code{LIST} display format
12377: First the screen number is displayed, then 16 lines of 64 characters,
12378: each line preceded by the line number.
12379:
12380: @item the length of a line affected by @code{\}:
12381: @cindex length of a line affected by @code{\}
12382: @cindex @code{\}, line length in blocks
12383: 64 characters.
12384: @end table
12385:
12386:
12387: @c ---------------------------------------------------------------------
12388: @node block-ambcond, block-other, block-idef, The optional Block word set
12389: @subsection Ambiguous conditions
12390: @c ---------------------------------------------------------------------
12391: @cindex block words, ambiguous conditions
12392: @cindex ambiguous conditions, block words
12393:
12394: @table @i
12395: @item correct block read was not possible:
12396: @cindex block read not possible
12397: Typically results in a @code{throw} of some OS-derived value (between
12398: -512 and -2048). If the blocks file was just not long enough, blanks are
12399: supplied for the missing portion.
12400:
12401: @item I/O exception in block transfer:
12402: @cindex I/O exception in block transfer
12403: @cindex block transfer, I/O exception
12404: Typically results in a @code{throw} of some OS-derived value (between
12405: -512 and -2048).
12406:
12407: @item invalid block number:
12408: @cindex invalid block number
12409: @cindex block number invalid
12410: @code{-35 throw} (Invalid block number)
12411:
12412: @item a program directly alters the contents of @code{BLK}:
12413: @cindex @code{BLK}, altering @code{BLK}
12414: The input stream is switched to that other block, at the same
12415: position. If the storing to @code{BLK} happens when interpreting
12416: non-block input, the system will get quite confused when the block ends.
12417:
12418: @item no current block buffer for @code{UPDATE}:
12419: @cindex @code{UPDATE}, no current block buffer
12420: @code{UPDATE} has no effect.
12421:
12422: @end table
12423:
12424: @c ---------------------------------------------------------------------
12425: @node block-other, , block-ambcond, The optional Block word set
12426: @subsection Other system documentation
12427: @c ---------------------------------------------------------------------
12428: @cindex other system documentation, block words
12429: @cindex block words, other system documentation
12430:
12431: @table @i
12432: @item any restrictions a multiprogramming system places on the use of buffer addresses:
12433: No restrictions (yet).
12434:
12435: @item the number of blocks available for source and data:
12436: depends on your disk space.
12437:
12438: @end table
12439:
12440:
12441: @c =====================================================================
12442: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
12443: @section The optional Double Number word set
12444: @c =====================================================================
12445: @cindex system documentation, double words
12446: @cindex double words, system documentation
12447:
12448: @menu
12449: * double-ambcond:: Ambiguous Conditions
12450: @end menu
12451:
12452:
12453: @c ---------------------------------------------------------------------
12454: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
12455: @subsection Ambiguous conditions
12456: @c ---------------------------------------------------------------------
12457: @cindex double words, ambiguous conditions
12458: @cindex ambiguous conditions, double words
12459:
12460: @table @i
12461: @item @i{d} outside of range of @i{n} in @code{D>S}:
12462: @cindex @code{D>S}, @i{d} out of range of @i{n}
12463: The least significant cell of @i{d} is produced.
12464:
12465: @end table
12466:
12467:
12468: @c =====================================================================
12469: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
12470: @section The optional Exception word set
12471: @c =====================================================================
12472: @cindex system documentation, exception words
12473: @cindex exception words, system documentation
12474:
12475: @menu
12476: * exception-idef:: Implementation Defined Options
12477: @end menu
12478:
12479:
12480: @c ---------------------------------------------------------------------
12481: @node exception-idef, , The optional Exception word set, The optional Exception word set
12482: @subsection Implementation Defined Options
12483: @c ---------------------------------------------------------------------
12484: @cindex implementation-defined options, exception words
12485: @cindex exception words, implementation-defined options
12486:
12487: @table @i
12488: @item @code{THROW}-codes used in the system:
12489: @cindex @code{THROW}-codes used in the system
12490: The codes -256@minus{}-511 are used for reporting signals. The mapping
12491: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
12492: codes -512@minus{}-2047 are used for OS errors (for file and memory
12493: allocation operations). The mapping from OS error numbers to throw codes
12494: is -512@minus{}@code{errno}. One side effect of this mapping is that
12495: undefined OS errors produce a message with a strange number; e.g.,
12496: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
12497: @end table
12498:
12499: @c =====================================================================
12500: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
12501: @section The optional Facility word set
12502: @c =====================================================================
12503: @cindex system documentation, facility words
12504: @cindex facility words, system documentation
12505:
12506: @menu
12507: * facility-idef:: Implementation Defined Options
12508: * facility-ambcond:: Ambiguous Conditions
12509: @end menu
12510:
12511:
12512: @c ---------------------------------------------------------------------
12513: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
12514: @subsection Implementation Defined Options
12515: @c ---------------------------------------------------------------------
12516: @cindex implementation-defined options, facility words
12517: @cindex facility words, implementation-defined options
12518:
12519: @table @i
12520: @item encoding of keyboard events (@code{EKEY}):
12521: @cindex keyboard events, encoding in @code{EKEY}
12522: @cindex @code{EKEY}, encoding of keyboard events
12523: Keys corresponding to ASCII characters are encoded as ASCII characters.
12524: Other keys are encoded with the constants @code{k-left}, @code{k-right},
12525: @code{k-up}, @code{k-down}, @code{k-home}, @code{k-end}, @code{k1},
12526: @code{k2}, @code{k3}, @code{k4}, @code{k5}, @code{k6}, @code{k7},
12527: @code{k8}, @code{k9}, @code{k10}, @code{k11}, @code{k12}.
12528:
12529:
12530: @item duration of a system clock tick:
12531: @cindex duration of a system clock tick
12532: @cindex clock tick duration
12533: System dependent. With respect to @code{MS}, the time is specified in
12534: microseconds. How well the OS and the hardware implement this, is
12535: another question.
12536:
12537: @item repeatability to be expected from the execution of @code{MS}:
12538: @cindex repeatability to be expected from the execution of @code{MS}
12539: @cindex @code{MS}, repeatability to be expected
12540: System dependent. On Unix, a lot depends on load. If the system is
12541: lightly loaded, and the delay is short enough that Gforth does not get
12542: swapped out, the performance should be acceptable. Under MS-DOS and
12543: other single-tasking systems, it should be good.
12544:
12545: @end table
12546:
12547:
12548: @c ---------------------------------------------------------------------
12549: @node facility-ambcond, , facility-idef, The optional Facility word set
12550: @subsection Ambiguous conditions
12551: @c ---------------------------------------------------------------------
12552: @cindex facility words, ambiguous conditions
12553: @cindex ambiguous conditions, facility words
12554:
12555: @table @i
12556: @item @code{AT-XY} can't be performed on user output device:
12557: @cindex @code{AT-XY} can't be performed on user output device
12558: Largely terminal dependent. No range checks are done on the arguments.
12559: No errors are reported. You may see some garbage appearing, you may see
12560: simply nothing happen.
12561:
12562: @end table
12563:
12564:
12565: @c =====================================================================
12566: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
12567: @section The optional File-Access word set
12568: @c =====================================================================
12569: @cindex system documentation, file words
12570: @cindex file words, system documentation
12571:
12572: @menu
12573: * file-idef:: Implementation Defined Options
12574: * file-ambcond:: Ambiguous Conditions
12575: @end menu
12576:
12577: @c ---------------------------------------------------------------------
12578: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
12579: @subsection Implementation Defined Options
12580: @c ---------------------------------------------------------------------
12581: @cindex implementation-defined options, file words
12582: @cindex file words, implementation-defined options
12583:
12584: @table @i
12585: @item file access methods used:
12586: @cindex file access methods used
12587: @code{R/O}, @code{R/W} and @code{BIN} work as you would
12588: expect. @code{W/O} translates into the C file opening mode @code{w} (or
12589: @code{wb}): The file is cleared, if it exists, and created, if it does
12590: not (with both @code{open-file} and @code{create-file}). Under Unix
12591: @code{create-file} creates a file with 666 permissions modified by your
12592: umask.
12593:
12594: @item file exceptions:
12595: @cindex file exceptions
12596: The file words do not raise exceptions (except, perhaps, memory access
12597: faults when you pass illegal addresses or file-ids).
12598:
12599: @item file line terminator:
12600: @cindex file line terminator
12601: System-dependent. Gforth uses C's newline character as line
12602: terminator. What the actual character code(s) of this are is
12603: system-dependent.
12604:
12605: @item file name format:
12606: @cindex file name format
12607: System dependent. Gforth just uses the file name format of your OS.
12608:
12609: @item information returned by @code{FILE-STATUS}:
12610: @cindex @code{FILE-STATUS}, returned information
12611: @code{FILE-STATUS} returns the most powerful file access mode allowed
12612: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
12613: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
12614: along with the returned mode.
12615:
12616: @item input file state after an exception when including source:
12617: @cindex exception when including source
12618: All files that are left via the exception are closed.
12619:
12620: @item @i{ior} values and meaning:
12621: @cindex @i{ior} values and meaning
12622: @cindex @i{wior} values and meaning
12623: The @i{ior}s returned by the file and memory allocation words are
12624: intended as throw codes. They typically are in the range
12625: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
12626: @i{ior}s is -512@minus{}@i{errno}.
12627:
12628: @item maximum depth of file input nesting:
12629: @cindex maximum depth of file input nesting
12630: @cindex file input nesting, maximum depth
12631: limited by the amount of return stack, locals/TIB stack, and the number
12632: of open files available. This should not give you troubles.
12633:
12634: @item maximum size of input line:
12635: @cindex maximum size of input line
12636: @cindex input line size, maximum
12637: @code{/line}. Currently 255.
12638:
12639: @item methods of mapping block ranges to files:
12640: @cindex mapping block ranges to files
12641: @cindex files containing blocks
12642: @cindex blocks in files
12643: By default, blocks are accessed in the file @file{blocks.fb} in the
12644: current working directory. The file can be switched with @code{USE}.
12645:
12646: @item number of string buffers provided by @code{S"}:
12647: @cindex @code{S"}, number of string buffers
12648: 1
12649:
12650: @item size of string buffer used by @code{S"}:
12651: @cindex @code{S"}, size of string buffer
12652: @code{/line}. currently 255.
12653:
12654: @end table
12655:
12656: @c ---------------------------------------------------------------------
12657: @node file-ambcond, , file-idef, The optional File-Access word set
12658: @subsection Ambiguous conditions
12659: @c ---------------------------------------------------------------------
12660: @cindex file words, ambiguous conditions
12661: @cindex ambiguous conditions, file words
12662:
12663: @table @i
12664: @item attempting to position a file outside its boundaries:
12665: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
12666: @code{REPOSITION-FILE} is performed as usual: Afterwards,
12667: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
12668:
12669: @item attempting to read from file positions not yet written:
12670: @cindex reading from file positions not yet written
12671: End-of-file, i.e., zero characters are read and no error is reported.
12672:
12673: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
12674: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
12675: An appropriate exception may be thrown, but a memory fault or other
12676: problem is more probable.
12677:
12678: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
12679: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
12680: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
12681: The @i{ior} produced by the operation, that discovered the problem, is
12682: thrown.
12683:
12684: @item named file cannot be opened (@code{INCLUDED}):
12685: @cindex @code{INCLUDED}, named file cannot be opened
12686: The @i{ior} produced by @code{open-file} is thrown.
12687:
12688: @item requesting an unmapped block number:
12689: @cindex unmapped block numbers
12690: There are no unmapped legal block numbers. On some operating systems,
12691: writing a block with a large number may overflow the file system and
12692: have an error message as consequence.
12693:
12694: @item using @code{source-id} when @code{blk} is non-zero:
12695: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
12696: @code{source-id} performs its function. Typically it will give the id of
12697: the source which loaded the block. (Better ideas?)
12698:
12699: @end table
12700:
12701:
12702: @c =====================================================================
12703: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
12704: @section The optional Floating-Point word set
12705: @c =====================================================================
12706: @cindex system documentation, floating-point words
12707: @cindex floating-point words, system documentation
12708:
12709: @menu
12710: * floating-idef:: Implementation Defined Options
12711: * floating-ambcond:: Ambiguous Conditions
12712: @end menu
12713:
12714:
12715: @c ---------------------------------------------------------------------
12716: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
12717: @subsection Implementation Defined Options
12718: @c ---------------------------------------------------------------------
12719: @cindex implementation-defined options, floating-point words
12720: @cindex floating-point words, implementation-defined options
12721:
12722: @table @i
12723: @item format and range of floating point numbers:
12724: @cindex format and range of floating point numbers
12725: @cindex floating point numbers, format and range
12726: System-dependent; the @code{double} type of C.
12727:
12728: @item results of @code{REPRESENT} when @i{float} is out of range:
12729: @cindex @code{REPRESENT}, results when @i{float} is out of range
12730: System dependent; @code{REPRESENT} is implemented using the C library
12731: function @code{ecvt()} and inherits its behaviour in this respect.
12732:
12733: @item rounding or truncation of floating-point numbers:
12734: @cindex rounding of floating-point numbers
12735: @cindex truncation of floating-point numbers
12736: @cindex floating-point numbers, rounding or truncation
12737: System dependent; the rounding behaviour is inherited from the hosting C
12738: compiler. IEEE-FP-based (i.e., most) systems by default round to
12739: nearest, and break ties by rounding to even (i.e., such that the last
12740: bit of the mantissa is 0).
12741:
12742: @item size of floating-point stack:
12743: @cindex floating-point stack size
12744: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
12745: the floating-point stack (in floats). You can specify this on startup
12746: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
12747:
12748: @item width of floating-point stack:
12749: @cindex floating-point stack width
12750: @code{1 floats}.
12751:
12752: @end table
12753:
12754:
12755: @c ---------------------------------------------------------------------
12756: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
12757: @subsection Ambiguous conditions
12758: @c ---------------------------------------------------------------------
12759: @cindex floating-point words, ambiguous conditions
12760: @cindex ambiguous conditions, floating-point words
12761:
12762: @table @i
12763: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
12764: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
12765: System-dependent. Typically results in a @code{-23 THROW} like other
12766: alignment violations.
12767:
12768: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
12769: @cindex @code{f@@} used with an address that is not float aligned
12770: @cindex @code{f!} used with an address that is not float aligned
12771: System-dependent. Typically results in a @code{-23 THROW} like other
12772: alignment violations.
12773:
12774: @item floating-point result out of range:
12775: @cindex floating-point result out of range
12776: System-dependent. Can result in a @code{-55 THROW} (Floating-point
12777: unidentified fault), or can produce a special value representing, e.g.,
12778: Infinity.
12779:
12780: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
12781: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
12782: System-dependent. Typically results in an alignment fault like other
12783: alignment violations.
12784:
12785: @item @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
12786: @cindex @code{base} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
12787: The floating-point number is converted into decimal nonetheless.
12788:
12789: @item Both arguments are equal to zero (@code{FATAN2}):
12790: @cindex @code{FATAN2}, both arguments are equal to zero
12791: System-dependent. @code{FATAN2} is implemented using the C library
12792: function @code{atan2()}.
12793:
12794: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
12795: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
12796: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
12797: because of small errors and the tan will be a very large (or very small)
12798: but finite number.
12799:
12800: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
12801: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
12802: The result is rounded to the nearest float.
12803:
12804: @item dividing by zero:
12805: @cindex dividing by zero, floating-point
12806: @cindex floating-point dividing by zero
12807: @cindex floating-point unidentified fault, FP divide-by-zero
12808: @code{-55 throw} (Floating-point unidentified fault)
12809:
12810: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
12811: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
12812: System dependent. On IEEE-FP based systems the number is converted into
12813: an infinity.
12814:
12815: @item @i{float}<1 (@code{FACOSH}):
12816: @cindex @code{FACOSH}, @i{float}<1
12817: @cindex floating-point unidentified fault, @code{FACOSH}
12818: @code{-55 throw} (Floating-point unidentified fault)
12819:
12820: @item @i{float}=<-1 (@code{FLNP1}):
12821: @cindex @code{FLNP1}, @i{float}=<-1
12822: @cindex floating-point unidentified fault, @code{FLNP1}
12823: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
12824: negative infinity is typically produced for @i{float}=-1.
12825:
12826: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
12827: @cindex @code{FLN}, @i{float}=<0
12828: @cindex @code{FLOG}, @i{float}=<0
12829: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
12830: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
12831: negative infinity is typically produced for @i{float}=0.
12832:
12833: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
12834: @cindex @code{FASINH}, @i{float}<0
12835: @cindex @code{FSQRT}, @i{float}<0
12836: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
12837: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
12838: produces values for these inputs on my Linux box (Bug in the C library?)
12839:
12840: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
12841: @cindex @code{FACOS}, |@i{float}|>1
12842: @cindex @code{FASIN}, |@i{float}|>1
12843: @cindex @code{FATANH}, |@i{float}|>1
12844: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
12845: @code{-55 throw} (Floating-point unidentified fault).
12846:
12847: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
12848: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
12849: @cindex floating-point unidentified fault, @code{F>D}
12850: @code{-55 throw} (Floating-point unidentified fault).
12851:
12852: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
12853: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
12854: This does not happen.
12855: @end table
12856:
12857: @c =====================================================================
12858: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
12859: @section The optional Locals word set
12860: @c =====================================================================
12861: @cindex system documentation, locals words
12862: @cindex locals words, system documentation
12863:
12864: @menu
12865: * locals-idef:: Implementation Defined Options
12866: * locals-ambcond:: Ambiguous Conditions
12867: @end menu
12868:
12869:
12870: @c ---------------------------------------------------------------------
12871: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
12872: @subsection Implementation Defined Options
12873: @c ---------------------------------------------------------------------
12874: @cindex implementation-defined options, locals words
12875: @cindex locals words, implementation-defined options
12876:
12877: @table @i
12878: @item maximum number of locals in a definition:
12879: @cindex maximum number of locals in a definition
12880: @cindex locals, maximum number in a definition
12881: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
12882: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
12883: characters. The number of locals in a definition is bounded by the size
12884: of locals-buffer, which contains the names of the locals.
12885:
12886: @end table
12887:
12888:
12889: @c ---------------------------------------------------------------------
12890: @node locals-ambcond, , locals-idef, The optional Locals word set
12891: @subsection Ambiguous conditions
12892: @c ---------------------------------------------------------------------
12893: @cindex locals words, ambiguous conditions
12894: @cindex ambiguous conditions, locals words
12895:
12896: @table @i
12897: @item executing a named local in interpretation state:
12898: @cindex local in interpretation state
12899: @cindex Interpreting a compile-only word, for a local
12900: Locals have no interpretation semantics. If you try to perform the
12901: interpretation semantics, you will get a @code{-14 throw} somewhere
12902: (Interpreting a compile-only word). If you perform the compilation
12903: semantics, the locals access will be compiled (irrespective of state).
12904:
12905: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
12906: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
12907: @cindex @code{TO} on non-@code{VALUE}s and non-locals
12908: @cindex Invalid name argument, @code{TO}
12909: @code{-32 throw} (Invalid name argument)
12910:
12911: @end table
12912:
12913:
12914: @c =====================================================================
12915: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
12916: @section The optional Memory-Allocation word set
12917: @c =====================================================================
12918: @cindex system documentation, memory-allocation words
12919: @cindex memory-allocation words, system documentation
12920:
12921: @menu
12922: * memory-idef:: Implementation Defined Options
12923: @end menu
12924:
12925:
12926: @c ---------------------------------------------------------------------
12927: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
12928: @subsection Implementation Defined Options
12929: @c ---------------------------------------------------------------------
12930: @cindex implementation-defined options, memory-allocation words
12931: @cindex memory-allocation words, implementation-defined options
12932:
12933: @table @i
12934: @item values and meaning of @i{ior}:
12935: @cindex @i{ior} values and meaning
12936: The @i{ior}s returned by the file and memory allocation words are
12937: intended as throw codes. They typically are in the range
12938: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
12939: @i{ior}s is -512@minus{}@i{errno}.
12940:
12941: @end table
12942:
12943: @c =====================================================================
12944: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
12945: @section The optional Programming-Tools word set
12946: @c =====================================================================
12947: @cindex system documentation, programming-tools words
12948: @cindex programming-tools words, system documentation
12949:
12950: @menu
12951: * programming-idef:: Implementation Defined Options
12952: * programming-ambcond:: Ambiguous Conditions
12953: @end menu
12954:
12955:
12956: @c ---------------------------------------------------------------------
12957: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
12958: @subsection Implementation Defined Options
12959: @c ---------------------------------------------------------------------
12960: @cindex implementation-defined options, programming-tools words
12961: @cindex programming-tools words, implementation-defined options
12962:
12963: @table @i
12964: @item ending sequence for input following @code{;CODE} and @code{CODE}:
12965: @cindex @code{;CODE} ending sequence
12966: @cindex @code{CODE} ending sequence
12967: @code{END-CODE}
12968:
12969: @item manner of processing input following @code{;CODE} and @code{CODE}:
12970: @cindex @code{;CODE}, processing input
12971: @cindex @code{CODE}, processing input
12972: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
12973: the input is processed by the text interpreter, (starting) in interpret
12974: state.
12975:
12976: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
12977: @cindex @code{ASSEMBLER}, search order capability
12978: The ANS Forth search order word set.
12979:
12980: @item source and format of display by @code{SEE}:
12981: @cindex @code{SEE}, source and format of output
12982: The source for @code{see} is the intermediate code used by the inner
12983: interpreter. The current @code{see} tries to output Forth source code
12984: as well as possible.
12985:
12986: @end table
12987:
12988: @c ---------------------------------------------------------------------
12989: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
12990: @subsection Ambiguous conditions
12991: @c ---------------------------------------------------------------------
12992: @cindex programming-tools words, ambiguous conditions
12993: @cindex ambiguous conditions, programming-tools words
12994:
12995: @table @i
12996:
12997: @item deleting the compilation word list (@code{FORGET}):
12998: @cindex @code{FORGET}, deleting the compilation word list
12999: Not implemented (yet).
13000:
13001: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
13002: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
13003: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
13004: @cindex control-flow stack underflow
13005: This typically results in an @code{abort"} with a descriptive error
13006: message (may change into a @code{-22 throw} (Control structure mismatch)
13007: in the future). You may also get a memory access error. If you are
13008: unlucky, this ambiguous condition is not caught.
13009:
13010: @item @i{name} can't be found (@code{FORGET}):
13011: @cindex @code{FORGET}, @i{name} can't be found
13012: Not implemented (yet).
13013:
13014: @item @i{name} not defined via @code{CREATE}:
13015: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
13016: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
13017: the execution semantics of the last defined word no matter how it was
13018: defined.
13019:
13020: @item @code{POSTPONE} applied to @code{[IF]}:
13021: @cindex @code{POSTPONE} applied to @code{[IF]}
13022: @cindex @code{[IF]} and @code{POSTPONE}
13023: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
13024: equivalent to @code{[IF]}.
13025:
13026: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
13027: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
13028: Continue in the same state of conditional compilation in the next outer
13029: input source. Currently there is no warning to the user about this.
13030:
13031: @item removing a needed definition (@code{FORGET}):
13032: @cindex @code{FORGET}, removing a needed definition
13033: Not implemented (yet).
13034:
13035: @end table
13036:
13037:
13038: @c =====================================================================
13039: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
13040: @section The optional Search-Order word set
13041: @c =====================================================================
13042: @cindex system documentation, search-order words
13043: @cindex search-order words, system documentation
13044:
13045: @menu
13046: * search-idef:: Implementation Defined Options
13047: * search-ambcond:: Ambiguous Conditions
13048: @end menu
13049:
13050:
13051: @c ---------------------------------------------------------------------
13052: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
13053: @subsection Implementation Defined Options
13054: @c ---------------------------------------------------------------------
13055: @cindex implementation-defined options, search-order words
13056: @cindex search-order words, implementation-defined options
13057:
13058: @table @i
13059: @item maximum number of word lists in search order:
13060: @cindex maximum number of word lists in search order
13061: @cindex search order, maximum depth
13062: @code{s" wordlists" environment? drop .}. Currently 16.
13063:
13064: @item minimum search order:
13065: @cindex minimum search order
13066: @cindex search order, minimum
13067: @code{root root}.
13068:
13069: @end table
13070:
13071: @c ---------------------------------------------------------------------
13072: @node search-ambcond, , search-idef, The optional Search-Order word set
13073: @subsection Ambiguous conditions
13074: @c ---------------------------------------------------------------------
13075: @cindex search-order words, ambiguous conditions
13076: @cindex ambiguous conditions, search-order words
13077:
13078: @table @i
13079: @item changing the compilation word list (during compilation):
13080: @cindex changing the compilation word list (during compilation)
13081: @cindex compilation word list, change before definition ends
13082: The word is entered into the word list that was the compilation word list
13083: at the start of the definition. Any changes to the name field (e.g.,
13084: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
13085: are applied to the latest defined word (as reported by @code{last} or
13086: @code{lastxt}), if possible, irrespective of the compilation word list.
13087:
13088: @item search order empty (@code{previous}):
13089: @cindex @code{previous}, search order empty
13090: @cindex vocstack empty, @code{previous}
13091: @code{abort" Vocstack empty"}.
13092:
13093: @item too many word lists in search order (@code{also}):
13094: @cindex @code{also}, too many word lists in search order
13095: @cindex vocstack full, @code{also}
13096: @code{abort" Vocstack full"}.
13097:
13098: @end table
13099:
13100: @c ***************************************************************
13101: @node Standard vs Extensions, Model, ANS conformance, Top
13102: @chapter Should I use Gforth extensions?
13103: @cindex Gforth extensions
13104:
13105: As you read through the rest of this manual, you will see documentation
13106: for @i{Standard} words, and documentation for some appealing Gforth
13107: @i{extensions}. You might ask yourself the question: @i{``Should I
13108: restrict myself to the standard, or should I use the extensions?''}
13109:
13110: The answer depends on the goals you have for the program you are working
13111: on:
13112:
13113: @itemize @bullet
13114:
13115: @item Is it just for yourself or do you want to share it with others?
13116:
13117: @item
13118: If you want to share it, do the others all use Gforth?
13119:
13120: @item
13121: If it is just for yourself, do you want to restrict yourself to Gforth?
13122:
13123: @end itemize
13124:
13125: If restricting the program to Gforth is ok, then there is no reason not
13126: to use extensions. It is still a good idea to keep to the standard
13127: where it is easy, in case you want to reuse these parts in another
13128: program that you want to be portable.
13129:
13130: If you want to be able to port the program to other Forth systems, there
13131: are the following points to consider:
13132:
13133: @itemize @bullet
13134:
13135: @item
13136: Most Forth systems that are being maintained support the ANS Forth
13137: standard. So if your program complies with the standard, it will be
13138: portable among many systems.
13139:
13140: @item
13141: A number of the Gforth extensions can be implemented in ANS Forth using
13142: public-domain files provided in the @file{compat/} directory. These are
13143: mentioned in the text in passing. There is no reason not to use these
13144: extensions, your program will still be ANS Forth compliant; just include
13145: the appropriate compat files with your program.
13146:
13147: @item
13148: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
13149: analyse your program and determine what non-Standard words it relies
13150: upon. However, it does not check whether you use standard words in a
13151: non-standard way.
13152:
13153: @item
13154: Some techniques are not standardized by ANS Forth, and are hard or
13155: impossible to implement in a standard way, but can be implemented in
13156: most Forth systems easily, and usually in similar ways (e.g., accessing
13157: word headers). Forth has a rich historical precedent for programmers
13158: taking advantage of implementation-dependent features of their tools
13159: (for example, relying on a knowledge of the dictionary
13160: structure). Sometimes these techniques are necessary to extract every
13161: last bit of performance from the hardware, sometimes they are just a
13162: programming shorthand.
13163:
13164: @item
13165: Does using a Gforth extension save more work than the porting this part
13166: to other Forth systems (if any) will cost?
13167:
13168: @item
13169: Is the additional functionality worth the reduction in portability and
13170: the additional porting problems?
13171:
13172: @end itemize
13173:
13174: In order to perform these consideratios, you need to know what's
13175: standard and what's not. This manual generally states if something is
13176: non-standard, but the authoritative source is the standard document.
13177: Appendix A of the Standard (@var{Rationale}) provides a valuable insight
13178: into the thought processes of the technical committee.
13179:
13180: Note also that portability between Forth systems is not the only
13181: portability issue; there is also the issue of portability between
13182: different platforms (processor/OS combinations).
13183:
13184: @c ***************************************************************
13185: @node Model, Integrating Gforth, Standard vs Extensions, Top
13186: @chapter Model
13187:
13188: This chapter has yet to be written. It will contain information, on
13189: which internal structures you can rely.
13190:
13191: @c ***************************************************************
13192: @node Integrating Gforth, Emacs and Gforth, Model, Top
13193: @chapter Integrating Gforth into C programs
13194:
13195: This is not yet implemented.
13196:
13197: Several people like to use Forth as scripting language for applications
13198: that are otherwise written in C, C++, or some other language.
13199:
13200: The Forth system ATLAST provides facilities for embedding it into
13201: applications; unfortunately it has several disadvantages: most
13202: importantly, it is not based on ANS Forth, and it is apparently dead
13203: (i.e., not developed further and not supported). The facilities
13204: provided by Gforth in this area are inspired by ATLAST's facilities, so
13205: making the switch should not be hard.
13206:
13207: We also tried to design the interface such that it can easily be
13208: implemented by other Forth systems, so that we may one day arrive at a
13209: standardized interface. Such a standard interface would allow you to
13210: replace the Forth system without having to rewrite C code.
13211:
13212: You embed the Gforth interpreter by linking with the library
13213: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
13214: global symbols in this library that belong to the interface, have the
13215: prefix @code{forth_}. (Global symbols that are used internally have the
13216: prefix @code{gforth_}).
13217:
13218: You can include the declarations of Forth types and the functions and
13219: variables of the interface with @code{#include <forth.h>}.
13220:
13221: Types.
13222:
13223: Variables.
13224:
13225: Data and FP Stack pointer. Area sizes.
13226:
13227: functions.
13228:
13229: forth_init(imagefile)
13230: forth_evaluate(string) exceptions?
13231: forth_goto(address) (or forth_execute(xt)?)
13232: forth_continue() (a corountining mechanism)
13233:
13234: Adding primitives.
13235:
13236: No checking.
13237:
13238: Signals?
13239:
13240: Accessing the Stacks
13241:
13242: @c ******************************************************************
13243: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
13244: @chapter Emacs and Gforth
13245: @cindex Emacs and Gforth
13246:
13247: @cindex @file{gforth.el}
13248: @cindex @file{forth.el}
13249: @cindex Rydqvist, Goran
13250: @cindex comment editing commands
13251: @cindex @code{\}, editing with Emacs
13252: @cindex debug tracer editing commands
13253: @cindex @code{~~}, removal with Emacs
13254: @cindex Forth mode in Emacs
13255: Gforth comes with @file{gforth.el}, an improved version of
13256: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
13257: improvements are:
13258:
13259: @itemize @bullet
13260: @item
13261: A better (but still not perfect) handling of indentation.
13262: @item
13263: Comment paragraph filling (@kbd{M-q})
13264: @item
13265: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
13266: @item
13267: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
13268: @item
13269: Support of the @code{info-lookup} feature for looking up the
13270: documentation of a word.
13271: @end itemize
13272:
13273: I left the stuff I do not use alone, even though some of it only makes
13274: sense for TILE. To get a description of these features, enter Forth mode
13275: and type @kbd{C-h m}.
13276:
13277: @cindex source location of error or debugging output in Emacs
13278: @cindex error output, finding the source location in Emacs
13279: @cindex debugging output, finding the source location in Emacs
13280: In addition, Gforth supports Emacs quite well: The source code locations
13281: given in error messages, debugging output (from @code{~~}) and failed
13282: assertion messages are in the right format for Emacs' compilation mode
13283: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
13284: Manual}) so the source location corresponding to an error or other
13285: message is only a few keystrokes away (@kbd{C-x `} for the next error,
13286: @kbd{C-c C-c} for the error under the cursor).
13287:
13288: @cindex @file{TAGS} file
13289: @cindex @file{etags.fs}
13290: @cindex viewing the source of a word in Emacs
13291: @cindex @code{require}, placement in files
13292: @cindex @code{include}, placement in files
13293: Also, if you @code{require} @file{etags.fs}, a new @file{TAGS} file will
13294: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
13295: contains the definitions of all words defined afterwards. You can then
13296: find the source for a word using @kbd{M-.}. Note that emacs can use
13297: several tags files at the same time (e.g., one for the Gforth sources
13298: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
13299: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
13300: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
13301: @file{/usr/local/share/gforth/0.2.0/TAGS}). To get the best behaviour
13302: with @file{etags.fs}, you should avoid putting definitions both before
13303: and after @code{require} etc., otherwise you will see the same file
13304: visited several times by commands like @code{tags-search}.
13305:
13306: @cindex viewing the documentation of a word in Emacs
13307: @cindex context-sensitive help
13308: Moreover, for words documented in this manual, you can look up the
13309: glossary entry quickly by using @kbd{C-h TAB}
13310: (@code{info-lookup-symbol}, see @pxref{Documentation, ,Documentation
13311: Commands, emacs, Emacs Manual}). This feature requires Emacs 20.3 or
13312: later and does not work for words containing @code{:}.
13313:
13314:
13315: @cindex @file{.emacs}
13316: To get all these benefits, add the following lines to your @file{.emacs}
13317: file:
13318:
13319: @example
13320: (autoload 'forth-mode "gforth.el")
13321: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
13322: @end example
13323:
13324: @c ******************************************************************
13325: @node Image Files, Engine, Emacs and Gforth, Top
13326: @chapter Image Files
13327: @cindex image file
13328: @cindex @file{.fi} files
13329: @cindex precompiled Forth code
13330: @cindex dictionary in persistent form
13331: @cindex persistent form of dictionary
13332:
13333: An image file is a file containing an image of the Forth dictionary,
13334: i.e., compiled Forth code and data residing in the dictionary. By
13335: convention, we use the extension @code{.fi} for image files.
13336:
13337: @menu
13338: * Image Licensing Issues:: Distribution terms for images.
13339: * Image File Background:: Why have image files?
13340: * Non-Relocatable Image Files:: don't always work.
13341: * Data-Relocatable Image Files:: are better.
13342: * Fully Relocatable Image Files:: better yet.
13343: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
13344: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
13345: * Modifying the Startup Sequence:: and turnkey applications.
13346: @end menu
13347:
13348: @node Image Licensing Issues, Image File Background, Image Files, Image Files
13349: @section Image Licensing Issues
13350: @cindex license for images
13351: @cindex image license
13352:
13353: An image created with @code{gforthmi} (@pxref{gforthmi}) or
13354: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
13355: original image; i.e., according to copyright law it is a derived work of
13356: the original image.
13357:
13358: Since Gforth is distributed under the GNU GPL, the newly created image
13359: falls under the GNU GPL, too. In particular, this means that if you
13360: distribute the image, you have to make all of the sources for the image
13361: available, including those you wrote. For details see @ref{License, ,
13362: GNU General Public License (Section 3)}.
13363:
13364: If you create an image with @code{cross} (@pxref{cross.fs}), the image
13365: contains only code compiled from the sources you gave it; if none of
13366: these sources is under the GPL, the terms discussed above do not apply
13367: to the image. However, if your image needs an engine (a gforth binary)
13368: that is under the GPL, you should make sure that you distribute both in
13369: a way that is at most a @emph{mere aggregation}, if you don't want the
13370: terms of the GPL to apply to the image.
13371:
13372: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
13373: @section Image File Background
13374: @cindex image file background
13375:
13376: Our Forth system consists not only of primitives, but also of
13377: definitions written in Forth. Since the Forth compiler itself belongs to
13378: those definitions, it is not possible to start the system with the
13379: primitives and the Forth source alone. Therefore we provide the Forth
13380: code as an image file in nearly executable form. When Gforth starts up,
13381: a C routine loads the image file into memory, optionally relocates the
13382: addresses, then sets up the memory (stacks etc.) according to
13383: information in the image file, and (finally) starts executing Forth
13384: code.
13385:
13386: The image file variants represent different compromises between the
13387: goals of making it easy to generate image files and making them
13388: portable.
13389:
13390: @cindex relocation at run-time
13391: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
13392: run-time. This avoids many of the complications discussed below (image
13393: files are data relocatable without further ado), but costs performance
13394: (one addition per memory access).
13395:
13396: @cindex relocation at load-time
13397: By contrast, the Gforth loader performs relocation at image load time. The
13398: loader also has to replace tokens that represent primitive calls with the
13399: appropriate code-field addresses (or code addresses in the case of
13400: direct threading).
13401:
13402: There are three kinds of image files, with different degrees of
13403: relocatability: non-relocatable, data-relocatable, and fully relocatable
13404: image files.
13405:
13406: @cindex image file loader
13407: @cindex relocating loader
13408: @cindex loader for image files
13409: These image file variants have several restrictions in common; they are
13410: caused by the design of the image file loader:
13411:
13412: @itemize @bullet
13413: @item
13414: There is only one segment; in particular, this means, that an image file
13415: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
13416: them). The contents of the stacks are not represented, either.
13417:
13418: @item
13419: The only kinds of relocation supported are: adding the same offset to
13420: all cells that represent data addresses; and replacing special tokens
13421: with code addresses or with pieces of machine code.
13422:
13423: If any complex computations involving addresses are performed, the
13424: results cannot be represented in the image file. Several applications that
13425: use such computations come to mind:
13426: @itemize @minus
13427: @item
13428: Hashing addresses (or data structures which contain addresses) for table
13429: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
13430: purpose, you will have no problem, because the hash tables are
13431: recomputed automatically when the system is started. If you use your own
13432: hash tables, you will have to do something similar.
13433:
13434: @item
13435: There's a cute implementation of doubly-linked lists that uses
13436: @code{XOR}ed addresses. You could represent such lists as singly-linked
13437: in the image file, and restore the doubly-linked representation on
13438: startup.@footnote{In my opinion, though, you should think thrice before
13439: using a doubly-linked list (whatever implementation).}
13440:
13441: @item
13442: The code addresses of run-time routines like @code{docol:} cannot be
13443: represented in the image file (because their tokens would be replaced by
13444: machine code in direct threaded implementations). As a workaround,
13445: compute these addresses at run-time with @code{>code-address} from the
13446: executions tokens of appropriate words (see the definitions of
13447: @code{docol:} and friends in @file{kernel.fs}).
13448:
13449: @item
13450: On many architectures addresses are represented in machine code in some
13451: shifted or mangled form. You cannot put @code{CODE} words that contain
13452: absolute addresses in this form in a relocatable image file. Workarounds
13453: are representing the address in some relative form (e.g., relative to
13454: the CFA, which is present in some register), or loading the address from
13455: a place where it is stored in a non-mangled form.
13456: @end itemize
13457: @end itemize
13458:
13459: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
13460: @section Non-Relocatable Image Files
13461: @cindex non-relocatable image files
13462: @cindex image file, non-relocatable
13463:
13464: These files are simple memory dumps of the dictionary. They are specific
13465: to the executable (i.e., @file{gforth} file) they were created
13466: with. What's worse, they are specific to the place on which the
13467: dictionary resided when the image was created. Now, there is no
13468: guarantee that the dictionary will reside at the same place the next
13469: time you start Gforth, so there's no guarantee that a non-relocatable
13470: image will work the next time (Gforth will complain instead of crashing,
13471: though).
13472:
13473: You can create a non-relocatable image file with
13474:
13475:
13476: doc-savesystem
13477:
13478:
13479: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
13480: @section Data-Relocatable Image Files
13481: @cindex data-relocatable image files
13482: @cindex image file, data-relocatable
13483:
13484: These files contain relocatable data addresses, but fixed code addresses
13485: (instead of tokens). They are specific to the executable (i.e.,
13486: @file{gforth} file) they were created with. For direct threading on some
13487: architectures (e.g., the i386), data-relocatable images do not work. You
13488: get a data-relocatable image, if you use @file{gforthmi} with a
13489: Gforth binary that is not doubly indirect threaded (@pxref{Fully
13490: Relocatable Image Files}).
13491:
13492: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
13493: @section Fully Relocatable Image Files
13494: @cindex fully relocatable image files
13495: @cindex image file, fully relocatable
13496:
13497: @cindex @file{kern*.fi}, relocatability
13498: @cindex @file{gforth.fi}, relocatability
13499: These image files have relocatable data addresses, and tokens for code
13500: addresses. They can be used with different binaries (e.g., with and
13501: without debugging) on the same machine, and even across machines with
13502: the same data formats (byte order, cell size, floating point
13503: format). However, they are usually specific to the version of Gforth
13504: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
13505: are fully relocatable.
13506:
13507: There are two ways to create a fully relocatable image file:
13508:
13509: @menu
13510: * gforthmi:: The normal way
13511: * cross.fs:: The hard way
13512: @end menu
13513:
13514: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
13515: @subsection @file{gforthmi}
13516: @cindex @file{comp-i.fs}
13517: @cindex @file{gforthmi}
13518:
13519: You will usually use @file{gforthmi}. If you want to create an
13520: image @i{file} that contains everything you would load by invoking
13521: Gforth with @code{gforth @i{options}}, you simply say:
13522: @example
13523: gforthmi @i{file} @i{options}
13524: @end example
13525:
13526: E.g., if you want to create an image @file{asm.fi} that has the file
13527: @file{asm.fs} loaded in addition to the usual stuff, you could do it
13528: like this:
13529:
13530: @example
13531: gforthmi asm.fi asm.fs
13532: @end example
13533:
13534: @file{gforthmi} is implemented as a sh script and works like this: It
13535: produces two non-relocatable images for different addresses and then
13536: compares them. Its output reflects this: first you see the output (if
13537: any) of the two Gforth invocations that produce the non-relocatable image
13538: files, then you see the output of the comparing program: It displays the
13539: offset used for data addresses and the offset used for code addresses;
13540: moreover, for each cell that cannot be represented correctly in the
13541: image files, it displays a line like this:
13542:
13543: @example
13544: 78DC BFFFFA50 BFFFFA40
13545: @end example
13546:
13547: This means that at offset $78dc from @code{forthstart}, one input image
13548: contains $bffffa50, and the other contains $bffffa40. Since these cells
13549: cannot be represented correctly in the output image, you should examine
13550: these places in the dictionary and verify that these cells are dead
13551: (i.e., not read before they are written).
13552:
13553: @cindex --application, @code{gforthmi} option
13554: If you insert the option @code{--application} in front of the image file
13555: name, you will get an image that uses the @code{--appl-image} option
13556: instead of the @code{--image-file} option (@pxref{Invoking
13557: Gforth}). When you execute such an image on Unix (by typing the image
13558: name as command), the Gforth engine will pass all options to the image
13559: instead of trying to interpret them as engine options.
13560:
13561: If you type @file{gforthmi} with no arguments, it prints some usage
13562: instructions.
13563:
13564: @cindex @code{savesystem} during @file{gforthmi}
13565: @cindex @code{bye} during @file{gforthmi}
13566: @cindex doubly indirect threaded code
13567: @cindex environment variables
13568: @cindex @code{GFORTHD} -- environment variable
13569: @cindex @code{GFORTH} -- environment variable
13570: @cindex @code{gforth-ditc}
13571: There are a few wrinkles: After processing the passed @i{options}, the
13572: words @code{savesystem} and @code{bye} must be visible. A special doubly
13573: indirect threaded version of the @file{gforth} executable is used for
13574: creating the non-relocatable images; you can pass the exact filename of
13575: this executable through the environment variable @code{GFORTHD}
13576: (default: @file{gforth-ditc}); if you pass a version that is not doubly
13577: indirect threaded, you will not get a fully relocatable image, but a
13578: data-relocatable image (because there is no code address offset). The
13579: normal @file{gforth} executable is used for creating the relocatable
13580: image; you can pass the exact filename of this executable through the
13581: environment variable @code{GFORTH}.
13582:
13583: @node cross.fs, , gforthmi, Fully Relocatable Image Files
13584: @subsection @file{cross.fs}
13585: @cindex @file{cross.fs}
13586: @cindex cross-compiler
13587: @cindex metacompiler
13588: @cindex target compiler
13589:
13590: You can also use @code{cross}, a batch compiler that accepts a Forth-like
13591: programming language (@pxref{Cross Compiler}).
13592:
13593: @code{cross} allows you to create image files for machines with
13594: different data sizes and data formats than the one used for generating
13595: the image file. You can also use it to create an application image that
13596: does not contain a Forth compiler. These features are bought with
13597: restrictions and inconveniences in programming. E.g., addresses have to
13598: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
13599: order to make the code relocatable.
13600:
13601:
13602: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
13603: @section Stack and Dictionary Sizes
13604: @cindex image file, stack and dictionary sizes
13605: @cindex dictionary size default
13606: @cindex stack size default
13607:
13608: If you invoke Gforth with a command line flag for the size
13609: (@pxref{Invoking Gforth}), the size you specify is stored in the
13610: dictionary. If you save the dictionary with @code{savesystem} or create
13611: an image with @file{gforthmi}, this size will become the default
13612: for the resulting image file. E.g., the following will create a
13613: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
13614:
13615: @example
13616: gforthmi gforth.fi -m 1M
13617: @end example
13618:
13619: In other words, if you want to set the default size for the dictionary
13620: and the stacks of an image, just invoke @file{gforthmi} with the
13621: appropriate options when creating the image.
13622:
13623: @cindex stack size, cache-friendly
13624: Note: For cache-friendly behaviour (i.e., good performance), you should
13625: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
13626: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
13627: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
13628:
13629: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
13630: @section Running Image Files
13631: @cindex running image files
13632: @cindex invoking image files
13633: @cindex image file invocation
13634:
13635: @cindex -i, invoke image file
13636: @cindex --image file, invoke image file
13637: You can invoke Gforth with an image file @i{image} instead of the
13638: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
13639: @example
13640: gforth -i @i{image}
13641: @end example
13642:
13643: @cindex executable image file
13644: @cindex image file, executable
13645: If your operating system supports starting scripts with a line of the
13646: form @code{#! ...}, you just have to type the image file name to start
13647: Gforth with this image file (note that the file extension @code{.fi} is
13648: just a convention). I.e., to run Gforth with the image file @i{image},
13649: you can just type @i{image} instead of @code{gforth -i @i{image}}.
13650: This works because every @code{.fi} file starts with a line of this
13651: format:
13652:
13653: @example
13654: #! /usr/local/bin/gforth-0.4.0 -i
13655: @end example
13656:
13657: The file and pathname for the Gforth engine specified on this line is
13658: the specific Gforth executable that it was built against; i.e. the value
13659: of the environment variable @code{GFORTH} at the time that
13660: @file{gforthmi} was executed.
13661:
13662: You can make use of the same shell capability to make a Forth source
13663: file into an executable. For example, if you place this text in a file:
13664:
13665: @example
13666: #! /usr/local/bin/gforth
13667:
13668: ." Hello, world" CR
13669: bye
13670: @end example
13671:
13672: @noindent
13673: and then make the file executable (chmod +x in Unix), you can run it
13674: directly from the command line. The sequence @code{#!} is used in two
13675: ways; firstly, it is recognised as a ``magic sequence'' by the operating
13676: system@footnote{The Unix kernel actually recognises two types of files:
13677: executable files and files of data, where the data is processed by an
13678: interpreter that is specified on the ``interpreter line'' -- the first
13679: line of the file, starting with the sequence #!. There may be a small
13680: limit (e.g., 32) on the number of characters that may be specified on
13681: the interpreter line.} secondly it is treated as a comment character by
13682: Gforth. Because of the second usage, a space is required between
13683: @code{#!} and the path to the executable.
13684:
13685: The disadvantage of this latter technique, compared with using
13686: @file{gforthmi}, is that it is slower; the Forth source code is compiled
13687: on-the-fly, each time the program is invoked.
13688:
13689:
13690: doc-#!
13691:
13692:
13693: @node Modifying the Startup Sequence, , Running Image Files, Image Files
13694: @section Modifying the Startup Sequence
13695: @cindex startup sequence for image file
13696: @cindex image file initialization sequence
13697: @cindex initialization sequence of image file
13698:
13699: You can add your own initialization to the startup sequence through the
13700: deferred word @code{'cold}. @code{'cold} is invoked just before the
13701: image-specific command line processing (by default, loading files and
13702: evaluating (@code{-e}) strings) starts.
13703:
13704: A sequence for adding your initialization usually looks like this:
13705:
13706: @example
13707: :noname
13708: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
13709: ... \ your stuff
13710: ; IS 'cold
13711: @end example
13712:
13713: @cindex turnkey image files
13714: @cindex image file, turnkey applications
13715: You can make a turnkey image by letting @code{'cold} execute a word
13716: (your turnkey application) that never returns; instead, it exits Gforth
13717: via @code{bye} or @code{throw}.
13718:
13719: @cindex command-line arguments, access
13720: @cindex arguments on the command line, access
13721: You can access the (image-specific) command-line arguments through the
13722: variables @code{argc} and @code{argv}. @code{arg} provides convenient
13723: access to @code{argv}.
13724:
13725: If @code{'cold} exits normally, Gforth processes the command-line
13726: arguments as files to be loaded and strings to be evaluated. Therefore,
13727: @code{'cold} should remove the arguments it has used in this case.
13728:
13729:
13730:
13731: doc-'cold
13732: doc-argc
13733: doc-argv
13734: doc-arg
13735:
13736:
13737:
13738: @c ******************************************************************
13739: @node Engine, Binding to System Library, Image Files, Top
13740: @chapter Engine
13741: @cindex engine
13742: @cindex virtual machine
13743:
13744: Reading this chapter is not necessary for programming with Gforth. It
13745: may be helpful for finding your way in the Gforth sources.
13746:
13747: The ideas in this section have also been published in Bernd Paysan,
13748: @cite{ANS fig/GNU/??? Forth} (in German), Forth-Tagung '93 and M. Anton
13749: Ertl, @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z, A
13750: Portable Forth Engine}}, EuroForth '93.
13751:
13752: @menu
13753: * Portability::
13754: * Threading::
13755: * Primitives::
13756: * Performance::
13757: @end menu
13758:
13759: @node Portability, Threading, Engine, Engine
13760: @section Portability
13761: @cindex engine portability
13762:
13763: An important goal of the Gforth Project is availability across a wide
13764: range of personal machines. fig-Forth, and, to a lesser extent, F83,
13765: achieved this goal by manually coding the engine in assembly language
13766: for several then-popular processors. This approach is very
13767: labor-intensive and the results are short-lived due to progress in
13768: computer architecture.
13769:
13770: @cindex C, using C for the engine
13771: Others have avoided this problem by coding in C, e.g., Mitch Bradley
13772: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
13773: particularly popular for UNIX-based Forths due to the large variety of
13774: architectures of UNIX machines. Unfortunately an implementation in C
13775: does not mix well with the goals of efficiency and with using
13776: traditional techniques: Indirect or direct threading cannot be expressed
13777: in C, and switch threading, the fastest technique available in C, is
13778: significantly slower. Another problem with C is that it is very
13779: cumbersome to express double integer arithmetic.
13780:
13781: @cindex GNU C for the engine
13782: @cindex long long
13783: Fortunately, there is a portable language that does not have these
13784: limitations: GNU C, the version of C processed by the GNU C compiler
13785: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
13786: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
13787: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
13788: threading possible, its @code{long long} type (@pxref{Long Long, ,
13789: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
13790: double numbers@footnote{Unfortunately, long longs are not implemented
13791: properly on all machines (e.g., on alpha-osf1, long longs are only 64
13792: bits, the same size as longs (and pointers), but they should be twice as
13793: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
13794: C Manual}). So, we had to implement doubles in C after all. Still, on
13795: most machines we can use long longs and achieve better performance than
13796: with the emulation package.}. GNU C is available for free on all
13797: important (and many unimportant) UNIX machines, VMS, 80386s running
13798: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
13799: on all these machines.
13800:
13801: Writing in a portable language has the reputation of producing code that
13802: is slower than assembly. For our Forth engine we repeatedly looked at
13803: the code produced by the compiler and eliminated most compiler-induced
13804: inefficiencies by appropriate changes in the source code.
13805:
13806: @cindex explicit register declarations
13807: @cindex --enable-force-reg, configuration flag
13808: @cindex -DFORCE_REG
13809: However, register allocation cannot be portably influenced by the
13810: programmer, leading to some inefficiencies on register-starved
13811: machines. We use explicit register declarations (@pxref{Explicit Reg
13812: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
13813: improve the speed on some machines. They are turned on by using the
13814: configuration flag @code{--enable-force-reg} (@code{gcc} switch
13815: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
13816: machine, but also on the compiler version: On some machines some
13817: compiler versions produce incorrect code when certain explicit register
13818: declarations are used. So by default @code{-DFORCE_REG} is not used.
13819:
13820: @node Threading, Primitives, Portability, Engine
13821: @section Threading
13822: @cindex inner interpreter implementation
13823: @cindex threaded code implementation
13824:
13825: @cindex labels as values
13826: GNU C's labels as values extension (available since @code{gcc-2.0},
13827: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
13828: makes it possible to take the address of @i{label} by writing
13829: @code{&&@i{label}}. This address can then be used in a statement like
13830: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
13831: @code{goto x}.
13832:
13833: @cindex @code{NEXT}, indirect threaded
13834: @cindex indirect threaded inner interpreter
13835: @cindex inner interpreter, indirect threaded
13836: With this feature an indirect threaded @code{NEXT} looks like:
13837: @example
13838: cfa = *ip++;
13839: ca = *cfa;
13840: goto *ca;
13841: @end example
13842: @cindex instruction pointer
13843: For those unfamiliar with the names: @code{ip} is the Forth instruction
13844: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
13845: execution token and points to the code field of the next word to be
13846: executed; The @code{ca} (code address) fetched from there points to some
13847: executable code, e.g., a primitive or the colon definition handler
13848: @code{docol}.
13849:
13850: @cindex @code{NEXT}, direct threaded
13851: @cindex direct threaded inner interpreter
13852: @cindex inner interpreter, direct threaded
13853: Direct threading is even simpler:
13854: @example
13855: ca = *ip++;
13856: goto *ca;
13857: @end example
13858:
13859: Of course we have packaged the whole thing neatly in macros called
13860: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
13861:
13862: @menu
13863: * Scheduling::
13864: * Direct or Indirect Threaded?::
13865: * DOES>::
13866: @end menu
13867:
13868: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
13869: @subsection Scheduling
13870: @cindex inner interpreter optimization
13871:
13872: There is a little complication: Pipelined and superscalar processors,
13873: i.e., RISC and some modern CISC machines can process independent
13874: instructions while waiting for the results of an instruction. The
13875: compiler usually reorders (schedules) the instructions in a way that
13876: achieves good usage of these delay slots. However, on our first tries
13877: the compiler did not do well on scheduling primitives. E.g., for
13878: @code{+} implemented as
13879: @example
13880: n=sp[0]+sp[1];
13881: sp++;
13882: sp[0]=n;
13883: NEXT;
13884: @end example
13885: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
13886: scheduling. After a little thought the problem becomes clear: The
13887: compiler cannot know that @code{sp} and @code{ip} point to different
13888: addresses (and the version of @code{gcc} we used would not know it even
13889: if it was possible), so it could not move the load of the cfa above the
13890: store to the TOS. Indeed the pointers could be the same, if code on or
13891: very near the top of stack were executed. In the interest of speed we
13892: chose to forbid this probably unused ``feature'' and helped the compiler
13893: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
13894: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
13895: @example
13896: n=sp[0]+sp[1];
13897: sp++;
13898: NEXT_P1;
13899: sp[0]=n;
13900: NEXT_P2;
13901: @end example
13902: This can be scheduled optimally by the compiler.
13903:
13904: This division can be turned off with the switch @code{-DCISC_NEXT}. This
13905: switch is on by default on machines that do not profit from scheduling
13906: (e.g., the 80386), in order to preserve registers.
13907:
13908: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
13909: @subsection Direct or Indirect Threaded?
13910: @cindex threading, direct or indirect?
13911:
13912: @cindex -DDIRECT_THREADED
13913: Both! After packaging the nasty details in macro definitions we
13914: realized that we could switch between direct and indirect threading by
13915: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
13916: defining a few machine-specific macros for the direct-threading case.
13917: On the Forth level we also offer access words that hide the
13918: differences between the threading methods (@pxref{Threading Words}).
13919:
13920: Indirect threading is implemented completely machine-independently.
13921: Direct threading needs routines for creating jumps to the executable
13922: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
13923: machine-dependent, but they do not amount to many source lines. Therefore,
13924: even porting direct threading to a new machine requires little effort.
13925:
13926: @cindex --enable-indirect-threaded, configuration flag
13927: @cindex --enable-direct-threaded, configuration flag
13928: The default threading method is machine-dependent. You can enforce a
13929: specific threading method when building Gforth with the configuration
13930: flag @code{--enable-direct-threaded} or
13931: @code{--enable-indirect-threaded}. Note that direct threading is not
13932: supported on all machines.
13933:
13934: @node DOES>, , Direct or Indirect Threaded?, Threading
13935: @subsection DOES>
13936: @cindex @code{DOES>} implementation
13937:
13938: @cindex @code{dodoes} routine
13939: @cindex @code{DOES>}-code
13940: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
13941: the chunk of code executed by every word defined by a
13942: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
13943: the Forth code to be executed, i.e. the code after the
13944: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
13945:
13946: In fig-Forth the code field points directly to the @code{dodoes} and the
13947: @code{DOES>}-code address is stored in the cell after the code address (i.e. at
13948: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
13949: the Forth-79 and all later standards, because in fig-Forth this address
13950: lies in the body (which is illegal in these standards). However, by
13951: making the code field larger for all words this solution becomes legal
13952: again. We use this approach for the indirect threaded version and for
13953: direct threading on some machines. Leaving a cell unused in most words
13954: is a bit wasteful, but on the machines we are targeting this is hardly a
13955: problem. The other reason for having a code field size of two cells is
13956: to avoid having different image files for direct and indirect threaded
13957: systems (direct threaded systems require two-cell code fields on many
13958: machines).
13959:
13960: @cindex @code{DOES>}-handler
13961: The other approach is that the code field points or jumps to the cell
13962: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
13963: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
13964: @code{DOES>}-code address by computing the code address, i.e., the address of
13965: the jump to @code{dodoes}, and add the length of that jump field. A variant of
13966: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
13967: return address (which can be found in the return register on RISCs) is
13968: the @code{DOES>}-code address. Since the two cells available in the code field
13969: are used up by the jump to the code address in direct threading on many
13970: architectures, we use this approach for direct threading on these
13971: architectures. We did not want to add another cell to the code field.
13972:
13973: @node Primitives, Performance, Threading, Engine
13974: @section Primitives
13975: @cindex primitives, implementation
13976: @cindex virtual machine instructions, implementation
13977:
13978: @menu
13979: * Automatic Generation::
13980: * TOS Optimization::
13981: * Produced code::
13982: @end menu
13983:
13984: @node Automatic Generation, TOS Optimization, Primitives, Primitives
13985: @subsection Automatic Generation
13986: @cindex primitives, automatic generation
13987:
13988: @cindex @file{prims2x.fs}
13989: Since the primitives are implemented in a portable language, there is no
13990: longer any need to minimize the number of primitives. On the contrary,
13991: having many primitives has an advantage: speed. In order to reduce the
13992: number of errors in primitives and to make programming them easier, we
13993: provide a tool, the primitive generator (@file{prims2x.fs}), that
13994: automatically generates most (and sometimes all) of the C code for a
13995: primitive from the stack effect notation. The source for a primitive
13996: has the following form:
13997:
13998: @cindex primitive source format
13999: @format
14000: @i{Forth-name} ( @i{stack-effect} ) @i{category} [@i{pronounc.}]
14001: [@code{""}@i{glossary entry}@code{""}]
14002: @i{C code}
14003: [@code{:}
14004: @i{Forth code}]
14005: @end format
14006:
14007: The items in brackets are optional. The category and glossary fields
14008: are there for generating the documentation, the Forth code is there
14009: for manual implementations on machines without GNU C. E.g., the source
14010: for the primitive @code{+} is:
14011: @example
14012: + ( n1 n2 -- n ) core plus
14013: n = n1+n2;
14014: @end example
14015:
14016: This looks like a specification, but in fact @code{n = n1+n2} is C
14017: code. Our primitive generation tool extracts a lot of information from
14018: the stack effect notations@footnote{We use a one-stack notation, even
14019: though we have separate data and floating-point stacks; The separate
14020: notation can be generated easily from the unified notation.}: The number
14021: of items popped from and pushed on the stack, their type, and by what
14022: name they are referred to in the C code. It then generates a C code
14023: prelude and postlude for each primitive. The final C code for @code{+}
14024: looks like this:
14025:
14026: @example
14027: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
14028: /* */ /* documentation */
14029: @{
14030: DEF_CA /* definition of variable ca (indirect threading) */
14031: Cell n1; /* definitions of variables */
14032: Cell n2;
14033: Cell n;
14034: n1 = (Cell) sp[1]; /* input */
14035: n2 = (Cell) TOS;
14036: sp += 1; /* stack adjustment */
14037: NAME("+") /* debugging output (with -DDEBUG) */
14038: @{
14039: n = n1+n2; /* C code taken from the source */
14040: @}
14041: NEXT_P1; /* NEXT part 1 */
14042: TOS = (Cell)n; /* output */
14043: NEXT_P2; /* NEXT part 2 */
14044: @}
14045: @end example
14046:
14047: This looks long and inefficient, but the GNU C compiler optimizes quite
14048: well and produces optimal code for @code{+} on, e.g., the R3000 and the
14049: HP RISC machines: Defining the @code{n}s does not produce any code, and
14050: using them as intermediate storage also adds no cost.
14051:
14052: There are also other optimizations that are not illustrated by this
14053: example: assignments between simple variables are usually for free (copy
14054: propagation). If one of the stack items is not used by the primitive
14055: (e.g. in @code{drop}), the compiler eliminates the load from the stack
14056: (dead code elimination). On the other hand, there are some things that
14057: the compiler does not do, therefore they are performed by
14058: @file{prims2x.fs}: The compiler does not optimize code away that stores
14059: a stack item to the place where it just came from (e.g., @code{over}).
14060:
14061: While programming a primitive is usually easy, there are a few cases
14062: where the programmer has to take the actions of the generator into
14063: account, most notably @code{?dup}, but also words that do not (always)
14064: fall through to @code{NEXT}.
14065:
14066: @node TOS Optimization, Produced code, Automatic Generation, Primitives
14067: @subsection TOS Optimization
14068: @cindex TOS optimization for primitives
14069: @cindex primitives, keeping the TOS in a register
14070:
14071: An important optimization for stack machine emulators, e.g., Forth
14072: engines, is keeping one or more of the top stack items in
14073: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
14074: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
14075: @itemize @bullet
14076: @item
14077: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
14078: due to fewer loads from and stores to the stack.
14079: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
14080: @i{y<n}, due to additional moves between registers.
14081: @end itemize
14082:
14083: @cindex -DUSE_TOS
14084: @cindex -DUSE_NO_TOS
14085: In particular, keeping one item in a register is never a disadvantage,
14086: if there are enough registers. Keeping two items in registers is a
14087: disadvantage for frequent words like @code{?branch}, constants,
14088: variables, literals and @code{i}. Therefore our generator only produces
14089: code that keeps zero or one items in registers. The generated C code
14090: covers both cases; the selection between these alternatives is made at
14091: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
14092: code for @code{+} is just a simple variable name in the one-item case,
14093: otherwise it is a macro that expands into @code{sp[0]}. Note that the
14094: GNU C compiler tries to keep simple variables like @code{TOS} in
14095: registers, and it usually succeeds, if there are enough registers.
14096:
14097: @cindex -DUSE_FTOS
14098: @cindex -DUSE_NO_FTOS
14099: The primitive generator performs the TOS optimization for the
14100: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
14101: operations the benefit of this optimization is even larger:
14102: floating-point operations take quite long on most processors, but can be
14103: performed in parallel with other operations as long as their results are
14104: not used. If the FP-TOS is kept in a register, this works. If
14105: it is kept on the stack, i.e., in memory, the store into memory has to
14106: wait for the result of the floating-point operation, lengthening the
14107: execution time of the primitive considerably.
14108:
14109: The TOS optimization makes the automatic generation of primitives a
14110: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
14111: @code{TOS} is not sufficient. There are some special cases to
14112: consider:
14113: @itemize @bullet
14114: @item In the case of @code{dup ( w -- w w )} the generator must not
14115: eliminate the store to the original location of the item on the stack,
14116: if the TOS optimization is turned on.
14117: @item Primitives with stack effects of the form @code{--}
14118: @i{out1}...@i{outy} must store the TOS to the stack at the start.
14119: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
14120: must load the TOS from the stack at the end. But for the null stack
14121: effect @code{--} no stores or loads should be generated.
14122: @end itemize
14123:
14124: @node Produced code, , TOS Optimization, Primitives
14125: @subsection Produced code
14126: @cindex primitives, assembly code listing
14127:
14128: @cindex @file{engine.s}
14129: To see what assembly code is produced for the primitives on your machine
14130: with your compiler and your flag settings, type @code{make engine.s} and
14131: look at the resulting file @file{engine.s}.
14132:
14133: @node Performance, , Primitives, Engine
14134: @section Performance
14135: @cindex performance of some Forth interpreters
14136: @cindex engine performance
14137: @cindex benchmarking Forth systems
14138: @cindex Gforth performance
14139:
14140: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
14141: impossible to write a significantly faster engine.
14142:
14143: On register-starved machines like the 386 architecture processors
14144: improvements are possible, because @code{gcc} does not utilize the
14145: registers as well as a human, even with explicit register declarations;
14146: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
14147: and hand-tuned it for the 486; this system is 1.19 times faster on the
14148: Sieve benchmark on a 486DX2/66 than Gforth compiled with
14149: @code{gcc-2.6.3} with @code{-DFORCE_REG}. The situation has improved
14150: with gcc-2.95 and gforth-0.4.9; now the most important virtual machine
14151: registers fit in real registers (and we can even afford to use the TOS
14152: optimization), resulting in a speedup of 1.14 on the sieve over the
14153: earlier results.
14154:
14155: @cindex Win32Forth performance
14156: @cindex NT Forth performance
14157: @cindex eforth performance
14158: @cindex ThisForth performance
14159: @cindex PFE performance
14160: @cindex TILE performance
14161: The potential advantage of assembly language implementations
14162: is not necessarily realized in complete Forth systems: We compared
14163: Gforth-0.4.9 (direct threaded, compiled with @code{gcc-2.95.1} and
14164: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
14165: 1994) and Eforth (with and without peephole (aka pinhole) optimization
14166: of the threaded code); all these systems were written in assembly
14167: language. We also compared Gforth with three systems written in C:
14168: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
14169: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
14170: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
14171: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
14172: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
14173: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
14174: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
14175: 486DX2/66 with similar memory performance under Windows NT. Marcel
14176: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
14177: added the peephole optimizer, ran the benchmarks and reported the
14178: results.
14179:
14180: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
14181: matrix multiplication come from the Stanford integer benchmarks and have
14182: been translated into Forth by Martin Fraeman; we used the versions
14183: included in the TILE Forth package, but with bigger data set sizes; and
14184: a recursive Fibonacci number computation for benchmarking calling
14185: performance. The following table shows the time taken for the benchmarks
14186: scaled by the time taken by Gforth (in other words, it shows the speedup
14187: factor that Gforth achieved over the other systems).
14188:
14189: @example
14190: relative Win32- NT eforth This-
14191: time Gforth Forth Forth eforth +opt PFE Forth TILE
14192: sieve 1.00 1.58 1.30 1.58 0.97 1.80 3.63 9.79
14193: bubble 1.00 1.55 1.67 1.75 1.04 1.78 4.59
14194: matmul 1.00 1.67 1.53 1.66 0.84 1.79 4.63
14195: fib 1.00 1.75 1.53 1.40 0.99 1.99 3.43 4.93
14196: @end example
14197:
14198: You may be quite surprised by the good performance of Gforth when
14199: compared with systems written in assembly language. One important reason
14200: for the disappointing performance of these other systems is probably
14201: that they are not written optimally for the 486 (e.g., they use the
14202: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
14203: but costly method for relocating the Forth image: like @code{cforth}, it
14204: computes the actual addresses at run time, resulting in two address
14205: computations per @code{NEXT} (@pxref{Image File Background}).
14206:
14207: Only Eforth with the peephole optimizer performs comparable to
14208: Gforth. The speedups achieved with peephole optimization of threaded
14209: code are quite remarkable. Adding a peephole optimizer to Gforth should
14210: cause similar speedups.
14211:
14212: The speedup of Gforth over PFE, ThisForth and TILE can be easily
14213: explained with the self-imposed restriction of the latter systems to
14214: standard C, which makes efficient threading impossible (however, the
14215: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
14216: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
14217: Moreover, current C compilers have a hard time optimizing other aspects
14218: of the ThisForth and the TILE source.
14219:
14220: The performance of Gforth on 386 architecture processors varies widely
14221: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
14222: allocate any of the virtual machine registers into real machine
14223: registers by itself and would not work correctly with explicit register
14224: declarations, giving a 1.5 times slower engine (on a 486DX2/66 running
14225: the Sieve) than the one measured above.
14226:
14227: Note that there have been several releases of Win32Forth since the
14228: release presented here, so the results presented above may have little
14229: predictive value for the performance of Win32Forth today (results for
14230: the current release on an i486DX2/66 are welcome).
14231:
14232: @cindex @file{Benchres}
14233: In
14234: @cite{@uref{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz,
14235: Translating Forth to Efficient C}} by M. Anton Ertl and Martin
14236: Maierhofer (presented at EuroForth '95), an indirect threaded version of
14237: Gforth is compared with Win32Forth, NT Forth, PFE, ThisForth, and
14238: several native code systems; that version of Gforth is slower on a 486
14239: than the direct threaded version used here. You can find a newer version
14240: of these measurements at
14241: @uref{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
14242: find numbers for Gforth on various machines in @file{Benchres}.
14243:
14244: @c ******************************************************************
14245: @node Binding to System Library, Cross Compiler, Engine, Top
14246: @chapter Binding to System Library
14247:
14248: @node Cross Compiler, Bugs, Binding to System Library, Top
14249: @chapter Cross Compiler
14250: @cindex @file{cross.fs}
14251: @cindex cross-compiler
14252: @cindex metacompiler
14253: @cindex target compiler
14254:
14255: The cross compiler is used to bootstrap a Forth kernel. Since Gforth is
14256: mostly written in Forth, including crucial parts like the outer
14257: interpreter and compiler, it needs compiled Forth code to get
14258: started. The cross compiler allows to create new images for other
14259: architectures, even running under another Forth system.
14260:
14261: @menu
14262: * Using the Cross Compiler::
14263: * How the Cross Compiler Works::
14264: @end menu
14265:
14266: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
14267: @section Using the Cross Compiler
14268:
14269: The cross compiler uses a language that resembles Forth, but isn't. The
14270: main difference is that you can execute Forth code after definition,
14271: while you usually can't execute the code compiled by cross, because the
14272: code you are compiling is typically for a different computer than the
14273: one you are compiling on.
14274:
14275: The Makefile is already set up to allow you to create kernels for new
14276: architectures with a simple make command. The generic kernels using the
14277: GCC compiled virtual machine are created in the normal build process
14278: with @code{make}. To create a embedded Gforth executable for e.g. the
14279: 8086 processor (running on a DOS machine), type
14280:
14281: @example
14282: make kernl-8086.fi
14283: @end example
14284:
14285: This will use the machine description from the @file{arch/8086}
14286: directory to create a new kernel. A machine file may look like that:
14287:
14288: @example
14289: \ Parameter for target systems 06oct92py
14290:
14291: 4 Constant cell \ cell size in bytes
14292: 2 Constant cell<< \ cell shift to bytes
14293: 5 Constant cell>bit \ cell shift to bits
14294: 8 Constant bits/char \ bits per character
14295: 8 Constant bits/byte \ bits per byte [default: 8]
14296: 8 Constant float \ bytes per float
14297: 8 Constant /maxalign \ maximum alignment in bytes
14298: false Constant bigendian \ byte order
14299: ( true=big, false=little )
14300:
14301: include machpc.fs \ feature list
14302: @end example
14303:
14304: This part is obligatory for the cross compiler itself, the feature list
14305: is used by the kernel to conditionally compile some features in and out,
14306: depending on whether the target supports these features.
14307:
14308: There are some optional features, if you define your own primitives,
14309: have an assembler, or need special, nonstandard preparation to make the
14310: boot process work. @code{asm-include} include an assembler,
14311: @code{prims-include} includes primitives, and @code{>boot} prepares for
14312: booting.
14313:
14314: @example
14315: : asm-include ." Include assembler" cr
14316: s" arch/8086/asm.fs" included ;
14317:
14318: : prims-include ." Include primitives" cr
14319: s" arch/8086/prim.fs" included ;
14320:
14321: : >boot ." Prepare booting" cr
14322: s" ' boot >body into-forth 1+ !" evaluate ;
14323: @end example
14324:
14325: These words are used as sort of macro during the cross compilation in
14326: the file @file{kernel/main.fs}. Instead of using this macros, it would
14327: be possible --- but more complicated --- to write a new kernel project
14328: file, too.
14329:
14330: @file{kernel/main.fs} expects the machine description file name on the
14331: stack; the cross compiler itself (@file{cross.fs}) assumes that either
14332: @code{mach-file} leaves a counted string on the stack, or
14333: @code{machine-file} leaves an address, count pair of the filename on the
14334: stack.
14335:
14336: The feature list is typically controlled using @code{SetValue}, generic
14337: files that are used by several projects can use @code{DefaultValue}
14338: instead. Both functions work like @code{Value}, when the value isn't
14339: defined, but @code{SetValue} works like @code{to} if the value is
14340: defined, and @code{DefaultValue} doesn't set anything, if the value is
14341: defined.
14342:
14343: @example
14344: \ generic mach file for pc gforth 03sep97jaw
14345:
14346: true DefaultValue NIL \ relocating
14347:
14348: >ENVIRON
14349:
14350: true DefaultValue file \ controls the presence of the
14351: \ file access wordset
14352: true DefaultValue OS \ flag to indicate a operating system
14353:
14354: true DefaultValue prims \ true: primitives are c-code
14355:
14356: true DefaultValue floating \ floating point wordset is present
14357:
14358: true DefaultValue glocals \ gforth locals are present
14359: \ will be loaded
14360: true DefaultValue dcomps \ double number comparisons
14361:
14362: true DefaultValue hash \ hashing primitives are loaded/present
14363:
14364: true DefaultValue xconds \ used together with glocals,
14365: \ special conditionals supporting gforths'
14366: \ local variables
14367: true DefaultValue header \ save a header information
14368:
14369: true DefaultValue backtrace \ enables backtrace code
14370:
14371: false DefaultValue ec
14372: false DefaultValue crlf
14373:
14374: cell 2 = [IF] &32 [ELSE] &256 [THEN] KB DefaultValue kernel-size
14375:
14376: &16 KB DefaultValue stack-size
14377: &15 KB &512 + DefaultValue fstack-size
14378: &15 KB DefaultValue rstack-size
14379: &14 KB &512 + DefaultValue lstack-size
14380: @end example
14381:
14382: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
14383: @section How the Cross Compiler Works
14384:
14385: @node Bugs, Origin, Cross Compiler, Top
14386: @appendix Bugs
14387: @cindex bug reporting
14388:
14389: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
14390:
14391: If you find a bug, please send a bug report to
14392: @email{bug-gforth@@gnu.org}. A bug report should include this
14393: information:
14394:
14395: @itemize @bullet
14396: @item
14397: The Gforth version used (it is announced at the start of an
14398: interactive Gforth session).
14399: @item
14400: The machine and operating system (on Unix
14401: systems @code{uname -a} will report this information).
14402: @item
14403: The installation options (send the file @file{config.status}).
14404: @item
14405: A complete list of changes (if any) you (or your installer) have made to the
14406: Gforth sources.
14407: @item
14408: A program (or a sequence of keyboard commands) that reproduces the bug.
14409: @item
14410: A description of what you think constitutes the buggy behaviour.
14411: @end itemize
14412:
14413: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
14414: to Report Bugs, gcc.info, GNU C Manual}.
14415:
14416:
14417: @node Origin, Forth-related information, Bugs, Top
14418: @appendix Authors and Ancestors of Gforth
14419:
14420: @section Authors and Contributors
14421: @cindex authors of Gforth
14422: @cindex contributors to Gforth
14423:
14424: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
14425: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
14426: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
14427: with their continuous feedback. Lennart Benshop contributed
14428: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
14429: support for calling C libraries. Helpful comments also came from Paul
14430: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
14431: Wavrik, Barrie Stott, Marc de Groot, Jorge Acerada, Bruce Hoyt, and
14432: Robert Epprecht. Since the release of Gforth-0.2.1 there were also
14433: helpful comments from many others; thank you all, sorry for not listing
14434: you here (but digging through my mailbox to extract your names is on my
14435: to-do list). Since the release of Gforth-0.4.0 Neal Crook worked on the
14436: manual.
14437:
14438: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
14439: and autoconf, among others), and to the creators of the Internet: Gforth
14440: was developed across the Internet, and its authors did not meet
14441: physically for the first 4 years of development.
14442:
14443: @section Pedigree
14444: @cindex pedigree of Gforth
14445:
14446: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
14447: Dirk Zoller) will cross-fertilize each other. Of course, a significant
14448: part of the design of Gforth was prescribed by ANS Forth.
14449:
14450: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
14451: 32 bit native code version of VolksForth for the Atari ST, written
14452: mostly by Dietrich Weineck.
14453:
14454: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
14455: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
14456: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
14457:
14458: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
14459: Forth-83 standard. !! Pedigree? When?
14460:
14461: A team led by Bill Ragsdale implemented fig-Forth on many processors in
14462: 1979. Robert Selzer and Bill Ragsdale developed the original
14463: implementation of fig-Forth for the 6502 based on microForth.
14464:
14465: The principal architect of microForth was Dean Sanderson. microForth was
14466: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
14467: the 1802, and subsequently implemented on the 8080, the 6800 and the
14468: Z80.
14469:
14470: All earlier Forth systems were custom-made, usually by Charles Moore,
14471: who discovered (as he puts it) Forth during the late 60s. The first full
14472: Forth existed in 1971.
14473:
14474: A part of the information in this section comes from @cite{The Evolution
14475: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
14476: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
14477: Notices 28(3), 1993. You can find more historical and genealogical
14478: information about Forth there.
14479:
14480: @node Forth-related information, Word Index, Origin, Top
14481: @appendix Other Forth-related information
14482: @cindex Forth-related information
14483:
14484: @menu
14485: * Internet resources::
14486: * Books::
14487: * The Forth Interest Group::
14488: * Conferences::
14489: @end menu
14490:
14491:
14492: @node Internet resources, Books, Forth-related information, Forth-related information
14493: @section Internet resources
14494: @cindex internet resources
14495:
14496: @cindex comp.lang.forth
14497: @cindex frequently asked questions
14498: There is an active news group (comp.lang.forth) discussing Forth and
14499: Forth-related issues. A frequently-asked-questions (FAQ) list
14500: is posted to the news group regularly, and archived at these sites:
14501:
14502: @itemize @bullet
14503: @item
14504: @uref{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
14505: @item
14506: @uref{ftp://ftp.forth.org/pub/Forth/FAQ/}
14507: @end itemize
14508:
14509: The FAQ list should be considered mandatory reading before posting to
14510: the news group.
14511:
14512: Here are some other web sites holding Forth-related material:
14513:
14514: @itemize @bullet
14515: @item
14516: @uref{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
14517: @item
14518: @uref{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
14519: @item
14520: @uref{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
14521: @item
14522: @uref{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
14523: Research page, including links to the Journal of Forth Application and
14524: Research (JFAR) and a searchable Forth bibliography.
14525: @end itemize
14526:
14527:
14528: @node Books, The Forth Interest Group, Internet resources, Forth-related information
14529: @section Books
14530: @cindex books on Forth
14531:
14532: As the Standard is relatively new, there are not many books out yet. It
14533: is not recommended to learn Forth by using Gforth and a book that is not
14534: written for ANS Forth, as you will not know your mistakes from the
14535: deviations of the book. However, books based on the Forth-83 standard
14536: should be ok, because ANS Forth is primarily an extension of Forth-83.
14537: Refer to the Forth FAQ for details of Forth-related books.
14538:
14539: @cindex standard document for ANS Forth
14540: @cindex ANS Forth document
14541: The definite reference if you want to write ANS Forth programs is, of
14542: course, the ANS Forth document. It is available in printed form from the
14543: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
14544: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
14545: $200. You can also get it from Global Engineering Documents (Tel.: USA
14546: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
14547:
14548: @cite{dpANS6}, the last draft of the standard, which was then submitted
14549: to ANSI for publication is available electronically and for free in some
14550: MS Word format, and it has been converted to HTML
14551: (@uref{http://www.taygeta.com/forth/dpans.html}; this HTML version also
14552: includes the answers to Requests for Interpretation (RFIs). Some
14553: pointers to these versions can be found through
14554: @*@uref{http://www.complang.tuwien.ac.at/projects/forth.html}.
14555:
14556:
14557: @node The Forth Interest Group, Conferences, Books, Forth-related information
14558: @section The Forth Interest Group
14559: @cindex Forth interest group (FIG)
14560:
14561: The Forth Interest Group (FIG) is a world-wide, non-profit,
14562: member-supported organisation. It publishes a regular magazine,
14563: @var{FORTH Dimensions}, and offers other benefits of membership. You can
14564: contact the FIG through their office email address:
14565: @email{office@@forth.org} or by visiting their web site at
14566: @uref{http://www.forth.org/}. This web site also includes links to FIG
14567: chapters in other countries and American cities
14568: (@uref{http://www.forth.org/chapters.html}).
14569:
14570: @node Conferences, , The Forth Interest Group, Forth-related information
14571: @section Conferences
14572: @cindex Conferences
14573:
14574: There are several regular conferences related to Forth. They are all
14575: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
14576: news group:
14577:
14578: @itemize @bullet
14579: @item
14580: FORML -- the Forth modification laboratory convenes every year near
14581: Monterey, California.
14582: @item
14583: The Rochester Forth Conference -- an annual conference traditionally
14584: held in Rochester, New York.
14585: @item
14586: EuroForth -- this European conference takes place annually.
14587: @end itemize
14588:
14589:
14590: @node Word Index, Name Index, Forth-related information, Top
14591: @unnumbered Word Index
14592:
14593: This index is a list of Forth words that have ``glossary'' entries
14594: within this manual. Each word is listed with its stack effect and
14595: wordset.
14596:
14597: @printindex fn
14598:
14599: @node Name Index, Concept Index, Word Index, Top
14600: @unnumbered Name Index
14601:
14602: This index is a list of Forth words that have ``glossary'' entries
14603: within this manual.
14604:
14605: @printindex ky
14606:
14607: @node Concept Index, , Name Index, Top
14608: @unnumbered Concept and Word Index
14609:
14610: Not all entries listed in this index are present verbatim in the
14611: text. This index also duplicates, in abbreviated form, all of the words
14612: listed in the Word Index (only the names are listed for the words here).
14613:
14614: @printindex cp
14615:
14616: @contents
14617: @bye
14618:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>