Annotation of gforth/doc/gforth.ds, revision 1.29
1.1 anton 1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
1.28 crook 3:
1.21 crook 4: @comment TODO: nac29jan99 - a list of things to add in the next edit:
1.28 crook 5: @comment 1. x-ref all ambiguous or implementation-defined features?
6: @comment 2. Describe the use of Auser Avariable AConstant A, etc.
7: @comment 3. words in miscellaneous section need a home.
8: @comment 4. search for TODO for other minor and major works required.
9: @comment 5. [rats] change all @var to @i in Forth source so that info
10: @comment file looks decent.
1.29 ! crook 11: @comment .. would be useful to have a word that identified all deferred words
! 12: @comment should semantics stuff in intro be moved to another section
! 13:
1.28 crook 14:
1.1 anton 15: @comment %**start of header (This is for running Texinfo on a region.)
16: @setfilename gforth.info
17: @settitle Gforth Manual
18: @dircategory GNU programming tools
19: @direntry
20: * Gforth: (gforth). A fast interpreter for the Forth language.
21: @end direntry
22: @comment @setchapternewpage odd
1.29 ! crook 23: @comment TODO this gets left in by HTML converter
1.12 anton 24: @macro progstyle {}
25: Programming style note:
1.3 anton 26: @end macro
1.1 anton 27: @comment %**end of header (This is for running Texinfo on a region.)
28:
1.29 ! crook 29:
! 30: @comment ----------------------------------------------------------
! 31: @comment macros for beautifying glossary entries
! 32: @comment if these are used, need to strip them out for HTML converter
! 33: @comment else they get repeated verbatim in HTML output.
! 34: @comment .. not working yet.
! 35:
! 36: @macro GLOSS-START {}
! 37: @iftex
! 38: @ninerm
! 39: @end iftex
! 40: @end macro
! 41:
! 42: @macro GLOSS-END {}
! 43: @iftex
! 44: @rm
! 45: @end iftex
! 46: @end macro
! 47:
! 48: @comment ----------------------------------------------------------
! 49:
! 50:
1.10 anton 51: @include version.texi
52:
1.1 anton 53: @ifinfo
1.11 anton 54: This file documents Gforth @value{VERSION}
1.1 anton 55:
1.26 crook 56: Copyright @copyright{} 1995-1999 Free Software Foundation, Inc.
1.1 anton 57:
58: Permission is granted to make and distribute verbatim copies of
59: this manual provided the copyright notice and this permission notice
60: are preserved on all copies.
61:
62: @ignore
63: Permission is granted to process this file through TeX and print the
64: results, provided the printed document carries a copying permission
65: notice identical to this one except for the removal of this paragraph
66: (this paragraph not being relevant to the printed manual).
67:
68: @end ignore
69: Permission is granted to copy and distribute modified versions of this
70: manual under the conditions for verbatim copying, provided also that the
71: sections entitled "Distribution" and "General Public License" are
72: included exactly as in the original, and provided that the entire
73: resulting derived work is distributed under the terms of a permission
74: notice identical to this one.
75:
76: Permission is granted to copy and distribute translations of this manual
77: into another language, under the above conditions for modified versions,
78: except that the sections entitled "Distribution" and "General Public
79: License" may be included in a translation approved by the author instead
80: of in the original English.
81: @end ifinfo
82:
83: @finalout
84: @titlepage
85: @sp 10
86: @center @titlefont{Gforth Manual}
87: @sp 2
1.11 anton 88: @center for version @value{VERSION}
1.1 anton 89: @sp 2
90: @center Anton Ertl
1.6 pazsan 91: @center Bernd Paysan
1.5 anton 92: @center Jens Wilke
1.23 crook 93: @center Neal Crook
1.1 anton 94: @sp 3
1.29 ! crook 95: @center This manual is permanently under construction and was last updated on 04-May-1999
1.1 anton 96:
97: @comment The following two commands start the copyright page.
98: @page
99: @vskip 0pt plus 1filll
1.29 ! crook 100: Copyright @copyright{} 1995--1999 Free Software Foundation, Inc.
1.1 anton 101:
102: @comment !! Published by ... or You can get a copy of this manual ...
103:
104: Permission is granted to make and distribute verbatim copies of
105: this manual provided the copyright notice and this permission notice
106: are preserved on all copies.
107:
108: Permission is granted to copy and distribute modified versions of this
109: manual under the conditions for verbatim copying, provided also that the
110: sections entitled "Distribution" and "General Public License" are
111: included exactly as in the original, and provided that the entire
112: resulting derived work is distributed under the terms of a permission
113: notice identical to this one.
114:
115: Permission is granted to copy and distribute translations of this manual
116: into another language, under the above conditions for modified versions,
117: except that the sections entitled "Distribution" and "General Public
118: License" may be included in a translation approved by the author instead
119: of in the original English.
120: @end titlepage
121:
122:
123: @node Top, License, (dir), (dir)
124: @ifinfo
125: Gforth is a free implementation of ANS Forth available on many
1.11 anton 126: personal machines. This manual corresponds to version @value{VERSION}.
1.1 anton 127: @end ifinfo
128:
129: @menu
1.21 crook 130: * License:: The GPL
1.26 crook 131: * Goals:: About the Gforth Project
1.29 ! crook 132: * Gforth Environment:: Starting (and exiting) Gforth
1.21 crook 133: * Introduction:: An introduction to ANS Forth
1.1 anton 134: * Words:: Forth words available in Gforth
1.24 anton 135: * Error messages:: How to interpret them
1.1 anton 136: * Tools:: Programming tools
137: * ANS conformance:: Implementation-defined options etc.
138: * Model:: The abstract machine of Gforth
139: * Integrating Gforth:: Forth as scripting language for applications
140: * Emacs and Gforth:: The Gforth Mode
141: * Image Files:: @code{.fi} files contain compiled code
142: * Engine:: The inner interpreter and the primitives
1.24 anton 143: * Binding to System Library::
1.13 pazsan 144: * Cross Compiler:: The Cross Compiler
1.1 anton 145: * Bugs:: How to report them
146: * Origin:: Authors and ancestors of Gforth
1.21 crook 147: * Forth-related information:: Books and places to look on the WWW
1.1 anton 148: * Word Index:: An item for each Forth word
149: * Concept Index:: A menu covering many topics
1.12 anton 150:
1.24 anton 151: @detailmenu --- The Detailed Node Listing ---
1.12 anton 152:
1.26 crook 153: Goals of Gforth
154:
155: * Gforth Extensions Sinful?::
156:
1.29 ! crook 157: Gforth Environment
! 158:
! 159: * Invoking Gforth::
! 160: * Leaving Gforth::
! 161: * Command-line editing::
! 162: * Upper and lower case::
! 163: * Environment variables::
! 164: * Gforth Files::
! 165:
1.24 anton 166: An Introduction to ANS Forth
167:
168: * Introducing the Text Interpreter::
169: * Stacks and Postfix notation::
170: * Your first definition::
171: * How does that work?::
172: * Forth is written in Forth::
173: * Review - elements of a Forth system::
1.29 ! crook 174: * Where to go next::
1.24 anton 175: * Exercises::
176:
1.12 anton 177: Forth Words
178:
179: * Notation::
1.21 crook 180: * Comments::
181: * Boolean Flags::
1.12 anton 182: * Arithmetic::
183: * Stack Manipulation::
184: * Memory::
185: * Control Structures::
186: * Defining Words::
1.21 crook 187: * The Text Interpreter::
1.12 anton 188: * Tokens for Words::
1.21 crook 189: * Word Lists::
190: * Environmental Queries::
1.12 anton 191: * Files::
192: * Blocks::
193: * Other I/O::
194: * Programming Tools::
195: * Assembler and Code Words::
196: * Threading Words::
1.26 crook 197: * Locals::
198: * Structures::
199: * Object-oriented Forth::
1.21 crook 200: * Passing Commands to the OS::
201: * Miscellaneous Words::
1.12 anton 202:
203: Arithmetic
204:
205: * Single precision::
206: * Bitwise operations::
1.21 crook 207: * Double precision:: Double-cell integer arithmetic
208: * Numeric comparison::
1.12 anton 209: * Mixed precision:: operations with single and double-cell integers
210: * Floating Point::
211:
212: Stack Manipulation
213:
214: * Data stack::
215: * Floating point stack::
216: * Return stack::
217: * Locals stack::
218: * Stack pointer manipulation::
219:
220: Memory
221:
1.27 crook 222: * Reserving Data Space::
1.12 anton 223: * Memory Access::
1.27 crook 224: * Address Arithmetic::
225: * Memory Blocks::
226: * Dynamic Allocation::
1.12 anton 227:
228: Control Structures
229:
230: * Selection::
231: * Simple Loops::
232: * Counted Loops::
233: * Arbitrary control structures::
234: * Calls and returns::
235: * Exception Handling::
236:
237: Defining Words
238:
239: * Simple Defining Words::
240: * Colon Definitions::
241: * User-defined Defining Words::
242: * Supplying names::
243: * Interpretation and Compilation Semantics::
244:
1.21 crook 245: The Text Interpreter
246:
1.29 ! crook 247: * Input Sources::
1.21 crook 248: * Number Conversion::
249: * Interpret/Compile states::
250: * Literals::
251: * Interpreter Directives::
252:
1.26 crook 253: Word Lists
254:
255: * Why use word lists?::
256: * Word list examples::
257:
258: Files
259:
260: * Forth source files::
261: * General files::
262: * Search Paths::
263: * Forth Search Paths::
264: * General Search Paths::
265:
266: Other I/O
267:
268: * Simple numeric output::
269: * Formatted numeric output::
270: * String Formats::
271: * Displaying characters and strings::
272: * Input::
273:
274: Programming Tools
275:
276: * Debugging:: Simple and quick.
277: * Assertions:: Making your programs self-checking.
278: * Singlestep Debugger:: Executing your program word by word.
279:
280: Locals
281:
282: * Gforth locals::
283: * ANS Forth locals::
284:
285: Gforth locals
286:
287: * Where are locals visible by name?::
288: * How long do locals live?::
289: * Programming Style::
290: * Implementation::
291:
1.12 anton 292: Structures
293:
294: * Why explicit structure support?::
295: * Structure Usage::
296: * Structure Naming Convention::
297: * Structure Implementation::
298: * Structure Glossary::
299:
300: Object-oriented Forth
301:
1.24 anton 302: * Why object-oriented programming?::
303: * Object-Oriented Terminology::
304: * Objects::
305: * OOF::
306: * Mini-OOF::
1.23 crook 307: * Comparison with other object models::
1.12 anton 308:
1.24 anton 309: The @file{objects.fs} model
1.12 anton 310:
311: * Properties of the Objects model::
312: * Basic Objects Usage::
1.23 crook 313: * The Objects base class::
1.12 anton 314: * Creating objects::
315: * Object-Oriented Programming Style::
316: * Class Binding::
317: * Method conveniences::
318: * Classes and Scoping::
319: * Object Interfaces::
320: * Objects Implementation::
321: * Objects Glossary::
322:
1.24 anton 323: The @file{oof.fs} model
1.12 anton 324:
325: * Properties of the OOF model::
326: * Basic OOF Usage::
1.23 crook 327: * The OOF base class::
1.12 anton 328: * Class Declaration::
329: * Class Implementation::
330:
1.24 anton 331: The @file{mini-oof.fs} model
1.23 crook 332:
333: * Basic Mini-OOF Usage::
334: * Mini-OOF Example::
335: * Mini-OOF Implementation::
336:
1.12 anton 337: Tools
338:
339: * ANS Report:: Report the words used, sorted by wordset.
340:
341: ANS conformance
342:
343: * The Core Words::
344: * The optional Block word set::
345: * The optional Double Number word set::
346: * The optional Exception word set::
347: * The optional Facility word set::
348: * The optional File-Access word set::
349: * The optional Floating-Point word set::
350: * The optional Locals word set::
351: * The optional Memory-Allocation word set::
352: * The optional Programming-Tools word set::
353: * The optional Search-Order word set::
354:
355: The Core Words
356:
357: * core-idef:: Implementation Defined Options
358: * core-ambcond:: Ambiguous Conditions
359: * core-other:: Other System Documentation
360:
361: The optional Block word set
362:
363: * block-idef:: Implementation Defined Options
364: * block-ambcond:: Ambiguous Conditions
365: * block-other:: Other System Documentation
366:
367: The optional Double Number word set
368:
369: * double-ambcond:: Ambiguous Conditions
370:
371: The optional Exception word set
372:
373: * exception-idef:: Implementation Defined Options
374:
375: The optional Facility word set
376:
377: * facility-idef:: Implementation Defined Options
378: * facility-ambcond:: Ambiguous Conditions
379:
380: The optional File-Access word set
381:
382: * file-idef:: Implementation Defined Options
383: * file-ambcond:: Ambiguous Conditions
384:
385: The optional Floating-Point word set
386:
387: * floating-idef:: Implementation Defined Options
388: * floating-ambcond:: Ambiguous Conditions
389:
390: The optional Locals word set
391:
392: * locals-idef:: Implementation Defined Options
393: * locals-ambcond:: Ambiguous Conditions
394:
395: The optional Memory-Allocation word set
396:
397: * memory-idef:: Implementation Defined Options
398:
399: The optional Programming-Tools word set
400:
401: * programming-idef:: Implementation Defined Options
402: * programming-ambcond:: Ambiguous Conditions
403:
404: The optional Search-Order word set
405:
406: * search-idef:: Implementation Defined Options
407: * search-ambcond:: Ambiguous Conditions
408:
409: Image Files
410:
1.24 anton 411: * Image Licensing Issues:: Distribution terms for images.
412: * Image File Background:: Why have image files?
413: * Non-Relocatable Image Files:: don't always work.
414: * Data-Relocatable Image Files:: are better.
1.12 anton 415: * Fully Relocatable Image Files:: better yet.
1.24 anton 416: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
417: * Running Image Files:: @code{gforth -i @var{file}} or @var{file}.
418: * Modifying the Startup Sequence:: and turnkey applications.
1.12 anton 419:
420: Fully Relocatable Image Files
421:
1.27 crook 422: * gforthmi:: The normal way
1.12 anton 423: * cross.fs:: The hard way
424:
425: Engine
426:
427: * Portability::
428: * Threading::
429: * Primitives::
430: * Performance::
431:
432: Threading
433:
434: * Scheduling::
435: * Direct or Indirect Threaded?::
436: * DOES>::
437:
438: Primitives
439:
440: * Automatic Generation::
441: * TOS Optimization::
442: * Produced code::
1.13 pazsan 443:
444: Cross Compiler
445:
446: * Using the Cross Compiler::
447: * How the Cross Compiler Works::
448:
1.24 anton 449: Other Forth-related information
1.21 crook 450:
451: * Internet resources::
452: * Books::
453: * The Forth Interest Group::
454: * Conferences::
455:
1.24 anton 456: @end detailmenu
1.1 anton 457: @end menu
458:
1.26 crook 459: @node License, Goals, Top, Top
1.1 anton 460: @unnumbered GNU GENERAL PUBLIC LICENSE
461: @center Version 2, June 1991
462:
463: @display
464: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
465: 675 Mass Ave, Cambridge, MA 02139, USA
466:
467: Everyone is permitted to copy and distribute verbatim copies
468: of this license document, but changing it is not allowed.
469: @end display
470:
471: @unnumberedsec Preamble
472:
473: The licenses for most software are designed to take away your
474: freedom to share and change it. By contrast, the GNU General Public
475: License is intended to guarantee your freedom to share and change free
476: software---to make sure the software is free for all its users. This
477: General Public License applies to most of the Free Software
478: Foundation's software and to any other program whose authors commit to
479: using it. (Some other Free Software Foundation software is covered by
480: the GNU Library General Public License instead.) You can apply it to
481: your programs, too.
482:
483: When we speak of free software, we are referring to freedom, not
484: price. Our General Public Licenses are designed to make sure that you
485: have the freedom to distribute copies of free software (and charge for
486: this service if you wish), that you receive source code or can get it
487: if you want it, that you can change the software or use pieces of it
488: in new free programs; and that you know you can do these things.
489:
490: To protect your rights, we need to make restrictions that forbid
491: anyone to deny you these rights or to ask you to surrender the rights.
492: These restrictions translate to certain responsibilities for you if you
493: distribute copies of the software, or if you modify it.
494:
495: For example, if you distribute copies of such a program, whether
496: gratis or for a fee, you must give the recipients all the rights that
497: you have. You must make sure that they, too, receive or can get the
498: source code. And you must show them these terms so they know their
499: rights.
500:
501: We protect your rights with two steps: (1) copyright the software, and
502: (2) offer you this license which gives you legal permission to copy,
503: distribute and/or modify the software.
504:
505: Also, for each author's protection and ours, we want to make certain
506: that everyone understands that there is no warranty for this free
507: software. If the software is modified by someone else and passed on, we
508: want its recipients to know that what they have is not the original, so
509: that any problems introduced by others will not reflect on the original
510: authors' reputations.
511:
512: Finally, any free program is threatened constantly by software
513: patents. We wish to avoid the danger that redistributors of a free
514: program will individually obtain patent licenses, in effect making the
515: program proprietary. To prevent this, we have made it clear that any
516: patent must be licensed for everyone's free use or not licensed at all.
517:
518: The precise terms and conditions for copying, distribution and
519: modification follow.
520:
521: @iftex
522: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
523: @end iftex
524: @ifinfo
525: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
526: @end ifinfo
527:
528: @enumerate 0
529: @item
530: This License applies to any program or other work which contains
531: a notice placed by the copyright holder saying it may be distributed
532: under the terms of this General Public License. The ``Program'', below,
533: refers to any such program or work, and a ``work based on the Program''
534: means either the Program or any derivative work under copyright law:
535: that is to say, a work containing the Program or a portion of it,
536: either verbatim or with modifications and/or translated into another
537: language. (Hereinafter, translation is included without limitation in
538: the term ``modification''.) Each licensee is addressed as ``you''.
539:
540: Activities other than copying, distribution and modification are not
541: covered by this License; they are outside its scope. The act of
542: running the Program is not restricted, and the output from the Program
543: is covered only if its contents constitute a work based on the
544: Program (independent of having been made by running the Program).
545: Whether that is true depends on what the Program does.
546:
547: @item
548: You may copy and distribute verbatim copies of the Program's
549: source code as you receive it, in any medium, provided that you
550: conspicuously and appropriately publish on each copy an appropriate
551: copyright notice and disclaimer of warranty; keep intact all the
552: notices that refer to this License and to the absence of any warranty;
553: and give any other recipients of the Program a copy of this License
554: along with the Program.
555:
556: You may charge a fee for the physical act of transferring a copy, and
557: you may at your option offer warranty protection in exchange for a fee.
558:
559: @item
560: You may modify your copy or copies of the Program or any portion
561: of it, thus forming a work based on the Program, and copy and
562: distribute such modifications or work under the terms of Section 1
563: above, provided that you also meet all of these conditions:
564:
565: @enumerate a
566: @item
567: You must cause the modified files to carry prominent notices
568: stating that you changed the files and the date of any change.
569:
570: @item
571: You must cause any work that you distribute or publish, that in
572: whole or in part contains or is derived from the Program or any
573: part thereof, to be licensed as a whole at no charge to all third
574: parties under the terms of this License.
575:
576: @item
577: If the modified program normally reads commands interactively
578: when run, you must cause it, when started running for such
579: interactive use in the most ordinary way, to print or display an
580: announcement including an appropriate copyright notice and a
581: notice that there is no warranty (or else, saying that you provide
582: a warranty) and that users may redistribute the program under
583: these conditions, and telling the user how to view a copy of this
584: License. (Exception: if the Program itself is interactive but
585: does not normally print such an announcement, your work based on
586: the Program is not required to print an announcement.)
587: @end enumerate
588:
589: These requirements apply to the modified work as a whole. If
590: identifiable sections of that work are not derived from the Program,
591: and can be reasonably considered independent and separate works in
592: themselves, then this License, and its terms, do not apply to those
593: sections when you distribute them as separate works. But when you
594: distribute the same sections as part of a whole which is a work based
595: on the Program, the distribution of the whole must be on the terms of
596: this License, whose permissions for other licensees extend to the
597: entire whole, and thus to each and every part regardless of who wrote it.
598:
599: Thus, it is not the intent of this section to claim rights or contest
600: your rights to work written entirely by you; rather, the intent is to
601: exercise the right to control the distribution of derivative or
602: collective works based on the Program.
603:
604: In addition, mere aggregation of another work not based on the Program
605: with the Program (or with a work based on the Program) on a volume of
606: a storage or distribution medium does not bring the other work under
607: the scope of this License.
608:
609: @item
610: You may copy and distribute the Program (or a work based on it,
611: under Section 2) in object code or executable form under the terms of
612: Sections 1 and 2 above provided that you also do one of the following:
613:
614: @enumerate a
615: @item
616: Accompany it with the complete corresponding machine-readable
617: source code, which must be distributed under the terms of Sections
618: 1 and 2 above on a medium customarily used for software interchange; or,
619:
620: @item
621: Accompany it with a written offer, valid for at least three
622: years, to give any third party, for a charge no more than your
623: cost of physically performing source distribution, a complete
624: machine-readable copy of the corresponding source code, to be
625: distributed under the terms of Sections 1 and 2 above on a medium
626: customarily used for software interchange; or,
627:
628: @item
629: Accompany it with the information you received as to the offer
630: to distribute corresponding source code. (This alternative is
631: allowed only for noncommercial distribution and only if you
632: received the program in object code or executable form with such
633: an offer, in accord with Subsection b above.)
634: @end enumerate
635:
636: The source code for a work means the preferred form of the work for
637: making modifications to it. For an executable work, complete source
638: code means all the source code for all modules it contains, plus any
639: associated interface definition files, plus the scripts used to
640: control compilation and installation of the executable. However, as a
641: special exception, the source code distributed need not include
642: anything that is normally distributed (in either source or binary
643: form) with the major components (compiler, kernel, and so on) of the
644: operating system on which the executable runs, unless that component
645: itself accompanies the executable.
646:
647: If distribution of executable or object code is made by offering
648: access to copy from a designated place, then offering equivalent
649: access to copy the source code from the same place counts as
650: distribution of the source code, even though third parties are not
651: compelled to copy the source along with the object code.
652:
653: @item
654: You may not copy, modify, sublicense, or distribute the Program
655: except as expressly provided under this License. Any attempt
656: otherwise to copy, modify, sublicense or distribute the Program is
657: void, and will automatically terminate your rights under this License.
658: However, parties who have received copies, or rights, from you under
659: this License will not have their licenses terminated so long as such
660: parties remain in full compliance.
661:
662: @item
663: You are not required to accept this License, since you have not
664: signed it. However, nothing else grants you permission to modify or
665: distribute the Program or its derivative works. These actions are
666: prohibited by law if you do not accept this License. Therefore, by
667: modifying or distributing the Program (or any work based on the
668: Program), you indicate your acceptance of this License to do so, and
669: all its terms and conditions for copying, distributing or modifying
670: the Program or works based on it.
671:
672: @item
673: Each time you redistribute the Program (or any work based on the
674: Program), the recipient automatically receives a license from the
675: original licensor to copy, distribute or modify the Program subject to
676: these terms and conditions. You may not impose any further
677: restrictions on the recipients' exercise of the rights granted herein.
678: You are not responsible for enforcing compliance by third parties to
679: this License.
680:
681: @item
682: If, as a consequence of a court judgment or allegation of patent
683: infringement or for any other reason (not limited to patent issues),
684: conditions are imposed on you (whether by court order, agreement or
685: otherwise) that contradict the conditions of this License, they do not
686: excuse you from the conditions of this License. If you cannot
687: distribute so as to satisfy simultaneously your obligations under this
688: License and any other pertinent obligations, then as a consequence you
689: may not distribute the Program at all. For example, if a patent
690: license would not permit royalty-free redistribution of the Program by
691: all those who receive copies directly or indirectly through you, then
692: the only way you could satisfy both it and this License would be to
693: refrain entirely from distribution of the Program.
694:
695: If any portion of this section is held invalid or unenforceable under
696: any particular circumstance, the balance of the section is intended to
697: apply and the section as a whole is intended to apply in other
698: circumstances.
699:
700: It is not the purpose of this section to induce you to infringe any
701: patents or other property right claims or to contest validity of any
702: such claims; this section has the sole purpose of protecting the
703: integrity of the free software distribution system, which is
704: implemented by public license practices. Many people have made
705: generous contributions to the wide range of software distributed
706: through that system in reliance on consistent application of that
707: system; it is up to the author/donor to decide if he or she is willing
708: to distribute software through any other system and a licensee cannot
709: impose that choice.
710:
711: This section is intended to make thoroughly clear what is believed to
712: be a consequence of the rest of this License.
713:
714: @item
715: If the distribution and/or use of the Program is restricted in
716: certain countries either by patents or by copyrighted interfaces, the
717: original copyright holder who places the Program under this License
718: may add an explicit geographical distribution limitation excluding
719: those countries, so that distribution is permitted only in or among
720: countries not thus excluded. In such case, this License incorporates
721: the limitation as if written in the body of this License.
722:
723: @item
724: The Free Software Foundation may publish revised and/or new versions
725: of the General Public License from time to time. Such new versions will
726: be similar in spirit to the present version, but may differ in detail to
727: address new problems or concerns.
728:
729: Each version is given a distinguishing version number. If the Program
730: specifies a version number of this License which applies to it and ``any
731: later version'', you have the option of following the terms and conditions
732: either of that version or of any later version published by the Free
733: Software Foundation. If the Program does not specify a version number of
734: this License, you may choose any version ever published by the Free Software
735: Foundation.
736:
737: @item
738: If you wish to incorporate parts of the Program into other free
739: programs whose distribution conditions are different, write to the author
740: to ask for permission. For software which is copyrighted by the Free
741: Software Foundation, write to the Free Software Foundation; we sometimes
742: make exceptions for this. Our decision will be guided by the two goals
743: of preserving the free status of all derivatives of our free software and
744: of promoting the sharing and reuse of software generally.
745:
746: @iftex
747: @heading NO WARRANTY
748: @end iftex
749: @ifinfo
750: @center NO WARRANTY
751: @end ifinfo
752:
753: @item
754: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
755: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
756: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
757: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
758: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
759: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
760: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
761: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
762: REPAIR OR CORRECTION.
763:
764: @item
765: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
766: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
767: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
768: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
769: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
770: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
771: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
772: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
773: POSSIBILITY OF SUCH DAMAGES.
774: @end enumerate
775:
776: @iftex
777: @heading END OF TERMS AND CONDITIONS
778: @end iftex
779: @ifinfo
780: @center END OF TERMS AND CONDITIONS
781: @end ifinfo
782:
783: @page
784: @unnumberedsec How to Apply These Terms to Your New Programs
785:
786: If you develop a new program, and you want it to be of the greatest
787: possible use to the public, the best way to achieve this is to make it
788: free software which everyone can redistribute and change under these terms.
789:
790: To do so, attach the following notices to the program. It is safest
791: to attach them to the start of each source file to most effectively
792: convey the exclusion of warranty; and each file should have at least
793: the ``copyright'' line and a pointer to where the full notice is found.
794:
795: @smallexample
796: @var{one line to give the program's name and a brief idea of what it does.}
797: Copyright (C) 19@var{yy} @var{name of author}
798:
799: This program is free software; you can redistribute it and/or modify
800: it under the terms of the GNU General Public License as published by
801: the Free Software Foundation; either version 2 of the License, or
802: (at your option) any later version.
803:
804: This program is distributed in the hope that it will be useful,
805: but WITHOUT ANY WARRANTY; without even the implied warranty of
806: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
807: GNU General Public License for more details.
808:
809: You should have received a copy of the GNU General Public License
810: along with this program; if not, write to the Free Software
811: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
812: @end smallexample
813:
814: Also add information on how to contact you by electronic and paper mail.
815:
816: If the program is interactive, make it output a short notice like this
817: when it starts in an interactive mode:
818:
819: @smallexample
820: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
821: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
822: type `show w'.
823: This is free software, and you are welcome to redistribute it
824: under certain conditions; type `show c' for details.
825: @end smallexample
826:
827: The hypothetical commands @samp{show w} and @samp{show c} should show
828: the appropriate parts of the General Public License. Of course, the
829: commands you use may be called something other than @samp{show w} and
830: @samp{show c}; they could even be mouse-clicks or menu items---whatever
831: suits your program.
832:
833: You should also get your employer (if you work as a programmer) or your
834: school, if any, to sign a ``copyright disclaimer'' for the program, if
835: necessary. Here is a sample; alter the names:
836:
837: @smallexample
838: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
839: `Gnomovision' (which makes passes at compilers) written by James Hacker.
840:
841: @var{signature of Ty Coon}, 1 April 1989
842: Ty Coon, President of Vice
843: @end smallexample
844:
845: This General Public License does not permit incorporating your program into
846: proprietary programs. If your program is a subroutine library, you may
847: consider it more useful to permit linking proprietary applications with the
848: library. If this is what you want to do, use the GNU Library General
849: Public License instead of this License.
850:
851: @iftex
852: @unnumbered Preface
853: @cindex Preface
1.21 crook 854: This manual documents Gforth. Some introductory material is provided for
855: readers who are unfamiliar with Forth or who are migrating to Gforth
856: from other Forth compilers. However, this manual is primarily a
857: reference manual.
1.1 anton 858: @end iftex
859:
1.28 crook 860: @comment TODO much more blurb here.
1.26 crook 861:
862: @c ******************************************************************
1.29 ! crook 863: @node Goals, Gforth Environment, License, Top
1.26 crook 864: @comment node-name, next, previous, up
865: @chapter Goals of Gforth
866: @cindex goals of the Gforth project
867: The goal of the Gforth Project is to develop a standard model for
868: ANS Forth. This can be split into several subgoals:
869:
870: @itemize @bullet
871: @item
872: Gforth should conform to the ANS Forth Standard.
873: @item
874: It should be a model, i.e. it should define all the
875: implementation-dependent things.
876: @item
877: It should become standard, i.e. widely accepted and used. This goal
878: is the most difficult one.
879: @end itemize
880:
881: To achieve these goals Gforth should be
882: @itemize @bullet
883: @item
884: Similar to previous models (fig-Forth, F83)
885: @item
886: Powerful. It should provide for all the things that are considered
887: necessary today and even some that are not yet considered necessary.
888: @item
889: Efficient. It should not get the reputation of being exceptionally
890: slow.
891: @item
892: Free.
893: @item
894: Available on many machines/easy to port.
895: @end itemize
896:
897: Have we achieved these goals? Gforth conforms to the ANS Forth
898: standard. It may be considered a model, but we have not yet documented
899: which parts of the model are stable and which parts we are likely to
900: change. It certainly has not yet become a de facto standard, but it
901: appears to be quite popular. It has some similarities to and some
902: differences from previous models. It has some powerful features, but not
903: yet everything that we envisioned. We certainly have achieved our
904: execution speed goals (@pxref{Performance}). It is free and available
905: on many machines.
906:
907: @menu
908: * Gforth Extensions Sinful?::
909: @end menu
910:
911: @node Gforth Extensions Sinful?, , Goals, Goals
912: @comment node-name, next, previous, up
913: @section Is it a Sin to use Gforth Extensions?
914: @cindex Gforth extensions
915:
916: If you've been paying attention, you will have realised that there is an
917: ANS (American National Standard) for Forth. As you read through the rest
1.29 ! crook 918: of this manual, you will see documentation for @i{Standard} words, and
! 919: documentation for some appealing Gforth @i{extensions}. You might ask
! 920: yourself the question: @i{``Given that there is a standard, would I be
1.26 crook 921: committing a sin to use (non-Standard) Gforth extensions?''}
922:
923: The answer to that question is somewhat pragmatic and somewhat
924: philosophical. Consider these points:
925:
926: @itemize @bullet
927: @item
928: A number of the Gforth extensions can be implemented in ANS Forth using
929: files provided in the @file{compat/} directory. These are mentioned in
930: the text in passing.
931: @item
932: Forth has a rich historical precedent for programmers taking advantage
933: of implementation-dependent features of their tools (for example,
934: relying on a knowledge of the dictionary structure). Sometimes these
935: techniques are necessary to extract every last bit of performance from
936: the hardware, sometimes they are just a programming shorthand.
937: @item
938: The best way to break the rules is to know what the rules are. To learn
939: the rules, there is no substitute for studying the text of the Standard
940: itself. In particular, Appendix A of the Standard (@var{Rationale})
941: provides a valuable insight into the thought processes of the technical
942: committee.
943: @item
944: The best reason to break a rule is because you have to; because it's
945: more productive to do that, because it makes your code run fast enough
946: or because you can see no Standard way to achieve what you want to
947: achieve.
948: @end itemize
949:
950: The tool @file{ans-report.fs} (@pxref{ANS Report}) makes it easy to
951: analyse your program and determine what non-Standard definitions it
952: relies upon.
953:
1.29 ! crook 954:
1.26 crook 955: @c ******************************************************************
1.29 ! crook 956: @node Gforth Environment, Introduction, Goals, Top
! 957: @chapter Gforth Environment
! 958: @cindex Gforth environment
1.21 crook 959:
1.29 ! crook 960: Note: ultimately, the gforth man page will be auto-generated from the
! 961: material in this chapter.
1.21 crook 962:
963: @menu
1.29 ! crook 964: * Invoking Gforth:: Getting in
! 965: * Leaving Gforth:: Getting out
! 966: * Command-line editing::
! 967: * Upper and lower case::
! 968: * Environment variables:: ..that affect how Gforth starts up
! 969: * Gforth Files:: What gets installed and where
1.21 crook 970: @end menu
971:
1.29 ! crook 972:
1.21 crook 973: @comment ----------------------------------------------
1.29 ! crook 974: @node Invoking Gforth, Leaving Gforth, ,Gforth Environment
! 975: @section Invoking Gforth
! 976: @cindex invoking Gforth
! 977: @cindex running Gforth
! 978: @cindex command-line options
! 979: @cindex options on the command line
! 980: @cindex flags on the command line
1.21 crook 981:
1.29 ! crook 982: Gforth is made up of two parts; an executable ``engine'' (named gforth)
! 983: and an image file. To start it, you will usually just say @code{gforth}
! 984: -- this automatically loads the default image file. In many other cases
! 985: the default Gforth image will be invoked like this:
1.21 crook 986: @example
1.29 ! crook 987: gforth [files] [-e forth-code]
1.21 crook 988: @end example
1.29 ! crook 989: @noindent
! 990: This interprets the contents of the files and the Forth code in the order they
! 991: are given.
1.21 crook 992:
1.29 ! crook 993: In general, the command line looks like this:
1.21 crook 994:
995: @example
1.29 ! crook 996: gforth [initialization options] [image-specific options]
1.21 crook 997: @end example
998:
1.29 ! crook 999: The initialization options must come before the rest of the command
! 1000: line. They are:
1.26 crook 1001:
1.29 ! crook 1002: @table @code
! 1003: @cindex -i, command-line option
! 1004: @cindex --image-file, command-line option
! 1005: @item --image-file @i{file}
! 1006: @itemx -i @i{file}
! 1007: Loads the Forth image @i{file} instead of the default
! 1008: @file{gforth.fi} (@pxref{Image Files}).
1.21 crook 1009:
1.29 ! crook 1010: @cindex --path, command-line option
! 1011: @cindex -p, command-line option
! 1012: @item --path @i{path}
! 1013: @itemx -p @i{path}
! 1014: Uses @i{path} for searching the image file and Forth source code files
! 1015: instead of the default in the environment variable @code{GFORTHPATH} or
! 1016: the path specified at installation time (e.g.,
! 1017: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
! 1018: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1.21 crook 1019:
1.29 ! crook 1020: @cindex --dictionary-size, command-line option
! 1021: @cindex -m, command-line option
! 1022: @cindex @i{size} parameters for command-line options
! 1023: @cindex size of the dictionary and the stacks
! 1024: @item --dictionary-size @i{size}
! 1025: @itemx -m @i{size}
! 1026: Allocate @i{size} space for the Forth dictionary space instead of
! 1027: using the default specified in the image (typically 256K). The
! 1028: @i{size} specification for this and subsequent options consists of
! 1029: an integer and a unit (e.g.,
! 1030: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
! 1031: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
! 1032: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
! 1033: @code{e} is used.
1.21 crook 1034:
1.29 ! crook 1035: @cindex --data-stack-size, command-line option
! 1036: @cindex -d, command-line option
! 1037: @item --data-stack-size @i{size}
! 1038: @itemx -d @i{size}
! 1039: Allocate @i{size} space for the data stack instead of using the
! 1040: default specified in the image (typically 16K).
1.21 crook 1041:
1.29 ! crook 1042: @cindex --return-stack-size, command-line option
! 1043: @cindex -r, command-line option
! 1044: @item --return-stack-size @i{size}
! 1045: @itemx -r @i{size}
! 1046: Allocate @i{size} space for the return stack instead of using the
! 1047: default specified in the image (typically 15K).
1.21 crook 1048:
1.29 ! crook 1049: @cindex --fp-stack-size, command-line option
! 1050: @cindex -f, command-line option
! 1051: @item --fp-stack-size @i{size}
! 1052: @itemx -f @i{size}
! 1053: Allocate @i{size} space for the floating point stack instead of
! 1054: using the default specified in the image (typically 15.5K). In this case
! 1055: the unit specifier @code{e} refers to floating point numbers.
1.21 crook 1056:
1.29 ! crook 1057: @cindex --locals-stack-size, command-line option
! 1058: @cindex -l, command-line option
! 1059: @item --locals-stack-size @i{size}
! 1060: @itemx -l @i{size}
! 1061: Allocate @i{size} space for the locals stack instead of using the
! 1062: default specified in the image (typically 14.5K).
1.21 crook 1063:
1.29 ! crook 1064: @cindex -h, command-line option
! 1065: @cindex --help, command-line option
! 1066: @item --help
! 1067: @itemx -h
! 1068: Print a message about the command-line options
1.21 crook 1069:
1.29 ! crook 1070: @cindex -v, command-line option
! 1071: @cindex --version, command-line option
! 1072: @item --version
! 1073: @itemx -v
! 1074: Print version and exit
1.21 crook 1075:
1.29 ! crook 1076: @cindex --debug, command-line option
! 1077: @item --debug
! 1078: Print some information useful for debugging on startup.
1.21 crook 1079:
1.29 ! crook 1080: @cindex --offset-image, command-line option
! 1081: @item --offset-image
! 1082: Start the dictionary at a slightly different position than would be used
! 1083: otherwise (useful for creating data-relocatable images,
! 1084: @pxref{Data-Relocatable Image Files}).
1.21 crook 1085:
1.29 ! crook 1086: @cindex --no-offset-im, command-line option
! 1087: @item --no-offset-im
! 1088: Start the dictionary at the normal position.
1.21 crook 1089:
1.29 ! crook 1090: @cindex --clear-dictionary, command-line option
! 1091: @item --clear-dictionary
! 1092: Initialize all bytes in the dictionary to 0 before loading the image
! 1093: (@pxref{Data-Relocatable Image Files}).
! 1094:
! 1095: @cindex --die-on-signal, command-line-option
! 1096: @item --die-on-signal
! 1097: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
! 1098: or the segmentation violation SIGSEGV) by translating it into a Forth
! 1099: @code{THROW}. With this option, Gforth exits if it receives such a
! 1100: signal. This option is useful when the engine and/or the image might be
! 1101: severely broken (such that it causes another signal before recovering
! 1102: from the first); this option avoids endless loops in such cases.
! 1103: @end table
! 1104:
! 1105: @cindex loading files at startup
! 1106: @cindex executing code on startup
! 1107: @cindex batch processing with Gforth
! 1108: As explained above, the image-specific command-line arguments for the
! 1109: default image @file{gforth.fi} consist of a sequence of filenames and
! 1110: @code{-e @var{forth-code}} options that are interpreted in the sequence
! 1111: in which they are given. The @code{-e @var{forth-code}} or
! 1112: @code{--evaluate @var{forth-code}} option evaluates the Forth
! 1113: code. This option takes only one argument; if you want to evaluate more
! 1114: Forth words, you have to quote them or use @code{-e} several times. To exit
! 1115: after processing the command line (instead of entering interactive mode)
! 1116: append @code{-e bye} to the command line.
! 1117:
! 1118: @cindex versions, invoking other versions of Gforth
! 1119: If you have several versions of Gforth installed, @code{gforth} will
! 1120: invoke the version that was installed last. @code{gforth-@i{version}}
! 1121: invokes a specific version. You may want to use the option
! 1122: @code{--path}, if your environment contains the variable
! 1123: @code{GFORTHPATH}.
! 1124:
! 1125: Not yet implemented:
! 1126: On startup the system first executes the system initialization file
! 1127: (unless the option @code{--no-init-file} is given; note that the system
! 1128: resulting from using this option may not be ANS Forth conformant). Then
! 1129: the user initialization file @file{.gforth.fs} is executed, unless the
! 1130: option @code{--no-rc} is given; this file is first searched in @file{.},
! 1131: then in @file{~}, then in the normal path (see above).
1.21 crook 1132:
1133:
1134:
1.29 ! crook 1135: @comment ----------------------------------------------
! 1136: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
! 1137: @section Leaving Gforth
! 1138: @cindex Gforth - leaving
! 1139: @cindex leaving Gforth
1.21 crook 1140:
1.29 ! crook 1141: You can leave Gforth by typing @code{bye} or Ctrl-D or (if you invoked
! 1142: Gforth with the @code{--die-on-signal} option) Ctrl-C. When you leave
! 1143: Gforth, all of your definitions and data are discarded. @xref{Image
! 1144: Files} for ways of saving the state of the system before leaving Gforth.
1.21 crook 1145:
1.29 ! crook 1146: doc-bye
1.21 crook 1147:
1.29 ! crook 1148: @comment ----------------------------------------------
! 1149: @node Command-line editing, Upper and lower case,Leaving Gforth,Gforth Environment
! 1150: @section Command-line editing
! 1151: @cindex command-line editing
1.21 crook 1152:
1.29 ! crook 1153: Gforth maintains a history file that records every line that you type to
! 1154: the text interpreter. This file is preserved between sessions, and is
! 1155: used to provide a command-line recall facility; if you type ctrl-P
! 1156: repeatedly you can recall successively older commands from this (or
! 1157: previous) session(s). The full list of command-line editing facilities is:
1.21 crook 1158:
1159: @itemize @bullet
1160: @item
1.29 ! crook 1161: ctrl-P (``previous'') (or up-arrow) to recall successively older
! 1162: commands from the history buffer.
! 1163: @item
! 1164: ctrl-N (``next'') (or down-arrow) to recall successively newer commands
! 1165: from the history buffer.
! 1166: @item
! 1167: ctrl-F (or right-arrow) to move the cursor right, non-destructively.
! 1168: @item
! 1169: ctrl-B (or left-arrow) to move the cursor left, non-destructively.
! 1170: @item
! 1171: ctrl-H (backspace) to delete the character to the left of the cursor,
! 1172: closing up the line.
! 1173: @item
! 1174: ctrl-K to delete (``kill'') from the cursor to the end of the line.
! 1175: @item
! 1176: ctrl-A to move the cursor to the start of the line.
1.21 crook 1177: @item
1.29 ! crook 1178: ctrl-E to move the cursor to the end of the line.
1.21 crook 1179: @item
1.29 ! crook 1180: carriage-return or line-feed (ctrl-J, ctrl-M) to submit the current
! 1181: line.
1.21 crook 1182: @item
1.29 ! crook 1183: tab to step through all possible full-word completions of the word
! 1184: currently being typed.
1.21 crook 1185: @item
1.29 ! crook 1186: ctrl-D to terminate Gforth (gracefully, using @code{bye}).
1.21 crook 1187: @end itemize
1188:
1.29 ! crook 1189: When editing, displayable characters are inserted to the left of the
! 1190: cursor position; the line is always in ``insert'' (as opposed to
! 1191: ``overstrike'') mode.
! 1192:
! 1193: @cindex history file
! 1194: @cindex @file{.gforth-history}
! 1195: On Unix systems, the history file is @file{~/.gforth-history} by
! 1196: default@footnote{i.e. it is stored in the user's home directory.}. You
! 1197: can find out the name and location of your history file using:
! 1198:
! 1199: @example
! 1200: history-file type \ Unix-class systems
1.21 crook 1201:
1.29 ! crook 1202: history-file type \ Other systems
! 1203: history-dir type
1.21 crook 1204: @end example
1205:
1.29 ! crook 1206: If you enter long definitions by hand, you can use a text editor to
! 1207: paste them out of the history file into a Forth source file for reuse at
! 1208: a later time.
! 1209:
! 1210: Gforth never trims the size of the history file, so you should do this
! 1211: periodically, if necessary.
! 1212:
! 1213: @comment this is all defined in history.fs
! 1214: @comment TODO the ctrl-D behaviour can either do a bye or a beep.. how is that option
! 1215: @comment chosen?
! 1216:
! 1217:
! 1218:
! 1219: @comment ----------------------------------------------
! 1220: @node Upper and lower case, Environment variables,Command-line editing,Gforth Environment
! 1221: @section Upper and lower case
! 1222: @cindex case-sensitivity
! 1223: @cindex upper and lower case
! 1224:
! 1225: Gforth is case-insensitive, so you can enter definitions and invoke
! 1226: Standard words using upper, lower or mixed case (however,
! 1227: @pxref{core-idef, Implementation-defined options, Implementation-defined
! 1228: options}).
! 1229:
! 1230: ANS Forth only @i{requires} implementations to recognise Standard words when
! 1231: they are typed entirely in upper case. Therefore, a Standard program
! 1232: must use upper case for all Standard words@footnote{You can use whatever
! 1233: case you like for words that you define.}.
! 1234:
! 1235:
! 1236: @comment ----------------------------------------------
! 1237: @node Environment variables, Gforth Files, Upper and lower case,Gforth Environment
! 1238: @section Environment variables
! 1239: @cindex environment variables
1.21 crook 1240:
1.29 ! crook 1241: Gforth uses these environment variables:
1.21 crook 1242:
1.29 ! crook 1243: @itemize @bullet
! 1244: @item
! 1245: @cindex GFORTHHIST - environment variable
! 1246: GFORTHHIST - (Unix systems only) specifies the directory in which to
! 1247: open/create the history file, @file{.gforth-history}. Default:
! 1248: @code{$HOME}.
1.21 crook 1249:
1.29 ! crook 1250: @item
! 1251: @cindex GFORTHPATH - environment variable
! 1252: GFORTHPATH - specifies the path used when searching for the gforth image file and
! 1253: for Forth source-code files.
1.21 crook 1254:
1.29 ! crook 1255: @item
! 1256: @cindex GFORTH - environment variable
! 1257: GFORTH - used by @file{gforthmi} @xref{gforthmi}.
1.26 crook 1258:
1.29 ! crook 1259: @item
! 1260: @cindex GFORTHD - environment variable
! 1261: GFORTHD - used by @file{gforthmi} @xref{gforthmi}.
1.21 crook 1262:
1.29 ! crook 1263: @item
! 1264: @cindex TMP, TEMP - environment variable
! 1265: TMP, TEMP - (non-Unix systems only) used as a potential location for the
! 1266: history file.
! 1267: @end itemize
1.21 crook 1268:
1.29 ! crook 1269: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
! 1270: @comment mentioning these.
1.21 crook 1271:
1.29 ! crook 1272: All the Gforth environment variables default to sensible values if they
! 1273: are not set.
1.21 crook 1274:
1275:
1.29 ! crook 1276: @comment ----------------------------------------------
! 1277: @node Gforth Files, ,Environment variables,Gforth Environment
! 1278: @section Gforth files
! 1279: @cindex Gforth files
1.21 crook 1280:
1.29 ! crook 1281: When Gforth is installed on a Unix system it leaves files in these
! 1282: locations:
1.21 crook 1283:
1.26 crook 1284: @itemize @bullet
1285: @item
1.29 ! crook 1286: @file{/usr/local/bin/gforth}
! 1287: @item
! 1288: @file{/usr/local/bin/gforthmi}
! 1289: @item
! 1290: @file{/usr/local/man/man1/gforth.1} - man page.
! 1291: @item
! 1292: @file{/usr/local/info} - the Info version of this manual.
! 1293: @item
! 1294: @file{/usr/local/lib/gforth/<version>/..} - Gforth @file{.fi} files.
! 1295: @item
! 1296: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1.26 crook 1297: @item
1.29 ! crook 1298: @file{/usr/local/share/gforth/<version>/..} - Gforth source files.
1.26 crook 1299: @item
1.29 ! crook 1300: @file{../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1.26 crook 1301: @end itemize
1.21 crook 1302:
1303:
1.29 ! crook 1304: @c ******************************************************************
! 1305: @node Introduction, Words, Gforth Environment, Top
! 1306: @comment node-name, next, previous, up
! 1307: @chapter An Introduction to ANS Forth
! 1308: @cindex Forth - an introduction
1.21 crook 1309:
1.29 ! crook 1310: The primary purpose of this manual is to document Gforth. However, since
! 1311: Forth is not a widely-known language and there is a lack of up-to-date
! 1312: teaching material, it seems worthwhile to provide some introductory
! 1313: material. @xref{Forth-related information} for other sources of Forth-related
! 1314: information.
1.21 crook 1315:
1.29 ! crook 1316: The examples in this section should work on any ANS Forth; the
! 1317: output shown was produced using Gforth. Each example attempts to
! 1318: reproduce the exact output that Gforth produces. If you try out the
! 1319: examples (and you should), what you should type is shown @kbd{like this}
! 1320: and Gforth's response is shown @code{like this}. The single exception is
! 1321: that, where the example shows @kbd{<return>} it means that you should
! 1322: press the ``carriage return'' key. Unfortunately, some output formats for
! 1323: this manual cannot show the difference between @kbd{this} and
! 1324: @code{this} which will make trying out the examples harder (but not
! 1325: impossible).
1.21 crook 1326:
1.29 ! crook 1327: Forth is an unusual language. It provides an interactive development
! 1328: environment which includes both an interpreter and compiler. Forth
! 1329: programming style encourages you to break a problem down into many
! 1330: @cindex factoring
! 1331: small fragments (@dfn{factoring}), and then to develop and test each
! 1332: fragment interactively. Forth advocates assert that breaking the
! 1333: edit-compile-test cycle used by conventional programming languages can
! 1334: lead to great productivity improvements.
1.21 crook 1335:
1.29 ! crook 1336: @menu
! 1337: * Introducing the Text Interpreter::
! 1338: * Stacks and Postfix notation::
! 1339: * Your first definition::
! 1340: * How does that work?::
! 1341: * Forth is written in Forth::
! 1342: * Review - elements of a Forth system::
! 1343: * Where to go next::
! 1344: * Exercises::
! 1345: @end menu
1.21 crook 1346:
1.29 ! crook 1347: @comment ----------------------------------------------
! 1348: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
! 1349: @section Introducing the Text Interpreter
! 1350: @cindex text interpreter
! 1351: @cindex outer interpreter
1.21 crook 1352:
1.29 ! crook 1353: When you invoke the Forth image, you will see a startup banner printed
! 1354: and nothing else (if you have Gforth installed on your system, try
! 1355: invoking it now, by typing @kbd{gforth<return>}). Forth is now running
! 1356: its command line interpreter, which is called the @dfn{Text Interpreter}
! 1357: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
! 1358: about the text interpreter as you read through this chapter,
! 1359: but @pxref{The Text Interpreter} for more detail).
1.21 crook 1360:
1.29 ! crook 1361: Although it's not obvious, Forth is actually waiting for your
! 1362: input. Type a number and press the <return> key:
1.21 crook 1363:
1.26 crook 1364: @example
1.29 ! crook 1365: @kbd{45<return>} ok
1.26 crook 1366: @end example
1.21 crook 1367:
1.29 ! crook 1368: Rather than give you a prompt to invite you to input something, the text
! 1369: interpreter prints a status message @i{after} it has processed a line
! 1370: of input. The status message in this case (``@code{ ok}'' followed by
! 1371: carriage-return) indicates that the text interpreter was able to process
! 1372: all of your input successfully. Now type something illegal:
! 1373:
! 1374: @example
! 1375: @kbd{qwer341<return>}
! 1376: :1: Undefined word
! 1377: qwer341
! 1378: ^^^^^^^
! 1379: $400D2BA8 Bounce
! 1380: $400DBDA8 no.extensions
! 1381: @end example
1.23 crook 1382:
1.29 ! crook 1383: The exact text, other than the ``Undefined word'' may differ slightly on
! 1384: your system, but the effect is the same; when the text interpreter
! 1385: detects an error, it discards any remaining text on a line, resets
! 1386: certain internal state and prints an error message.
1.23 crook 1387:
1.29 ! crook 1388: The text interpreter waits for you to press carriage-return, and then
! 1389: processes your input line. Starting at the beginning of the line, it
! 1390: breaks the line into groups of characters separated by spaces. For each
! 1391: group of characters in turn, it makes two attempts to do something:
1.23 crook 1392:
1.29 ! crook 1393: @itemize @bullet
! 1394: @item
! 1395: It tries to treat it as a command. It does this by searching a @dfn{name
! 1396: dictionary}. If the group of characters matches an entry in the name
! 1397: dictionary, the name dictionary provides the text interpreter with
! 1398: information that allows the text interpreter perform some actions. In
! 1399: Forth jargon, we say that the group
! 1400: @cindex word
! 1401: @cindex definition
! 1402: @cindex execution token
! 1403: @cindex xt
! 1404: of characters names a @dfn{word}, that the dictionary search returns an
! 1405: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
! 1406: word, and that the text interpreter executes the xt. Often, the terms
! 1407: @dfn{word} and @dfn{definition} are used interchangeably.
! 1408: @item
! 1409: If the text interpreter fails to find a match in the name dictionary, it
! 1410: tries to treat the group of characters as a number in the current number
! 1411: base (when you start up Forth, the current number base is base 10). If
! 1412: the group of characters legitimately represents a number, the text
! 1413: interpreter pushes the number onto a stack (we'll learn more about that
! 1414: in the next section).
! 1415: @end itemize
1.23 crook 1416:
1.29 ! crook 1417: If the text interpreter is unable to do either of these things with any
! 1418: group of characters, it discards the group of characters and the rest of
! 1419: the line, then prints an error message. If the text interpreter reaches
! 1420: the end of the line without error, it prints the status message ``@code{ ok}''
! 1421: followed by carriage-return.
1.21 crook 1422:
1.29 ! crook 1423: This is the simplest command we can give to the text interpreter:
1.23 crook 1424:
1425: @example
1.29 ! crook 1426: @kbd{<return>} ok
1.23 crook 1427: @end example
1.21 crook 1428:
1.29 ! crook 1429: The text interpreter did everything we asked it to do (nothing) without
! 1430: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
! 1431: command:
1.21 crook 1432:
1.23 crook 1433: @example
1.29 ! crook 1434: @kbd{12 dup fred dup<return>}
! 1435: :1: Undefined word
! 1436: 12 dup fred dup
! 1437: ^^^^
! 1438: $400D2BA8 Bounce
! 1439: $400DBDA8 no.extensions
1.23 crook 1440: @end example
1.21 crook 1441:
1.29 ! crook 1442: When you press the carriage-return key, the text interpreter starts to
! 1443: work its way along the line:
1.21 crook 1444:
1.29 ! crook 1445: @itemize @bullet
! 1446: @item
! 1447: When it gets to the space after the @code{2}, it takes the group of
! 1448: characters @code{12} and looks them up in the name
! 1449: dictionary@footnote{We can't tell if it found them or not, but assume
! 1450: for now that it did not}. There is no match for this group of characters
! 1451: in the name dictionary, so it tries to treat them as a number. It is
! 1452: able to do this successfully, so it puts the number, 12, ``on the stack''
! 1453: (whatever that means).
! 1454: @item
! 1455: The text interpreter resumes scanning the line and gets the next group
! 1456: of characters, @code{dup}. It looks it up in the name dictionary and
! 1457: (you'll have to take my word for this) finds it, and executes the word
! 1458: @code{dup} (whatever that means).
! 1459: @item
! 1460: Once again, the text interpreter resumes scanning the line and gets the
! 1461: group of characters @code{fred}. It looks them up in the name
! 1462: dictionary, but can't find them. It tries to treat them as a number, but
! 1463: they don't represent any legal number.
! 1464: @end itemize
1.21 crook 1465:
1.29 ! crook 1466: At this point, the text interpreter gives up and prints an error
! 1467: message. The error message shows exactly how far the text interpreter
! 1468: got in processing the line. In particular, it shows that the text
! 1469: interpreter made no attempt to do anything with the final character
! 1470: group, @code{dup}, even though we have good reason to believe that the
! 1471: text interpreter would have no problem looking that word up and
! 1472: executing it a second time.
1.21 crook 1473:
1474:
1.29 ! crook 1475: @comment ----------------------------------------------
! 1476: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
! 1477: @section Stacks, postfix notation and parameter passing
! 1478: @cindex text interpreter
! 1479: @cindex outer interpreter
1.21 crook 1480:
1.29 ! crook 1481: In procedural programming languages (like C and Pascal), the
! 1482: building-block of programs is the @dfn{function} or @dfn{procedure}. These
! 1483: functions or procedures are called with @dfn{explicit parameters}. For
! 1484: example, in C we might write:
1.21 crook 1485:
1.23 crook 1486: @example
1.29 ! crook 1487: total = total + new_volume(length,height,depth);
1.23 crook 1488: @end example
1.21 crook 1489:
1.23 crook 1490: @noindent
1.29 ! crook 1491: where new_volume is a function-call to another piece of code, and total,
! 1492: length, height and depth are all variables. length, height and depth are
! 1493: parameters to the function-call.
1.21 crook 1494:
1.29 ! crook 1495: In Forth, the equivalent of the function or procedure is the
! 1496: @dfn{definition} and parameters are implicitly passed between
! 1497: definitions using a shared stack that is visible to the
! 1498: programmer. Although Forth does support variables, the existence of the
! 1499: stack means that they are used far less often than in most other
! 1500: programming languages. When the text interpreter encounters a number, it
! 1501: will place (@dfn{push}) it on the stack. There are several stacks (the
! 1502: actual number is implementation-dependent ..) and the particular stack
! 1503: used for any operation is implied unambiguously by the operation being
! 1504: performed. The stack used for all integer operations is called the @dfn{data
! 1505: stack} and, since this is the stack used most commonly, references to
! 1506: ``the data stack'' are often abbreviated to ``the stack''.
1.21 crook 1507:
1.29 ! crook 1508: The stacks have a last-in, first-out (LIFO) organisation. If you type:
1.21 crook 1509:
1.23 crook 1510: @example
1.29 ! crook 1511: @kbd{1 2 3<return>} ok
1.23 crook 1512: @end example
1.21 crook 1513:
1.29 ! crook 1514: Then this instructs the text interpreter to placed three numbers on the
! 1515: (data) stack. An analogy for the behaviour of the stack is to take a
! 1516: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
! 1517: the table. The 3 was the last card onto the pile (``last-in'') and if
! 1518: you take a card off the pile then, unless you're prepared to fiddle a
! 1519: bit, the card that you take off will be the 3 (``first-out''). The
! 1520: number that will be first-out of the stack is called the @dfn{top of
! 1521: stack}, which
! 1522: @cindex TOS definition
! 1523: is often abbreviated to @dfn{TOS}.
1.21 crook 1524:
1.29 ! crook 1525: To understand how parameters are passed in Forth, consider the
! 1526: behaviour of the definition @code{+} (pronounced ``plus''). You will not
! 1527: be surprised to learn that this definition performs addition. More
! 1528: precisely, it adds two number together and produces a result. Where does
! 1529: it get the two numbers from? It takes the top two numbers off the
! 1530: stack. Where does it place the result? On the stack. You can act-out the
! 1531: behaviour of @code{+} with your playing cards like this:
1.21 crook 1532:
1533: @itemize @bullet
1534: @item
1.29 ! crook 1535: Pick up two cards from the stack on the table
1.21 crook 1536: @item
1.29 ! crook 1537: Stare at them intently and ask yourself ``what @i{is} the sum of these two
! 1538: numbers''
1.21 crook 1539: @item
1.29 ! crook 1540: Decide that the answer is 5
1.21 crook 1541: @item
1.29 ! crook 1542: Shuffle the two cards back into the pack and find a 5
1.21 crook 1543: @item
1.29 ! crook 1544: Put a 5 on the remaining ace that's on the table.
1.21 crook 1545: @end itemize
1546:
1.29 ! crook 1547: If you don't have a pack of cards handy but you do have Forth running,
! 1548: you can use the definition @code{.s} to show the current state of the stack,
! 1549: without affecting the stack. Type:
1.21 crook 1550:
1551: @example
1.29 ! crook 1552: @kbd{clearstack 1 2 3<return>} ok
! 1553: @kbd{.s<return>} <3> 1 2 3 ok
1.23 crook 1554: @end example
1555:
1.29 ! crook 1556: The text interpreter looks up the word @code{clearstack} and executes
! 1557: it; it tidies up the stack and removes any entries that may have been
! 1558: left on it by earlier examples. The text interpreter pushes each of the
! 1559: three numbers in turn onto the stack. Finally, the text interpreter
! 1560: looks up the word @code{.s} and executes it. The effect of executing
! 1561: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
! 1562: followed by a list of all the items on the stack; the item on the far
! 1563: right-hand side is the TOS.
1.21 crook 1564:
1.29 ! crook 1565: You can now type:
1.21 crook 1566:
1.29 ! crook 1567: @example
! 1568: @kbd{+ .s<return>} <2> 1 5 ok
! 1569: @end example
1.21 crook 1570:
1.29 ! crook 1571: @noindent
! 1572: which is correct; there are now 2 items on the stack and the result of
! 1573: the addition is 5.
1.23 crook 1574:
1.29 ! crook 1575: If you're playing with cards, try doing a second addition: pick up the
! 1576: two cards, work out that their sum is 6, shuffle them into the pack,
! 1577: look for a 6 and place that on the table. You now have just one item on
! 1578: the stack. What happens if you try to do a third addition? Pick up the
! 1579: first card, pick up the second card -- ah! There is no second card. This
! 1580: is called a @dfn{stack underflow} and consitutes an error. If you try to
! 1581: do the same thing with Forth it will report an error (probably a Stack
! 1582: Underflow or an Invalid Memory Address error).
1.23 crook 1583:
1.29 ! crook 1584: The opposite situation to a stack underflow is a @dfn{stack overflow},
! 1585: which simply accepts that there is a finite amount of storage space
! 1586: reserved for the stack. To stretch the playing card analogy, if you had
! 1587: enough packs of cards and you piled the cards up on the table, you would
! 1588: eventually be unable to add another card; you'd hit the ceiling. Gforth
! 1589: allows you to set the maximum size of the stacks. In general, the only
! 1590: time that you will get a stack overflow is because a definition has a
! 1591: bug in it and is generating data on the stack uncontrollably.
1.23 crook 1592:
1.29 ! crook 1593: There's one final use for the playing card analogy. If you model your
! 1594: stack using a pack of playing cards, the maximum number of items on
! 1595: your stack will be 52 (I assume you didn't use the Joker). The maximum
! 1596: @i{value} of any item on the stack is 13 (the King). In fact, the only
! 1597: possible numbers are positive integer numbers 1 through 13; you can't
! 1598: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
! 1599: think about some of the cards, you can accommodate different
! 1600: numbers. For example, you could think of the Jack as representing 0,
! 1601: the Queen as representing -1 and the King as representing -2. Your
! 1602: *range* remains unchanged (you can still only represent a total of 13
! 1603: numbers) but the numbers that you can represent are -2 through 10.
1.28 crook 1604:
1.29 ! crook 1605: In that analogy, the limit was the amount of information that a single
! 1606: stack entry could hold, and Forth has a similar limit. In Forth, the
! 1607: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
! 1608: implementation dependent and affects the maximum value that a stack
! 1609: entry can hold. A Standard Forth provides a cell size of at least
! 1610: 16-bits, and most desktop systems use a cell size of 32-bits.
1.21 crook 1611:
1.29 ! crook 1612: Forth does not do any type checking for you, so you are free to
! 1613: manipulate and combine stack items in any way you wish. A convenient way
! 1614: of treating stack items is as 2's complement signed integers, and that
! 1615: is what Standard words like @code{+} do. Therefore you can type:
1.21 crook 1616:
1.29 ! crook 1617: @example
! 1618: @kbd{-5 12 + .s<return>} <1> 7 ok
! 1619: @end example
1.21 crook 1620:
1.29 ! crook 1621: If you use numbers and definitions like @code{+} in order to turn Forth
! 1622: into a great big pocket calculator, you will realise that it's rather
! 1623: different from a normal calculator. Rather than typing 2 + 3 = you had
! 1624: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
! 1625: result). The terminology used to describe this difference is to say that
! 1626: your calculator uses @dfn{Infix Notation} (parameters and operators are
! 1627: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
! 1628: operators are separate), also called @dfn{Reverse Polish Notation}.
1.21 crook 1629:
1.29 ! crook 1630: Whilst postfix notation might look confusing to begin with, it has
! 1631: several important advantages:
1.21 crook 1632:
1.23 crook 1633: @itemize @bullet
1634: @item
1.29 ! crook 1635: it is unambiguous
1.23 crook 1636: @item
1.29 ! crook 1637: it is more concise
1.23 crook 1638: @item
1.29 ! crook 1639: it fits naturally with a stack-based system
1.23 crook 1640: @end itemize
1.21 crook 1641:
1.29 ! crook 1642: To examine these claims in more detail, consider these sums:
1.21 crook 1643:
1.29 ! crook 1644: @example
! 1645: 6 + 5 * 4 =
! 1646: 4 * 5 + 6 =
! 1647: @end example
1.21 crook 1648:
1.29 ! crook 1649: If you're just learning maths or your maths is very rusty, you will
! 1650: probably come up with the answer 44 for the first and 26 for the
! 1651: second. If you are a bit of a whizz at maths you will remember the
! 1652: @i{convention} that multiplication takes precendence over addition, and
! 1653: you'd come up with the answer 26 both times. To explain the answer 26
! 1654: to someone who got the answer 44, you'd probably rewrite the first sum
! 1655: like this:
1.21 crook 1656:
1.29 ! crook 1657: @example
! 1658: 6 + (5 * 4) =
! 1659: @end example
1.21 crook 1660:
1.29 ! crook 1661: If what you really wanted was to perform the addition before the
! 1662: multiplication, you would have to use parentheses to force it.
1.21 crook 1663:
1.29 ! crook 1664: If you did the first two sums on a pocket calculator you would probably
! 1665: get the right answers, unless you were very cautious and entered them using
! 1666: these keystroke sequences:
1.21 crook 1667:
1.29 ! crook 1668: 6 + 5 = * 4 =
! 1669: 4 * 5 = + 6 =
1.21 crook 1670:
1.29 ! crook 1671: Postfix notation is unambiguous because the order that the operators
! 1672: are applied is always explicit; that also means that parentheses are
! 1673: never required. The operators are @i{active} (the act of quoting the
! 1674: operator makes the operation occur) which removes the need for ``=''.
1.28 crook 1675:
1.29 ! crook 1676: The sum 6 + 5 * 4 can be written (in postfix notation) in two
! 1677: equivalent ways:
1.26 crook 1678:
1679: @example
1.29 ! crook 1680: 6 5 4 * + or:
! 1681: 5 4 * 6 +
1.26 crook 1682: @end example
1.23 crook 1683:
1.29 ! crook 1684: An important thing that you should notice about this notation is that
! 1685: the @i{order} of the numbers does not change; if you want to subtract
! 1686: 2 from 10 you type @code{10 2 -}.
1.1 anton 1687:
1.29 ! crook 1688: The reason that Forth uses postfix notation is very simple to explain: it
! 1689: makes the implementation extremely simple, and it follows naturally from
! 1690: using the stack as a mechanism for passing parameters. Another way of
! 1691: thinking about this is to realise that all Forth definitions are
! 1692: @i{active}; they execute as they are encountered by the text
! 1693: interpreter. The result of this is that the syntax of Forth is trivially
! 1694: simple.
1.1 anton 1695:
1696:
1697:
1.29 ! crook 1698: @comment ----------------------------------------------
! 1699: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
! 1700: @section Your first Forth definition
! 1701: @cindex first definition
1.1 anton 1702:
1.29 ! crook 1703: Until now, the examples we've seen have been trivial; we've just been
! 1704: using Forth as a bigger-than-pocket calculator. Also, each calculation
! 1705: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
! 1706: again@footnote{That's not quite true. If you press the up-arrow key on
! 1707: your keyboard you should be able to scroll back to any earlier command,
! 1708: edit it and re-enter it.} In this section we'll see how to add new
! 1709: words to Forth's vocabulary.
1.1 anton 1710:
1.29 ! crook 1711: The easiest way to create a new word is to use a @dfn{colon
! 1712: definition}. We'll define a few and try them out before worrying too
! 1713: much about how they work. Try typing in these examples; be careful to
! 1714: copy the spaces accurately:
1.1 anton 1715:
1.29 ! crook 1716: @example
! 1717: : add-two 2 + . ;
! 1718: : greet ." Hello and welcome" ;
! 1719: : demo 5 add-two ;
! 1720: @end example
1.1 anton 1721:
1.29 ! crook 1722: @noindent
! 1723: Now try them out:
1.1 anton 1724:
1.29 ! crook 1725: @example
! 1726: @kbd{greet<return>} Hello and welcome ok
! 1727: @kbd{greet greet<return>} Hello and welcomeHello and welcome ok
! 1728: @kbd{4 add-two<return>} 6 ok
! 1729: @kbd{demo<return>} 7 ok
! 1730: @kbd{9 greet demo add-two<return>} Hello and welcome7 11 ok
! 1731: @end example
1.1 anton 1732:
1.29 ! crook 1733: The first new thing that we've introduced here is the pair of words
! 1734: @code{:} and @code{;}. These are used to start and terminate a new
! 1735: definition, respectively. The first word after the @code{:} is the name
! 1736: for the new definition.
1.1 anton 1737:
1.29 ! crook 1738: As you can see from the examples, a definition is built up of words that
! 1739: have already been defined; Forth makes no distinction between
! 1740: definitions that existed when you started the system up, and those that
! 1741: you define yourself.
1.1 anton 1742:
1.29 ! crook 1743: The examples also introduce the words @code{.} (dot), @code{."}
! 1744: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
! 1745: the stack and displays it. It's like @code{.s} except that it only
! 1746: displays the top item of the stack and it is destructive; after it has
! 1747: executed, the number is no longer on the stack. There is always one
! 1748: space printed after the number, and no spaces before it. Dot-quote
! 1749: defines a string (a sequence of characters) that will be printed when
! 1750: the word is executed. The string can contain any printable characters
! 1751: except @code{"}. A @code{"} has a special function; it is not a Forth
! 1752: word but it acts as a delimiter (the way that delimiters work is
! 1753: described in the next section). Finally, @code{dup} duplicates the value
! 1754: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
1.1 anton 1755:
1.29 ! crook 1756: We already know that the text interpreter searches through the
! 1757: dictionary to locate names. If you've followed the examples earlier, you
! 1758: will already have a definition called @code{add-two}. Lets try modifying
! 1759: it by typing in a new definition:
1.1 anton 1760:
1.29 ! crook 1761: @example
! 1762: @kbd{: add-two dup . ." + 2 =" 2 + . ;<return>} redefined add-two ok
! 1763: @end example
1.5 anton 1764:
1.29 ! crook 1765: Forth recognised that we were defining a word that already exists, and
! 1766: printed a message to warn us of that fact. Let's try out the new
! 1767: definition:
1.5 anton 1768:
1.29 ! crook 1769: @example
! 1770: @kbd{9 add-two<return>} 9 + 2 =11 ok
! 1771: @end example
1.1 anton 1772:
1.29 ! crook 1773: @noindent
! 1774: All that we've actually done here, though, is to create a new
! 1775: definition, with a particular name. The fact that there was already a
! 1776: definition with the same name did not make any difference to the way
! 1777: that the new definition was created (except that Forth printed a warning
! 1778: message). The old definition of add-two still exists (try @code{demo}
! 1779: again to see that this is true). Any new definition will use the new
! 1780: definition of @code{add-two}, but old definitions continue to use the
! 1781: version that already existed at the time that they were @code{compiled}.
1.1 anton 1782:
1.29 ! crook 1783: Before you go on to the next section, try defining and redefining some
! 1784: words of your own.
1.1 anton 1785:
1.29 ! crook 1786: @comment ----------------------------------------------
! 1787: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
! 1788: @section How does that work?
! 1789: @cindex parsing words
1.1 anton 1790:
1.29 ! crook 1791: Now we're going to take another look at the definition of @code{add-two}
! 1792: from the previous section. From our knowledge of the way that the text
! 1793: interpreter works, we would have expected this result when we tried to
! 1794: define @code{add-two}:
1.21 crook 1795:
1.29 ! crook 1796: @example
! 1797: @kbd{: add-two 2 + . " ;<return>}
! 1798: ^^^^^^^
! 1799: Error: Undefined word
! 1800: @end example
1.28 crook 1801:
1.29 ! crook 1802: The reason that this didn't happen is bound up in the way that @code{:}
! 1803: works. The word @code{:} does two special things. The first special
! 1804: thing that it does prevents the text interpreter from ever seeing the
! 1805: characters @code{add-two}. The text interpreter uses a variable called
! 1806: @cindex modifying >IN
! 1807: @code{>IN} (pronounced ''to-in'') to keep track of where it is in the
! 1808: input line. When it encounters the word @code{:} it behaves in exactly
! 1809: the same way as it does for any other word; it looks it up in the name
! 1810: dictionary, finds its xt and executes it. When @code{:} executes, it
! 1811: looks at the input buffer, finds the word @code{add-two} and advances the
! 1812: value of @code{>IN} to point past it. It then does some other stuff
! 1813: associated with creating the new definition (including creating an entry
! 1814: for @code{add-two} in the name dictionary). When the execution of @code{:}
! 1815: completes, control returns to the text interpreter, which is oblivious
! 1816: to the fact that it has been tricked into ignoring part of the input
! 1817: line.
1.21 crook 1818:
1.29 ! crook 1819: @cindex parsing words
! 1820: Words like @code{:} -- words that advance the value of @code{>IN} and so
! 1821: prevent the text interpreter from acting on the whole of the input line
! 1822: -- are called @dfn{parsing words}.
1.21 crook 1823:
1.29 ! crook 1824: @cindex @code{state} - effect on the text interpreter
! 1825: @cindex text interpreter - effect of state
! 1826: The second special thing that @code{:} does is change the value of a
! 1827: variable called @code{state}, which affects the way that the text
! 1828: interpreter behaves. When Gforth starts up, @code{state} has the value
! 1829: 0, and the text interpreter is said to be @dfn{interpreting}. During a
! 1830: colon definition (started with @code{:}), @code{state} is set to -1 and
! 1831: the text interpreter is said to be @dfn{compiling}. The word @code{;}
! 1832: ends the definition -- one of the things that it does is to change the
! 1833: value of @code{state} back to 0.
1.21 crook 1834:
1.29 ! crook 1835: We have already seen how the text interpreter behaves when it is
! 1836: interpreting; it looks for each character sequence in the dictionary,
! 1837: finds its xt and executes it, or it converts it to a number and pushes
! 1838: it onto the stack, or it fails to do either and generates an error.
1.21 crook 1839:
1.29 ! crook 1840: When the text interpreter is compiling, its behaviour is slightly
! 1841: different; it still looks for each character sequence in the dictionary
! 1842: and finds its xt, or converts it to a number, or fails to do either and
! 1843: generates an error. However, instead of executing the xt or pushing the
! 1844: number onto the stack it lays down (@dfn{compiles}) some magic to make
! 1845: that xt or number get executed or pushed at a later time; at the time
! 1846: that @code{add-two} is @dfn{executed}. Therefore, when you execute
! 1847: @code{add-two} its @dfn{run-time effect} is exactly the same as if you
! 1848: had typed @code{2 + .} outside of a definition, and pressed
! 1849: carriage-return.
1.28 crook 1850:
1.29 ! crook 1851: In Forth, every word or number can be described in terms of three
! 1852: properties:
1.28 crook 1853:
1854: @itemize @bullet
1855: @item
1.29 ! crook 1856: Its behaviour at @dfn{compile} time
1.28 crook 1857: @item
1.29 ! crook 1858: Its behaviour at @dfn{interpret} time
1.28 crook 1859: @item
1.29 ! crook 1860: Its behaviour at @dfn{execution} time.
! 1861: @end itemize
! 1862:
! 1863: These behaviours are called the @dfn{semantics} of the word or
! 1864: number. The value of @code{state} determines whether the text
! 1865: interpreter will use the compilation or interpretation semantics of a
! 1866: word or number that it encounters.
! 1867:
! 1868: @itemize @bullet
1.28 crook 1869: @item
1.29 ! crook 1870: @cindex interpretation semantics
! 1871: When the text interpreter encounters a word or number in @dfn{interpret}
! 1872: state, it performs the @dfn{interpretation semantics} of the word or
! 1873: number.
1.28 crook 1874: @item
1.29 ! crook 1875: @cindex compilation semantics
! 1876: When the text interpreter encounters a word or number in @dfn{compile}
! 1877: state, it performs the @dfn{compilation semantics} of the word or
! 1878: number.
! 1879: @end itemize
! 1880:
! 1881: @noindent
! 1882: Numbers are always treated in a fixed way:
! 1883:
! 1884: @itemize @bullet
1.28 crook 1885: @item
1.29 ! crook 1886: When the number is @dfn{compiled}, it is appended to the current
! 1887: definition so that its run-time behaviour is to execute. (In other
! 1888: words, the compilation semantics of a number are to postpone its
! 1889: execution semantics until the run-time of the definition that it is
! 1890: being compiled into.)
1.28 crook 1891: @item
1.29 ! crook 1892: When the number is @dfn{interpreted}, its behaviour is to execute. (In
! 1893: other words, the interpretation semantics of a number are to perform its
! 1894: execution semantics.)
1.28 crook 1895: @item
1.29 ! crook 1896: @cindex execution semantics
! 1897: When the number is @dfn{executed}, its behaviour is to push its value
! 1898: onto the stack. (In other words, the execution semantics of a number are
! 1899: to push its value onto the stack.)
! 1900: @end itemize
! 1901:
! 1902:
! 1903: The behaviour of a word is not so regular, but most have @i{default
! 1904: semantics} which means that they behave like this:
! 1905:
! 1906: @itemize @bullet
1.28 crook 1907: @item
1.29 ! crook 1908: The @dfn{compilation semantics} of the word are to append its
! 1909: @dfn{execution semantics} to the current definition (so that its
! 1910: run-time behaviour is to execute).
1.28 crook 1911: @item
1.29 ! crook 1912: The @dfn{interpretation semantics} of the word are to execute.
1.28 crook 1913: @item
1.29 ! crook 1914: The @dfn{execution semantics} of the word are to do something useful.
1.28 crook 1915: @end itemize
1916:
1917:
1.29 ! crook 1918: The actual behaviour of any particular word depends upon the way in
! 1919: which it was defined. When the text interpreter finds the word in the
! 1920: name dictionary, it not only retrieves the xt for the word, it also
! 1921: retrieves some flags: the @dfn{compile-only} flag and the @dfn{immediate
! 1922: flag}. The compile-only flag indicates that the word has no
! 1923: interpretation semantics; any attempt to interpret a word that has the
! 1924: compile-only flag set will generate an error (for example, @code{IF} has
! 1925: no interpretation semantics). The immediate flag changes the compilation
! 1926: semantics of the word; if it is set, the text interpreter will
! 1927: @dfn{execute} the word rather than @dfn{compiling}
! 1928: @cindex immediate words
! 1929: it. In other words, these so-called @dfn{immediate} words behave like
! 1930: this:
! 1931:
! 1932: @itemize @bullet
! 1933: @item
! 1934: The @dfn{compilation semantics} of the word are to perform its
! 1935: @dfn{execution semantics} (so that its compile-time behaviour is to
! 1936: execute).
! 1937: @item
! 1938: The @dfn{interpretation semantics} of the word are to execute.
! 1939: @item
! 1940: The @dfn{execution semantics} of the word are to do something useful.
! 1941: @end itemize
1.28 crook 1942:
1.29 ! crook 1943: This example shows the difference between an immediate and a
! 1944: non-immediate word:
1.28 crook 1945:
1.29 ! crook 1946: @example
! 1947: : show-state state @@ . ;
! 1948: : show-state-now show-state ; immediate
! 1949: : word1 show-state ;
! 1950: : word2 show-state-now ;
1.28 crook 1951: @end example
1.23 crook 1952:
1.29 ! crook 1953: The word @code{immediate} after the definition of @code{show-state-now}
! 1954: makes that word an immediate word. These definitions introduce a new
! 1955: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
! 1956: variable, and leaves it on the stack. Therefore, the behaviour of
! 1957: @code{show-state} is to print a number that represents the current value
! 1958: of @code{state}.
1.28 crook 1959:
1.29 ! crook 1960: When you execute @code{word1}, it prints the number 0, indicating that
! 1961: the system is interpreting. When the text interpreter compiled the
! 1962: definition of @code{word1}, it encountered @code{show-state} whose
! 1963: compilation semantics are to append its execution semantics to the
! 1964: current definition. When you execute @code{word1}, it performs the
! 1965: execution semantics of @code{show-state}. At the time that @code{word1}
! 1966: (and therefore @code{show-state}) are executed, the system is
! 1967: interpreting.
1.28 crook 1968:
1.29 ! crook 1969: When you pressed <return> after entering the definition of @code{word2},
! 1970: you should have seen the number -1 printed, followed by ``@code{
! 1971: ok}''. When the text interpreter compiled the definition of
! 1972: @code{word2}, it encountered @code{show-state-now}, an immediate word,
! 1973: whose compilation semantics are therefore to perform its execution
! 1974: semantics. It is executed straight away (even before the text
! 1975: interpreter has moved on to process another group of characters; the
! 1976: @code{;} in this example). The effect of executing it are to display the
! 1977: value of @code{state} @i{at the time that the definition of}
! 1978: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
! 1979: system is compiling at this time. If you execute @code{word2} it does
! 1980: nothing at all.
1.28 crook 1981:
1.29 ! crook 1982: @cindex @code{."}, how it works
! 1983: Before leaving the subject of immediate words, consider the behaviour of
! 1984: @code{."} in the definition of @code{greet}, in the previous
! 1985: section. This word is both a parsing word and an immediate word. Notice
! 1986: that there is a space between @code{."} and the start of the text
! 1987: @code{Hello and welcome}, but that there is no space between the last
! 1988: letter of @code{welcome} and the @code{"} character. The reason for this
! 1989: is that @code{."} is a Forth word; it must have a space after it so that
! 1990: the text interpreter can identify it. The @code{"} is not a Forth word;
! 1991: it is a @dfn{delimiter}. The examples earlier show that, when the string
! 1992: is displayed, there is neither a space before the @code{H} nor after the
! 1993: @code{e}. Since @code{."} is an immediate word, it executes at the time
! 1994: that @code{greet} is defined. When it executes, its behaviour is to
! 1995: search forward in the input line looking for the delimiter. When it
! 1996: finds the delimiter, it updates @code{>IN} to point past the
! 1997: delimiter. It also compiles some magic code into the definition of
! 1998: @code{greet}; the xt of a run-time routine that prints a text string. It
! 1999: compiles the string @code{Hello and welcome} into memory so that it is
! 2000: available to be printed later. When the text interpreter gains control,
! 2001: the next word it finds in the input stream is @code{;} and so it
! 2002: terminates the definition of @code{greet}.
1.28 crook 2003:
2004:
2005: @comment ----------------------------------------------
1.29 ! crook 2006: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
! 2007: @section Forth is written in Forth
! 2008: @cindex structure of Forth programs
! 2009:
! 2010: When you start up a Forth compiler, a large number of definitions
! 2011: already exist. In Forth, you develop a new application using bottom-up
! 2012: programming techniques to create new definitions that are defined in
! 2013: terms of existing definitions. As you create each definition you can
! 2014: test and debug it interactively.
! 2015:
! 2016: If you have tried out the examples in this section, you will probably
! 2017: have typed them in by hand; when you leave Gforth, your definitions will
! 2018: be lost. You can avoid this by using a text editor to enter Forth source
! 2019: code into a file, and then loading code from the file using
! 2020: @code{include} (@xref{Forth source files}). A Forth source file is
! 2021: processed by the text interpreter, just as though you had typed it in by
! 2022: hand@footnote{Actually, there are some subtle differences -- see
! 2023: @ref{The Text Interpreter}.}.
! 2024:
! 2025: Gforth also supports the traditional Forth alternative to using text
! 2026: files for program entry (@xref{Blocks}).
1.28 crook 2027:
1.29 ! crook 2028: In common with many, if not most, Forth compilers, most of Gforth is
! 2029: actually written in Forth. All of the @file{.fs} files in the
! 2030: installation directory@footnote{For example,
! 2031: @file{/usr/local/share/gforth..}} are Forth source files, which you can
! 2032: study to see examples of Forth programming.
1.28 crook 2033:
1.29 ! crook 2034: Gforth maintains a history file that records every line that you type to
! 2035: the text interpreter. This file is preserved between sessions, and is
! 2036: used to provide a command-line recall facility. If you enter long
! 2037: definitions by hand, you can use a text editor to paste them out of the
! 2038: history file into a Forth source file for reuse at a later time
! 2039: (@pxref{Command-line editing} for more information).
1.28 crook 2040:
2041:
2042: @comment ----------------------------------------------
1.29 ! crook 2043: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
! 2044: @section Review - elements of a Forth system
! 2045: @cindex elements of a Forth system
1.28 crook 2046:
1.29 ! crook 2047: To summarise this chapter:
1.28 crook 2048:
2049: @itemize @bullet
2050: @item
1.29 ! crook 2051: Forth programs use @dfn{factoring} to break a problem down into small
! 2052: fragments called @dfn{words} or @dfn{definitions}.
! 2053: @item
! 2054: Forth program development is an interactive process.
! 2055: @item
! 2056: The main command loop that accepts input, and controls both
! 2057: interpretation and compilation, is called the @dfn{text interpreter}
! 2058: (also known as the @dfn{outer interpreter}).
! 2059: @item
! 2060: Forth has a very simple syntax, consisting of words and numbers
! 2061: separated by spaces or carriage-return characters. Any additional syntax
! 2062: is imposed by @dfn{parsing words}.
! 2063: @item
! 2064: Forth uses a stack to pass parameters between words. As a result, it
! 2065: uses postfix notation.
! 2066: @item
! 2067: To use a word that has previously been defined, the text interpreter
! 2068: searches for the word in the @dfn{name dictionary}.
! 2069: @item
! 2070: Words have @dfn{interpretation semantics}, @dfn{compilation semantics}
! 2071: and @dfn{execution semantics}.
1.28 crook 2072: @item
1.29 ! crook 2073: The text interpreter uses the value of @code{state} to select between
! 2074: the use of the @dfn{interpretation semantics} and the @dfn{compilation
! 2075: semantics} of a word that it encounters.
1.28 crook 2076: @item
1.29 ! crook 2077: The relationship between the @dfn{interpretation semantics},
! 2078: @dfn{compilation semantics} and @dfn{execution semantics} for a word
! 2079: depend upon the way in which the word was defined (for example, whether
! 2080: it is an @dfn{immediate} word).
1.28 crook 2081: @item
1.29 ! crook 2082: Forth definitions can be implemented in Forth (called @dfn{high-level
! 2083: definitions}) or in some other way (usually a lower-level language and
! 2084: as a result often called @dfn{low-level definitions}, @dfn{code
! 2085: definitions} or @dfn{primitives}).
1.28 crook 2086: @item
1.29 ! crook 2087: Many Forth systems are implemented mainly in Forth.
1.28 crook 2088: @end itemize
2089:
2090:
1.29 ! crook 2091: @comment ----------------------------------------------
! 2092: @node Where to go next,Exercises,Review - elements of a Forth system, Introduction
! 2093: @section Where To Go Next
! 2094: @cindex where to go next
1.28 crook 2095:
1.29 ! crook 2096: Amazing as it may seem, if you have read (and understood) this far, you
! 2097: know almost all the fundamentals about the inner workings of a Forth
! 2098: system. You certainly know enough to be able to read and understand the
! 2099: rest of this manual and the ANS Forth document, to learn more about the
! 2100: facilities that Forth in general and Gforth in particular provide. Even
! 2101: scarier, you know almost enough to implement your own Forth system.
! 2102: However, that's not a good idea just yet.. better to try writing some
! 2103: programs in Gforth.
1.28 crook 2104:
1.29 ! crook 2105: Forth has such a rich vocabulary that it can be hard to know where to
! 2106: start in learning it. This section suggests a few sets of words that are
! 2107: enough to write small but useful programs. Use the word index in this
! 2108: document to learn more about each word, then try it out and try to write
! 2109: small definitions using it. Start by experimenting with these words:
1.28 crook 2110:
2111: @itemize @bullet
2112: @item
1.29 ! crook 2113: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
! 2114: @item
! 2115: Comparison: @code{MIN MAX =}
! 2116: @item
! 2117: Logic: @code{AND OR XOR NOT}
! 2118: @item
! 2119: Stack manipulation: @code{DUP DROP SWAP OVER}
1.28 crook 2120: @item
1.29 ! crook 2121: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
1.28 crook 2122: @item
1.29 ! crook 2123: Input/Output: @code{. ." EMIT CR KEY}
1.28 crook 2124: @item
1.29 ! crook 2125: Defining words: @code{: ; CREATE}
1.28 crook 2126: @item
1.29 ! crook 2127: Memory allocation words: @code{ALLOT ,}
1.28 crook 2128: @item
1.29 ! crook 2129: Tools: @code{SEE WORDS .S MARKER}
! 2130: @end itemize
! 2131:
! 2132: When you have mastered those, go on to:
! 2133:
! 2134: @itemize @bullet
1.28 crook 2135: @item
1.29 ! crook 2136: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
1.28 crook 2137: @item
1.29 ! crook 2138: Memory access: @code{@@ !}
1.28 crook 2139: @end itemize
1.23 crook 2140:
1.29 ! crook 2141: When you have mastered these, there's nothing for it but to read through
! 2142: the whole of this manual and find out what you've missed.
! 2143:
! 2144: @comment ----------------------------------------------
! 2145: @node Exercises, ,Where to go next, Introduction
! 2146: @section Exercises
! 2147: @cindex exercises
! 2148:
! 2149: TODO: provide a set of programming excercises linked into the stuff done
! 2150: already and into other sections of the manual. Provide solutions to all
! 2151: the exercises in a .fs file in the distribution.
! 2152:
! 2153: @c Get some inspiration from Starting Forth and Kelly&Spies.
! 2154:
! 2155: @c excercises:
! 2156: @c 1. take inches and convert to feet and inches.
! 2157: @c 2. take temperature and convert from fahrenheight to celcius;
! 2158: @c may need to care about symmetric vs floored??
! 2159: @c 3. take input line and do character substitution
! 2160: @c to encipher or decipher
! 2161: @c 4. as above but work on a file for in and out
! 2162: @c 5. take input line and convert to pig-latin
! 2163: @c
! 2164: @c thing of sets of things to exercise then come up with
! 2165: @c problems that need those things.
! 2166:
! 2167:
1.26 crook 2168: @c ******************************************************************
1.29 ! crook 2169: @node Words, Error messages, Introduction, Top
1.1 anton 2170: @chapter Forth Words
1.26 crook 2171: @cindex words
1.1 anton 2172:
2173: @menu
2174: * Notation::
1.21 crook 2175: * Comments::
2176: * Boolean Flags::
1.1 anton 2177: * Arithmetic::
2178: * Stack Manipulation::
1.5 anton 2179: * Memory::
1.1 anton 2180: * Control Structures::
2181: * Defining Words::
1.21 crook 2182: * The Text Interpreter::
1.12 anton 2183: * Tokens for Words::
1.21 crook 2184: * Word Lists::
2185: * Environmental Queries::
1.12 anton 2186: * Files::
2187: * Blocks::
2188: * Other I/O::
2189: * Programming Tools::
2190: * Assembler and Code Words::
2191: * Threading Words::
1.26 crook 2192: * Locals::
2193: * Structures::
2194: * Object-oriented Forth::
1.21 crook 2195: * Passing Commands to the OS::
2196: * Miscellaneous Words::
1.1 anton 2197: @end menu
2198:
1.21 crook 2199: @node Notation, Comments, Words, Words
1.1 anton 2200: @section Notation
2201: @cindex notation of glossary entries
2202: @cindex format of glossary entries
2203: @cindex glossary notation format
2204: @cindex word glossary entry format
2205:
2206: The Forth words are described in this section in the glossary notation
2207: that has become a de-facto standard for Forth texts, i.e.,
2208:
2209: @format
1.29 ! crook 2210: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
1.1 anton 2211: @end format
1.29 ! crook 2212: @i{Description}
1.1 anton 2213:
2214: @table @var
2215: @item word
1.28 crook 2216: The name of the word.
1.1 anton 2217:
2218: @item Stack effect
2219: @cindex stack effect
1.29 ! crook 2220: The stack effect is written in the notation @code{@i{before} --
! 2221: @i{after}}, where @i{before} and @i{after} describe the top of
1.1 anton 2222: stack entries before and after the execution of the word. The rest of
2223: the stack is not touched by the word. The top of stack is rightmost,
2224: i.e., a stack sequence is written as it is typed in. Note that Gforth
2225: uses a separate floating point stack, but a unified stack
1.29 ! crook 2226: notation. Also, return stack effects are not shown in @i{stack
! 2227: effect}, but in @i{Description}. The name of a stack item describes
1.1 anton 2228: the type and/or the function of the item. See below for a discussion of
2229: the types.
2230:
2231: All words have two stack effects: A compile-time stack effect and a
2232: run-time stack effect. The compile-time stack-effect of most words is
1.29 ! crook 2233: @i{ -- }. If the compile-time stack-effect of a word deviates from
1.1 anton 2234: this standard behaviour, or the word does other unusual things at
2235: compile time, both stack effects are shown; otherwise only the run-time
2236: stack effect is shown.
2237:
2238: @cindex pronounciation of words
2239: @item pronunciation
2240: How the word is pronounced.
2241:
2242: @cindex wordset
2243: @item wordset
1.21 crook 2244: The ANS Forth standard is divided into several word sets. A standard
2245: system need not support all of them. Therefore, in theory, the fewer
2246: word sets your program uses the more portable it will be. However, we
2247: suspect that most ANS Forth systems on personal machines will feature
1.26 crook 2248: all word sets. Words that are not defined in ANS Forth have
1.21 crook 2249: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
1.1 anton 2250: describes words that will work in future releases of Gforth;
2251: @code{gforth-internal} words are more volatile. Environmental query
2252: strings are also displayed like words; you can recognize them by the
1.21 crook 2253: @code{environment} in the word set field.
1.1 anton 2254:
2255: @item Description
2256: A description of the behaviour of the word.
2257: @end table
2258:
2259: @cindex types of stack items
2260: @cindex stack item types
2261: The type of a stack item is specified by the character(s) the name
2262: starts with:
2263:
2264: @table @code
2265: @item f
2266: @cindex @code{f}, stack item type
2267: Boolean flags, i.e. @code{false} or @code{true}.
2268: @item c
2269: @cindex @code{c}, stack item type
2270: Char
2271: @item w
2272: @cindex @code{w}, stack item type
2273: Cell, can contain an integer or an address
2274: @item n
2275: @cindex @code{n}, stack item type
2276: signed integer
2277: @item u
2278: @cindex @code{u}, stack item type
2279: unsigned integer
2280: @item d
2281: @cindex @code{d}, stack item type
2282: double sized signed integer
2283: @item ud
2284: @cindex @code{ud}, stack item type
2285: double sized unsigned integer
2286: @item r
2287: @cindex @code{r}, stack item type
2288: Float (on the FP stack)
1.21 crook 2289: @item a-
1.1 anton 2290: @cindex @code{a_}, stack item type
2291: Cell-aligned address
1.21 crook 2292: @item c-
1.1 anton 2293: @cindex @code{c_}, stack item type
2294: Char-aligned address (note that a Char may have two bytes in Windows NT)
1.21 crook 2295: @item f-
1.1 anton 2296: @cindex @code{f_}, stack item type
2297: Float-aligned address
1.21 crook 2298: @item df-
1.1 anton 2299: @cindex @code{df_}, stack item type
2300: Address aligned for IEEE double precision float
1.21 crook 2301: @item sf-
1.1 anton 2302: @cindex @code{sf_}, stack item type
2303: Address aligned for IEEE single precision float
2304: @item xt
2305: @cindex @code{xt}, stack item type
2306: Execution token, same size as Cell
2307: @item wid
2308: @cindex @code{wid}, stack item type
1.21 crook 2309: Word list ID, same size as Cell
1.1 anton 2310: @item f83name
2311: @cindex @code{f83name}, stack item type
2312: Pointer to a name structure
2313: @item "
2314: @cindex @code{"}, stack item type
1.12 anton 2315: string in the input stream (not on the stack). The terminating character
2316: is a blank by default. If it is not a blank, it is shown in @code{<>}
1.1 anton 2317: quotes.
2318: @end table
2319:
1.21 crook 2320: @node Comments, Boolean Flags, Notation, Words
2321: @section Comments
1.26 crook 2322: @cindex comments
1.21 crook 2323:
1.29 ! crook 2324: Forth supports two styles of comment; the traditional @i{in-line} comment,
! 2325: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
1.21 crook 2326:
1.23 crook 2327: doc-(
1.21 crook 2328: doc-\
1.23 crook 2329: doc-\G
1.21 crook 2330:
2331: @node Boolean Flags, Arithmetic, Comments, Words
2332: @section Boolean Flags
1.26 crook 2333: @cindex Boolean flags
1.21 crook 2334:
2335: A Boolean flag is cell-sized. A cell with all bits clear represents the
2336: flag @code{false} and a flag with all bits set represents the flag
1.26 crook 2337: @code{true}. Words that check a flag (for example, @code{IF}) will treat
1.29 ! crook 2338: a cell that has @i{any} bit set as @code{true}.
1.21 crook 2339:
2340: doc-true
2341: doc-false
1.29 ! crook 2342: doc-on
! 2343: doc-off
1.21 crook 2344:
2345: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
1.1 anton 2346: @section Arithmetic
2347: @cindex arithmetic words
2348:
2349: @cindex division with potentially negative operands
2350: Forth arithmetic is not checked, i.e., you will not hear about integer
2351: overflow on addition or multiplication, you may hear about division by
2352: zero if you are lucky. The operator is written after the operands, but
2353: the operands are still in the original order. I.e., the infix @code{2-1}
2354: corresponds to @code{2 1 -}. Forth offers a variety of division
2355: operators. If you perform division with potentially negative operands,
2356: you do not want to use @code{/} or @code{/mod} with its undefined
2357: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
2358: former, @pxref{Mixed precision}).
1.26 crook 2359: @comment TODO discuss the different division forms and the std approach
1.1 anton 2360:
2361: @menu
2362: * Single precision::
2363: * Bitwise operations::
1.21 crook 2364: * Double precision:: Double-cell integer arithmetic
2365: * Numeric comparison::
1.29 ! crook 2366: * Mixed precision:: Operations with single and double-cell integers
1.1 anton 2367: * Floating Point::
2368: @end menu
2369:
2370: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
2371: @subsection Single precision
2372: @cindex single precision arithmetic words
2373:
1.21 crook 2374: By default, numbers in Forth are single-precision integers that are 1
1.26 crook 2375: cell in size. They can be signed or unsigned, depending upon how you
1.21 crook 2376: treat them. @xref{Number Conversion} for the rules used by the text
2377: interpreter for recognising single-precision integers.
2378:
1.1 anton 2379: doc-+
1.21 crook 2380: doc-1+
1.1 anton 2381: doc--
1.21 crook 2382: doc-1-
1.1 anton 2383: doc-*
2384: doc-/
2385: doc-mod
2386: doc-/mod
2387: doc-negate
2388: doc-abs
2389: doc-min
2390: doc-max
1.21 crook 2391: doc-d>s
1.27 crook 2392: doc-floored
1.1 anton 2393:
1.21 crook 2394: @node Bitwise operations, Double precision, Single precision, Arithmetic
1.1 anton 2395: @subsection Bitwise operations
2396: @cindex bitwise operation words
2397:
2398: doc-and
2399: doc-or
2400: doc-xor
2401: doc-invert
1.21 crook 2402: doc-lshift
2403: doc-rshift
1.1 anton 2404: doc-2*
1.21 crook 2405: doc-d2*
1.1 anton 2406: doc-2/
1.21 crook 2407: doc-d2/
2408:
2409: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
2410: @subsection Double precision
2411: @cindex double precision arithmetic words
2412:
2413: @xref{Number Conversion} for the rules used by the text interpreter for
2414: recognising double-precision integers.
2415:
2416: A double precision number is represented by a cell pair, with the most
1.26 crook 2417: significant digit at the TOS. It is trivial to convert an unsigned
2418: single to an (unsigned) double; simply push a @code{0} onto the
2419: TOS. Since numbers are represented by Gforth using 2's complement
2420: arithmetic, converting a signed single to a (signed) double requires
2421: sign-extension across the most significant digit. This can be achieved
2422: using @code{s>d}. The moral of the story is that you cannot convert a
2423: number without knowing whether it represents an unsigned or a
2424: signed number.
1.21 crook 2425:
2426: doc-s>d
2427: doc-d+
2428: doc-d-
2429: doc-dnegate
2430: doc-dabs
2431: doc-dmin
2432: doc-dmax
2433:
2434: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
2435: @subsection Numeric comparison
2436: @cindex numeric comparison words
2437:
1.28 crook 2438: doc-<
2439: doc-<=
2440: doc-<>
2441: doc-=
2442: doc->
2443: doc->=
2444:
1.21 crook 2445: doc-0<
1.23 crook 2446: doc-0<=
1.21 crook 2447: doc-0<>
2448: doc-0=
1.23 crook 2449: doc-0>
2450: doc-0>=
1.28 crook 2451:
2452: doc-u<
2453: doc-u<=
2454: @comment TODO why u<> and u= .. they are the same as <> and =
2455: doc-u<>
2456: doc-u=
2457: doc-u>
2458: doc-u>=
2459:
2460: doc-within
2461:
2462: doc-d<
2463: doc-d<=
2464: doc-d<>
2465: doc-d=
2466: doc-d>
2467: doc-d>=
1.23 crook 2468:
1.21 crook 2469: doc-d0<
1.23 crook 2470: doc-d0<=
2471: doc-d0<>
1.21 crook 2472: doc-d0=
1.23 crook 2473: doc-d0>
2474: doc-d0>=
2475:
1.21 crook 2476: doc-du<
1.28 crook 2477: doc-du<=
2478: doc-du<>
2479: doc-du=
2480: doc-du>
2481: doc-du>=
1.1 anton 2482:
1.21 crook 2483: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
1.1 anton 2484: @subsection Mixed precision
2485: @cindex mixed precision arithmetic words
2486:
2487: doc-m+
2488: doc-*/
2489: doc-*/mod
2490: doc-m*
2491: doc-um*
2492: doc-m*/
2493: doc-um/mod
2494: doc-fm/mod
2495: doc-sm/rem
2496:
1.21 crook 2497: @node Floating Point, , Mixed precision, Arithmetic
1.1 anton 2498: @subsection Floating Point
2499: @cindex floating point arithmetic words
2500:
1.21 crook 2501: @xref{Number Conversion} for the rules used by the text interpreter for
2502: recognising floating-point numbers.
1.1 anton 2503:
2504: @cindex angles in trigonometric operations
2505: @cindex trigonometric operations
2506: Angles in floating point operations are given in radians (a full circle
1.26 crook 2507: has 2 pi radians). Gforth has a separate floating point
2508: stack, but the documentation uses the unified notation.
1.1 anton 2509:
2510: @cindex floating-point arithmetic, pitfalls
2511: Floating point numbers have a number of unpleasant surprises for the
2512: unwary (e.g., floating point addition is not associative) and even a few
2513: for the wary. You should not use them unless you know what you are doing
2514: or you don't care that the results you get are totally bogus. If you
2515: want to learn about the problems of floating point numbers (and how to
2516: avoid them), you might start with @cite{David Goldberg, What Every
2517: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
1.17 anton 2518: Computing Surveys 23(1):5@minus{}48, March 1991}
2519: (@url{http://www.validgh.com/goldberg/paper.ps}).
1.1 anton 2520:
1.21 crook 2521: doc-d>f
2522: doc-f>d
1.1 anton 2523: doc-f+
2524: doc-f-
2525: doc-f*
2526: doc-f/
2527: doc-fnegate
2528: doc-fabs
2529: doc-fmax
2530: doc-fmin
2531: doc-floor
2532: doc-fround
2533: doc-f**
2534: doc-fsqrt
2535: doc-fexp
2536: doc-fexpm1
2537: doc-fln
2538: doc-flnp1
2539: doc-flog
2540: doc-falog
2541: doc-fsin
2542: doc-fcos
2543: doc-fsincos
2544: doc-ftan
2545: doc-fasin
2546: doc-facos
2547: doc-fatan
2548: doc-fatan2
2549: doc-fsinh
2550: doc-fcosh
2551: doc-ftanh
2552: doc-fasinh
2553: doc-facosh
2554: doc-fatanh
1.21 crook 2555: doc-pi
1.28 crook 2556:
1.21 crook 2557: doc-f0<
1.28 crook 2558: doc-f0<=
2559: doc-f0<>
1.21 crook 2560: doc-f0=
1.28 crook 2561: doc-f0>
2562: doc-f0>=
2563:
1.21 crook 2564: doc-f<
2565: doc-f<=
2566: doc-f<>
2567: doc-f=
2568: doc-f>
2569: doc-f>=
1.28 crook 2570:
1.21 crook 2571: doc-f2*
2572: doc-f2/
2573: doc-1/f
2574: doc-f~
2575: doc-precision
2576: doc-set-precision
1.1 anton 2577:
2578: @node Stack Manipulation, Memory, Arithmetic, Words
2579: @section Stack Manipulation
2580: @cindex stack manipulation words
2581:
2582: @cindex floating-point stack in the standard
1.21 crook 2583: Gforth maintains a number of separate stacks:
2584:
1.29 ! crook 2585: @cindex data stack
! 2586: @cindex parameter stack
1.21 crook 2587: @itemize @bullet
2588: @item
1.29 ! crook 2589: A data stack (also known as the @dfn{parameter stack}) -- for
! 2590: characters, cells, addresses, and double cells.
1.21 crook 2591:
1.29 ! crook 2592: @cindex floating-point stack
1.21 crook 2593: @item
2594: A floating point stack -- for floating point numbers.
2595:
1.29 ! crook 2596: @cindex return stack
1.21 crook 2597: @item
2598: A return stack -- for storing the return addresses of colon
2599: definitions and other data.
2600:
1.29 ! crook 2601: @cindex locals stack
1.21 crook 2602: @item
2603: A locals stack for storing local variables.
2604: @end itemize
2605:
2606: Whilst every sane Forth has a separate floating-point stack, it is not
2607: strictly required; an ANS Forth system could theoretically keep
2608: floating-point numbers on the data stack. As an additional difficulty,
2609: you don't know how many cells a floating-point number takes. It is
2610: reportedly possible to write words in a way that they work also for a
2611: unified stack model, but we do not recommend trying it. Instead, just
2612: say that your program has an environmental dependency on a separate
2613: floating-point stack.
2614:
2615: doc-floating-stack
1.1 anton 2616:
2617: @cindex return stack and locals
2618: @cindex locals and return stack
1.21 crook 2619: A Forth system is allowed to keep local variables on the
1.1 anton 2620: return stack. This is reasonable, as local variables usually eliminate
2621: the need to use the return stack explicitly. So, if you want to produce
1.21 crook 2622: a standard compliant program and you are using local variables in a
2623: word, forget about return stack manipulations in that word (refer to the
1.1 anton 2624: standard document for the exact rules).
2625:
2626: @menu
2627: * Data stack::
2628: * Floating point stack::
2629: * Return stack::
2630: * Locals stack::
2631: * Stack pointer manipulation::
2632: @end menu
2633:
2634: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
2635: @subsection Data stack
2636: @cindex data stack manipulation words
2637: @cindex stack manipulations words, data stack
2638:
2639: doc-drop
2640: doc-nip
2641: doc-dup
2642: doc-over
2643: doc-tuck
2644: doc-swap
1.21 crook 2645: doc-pick
1.1 anton 2646: doc-rot
2647: doc--rot
2648: doc-?dup
2649: doc-roll
2650: doc-2drop
2651: doc-2nip
2652: doc-2dup
2653: doc-2over
2654: doc-2tuck
2655: doc-2swap
2656: doc-2rot
2657:
2658: @node Floating point stack, Return stack, Data stack, Stack Manipulation
2659: @subsection Floating point stack
2660: @cindex floating-point stack manipulation words
2661: @cindex stack manipulation words, floating-point stack
2662:
2663: doc-fdrop
2664: doc-fnip
2665: doc-fdup
2666: doc-fover
2667: doc-ftuck
2668: doc-fswap
1.21 crook 2669: doc-fpick
1.1 anton 2670: doc-frot
2671:
2672: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
2673: @subsection Return stack
2674: @cindex return stack manipulation words
2675: @cindex stack manipulation words, return stack
2676:
2677: doc->r
2678: doc-r>
2679: doc-r@
2680: doc-rdrop
2681: doc-2>r
2682: doc-2r>
2683: doc-2r@
2684: doc-2rdrop
2685:
2686: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
2687: @subsection Locals stack
2688:
1.26 crook 2689: @comment TODO
1.21 crook 2690:
1.1 anton 2691: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
2692: @subsection Stack pointer manipulation
2693: @cindex stack pointer manipulation words
2694:
1.21 crook 2695: doc-sp0
2696: doc-s0
1.1 anton 2697: doc-sp@
2698: doc-sp!
1.21 crook 2699: doc-fp0
1.1 anton 2700: doc-fp@
2701: doc-fp!
1.21 crook 2702: doc-rp0
2703: doc-r0
1.1 anton 2704: doc-rp@
2705: doc-rp!
1.21 crook 2706: doc-lp0
2707: doc-l0
1.1 anton 2708: doc-lp@
2709: doc-lp!
2710:
2711: @node Memory, Control Structures, Stack Manipulation, Words
2712: @section Memory
1.26 crook 2713: @cindex memory words
1.1 anton 2714:
1.27 crook 2715: @cindex dictionary
2716: Forth definitions are organised in memory structures that are
1.29 ! crook 2717: collectively called the @dfn{dictionary}. The dictionary can be
1.27 crook 2718: considered as three logical memory regions:
2719:
2720: @itemize @bullet
2721: @item
2722: @cindex code space
2723: @cindex code dictionary
1.29 ! crook 2724: Code space, also known as the @dfn{code dictionary}.
1.27 crook 2725: @item
2726: @cindex name space
2727: @cindex name dictionary
1.29 ! crook 2728: Name space, also known as the @dfn{name dictionary}@footnote{Sometimes,
! 2729: the term @dfn{dictionary} is used simply to refer to the name
1.27 crook 2730: dictionary, because it is the one region that is used for looking up
2731: names, just as you would in a conventional dictionary.}.
2732: @item
2733: @cindex data space
2734: Data space
2735: @end itemize
2736:
1.29 ! crook 2737: When you create a colon definition, the text interpreter compiles the
! 2738: code for the definition into the code dictionary and compiles the name
1.27 crook 2739: of the definition into the name dictionary, together with other
2740: information about the definition (such as its execution token).
2741:
2742: When you create a variable, the execution of @code{variable} will
2743: compile some code, assign once cell in data space, and compile the name
2744: of the variable into the name dictionary.
2745:
2746: @cindex memory regions - relationship between them
2747: ANS Forth does not specify the relationship between the three memory
2748: regions, and specifies that a Standard program must not access code or
2749: data space directly -- it may only access data space directly. In
2750: addition, the Standard defines what relationships you may and may not
2751: rely on when allocating regions in data space. These constraints are
2752: simply a reflection of the many diverse techniques that are used to
2753: implement Forth systems; understanding and following the requirements of
2754: the Standard allows you to write portable programs -- programs that run
2755: in the same way on any of these diverse systems. Another way of looking
2756: at this is to say that ANS Forth was designed to permit compliant Forth
2757: systems to be implemented in many diverse ways.
2758:
2759: @cindex memory regions - how they are assigned
1.29 ! crook 2760: Here are some examples of ways in which name, code and data spaces
! 2761: might be assigned in different Forth implementations:
1.27 crook 2762:
2763: @itemize @bullet
2764: @item
2765: For a Forth system that runs from RAM under a general-purpose operating
2766: system, it can be convenient to interleave name, code and data spaces in
2767: a single contiguous memory region. This organisation can be
2768: memory-efficient (for example, because the relationship between the name
2769: dictionary entry and the associated code dictionary entry can be
2770: implicit, rather than requiring an explicit memory pointer to reference
2771: from the name dictionary and the code dictionary). This is the
2772: organisation used by Gforth, as this example@footnote{The addresses
2773: in the example have been truncated to fit it onto the page, and the
2774: addresses and data shown will not match the output from your system} shows:
2775: @example
2776: hex
2777: variable fred 123456 fred !
2778: variable jim abcd jim !
2779: : foo + / - ;
2780: ' fred 10 - 50 dump
2781: ..80: 5C 46 0E 40 84 66 72 65 - 64 20 20 20 20 20 20 20 \F.@.fred
2782: ..90: D0 9B 04 08 00 00 00 00 - 56 34 12 00 80 46 0E 40 ........V4...F.@
2783: ..A0: 83 6A 69 6D 20 20 20 20 - D0 9B 04 08 00 00 00 00 .jim ........
2784: ..B0: CD AB 00 00 9C 46 0E 40 - 83 66 6F 6F 20 20 20 20 .....F.@.foo
2785: ..C0: 80 9B 04 08 00 00 00 00 - E4 2E 05 08 0C 2F 05 08 ............./..
2786: @end example
2787:
2788: @item
2789: For a high-performance system running on a modern RISC processor with a
2790: modified Harvard architecture (one that has a unified main memory but
2791: separate instruction and data caches), it is desirable to separate
2792: processor instructions from processor data. This encourages a high cache
2793: density and therefore a high cache hit rate. The Forth code dictionary
2794: is not necessarily made up entirely of processor instructions; its
2795: nature is dependent upon the Forth implementation.
2796:
2797: @item
2798: A Forth compiler that runs on a segmented 8086 processor could be
2799: designed to interleave the name, code and data spaces within a single
2800: 64Kbyte segment. A more common implementation choice is to use a
2801: separate 64Kbyte segment for each region, which provides more memory
2802: overall but provides an address map in which only the data space is
2803: accessible.
2804:
2805: @item
2806: Microprocessors exist that run Forth (or many of the primitives required
2807: to implement the Forth virtual machine efficiently) directly. On these
2808: processors, the relationship between name, code and data spaces may be
2809: imposed as a side-effect of the microarchitecture of the processor.
2810:
2811: @item
2812: A Forth compiler that executes from ROM on an embedded system needs its
2813: data space separated from the name and code spaces so that the data
2814: space can be mapped to a RAM area.
2815:
2816: @item
2817: A Forth compiler that runs on an embedded system may have a requirement
2818: for a small memory footprint. On such a system it can be useful to
2819: separate the name space from the data and code spaces; once the
2820: application has been compiled, the name dictionary is no longer
2821: required@footnote{more strictly speaking, most applications can be
2822: designed so that this is the case}. The name dictionary can be deleted
1.29 ! crook 2823: entirely, or could be stored in memory on a remote @i{host} system for
1.27 crook 2824: debug and development purposes. In the latter case, the compiler running
1.29 ! crook 2825: on the @i{target} system could implement a protocol across a
1.27 crook 2826: communication link that would allow it to interrogate the name dictionary.
2827: @end itemize
2828:
1.1 anton 2829: @menu
1.27 crook 2830: * Reserving Data Space::
2831: * Memory Access::
2832: * Address Arithmetic::
2833: * Memory Blocks::
2834: * Dynamic Allocation::
1.1 anton 2835: @end menu
2836:
1.27 crook 2837:
2838: @node Reserving Data Space, Memory Access, Memory, Memory
2839: @subsection Reserving Data Space
2840: @cindex reserving data space
2841: @cindex data space - reserving some
2842:
2843: @cindex data space pointer - contiguous regions
1.29 ! crook 2844: Data space may be reserved as individual chars or cells or in contiguous
! 2845: regions. These are the rules for reserving contiguous regions in a
! 2846: Standard (i.e., portable) way:
1.27 crook 2847: @itemize @bullet
2848: @item
2849: The value of the data-space pointer, @code{here}, always defines the
2850: beginning of a contiguous region of data space.
2851:
2852: @item
2853: @code{CREATE} establishes the beginning of a contiguous region of data
2854: space (the @code{CREATE}d definition returns the initial address of the
2855: region).
2856:
2857: @item
1.29 ! crook 2858: @code{variable} does @i{not} establish the beginning of a contiguous
1.27 crook 2859: region in data space; @code{variable} followed by @code{allot} is not
2860: guaranteed to allocate data space region that is contiguous with the
2861: storage allocated by @code{variable}. Instead, use @code{create} --
2862: @xref{Simple Defining Words} for examples.
2863:
2864: @item
2865: Successive calls to @code{allot}, @code{,} (comma), @code{2,} (2-comma),
2866: @code{c,} (c-comma) and @code{align} reserve a single contiguous region
2867: in data space. The contiguity of the region is interrupted by compiling
2868: (or removing) definitions from the dictionary.
2869:
2870: @item
2871: The most recently reserved contiguous region may be released by calling
2872: @code{allot} with a negative argument, provided that the region has not
2873: been interrupted by compiling (or removing) definitions from the
2874: dictionary.
2875: @end itemize
2876:
1.29 ! crook 2877: @cindex data space pointer - alignment
! 2878: These factors affect the alignment of @code{here}, the data
! 2879: space pointer:
! 2880:
! 2881: @itemize @bullet
! 2882: @item
! 2883: If the data-space pointer is aligned@footnote{In ANS Forth-speak,
! 2884: @i{aligned} implictly means @code{CELL}-aligned.} before an
! 2885: @code{allot}, and a whole number of characters are reserved or released, it
! 2886: will remain aligned after the @code{allot}.
! 2887:
! 2888: @item
! 2889: If the data-space pointer is character-aligned before an @code{allot},
! 2890: and a whole number of cells are reserved or released, it will remain
! 2891: character-aligned after the @code{allot}.
! 2892:
! 2893: @item
! 2894: The initial contents of data space reserved using @code{allot} is
! 2895: undefined.
! 2896:
! 2897: @item
! 2898: Definitions created by @code{create}, @code{variable}, @code{2variable}
! 2899: return aligned addresses.
! 2900:
! 2901: @item
! 2902: After a definition is compiled or @code{align} is executed, the data
! 2903: space pointer is guaranteed to be aligned.
! 2904: @end itemize
! 2905:
1.27 crook 2906: doc-here
2907: doc-unused
2908: doc-allot
2909: doc-c,
1.29 ! crook 2910: doc-f,
1.27 crook 2911: doc-,
2912: doc-2,
1.29 ! crook 2913: @cindex user space
! 2914: doc-udp
! 2915: doc-uallot
1.27 crook 2916:
2917:
2918: @node Memory Access, Address Arithmetic, Reserving Data Space, Memory
1.1 anton 2919: @subsection Memory Access
2920: @cindex memory access words
2921:
2922: doc-@
2923: doc-!
2924: doc-+!
2925: doc-c@
2926: doc-c!
2927: doc-2@
2928: doc-2!
2929: doc-f@
2930: doc-f!
2931: doc-sf@
2932: doc-sf!
2933: doc-df@
2934: doc-df!
2935:
1.27 crook 2936: @node Address Arithmetic, Memory Blocks, Memory Access, Memory
2937: @subsection Address Arithmetic
1.1 anton 2938: @cindex address arithmetic words
2939:
2940: ANS Forth does not specify the sizes of the data types. Instead, it
2941: offers a number of words for computing sizes and doing address
1.29 ! crook 2942: arithmetic. Address arithmetic is performed in terms of address units
! 2943: (aus); on most systems the address unit is one byte. Note that a
! 2944: character may have more than one au, so @code{chars} is no noop (on
! 2945: systems where it is a noop, it compiles to nothing).
1.1 anton 2946:
2947: @cindex alignment of addresses for types
2948: ANS Forth also defines words for aligning addresses for specific
2949: types. Many computers require that accesses to specific data types
2950: must only occur at specific addresses; e.g., that cells may only be
2951: accessed at addresses divisible by 4. Even if a machine allows unaligned
2952: accesses, it can usually perform aligned accesses faster.
2953:
2954: For the performance-conscious: alignment operations are usually only
2955: necessary during the definition of a data structure, not during the
2956: (more frequent) accesses to it.
2957:
2958: ANS Forth defines no words for character-aligning addresses. This is not
2959: an oversight, but reflects the fact that addresses that are not
2960: char-aligned have no use in the standard and therefore will not be
2961: created.
2962:
2963: @cindex @code{CREATE} and alignment
1.29 ! crook 2964: ANS Forth guarantees that addresses returned by @code{CREATE}d words
1.1 anton 2965: are cell-aligned; in addition, Gforth guarantees that these addresses
2966: are aligned for all purposes.
2967:
1.26 crook 2968: Note that the ANS Forth word @code{char} has nothing to do with address
2969: arithmetic.
1.1 anton 2970:
2971: doc-chars
2972: doc-char+
2973: doc-cells
2974: doc-cell+
2975: doc-cell
2976: doc-align
2977: doc-aligned
2978: doc-floats
2979: doc-float+
2980: doc-float
2981: doc-falign
2982: doc-faligned
2983: doc-sfloats
2984: doc-sfloat+
2985: doc-sfalign
2986: doc-sfaligned
2987: doc-dfloats
2988: doc-dfloat+
2989: doc-dfalign
2990: doc-dfaligned
2991: doc-maxalign
2992: doc-maxaligned
2993: doc-cfalign
2994: doc-cfaligned
2995: doc-address-unit-bits
2996:
1.27 crook 2997: @node Memory Blocks, Dynamic Allocation, Address Arithmetic, Memory
1.1 anton 2998: @subsection Memory Blocks
2999: @cindex memory block words
1.27 crook 3000: @cindex character strings - moving and copying
3001:
3002: Memory blocks often represent character strings; @xref{String Formats}
3003: for ways of storing character strings in memory. @xref{Displaying
3004: characters and strings} for other string-processing words.
1.1 anton 3005:
1.21 crook 3006: Some of these words work on address units (increments of @code{CELL}),
3007: and expect a @code{CELL}-aligned address. Others work on character units
3008: (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
3009: address. Choose the correct operation depending upon your data type. If
3010: you are moving a block of memory (for example, a region reserved by
3011: @code{allot}) it is safe to use @code{move}, and it should be faster
3012: than using @code{cmove}. If you are moving (for example) a string
3013: compiled using @code{S"}, it is not portable to use @code{move}; the
3014: alignment of the string in memory could change, and the relationship
3015: between @code{CELL} and @code{CHAR} could change.
3016:
3017: When copying characters between overlapping memory regions, choose
3018: carefully between @code{cmove} and @code{cmove>}.
3019:
1.29 ! crook 3020: You can only use any of these words @i{portably} to access data space.
1.21 crook 3021:
1.27 crook 3022: @comment TODO - think the naming of the arguments is wrong for move
1.29 ! crook 3023: @comment well, really it seems to be the Standard that's wrong; it
! 3024: @comment describes MOVE as a word that requires a CELL-aligned source
! 3025: @comment and destination address but a xtranfer count that need not
! 3026: @comment be a multiple of CELL.
1.1 anton 3027: doc-move
3028: doc-erase
3029: doc-cmove
3030: doc-cmove>
3031: doc-fill
3032: doc-blank
1.21 crook 3033: doc-compare
3034: doc-search
1.27 crook 3035: doc--trailing
3036: doc-/string
3037:
3038: @comment TODO examples
3039:
3040: @node Dynamic Allocation, ,Memory Blocks, Memory
3041: @subsection Dynamic Allocation of Memory
3042: @cindex dynamic allocation of memory
3043: @cindex memory-allocation word set
3044:
3045: The ANS Forth memory-allocation word set allows memory regions to be
3046: dynamically assigned, resized and released without affecting the data
3047: space pointer. In Gforth, these words are implemented using
3048: the standard C library calls malloc(), free() and resize().
3049:
3050: doc-allocate
3051: doc-free
3052: doc-resize
3053:
1.1 anton 3054:
1.26 crook 3055: @node Control Structures, Defining Words, Memory, Words
1.1 anton 3056: @section Control Structures
3057: @cindex control structures
3058:
3059: Control structures in Forth cannot be used in interpret state, only in
1.29 ! crook 3060: compile state@footnote{To be precise, they have no interpretation
! 3061: semantics (@pxref{Interpretation and Compilation Semantics}).}, i.e., in
1.1 anton 3062: a colon definition. We do not like this limitation, but have not seen a
3063: satisfying way around it yet, although many schemes have been proposed.
3064:
3065: @menu
1.29 ! crook 3066: * Selection:: IF.. ELSE.. ENDIF
! 3067: * Simple Loops:: BEGIN..
! 3068: * Counted Loops:: DO
! 3069: * Arbitrary control structures::
! 3070: * Calls and returns::
1.1 anton 3071: * Exception Handling::
3072: @end menu
3073:
3074: @node Selection, Simple Loops, Control Structures, Control Structures
3075: @subsection Selection
3076: @cindex selection control structures
3077: @cindex control structures for selection
3078:
3079: @cindex @code{IF} control structure
3080: @example
1.29 ! crook 3081: @i{flag}
1.1 anton 3082: IF
1.29 ! crook 3083: @i{code}
1.1 anton 3084: ENDIF
3085: @end example
1.21 crook 3086: @noindent
1.1 anton 3087: or
3088: @example
1.29 ! crook 3089: @i{flag}
1.1 anton 3090: IF
1.29 ! crook 3091: @i{code1}
1.1 anton 3092: ELSE
1.29 ! crook 3093: @i{code2}
1.1 anton 3094: ENDIF
3095: @end example
3096:
3097: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
3098: standard, and @code{ENDIF} is not, although it is quite popular. We
3099: recommend using @code{ENDIF}, because it is less confusing for people
3100: who also know other languages (and is not prone to reinforcing negative
3101: prejudices against Forth in these people). Adding @code{ENDIF} to a
3102: system that only supplies @code{THEN} is simple:
3103: @example
1.21 crook 3104: : ENDIF POSTPONE THEN ; immediate
1.1 anton 3105: @end example
3106:
3107: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
3108: (adv.)} has the following meanings:
3109: @quotation
3110: ... 2b: following next after in order ... 3d: as a necessary consequence
3111: (if you were there, then you saw them).
3112: @end quotation
3113: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
3114: and many other programming languages has the meaning 3d.]
3115:
1.21 crook 3116: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
1.1 anton 3117: you can avoid using @code{?dup}. Using these alternatives is also more
1.26 crook 3118: efficient than using @code{?dup}. Definitions in ANS Forth
1.1 anton 3119: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
3120: @file{compat/control.fs}.
3121:
3122: @cindex @code{CASE} control structure
3123: @example
1.29 ! crook 3124: @i{n}
1.1 anton 3125: CASE
1.29 ! crook 3126: @i{n1} OF @i{code1} ENDOF
! 3127: @i{n2} OF @i{code2} ENDOF
1.1 anton 3128: @dots{}
3129: ENDCASE
3130: @end example
3131:
1.29 ! crook 3132: Executes the first @i{codei}, where the @i{ni} is equal to
! 3133: @i{n}. A default case can be added by simply writing the code after
! 3134: the last @code{ENDOF}. It may use @i{n}, which is on top of the stack,
1.1 anton 3135: but must not consume it.
3136:
3137: @node Simple Loops, Counted Loops, Selection, Control Structures
3138: @subsection Simple Loops
3139: @cindex simple loops
3140: @cindex loops without count
3141:
3142: @cindex @code{WHILE} loop
3143: @example
3144: BEGIN
1.29 ! crook 3145: @i{code1}
! 3146: @i{flag}
1.1 anton 3147: WHILE
1.29 ! crook 3148: @i{code2}
1.1 anton 3149: REPEAT
3150: @end example
3151:
1.29 ! crook 3152: @i{code1} is executed and @i{flag} is computed. If it is true,
! 3153: @i{code2} is executed and the loop is restarted; If @i{flag} is
1.1 anton 3154: false, execution continues after the @code{REPEAT}.
3155:
3156: @cindex @code{UNTIL} loop
3157: @example
3158: BEGIN
1.29 ! crook 3159: @i{code}
! 3160: @i{flag}
1.1 anton 3161: UNTIL
3162: @end example
3163:
1.29 ! crook 3164: @i{code} is executed. The loop is restarted if @code{flag} is false.
1.1 anton 3165:
3166: @cindex endless loop
3167: @cindex loops, endless
3168: @example
3169: BEGIN
1.29 ! crook 3170: @i{code}
1.1 anton 3171: AGAIN
3172: @end example
3173:
3174: This is an endless loop.
3175:
3176: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
3177: @subsection Counted Loops
3178: @cindex counted loops
3179: @cindex loops, counted
3180: @cindex @code{DO} loops
3181:
3182: The basic counted loop is:
3183: @example
1.29 ! crook 3184: @i{limit} @i{start}
1.1 anton 3185: ?DO
1.29 ! crook 3186: @i{body}
1.1 anton 3187: LOOP
3188: @end example
3189:
1.29 ! crook 3190: This performs one iteration for every integer, starting from @i{start}
! 3191: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
1.21 crook 3192: accessed with @code{i}. For example, the loop:
1.1 anton 3193: @example
3194: 10 0 ?DO
3195: i .
3196: LOOP
3197: @end example
1.21 crook 3198: @noindent
3199: prints @code{0 1 2 3 4 5 6 7 8 9}
3200:
1.1 anton 3201: The index of the innermost loop can be accessed with @code{i}, the index
3202: of the next loop with @code{j}, and the index of the third loop with
3203: @code{k}.
3204:
3205: doc-i
3206: doc-j
3207: doc-k
3208:
3209: The loop control data are kept on the return stack, so there are some
1.21 crook 3210: restrictions on mixing return stack accesses and counted loop words. In
3211: particuler, if you put values on the return stack outside the loop, you
3212: cannot read them inside the loop@footnote{well, not in a way that is
3213: portable.}. If you put values on the return stack within a loop, you
3214: have to remove them before the end of the loop and before accessing the
3215: index of the loop.
1.1 anton 3216:
3217: There are several variations on the counted loop:
3218:
1.21 crook 3219: @itemize @bullet
3220: @item
3221: @code{LEAVE} leaves the innermost counted loop immediately; execution
3222: continues after the associated @code{LOOP} or @code{NEXT}. For example:
3223:
3224: @example
3225: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
3226: @end example
3227: prints @code{0 1 2 3}
3228:
1.1 anton 3229:
1.21 crook 3230: @item
3231: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
3232: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
3233: return stack so @code{EXIT} can get to its return address. For example:
3234:
3235: @example
3236: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
3237: @end example
3238: prints @code{0 1 2 3}
3239:
3240:
3241: @item
1.29 ! crook 3242: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
1.1 anton 3243: (and @code{LOOP} iterates until they become equal by wrap-around
3244: arithmetic). This behaviour is usually not what you want. Therefore,
3245: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1.29 ! crook 3246: @code{?DO}), which do not enter the loop if @i{start} is greater than
! 3247: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1.1 anton 3248: unsigned loop parameters.
3249:
1.21 crook 3250: @item
3251: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
3252: the loop, independent of the loop parameters. Do not use @code{DO}, even
3253: if you know that the loop is entered in any case. Such knowledge tends
3254: to become invalid during maintenance of a program, and then the
3255: @code{DO} will make trouble.
3256:
3257: @item
1.29 ! crook 3258: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
! 3259: index by @i{n} instead of by 1. The loop is terminated when the border
! 3260: between @i{limit-1} and @i{limit} is crossed. E.g.:
1.1 anton 3261:
1.21 crook 3262: @example
3263: 4 0 +DO i . 2 +LOOP
3264: @end example
3265: @noindent
3266: prints @code{0 2}
3267:
3268: @example
3269: 4 1 +DO i . 2 +LOOP
3270: @end example
3271: @noindent
3272: prints @code{1 3}
1.1 anton 3273:
3274:
3275: @cindex negative increment for counted loops
3276: @cindex counted loops with negative increment
1.29 ! crook 3277: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
1.1 anton 3278:
1.21 crook 3279: @example
3280: -1 0 ?DO i . -1 +LOOP
3281: @end example
3282: @noindent
3283: prints @code{0 -1}
1.1 anton 3284:
1.21 crook 3285: @example
3286: 0 0 ?DO i . -1 +LOOP
3287: @end example
3288: prints nothing.
1.1 anton 3289:
1.29 ! crook 3290: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
! 3291: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
! 3292: index by @i{u} each iteration. The loop is terminated when the border
! 3293: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
1.1 anton 3294: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
3295:
1.21 crook 3296: @example
3297: -2 0 -DO i . 1 -LOOP
3298: @end example
3299: @noindent
3300: prints @code{0 -1}
1.1 anton 3301:
1.21 crook 3302: @example
3303: -1 0 -DO i . 1 -LOOP
3304: @end example
3305: @noindent
3306: prints @code{0}
3307:
3308: @example
3309: 0 0 -DO i . 1 -LOOP
3310: @end example
3311: @noindent
3312: prints nothing.
1.1 anton 3313:
1.21 crook 3314: @end itemize
1.1 anton 3315:
3316: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1.26 crook 3317: @code{-LOOP} are not defined in ANS Forth. However, an implementation
3318: for these words that uses only standard words is provided in
3319: @file{compat/loops.fs}.
1.1 anton 3320:
3321:
3322: @cindex @code{FOR} loops
1.26 crook 3323: Another counted loop is:
1.1 anton 3324: @example
1.29 ! crook 3325: @i{n}
1.1 anton 3326: FOR
1.29 ! crook 3327: @i{body}
1.1 anton 3328: NEXT
3329: @end example
3330: This is the preferred loop of native code compiler writers who are too
1.26 crook 3331: lazy to optimize @code{?DO} loops properly. This loop structure is not
1.29 ! crook 3332: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
! 3333: @code{i} produces values starting with @i{n} and ending with 0. Other
1.26 crook 3334: Forth systems may behave differently, even if they support @code{FOR}
3335: loops. To avoid problems, don't use @code{FOR} loops.
1.1 anton 3336:
3337: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
3338: @subsection Arbitrary control structures
3339: @cindex control structures, user-defined
3340:
3341: @cindex control-flow stack
3342: ANS Forth permits and supports using control structures in a non-nested
3343: way. Information about incomplete control structures is stored on the
3344: control-flow stack. This stack may be implemented on the Forth data
3345: stack, and this is what we have done in Gforth.
3346:
3347: @cindex @code{orig}, control-flow stack item
3348: @cindex @code{dest}, control-flow stack item
3349: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
3350: entry represents a backward branch target. A few words are the basis for
3351: building any control structure possible (except control structures that
3352: need storage, like calls, coroutines, and backtracking).
3353:
3354: doc-if
3355: doc-ahead
3356: doc-then
3357: doc-begin
3358: doc-until
3359: doc-again
3360: doc-cs-pick
3361: doc-cs-roll
3362:
1.21 crook 3363: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
3364: manipulate the control-flow stack in a portable way. Without them, you
3365: would need to know how many stack items are occupied by a control-flow
3366: entry (many systems use one cell. In Gforth they currently take three,
3367: but this may change in the future).
3368:
1.1 anton 3369: Some standard control structure words are built from these words:
3370:
3371: doc-else
3372: doc-while
3373: doc-repeat
3374:
3375: Gforth adds some more control-structure words:
3376:
3377: doc-endif
3378: doc-?dup-if
3379: doc-?dup-0=-if
3380:
3381: Counted loop words constitute a separate group of words:
3382:
3383: doc-?do
3384: doc-+do
3385: doc-u+do
3386: doc--do
3387: doc-u-do
3388: doc-do
3389: doc-for
3390: doc-loop
3391: doc-+loop
3392: doc--loop
3393: doc-next
3394: doc-leave
3395: doc-?leave
3396: doc-unloop
3397: doc-done
3398:
1.21 crook 3399: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
3400: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
1.1 anton 3401: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
3402: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
3403: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
3404: resolved (by using one of the loop-ending words or @code{DONE}).
3405:
1.26 crook 3406: Another group of control structure words are:
1.1 anton 3407:
3408: doc-case
3409: doc-endcase
3410: doc-of
3411: doc-endof
3412:
1.21 crook 3413: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
3414: @code{CS-ROLL}.
1.1 anton 3415:
3416: @subsubsection Programming Style
3417:
3418: In order to ensure readability we recommend that you do not create
3419: arbitrary control structures directly, but define new control structure
3420: words for the control structure you want and use these words in your
1.26 crook 3421: program. For example, instead of writing:
1.1 anton 3422:
3423: @example
1.26 crook 3424: BEGIN
1.1 anton 3425: ...
1.26 crook 3426: IF [ 1 CS-ROLL ]
1.1 anton 3427: ...
1.26 crook 3428: AGAIN THEN
1.1 anton 3429: @end example
3430:
1.21 crook 3431: @noindent
1.1 anton 3432: we recommend defining control structure words, e.g.,
3433:
3434: @example
1.26 crook 3435: : WHILE ( DEST -- ORIG DEST )
3436: POSTPONE IF
3437: 1 CS-ROLL ; immediate
3438:
3439: : REPEAT ( orig dest -- )
3440: POSTPONE AGAIN
3441: POSTPONE THEN ; immediate
1.1 anton 3442: @end example
3443:
1.21 crook 3444: @noindent
1.1 anton 3445: and then using these to create the control structure:
3446:
3447: @example
1.26 crook 3448: BEGIN
1.1 anton 3449: ...
1.26 crook 3450: WHILE
1.1 anton 3451: ...
1.26 crook 3452: REPEAT
1.1 anton 3453: @end example
3454:
3455: That's much easier to read, isn't it? Of course, @code{REPEAT} and
3456: @code{WHILE} are predefined, so in this example it would not be
3457: necessary to define them.
3458:
3459: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
3460: @subsection Calls and returns
3461: @cindex calling a definition
3462: @cindex returning from a definition
3463:
1.3 anton 3464: @cindex recursive definitions
3465: A definition can be called simply be writing the name of the definition
1.26 crook 3466: to be called. Normally a definition is invisible during its own
1.3 anton 3467: definition. If you want to write a directly recursive definition, you
1.26 crook 3468: can use @code{recursive} to make the current definition visible, or
3469: @code{recurse} to call the current definition directly.
1.3 anton 3470:
3471: doc-recursive
3472: doc-recurse
3473:
1.21 crook 3474: @comment TODO add example of the two recursion methods
1.12 anton 3475: @quotation
3476: @progstyle
3477: I prefer using @code{recursive} to @code{recurse}, because calling the
3478: definition by name is more descriptive (if the name is well-chosen) than
3479: the somewhat cryptic @code{recurse}. E.g., in a quicksort
3480: implementation, it is much better to read (and think) ``now sort the
3481: partitions'' than to read ``now do a recursive call''.
3482: @end quotation
1.3 anton 3483:
1.29 ! crook 3484: For mutual recursion, use @code{Defer}red words, like this:
1.3 anton 3485:
3486: @example
1.28 crook 3487: Defer foo
1.3 anton 3488:
3489: : bar ( ... -- ... )
3490: ... foo ... ;
3491:
3492: :noname ( ... -- ... )
3493: ... bar ... ;
3494: IS foo
3495: @end example
3496:
1.26 crook 3497: The current definition returns control to the calling definition when
1.29 ! crook 3498: the end of the definition is reached or @code{EXIT} is
! 3499: encountered. Deferred words are discussed in more detail in @ref{Simple
! 3500: Defining Words}.
1.1 anton 3501:
3502: doc-exit
3503: doc-;s
3504:
3505: @node Exception Handling, , Calls and returns, Control Structures
3506: @subsection Exception Handling
1.26 crook 3507: @cindex exceptions
1.1 anton 3508:
1.26 crook 3509: If your program detects a fatal error condition, the simplest action
3510: that it can take is to @code{quit}. This resets the return stack and
3511: restarts the text interpreter, but does not print any error message.
1.21 crook 3512:
1.26 crook 3513: The next stage in severity is to execute @code{abort}, which has the
3514: same effect as @code{quit}, with the addition that it resets the data
3515: stack.
1.1 anton 3516:
1.26 crook 3517: A slightly more sophisticated approach is use use @code{abort"}, which
3518: compiles a string to be used as an error message and does a conditional
3519: @code{abort} at run-time. For example:
1.1 anton 3520:
1.26 crook 3521: @example
3522: @kbd{: checker abort" That flag was true" ." A false flag" ;<return>} ok
3523: @kbd{0 checker<return>} A false flag ok
3524: @kbd{1 checker<return>}
3525: :1: That flag was true
3526: 1 checker
3527: ^^^^^^^
3528: $400D1648 throw
3529: $400E4660
3530: @end example
1.1 anton 3531:
1.26 crook 3532: These simple techniques allow a program to react to a fatal error
3533: condition, but they are not exactly user-friendly. The ANS Forth
3534: Exception word set provides the pair of words @code{throw} and
3535: @code{catch}, which can be used to provide sophisticated error-handling.
1.1 anton 3536:
1.26 crook 3537: @code{catch} has a similar behaviour to @code{execute}, in that it takes
1.29 ! crook 3538: an @i{xt} as a parameter and starts execution of the xt. However,
1.26 crook 3539: before passing control to the xt, @code{catch} pushes an
1.29 ! crook 3540: @dfn{exception frame} onto the @dfn{exception stack}. This exception
1.26 crook 3541: frame is used to restore the system to a known state if a detected error
3542: occurs during the execution of the xt. A typical way to use @code{catch}
3543: would be:
1.1 anton 3544:
1.26 crook 3545: @example
3546: ... ['] foo catch IF ...
3547: @end example
1.1 anton 3548:
1.26 crook 3549: Whilst @code{foo} executes, it can call other words to any level of
3550: nesting, as usual. If @code{foo} (and all the words that it calls)
3551: execute successfully, control will ultimately passes to the word following
3552: the @code{catch}, and there will be a @code{true} flag (0) at
3553: TOS. However, if any word detects an error, it can terminate the
3554: execution of @code{foo} by pushing an error code onto the stack and then
3555: performing a @code{throw}. The execution of @code{throw} will pass
3556: control to the word following the @code{catch}, but this time the TOS
3557: will hold the error code. Therefore, the @code{IF} in the example
3558: can be used to determine whether @code{foo} executed successfully.
1.1 anton 3559:
1.26 crook 3560: This simple example shows how you can use @code{throw} and @code{catch}
3561: to ``take over'' exception handling from the system:
1.1 anton 3562: @example
1.26 crook 3563: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
1.1 anton 3564: @end example
3565:
1.26 crook 3566: The next example is more sophisticated and shows a multi-level
3567: @code{throw} and @code{catch}. To understand this example, start at the
3568: definition of @code{top-level} and work backwards:
3569:
1.1 anton 3570: @example
1.26 crook 3571: : lowest-level ( -- c )
3572: key dup 27 = if
3573: 1 throw \ ESCAPE key pressed
3574: else
3575: ." lowest-level successfull" CR
3576: then
3577: ;
3578:
3579: : lower-level ( -- c )
3580: lowest-level
3581: \ at this level consider a CTRL-U to be a fatal error
3582: dup 21 = if \ CTRL-U
3583: 2 throw
3584: else
3585: ." lower-level successfull" CR
3586: then
3587: ;
3588:
3589: : low-level ( -- c )
3590: ['] lower-level catch
3591: ?dup if
3592: \ error occurred - do we recognise it?
3593: dup 1 = if
3594: \ ESCAPE key pressed.. pretend it was an E
3595: [char] E
3596: else throw \ propogate the error upwards
3597: then
3598: then
3599: ." low-level successfull" CR
3600: ;
3601:
3602: : top-level ( -- )
3603: CR ['] low-level catch \ CATCH is used like EXECUTE
3604: ?dup if \ error occurred..
3605: ." Error " . ." occurred - contact your supplier"
3606: else
3607: ." The '" emit ." ' key was pressed" CR
3608: then
3609: ;
1.1 anton 3610: @end example
3611:
1.26 crook 3612: The ANS Forth document assigns @code{throw} codes thus:
1.1 anton 3613:
1.26 crook 3614: @itemize @bullet
3615: @item
3616: codes in the range -1 -- -255 are reserved to be assigned by the
3617: Standard. Assignments for codes in the range -1 -- -58 are currently
3618: documented in the Standard. In particular, @code{-1 throw} is equivalent
3619: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
3620: @item
3621: codes in the range -256 -- -4095 are reserved to be assigned by the system.
3622: @item
3623: all other codes may be assigned by programs.
3624: @end itemize
1.1 anton 3625:
1.26 crook 3626: Gforth provides the word @code{exception} as a mechanism for assigning
3627: system throw codes to applications. This allows multiple applications to
3628: co-exist in memory without any clash of @code{throw} codes. A definition
3629: of @code{exception} in ANS Forth is provided in
3630: @file{compat/exception.fs}.
1.1 anton 3631:
1.26 crook 3632: doc-quit
3633: doc-abort
3634: doc-abort"
1.1 anton 3635:
1.26 crook 3636: doc-catch
1.29 ! crook 3637: doc-throw
! 3638: doc---exception-exception
! 3639:
! 3640:
! 3641: @c -------------------------------------------------------------
! 3642: @node Defining Words, The Text Interpreter, Control Structures, Words
! 3643: @section Defining Words
! 3644: @cindex defining words
! 3645:
! 3646: @menu
! 3647: * Simple Defining Words:: Variables, values and constants
! 3648: * Colon Definitions::
! 3649: * User-defined Defining Words::
! 3650: * Supplying names::
! 3651: * Interpretation and Compilation Semantics::
! 3652: @end menu
! 3653:
! 3654: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
! 3655: @subsection Simple Defining Words
! 3656: @cindex simple defining words
! 3657: @cindex defining words, simple
! 3658:
! 3659: Defining words are used to create new entries in the dictionary. The
! 3660: simplest defining word is @code{CREATE}. @code{CREATE} is used like
! 3661: this:
! 3662:
! 3663: @example
! 3664: CREATE new-word1
! 3665: @end example
! 3666:
! 3667: @code{CREATE} is a parsing word that generates a dictionary entry for
! 3668: @code{new-word1}. When @code{new-word1} is executed, all that it does is
! 3669: leave an address on the stack. The address represents the value of
! 3670: the data space pointer (@code{HERE}) at the time that @code{new-word1}
! 3671: was defined. Therefore, @code{CREATE} is a way of associating a name
! 3672: with the address of a region of memory.
! 3673:
! 3674: By extending this example to reserve some memory in data space, we end
! 3675: up with a @i{variable}. Here are two different ways to do it:
! 3676:
! 3677: @example
! 3678: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
! 3679: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
! 3680: @end example
! 3681:
! 3682: The variable can be examined and modified using @code{@@} (``fetch'') and
! 3683: @code{!} (``store'') like this:
! 3684:
! 3685: @example
! 3686: new-word2 @@ . \ get address, fetch from it and display
! 3687: 1234 new-word2 ! \ new value, get address, store to it
! 3688: @end example
! 3689:
! 3690: As a final refinement, the whole code sequence can be wrapped up in a
! 3691: defining word (pre-empting the subject of the next section), making it
! 3692: easier to create new variables:
! 3693:
! 3694: @example
! 3695: : myvariable ( "name" -- a-addr ) CREATE 1 cells allot ;
! 3696:
! 3697: myvariable foo
! 3698: myvariable joe
! 3699:
! 3700: 45 3 * foo ! \ set foo to 135
! 3701: 1234 joe ! \ set joe to 1234
! 3702: 3 joe +! \ increment joe by 3.. to 1237
! 3703: @end example
! 3704:
! 3705: Not surprisingly, there is no need to define @code{myvariable}, since
! 3706: Forth already has a definition @code{Variable}. It behaves in exactly
! 3707: the same way as @code{myvariable} but it is implemented in an optimised
! 3708: way. Forth also provides @code{2Variable} and @code{fvariable} for
! 3709: double and floating-point variables, respectively.
! 3710:
! 3711: @cindex arrays
! 3712: A similar mechanism can be used to create arrays. For example, an
! 3713: 80-character text input buffer:
! 3714:
! 3715: @example
! 3716: CREATE text-buf 80 chars allot
! 3717:
! 3718: text-buf 0 chars c@@ \ the 1st character (offset 0)
! 3719: text-buf 3 chars c@@ \ the 4th character (offset 3)
! 3720: @end example
! 3721:
! 3722: You can build arbitrarily complex data structures by allocating
! 3723: appropriate areas of memory. @xref{Structures} for further discussions
! 3724: of this, and to learn about some Gforth tools that make it easier.
! 3725:
! 3726: @cindex user variables
! 3727: @cindex user space
! 3728: The defining word @code{User} behaves in the same way as @code{Variable}.
! 3729: The difference is that it reserves space in @i{user (data) space} rather
! 3730: than normal data space. In a Forth system that has a multi-tasker, each
! 3731: task has its own set of user variables.
! 3732:
! 3733: @comment TODO is that stuff about user variables strictly correct? Is it
! 3734: @comment just terminal tasks that have user variables?
! 3735: @comment should document tasker.fs (with some examples) elsewhere
! 3736: @comment in this manual, then expand on user space and user variables.
! 3737:
! 3738: After @code{CREATE} and @code{Variable}s, the next defining word to
! 3739: consider is @code{Constant}. @code{Constant} allows you to declare a
! 3740: fixed value and refer to it by name. For example:
! 3741:
! 3742: @example
! 3743: 12 Constant INCHES-PER-FOOT
! 3744: 3E+08 fconstant SPEED-O-LIGHT
! 3745: @end example
! 3746:
! 3747: A @code{Variable} can be both read and written, so its run-time
! 3748: behaviour is to supply an address through which its current value can be
! 3749: manipulated. In contrast, the value of a @code{Constant} cannot be
! 3750: changed once it has been declared@footnote{Well, often it can be -- but
! 3751: not in a Standard, portable way. It's safer to use a @code{Value} (read
! 3752: on).} so it's not necessary to supply the address -- it is more
! 3753: efficient to return the value of the constant directly. That's exactly
! 3754: what happens; the run-time effect of a constant is to put its value on
! 3755: the top of the stack (@ref{User-defined Defining Words} describes one
! 3756: way of implementing @code{Constant}).
! 3757:
! 3758: Gforth also provides @code{2Constant} and @code{fconstant} for defining
! 3759: double and floating-point constants, respectively.
! 3760:
! 3761: Constants in Forth behave differently from their equivalents in other
! 3762: programming languages. In other languages, a constant (such as an EQU in
! 3763: assembler or a #define in C) only exists at compile-time; in the
! 3764: executable program the constant has been translated into an absolute
! 3765: number and, unless you are using a symbolic debugger, it's impossible to
! 3766: know what abstract thing that number represents. In Forth a constant has
! 3767: an entry in the name dictionary and remains there after the code that
! 3768: uses it has been defined. In fact, it must remain in the dictionary
! 3769: since it has run-time duties to perform. For example:
! 3770:
! 3771: @example
! 3772: 12 Constant INCHES-PER-FOOT
! 3773: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
! 3774: @end example
! 3775:
! 3776: @cindex in-lining of constants
! 3777: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
! 3778: associated with the constant @code{INCHES-PER-FOOT}. If you use
! 3779: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
! 3780: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
! 3781: attempt to optimise constants by in-lining them where they are used. You
! 3782: can force Gforth to in-line a constant like this:
! 3783:
! 3784: @example
! 3785: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
! 3786: @end example
! 3787:
! 3788: If you use @code{see} to decompile @i{this} version of
! 3789: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
! 3790: longer present. @xref{Interpret/Compile states} and @xref{Literals}
! 3791: explain to this works.
! 3792:
! 3793: In-lining constants in this way might improve execution time
! 3794: fractionally, and can ensure that a constant is now only referenced at
! 3795: compile-time. However, the definition of the constant still remains in
! 3796: the dictionary. Some Forth compilers provide a mechanism for controlling
! 3797: a second dictionary for holding transient words such that this second
! 3798: dictionary can be deleted later in order to recover memory
! 3799: space. However, there is no standard way of doing this.
! 3800:
! 3801: One aspect of constants and variables that can sometimes be confusing is
! 3802: that they have different stack effects; one returns its value whilst the
! 3803: other returns the address of its value. The defining word @code{Value}
! 3804: provides an alternative to @code{Variable}, and has the same stack
! 3805: effect as a constant. A @code{Value} needs an additional word, @code{TO}
! 3806: to allow its value to be changed. Here are some examples:
! 3807:
! 3808: @example
! 3809: 12 Value APPLES \ a Value is initialised when it is declared.. like a
! 3810: \ constant but unlike a variable
! 3811: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
! 3812: APPLES \ puts 34 on the top of the stack.
! 3813: @end example
! 3814:
! 3815: The defining word @code{Defer} allows you to define a word by name
! 3816: without defining its behaviour; the definition of its behaviour is
! 3817: deferred. Here are two situation where this can be useful:
! 3818:
! 3819: @itemize @bullet
! 3820: @item
! 3821: Where you want to allow the behaviour of a word to be altered later, and
! 3822: for all precompiled references to the word to change when its behaviour
! 3823: is changed.
! 3824: @item
! 3825: For mutual recursion; @xref{Calls and returns}.
! 3826: @end itemize
! 3827:
! 3828: In the following example, @code{foo} always invokes the version of
! 3829: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
! 3830: always invokes the version that prints ``@code{Hello}''. There is no way
! 3831: of getting @code{foo} to use the later version without re-ordering the
! 3832: source code and recompilng it.
! 3833:
! 3834: @example
! 3835: : greet ." Good morning" ;
! 3836: : foo ... greet ... ;
! 3837: : greet ." Hello" ;
! 3838: : bar ... greet ... ;
! 3839: @end example
! 3840:
! 3841: This problem can be solved by defining @code{greet} as a @code{Defer}red
! 3842: word. The behaviour of a @code{Defer}red word can be defined and
! 3843: redefined at any time by using @code{IS} to associate the xt of a
! 3844: previously-defined word with it. The previous example becomes:
! 3845:
! 3846: @example
! 3847: Defer greet
! 3848: : foo ... greet ... ;
! 3849: : bar ... greet ... ;
! 3850: : greet1 ." Good morning" ;
! 3851: : greet2 ." Hello" ;
! 3852: ' greet2 IS greet \ make greet behave like greet2
! 3853: @end example
! 3854:
! 3855: A deferred word can only inherit default semantics from the xt (because
! 3856: that is all that an xt can represent -- @pxref{Tokens for Words} for
! 3857: more discussion of this). However, the semantics of the deferred word
! 3858: itself can be modified at the time that it is defined. For example:
! 3859:
! 3860: @example
! 3861: : bar .... ; compile-only
! 3862: Defer fred immediate
! 3863: Defer jim
! 3864:
! 3865: ' bar IS jim \ jim has default semantics
! 3866: ' bar IS fred \ fred is immediate
! 3867: @end example
1.1 anton 3868:
1.29 ! crook 3869: The defining word @code{Alias} allows you to define a word by name that
! 3870: has the same behaviour as some other word. Here are two situation where
! 3871: this can be useful:
1.1 anton 3872:
1.29 ! crook 3873: @itemize @bullet
! 3874: @item
! 3875: When you want access to a word's definition from a different word list
! 3876: (for an example of this, see the definition of the @code{Root} word list
! 3877: in the Gforth source).
! 3878: @item
! 3879: When you want to create a synonym; a definition that can be known by
! 3880: either of two names (for example, @code{THEN} and @code{ENDIF} are
! 3881: aliases).
! 3882: @end itemize
1.1 anton 3883:
1.29 ! crook 3884: The word whose behaviour the alias is to inherit is represented by an
! 3885: xt. Therefore, the alias can only inherits default semantics from its
! 3886: ancestor. The semantics of the alias itself can be modified at the time
! 3887: that it is defined. For example:
1.1 anton 3888:
1.29 ! crook 3889: @example
! 3890: : foo ... ; immediate
1.1 anton 3891:
1.29 ! crook 3892: ' foo Alias bar \ bar is not an immediate word
! 3893: ' foo Alias fooby immediate \ fooby is an immediate word
! 3894: @end example
1.26 crook 3895:
1.29 ! crook 3896: Words that are aliases have the same xt. Their semantics can differ
! 3897: because the rules about a word's semantics are stored in the name
! 3898: dictionary, and the aliases each have their own dictionary entry. It
! 3899: follows that words that are aliases have different name tokens and may
! 3900: have the same or different compilation tokens. Once again, see
! 3901: @ref{Tokens for Words} for more discussions of this.
1.27 crook 3902:
1.29 ! crook 3903: doc-create
1.26 crook 3904: doc-variable
3905: doc-2variable
3906: doc-fvariable
3907: doc-user
1.29 ! crook 3908: doc-constant
! 3909: doc-2constant
! 3910: doc-fconstant
1.26 crook 3911: doc-value
3912: doc-to
3913: doc-defer
3914: doc-is
1.29 ! crook 3915: doc-alias
! 3916: @comment TODO document these: what's defers <is> [is]
! 3917: doc-what's
1.28 crook 3918: doc-defers
1.26 crook 3919:
3920: Definitions in ANS Forth for @code{defer}, @code{<is>} and
3921: @code{[is]} are provided in @file{compat/defer.fs}.
1.29 ! crook 3922:
1.1 anton 3923:
1.26 crook 3924: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
3925: @subsection Colon Definitions
3926: @cindex colon definitions
1.1 anton 3927:
1.26 crook 3928: @example
3929: : name ( ... -- ... )
3930: word1 word2 word3 ;
3931: @end example
1.1 anton 3932:
1.29 ! crook 3933: @noindent
! 3934: Creates a word called @code{name} that, upon execution, executes
1.26 crook 3935: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
1.1 anton 3936:
1.29 ! crook 3937: The explanation above is somewhat superficial. @xref{Your first
! 3938: definition} for simple examples of colon definitions, then
! 3939: @xref{Interpretation and Compilation Semantics} for an in-depth
! 3940: discussion of some of the issues involved.
1.26 crook 3941:
3942: doc-:
3943: doc-;
1.1 anton 3944:
1.26 crook 3945: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
3946: @subsection User-defined Defining Words
3947: @cindex user-defined defining words
3948: @cindex defining words, user-defined
1.1 anton 3949:
1.29 ! crook 3950: You can create a new defining word by wrapping defining-time code around
! 3951: an existing defining word and putting the sequence in a colon
! 3952: definition. For example, suppose that you have a word @code{stats} that
! 3953: gathers statistics about colon definitions given the @i{xt} of the
! 3954: definition, and you want every colon definition in your application to
! 3955: make a call to @code{stats}. You can define and use a new version of
! 3956: @code{:} like this:
! 3957:
! 3958: @example
! 3959: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
! 3960: ... ; \ other code
! 3961:
! 3962: : my: : lastxt postpone literal ['] stats compile, ;
! 3963:
! 3964: my: foo + - ;
! 3965: @end example
! 3966:
! 3967: When @code{foo} is defined using @code{my:} these steps occur:
! 3968:
! 3969: @itemize @bullet
! 3970: @item
! 3971: @code{my:} is executed.
! 3972: @item
! 3973: The @code{:} within the definition (the one between @code{my:} and
! 3974: @code{lastxt}) is executed, and does just what it always does; it parses
! 3975: the input stream for a name, builds a dictionary header for the name
! 3976: @code{foo} and switches @code{state} from interpret to compile.
! 3977: @item
! 3978: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
! 3979: being defined -- @code{foo} -- onto the stack.
! 3980: @item
! 3981: The code that was produced by @code{postpone literal} is executed; this
! 3982: causes the value on the stack to be compiled as a literal in the code
! 3983: area of @code{foo}.
! 3984: @item
! 3985: The code @code{['] stats} compiles a literal into the definition of
! 3986: @code{my:}. When @code{compile,} is executed, that literal -- the
! 3987: execution token for @code{stats} -- is layed down in the code area of
! 3988: @code{foo} , following the literal@footnote{Strictly speaking, the
! 3989: mechanism that @code{compile,} uses to convert an @i{xt} into something
! 3990: in the code area is implementation-dependent. A threaded implementation
! 3991: might spit out the execution token directly whilst another
! 3992: implementation might spit out a native code sequence.}.
! 3993: @item
! 3994: At this point, the execution of @code{my:} is complete, and control
! 3995: returns to the text interpreter. The text interpreter is in compile
! 3996: state, so subsequent text @code{+ -} is compiled into the definition of
! 3997: @code{foo} and the @code{;} terminates the definition as always.
! 3998: @end itemize
! 3999:
! 4000: You can use @code{see} to decompile a word that was defined using
! 4001: @code{my:} and see how it is different from a normal @code{:}
! 4002: definition. For example:
! 4003:
! 4004: @example
! 4005: : bar + - ; \ like foo but using : rather than my:
! 4006: see bar
! 4007: : bar
! 4008: + - ;
! 4009: see foo
! 4010: : foo
! 4011: 107645672 stats + - ;
! 4012:
! 4013: \ use ' stats . to show that 107645672 is the xt for stats
! 4014: @end example
! 4015:
! 4016:
! 4017: Rather than edit your application's source code to change every @code{:}
! 4018: to a @code{my:}, use a deferred word:
! 4019:
! 4020: @example
! 4021: : real: : ; \ retain access to the original
! 4022: defer : \ redefine as a deferred word
! 4023: ' my: IS : \ use special version of :
! 4024: \
! 4025: \ load application here
! 4026: \
! 4027: ' real: IS : \ go back to the original
! 4028: @end example
! 4029:
! 4030: You can use techniques like this to make new defining words in terms of
! 4031: @i{any} existing defining word.
1.1 anton 4032:
4033:
1.29 ! crook 4034: @cindex defining defining words
1.26 crook 4035: @cindex @code{CREATE} ... @code{DOES>}
4036: If you want the words defined with your defining words to behave
4037: differently from words defined with standard defining words, you can
4038: write your defining word like this:
1.1 anton 4039:
4040: @example
1.26 crook 4041: : def-word ( "name" -- )
1.29 ! crook 4042: CREATE @i{code1}
1.26 crook 4043: DOES> ( ... -- ... )
1.29 ! crook 4044: @i{code2} ;
1.26 crook 4045:
4046: def-word name
1.1 anton 4047: @end example
4048:
1.29 ! crook 4049: @cindex child words
! 4050: This fragment defines a @dfn{defining word} @code{def-word} and then
! 4051: executes it. When @code{def-word} executes, it @code{CREATE}s a new
! 4052: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
! 4053: is not executed at this time. The word @code{name} is sometimes called a
! 4054: @dfn{child} of @code{def-word}.
! 4055:
! 4056: When you execute @code{name}, the address of the body of @code{name} is
! 4057: put on the data stack and @i{code2} is executed (the address of the body
! 4058: of @code{name} is the address @code{HERE} returns immediately after the
! 4059: @code{CREATE}).
! 4060:
! 4061: @cindex atavism in child words
! 4062: You can use @code{def-word} to define a set of child word that behave
! 4063: differently, though atavistically; they all have a common run-time
! 4064: behaviour determined by @i{code2}. Typically, the @i{code1} sequence
! 4065: builds a data area in the body of the child word. The structure of the
! 4066: data is common to all children of @code{def-word}, but the data values
! 4067: are specific -- and private -- to each child word. When a child word is
! 4068: executed, the address of its private data area is passed as a parameter
! 4069: on TOS to be used and manipulated@footnote{It is legitimate both to read
! 4070: and write to this data area.} by @i{code2}.
! 4071:
! 4072: The two fragments of code that make up the defining words act (are
! 4073: executed) at two completely separate times:
1.1 anton 4074:
1.29 ! crook 4075: @itemize @bullet
! 4076: @item
! 4077: At @i{define time}, the defining word executes @i{code1} to generate a
! 4078: child word
! 4079: @item
! 4080: At @i{child execution time}, when a child word is invoked, @i{code2}
! 4081: is executed, using parameters (data) that are private and specific to
! 4082: the child word.
! 4083: @end itemize
! 4084:
! 4085: @c NAC I think this is a really bad example, because it diminishes
! 4086: @c rather than emphasising the fact that some important stuff happens
! 4087: @c at define time, and other important stuff happens at child-invocation
! 4088: @c time, and that those two times are potentially very different.
! 4089: @c
! 4090: @c In other words, if you make the following definitions:
! 4091: @c @example
! 4092: @c : def-word1 ( "name" -- )
! 4093: @c CREATE @i{code1} ;
! 4094: @c
! 4095: @c : action1 ( ... -- ... )
! 4096: @c @i{code2} ;
! 4097: @c
! 4098: @c def-word1 name1
! 4099: @c @end example
! 4100: @c
! 4101: @c Using @code{name1 action1} is equivalent to using @code{name}.
1.1 anton 4102:
1.29 ! crook 4103: The classic example is that you can define @code{CONSTANT} in this way:
1.26 crook 4104:
1.1 anton 4105: @example
1.29 ! crook 4106: : CONSTANT ( w "name" -- )
! 4107: CREATE ,
1.26 crook 4108: DOES> ( -- w )
4109: @@ ;
1.1 anton 4110: @end example
4111:
1.29 ! crook 4112: @comment There is a beautiful description of how this works and what
! 4113: @comment it does in the Forthwrite 100th edition.. as well as an elegant
! 4114: @comment commentary on the Counting Fruits problem.
! 4115:
! 4116: When you create a constant with @code{5 CONSTANT five}, a set of
! 4117: define-time actions take place; first a new word @code{five} is created,
! 4118: then the value 5 is laid down in the body of @code{five} with
! 4119: @code{,}. When @code{five} is invoked, the address of the body is put on
! 4120: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
! 4121: no code of its own; it simply contains a data field and a pointer to the
! 4122: code that follows @code{DOES>} in its defining word. That makes words
! 4123: created in this way very compact.
! 4124:
! 4125: The final example in this section is intended to remind you that space
! 4126: reserved in @code{CREATE}d words is @i{data} space and therefore can be
! 4127: both read and written by a Standard program@footnote{Exercise: use this
! 4128: example as a starting point for your own implementation of @code{Value}
! 4129: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
! 4130: @code{[']}.}:
! 4131:
! 4132: @example
! 4133: : foo ( "name" -- )
! 4134: CREATE -1 ,
! 4135: DOES> ( -- )
! 4136: @@ .;
! 4137:
! 4138: foo first-word
! 4139: foo second-word
! 4140:
! 4141: 123 ' first-word >BODY !
! 4142: @end example
! 4143:
! 4144: If @code{first-word} had been a @code{CREATE}d word, we could simply
! 4145: have executed it to get the address of its data field. However, since it
! 4146: was defined to have @code{DOES>} actions, its execution semantics are to
! 4147: perform those @code{DOES>} actions. To get the address of its data field
! 4148: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
! 4149: translate the xt into the address of the data field. When you execute
! 4150: @code{first-word}, it will display @code{123}. When you execute
! 4151: @code{second-word} it will display @code{-1}.
1.26 crook 4152:
4153: @cindex stack effect of @code{DOES>}-parts
4154: @cindex @code{DOES>}-parts, stack effect
1.29 ! crook 4155: In the examples above the stack comment after the @code{DOES>} specifies
1.26 crook 4156: the stack effect of the defined words, not the stack effect of the
4157: following code (the following code expects the address of the body on
4158: the top of stack, which is not reflected in the stack comment). This is
4159: the convention that I use and recommend (it clashes a bit with using
4160: locals declarations for stack effect specification, though).
1.1 anton 4161:
1.26 crook 4162: @subsubsection Applications of @code{CREATE..DOES>}
4163: @cindex @code{CREATE} ... @code{DOES>}, applications
1.1 anton 4164:
1.26 crook 4165: You may wonder how to use this feature. Here are some usage patterns:
1.1 anton 4166:
1.26 crook 4167: @cindex factoring similar colon definitions
4168: When you see a sequence of code occurring several times, and you can
4169: identify a meaning, you will factor it out as a colon definition. When
4170: you see similar colon definitions, you can factor them using
4171: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
4172: that look very similar:
1.1 anton 4173: @example
1.26 crook 4174: : ori, ( reg-target reg-source n -- )
4175: 0 asm-reg-reg-imm ;
4176: : andi, ( reg-target reg-source n -- )
4177: 1 asm-reg-reg-imm ;
1.1 anton 4178: @end example
4179:
1.26 crook 4180: @noindent
4181: This could be factored with:
4182: @example
4183: : reg-reg-imm ( op-code -- )
4184: CREATE ,
4185: DOES> ( reg-target reg-source n -- )
4186: @@ asm-reg-reg-imm ;
4187:
4188: 0 reg-reg-imm ori,
4189: 1 reg-reg-imm andi,
4190: @end example
1.1 anton 4191:
1.26 crook 4192: @cindex currying
4193: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
4194: supply a part of the parameters for a word (known as @dfn{currying} in
4195: the functional language community). E.g., @code{+} needs two
4196: parameters. Creating versions of @code{+} with one parameter fixed can
4197: be done like this:
1.1 anton 4198: @example
1.26 crook 4199: : curry+ ( n1 -- )
4200: CREATE ,
4201: DOES> ( n2 -- n1+n2 )
4202: @@ + ;
4203:
4204: 3 curry+ 3+
4205: -2 curry+ 2-
1.1 anton 4206: @end example
4207:
1.26 crook 4208: @subsubsection The gory details of @code{CREATE..DOES>}
4209: @cindex @code{CREATE} ... @code{DOES>}, details
1.1 anton 4210:
1.26 crook 4211: doc-does>
1.1 anton 4212:
1.26 crook 4213: @cindex @code{DOES>} in a separate definition
4214: This means that you need not use @code{CREATE} and @code{DOES>} in the
4215: same definition; you can put the @code{DOES>}-part in a separate
1.29 ! crook 4216: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
1.26 crook 4217: @example
4218: : does1
4219: DOES> ( ... -- ... )
4220: ... ;
1.1 anton 4221:
1.26 crook 4222: : does2
4223: DOES> ( ... -- ... )
4224: ... ;
1.1 anton 4225:
1.26 crook 4226: : def-word ( ... -- ... )
4227: create ...
4228: IF
4229: does1
4230: ELSE
4231: does2
4232: ENDIF ;
4233: @end example
1.1 anton 4234:
1.26 crook 4235: In this example, the selection of whether to use @code{does1} or
4236: @code{does2} is made at compile-time; at the time that the child word is
1.29 ! crook 4237: @code{CREATE}d.
1.1 anton 4238:
1.26 crook 4239: @cindex @code{DOES>} in interpretation state
4240: In a standard program you can apply a @code{DOES>}-part only if the last
4241: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
4242: will override the behaviour of the last word defined in any case. In a
4243: standard program, you can use @code{DOES>} only in a colon
4244: definition. In Gforth, you can also use it in interpretation state, in a
4245: kind of one-shot mode; for example:
1.1 anton 4246: @example
1.26 crook 4247: CREATE name ( ... -- ... )
1.29 ! crook 4248: @i{initialization}
1.26 crook 4249: DOES>
1.29 ! crook 4250: @i{code} ;
1.1 anton 4251: @end example
4252:
1.26 crook 4253: @noindent
4254: is equivalent to the standard:
1.1 anton 4255: @example
1.26 crook 4256: :noname
4257: DOES>
1.29 ! crook 4258: @i{code} ;
1.26 crook 4259: CREATE name EXECUTE ( ... -- ... )
1.29 ! crook 4260: @i{initialization}
1.1 anton 4261: @end example
4262:
1.26 crook 4263: You can get the address of the body of a word with:
4264:
4265: doc->body
1.1 anton 4266:
1.26 crook 4267: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
1.29 ! crook 4268: @subsection Supplying the name of a defined word
1.26 crook 4269: @cindex names for defined words
4270: @cindex defining words, name parameter
1.1 anton 4271:
1.26 crook 4272: @cindex defining words, name given in a string
1.29 ! crook 4273: By default, a defining word takes the name for the defined word from the
1.26 crook 4274: input stream. Sometimes you want to supply the name from a string. You
4275: can do this with:
1.1 anton 4276:
1.26 crook 4277: doc-nextname
1.1 anton 4278:
1.26 crook 4279: For example:
1.1 anton 4280:
1.26 crook 4281: @example
4282: s" foo" nextname create
4283: @end example
4284: @noindent
4285: is equivalent to:
4286: @example
4287: create foo
4288: @end example
1.1 anton 4289:
1.26 crook 4290: @cindex defining words without name
1.29 ! crook 4291: Sometimes you want to define an @dfn{anonymous word}; a word without a
1.26 crook 4292: name. You can do this with:
1.1 anton 4293:
1.26 crook 4294: doc-:noname
1.1 anton 4295:
1.26 crook 4296: This leaves the execution token for the word on the stack after the
4297: closing @code{;}. Here's an example in which a deferred word is
4298: initialised with an @code{xt} from an anonymous colon definition:
4299: @example
4300: Defer deferred
4301: :noname ( ... -- ... )
4302: ... ;
4303: IS deferred
4304: @end example
1.1 anton 4305:
1.29 ! crook 4306: @noindent
1.26 crook 4307: Gforth provides an alternative way of doing this, using two separate
4308: words:
1.1 anton 4309:
1.26 crook 4310: doc-noname
4311: @cindex execution token of last defined word
4312: doc-lastxt
1.1 anton 4313:
1.29 ! crook 4314: @noindent
1.26 crook 4315: The previous example can be rewritten using @code{noname} and
4316: @code{lastxt}:
1.1 anton 4317:
1.26 crook 4318: @example
4319: Defer deferred
4320: noname : ( ... -- ... )
4321: ... ;
4322: lastxt IS deferred
4323: @end example
1.1 anton 4324:
1.29 ! crook 4325: @noindent
1.26 crook 4326: @code{lastxt} also works when the last word was not defined as
1.29 ! crook 4327: @code{noname}. It also has the useful property that is is valid as soon
! 4328: as the header for a definition has been build. Thus:
! 4329:
! 4330: @example
! 4331: lastxt . : foo [ lastxt . ] ; ' foo .
! 4332: @end example
! 4333:
! 4334: @noindent
! 4335: prints 3 numbers; the last two are the same.
1.1 anton 4336:
4337:
1.26 crook 4338: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
4339: @subsection Interpretation and Compilation Semantics
4340: @cindex semantics, interpretation and compilation
1.1 anton 4341:
1.26 crook 4342: @cindex interpretation semantics
4343: The @dfn{interpretation semantics} of a word are what the text
4344: interpreter does when it encounters the word in interpret state. It also
4345: appears in some other contexts, e.g., the execution token returned by
1.29 ! crook 4346: @code{' @i{word}} identifies the interpretation semantics of
! 4347: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
! 4348: interpret-state text interpretation of @code{@i{word}}).
1.1 anton 4349:
1.26 crook 4350: @cindex compilation semantics
4351: The @dfn{compilation semantics} of a word are what the text interpreter
4352: does when it encounters the word in compile state. It also appears in
1.29 ! crook 4353: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
1.26 crook 4354: standard terminology, ``appends to the current definition''.} the
1.29 ! crook 4355: compilation semantics of @i{word}.
1.1 anton 4356:
1.26 crook 4357: @cindex execution semantics
4358: The standard also talks about @dfn{execution semantics}. They are used
4359: only for defining the interpretation and compilation semantics of many
4360: words. By default, the interpretation semantics of a word are to
4361: @code{execute} its execution semantics, and the compilation semantics of
4362: a word are to @code{compile,} its execution semantics.@footnote{In
4363: standard terminology: The default interpretation semantics are its
4364: execution semantics; the default compilation semantics are to append its
4365: execution semantics to the execution semantics of the current
4366: definition.}
4367:
4368: @comment TODO expand, make it co-operate with new sections on text interpreter.
4369:
4370: @cindex immediate words
4371: @cindex compile-only words
4372: You can change the semantics of the most-recently defined word:
4373:
4374: doc-immediate
4375: doc-compile-only
4376: doc-restrict
4377:
4378: Note that ticking (@code{'}) a compile-only word gives an error
4379: (``Interpreting a compile-only word'').
1.1 anton 4380:
1.26 crook 4381: Gforth also allows you to define words with arbitrary combinations of
4382: interpretation and compilation semantics.
1.1 anton 4383:
1.26 crook 4384: doc-interpret/compile:
1.1 anton 4385:
1.26 crook 4386: This feature was introduced for implementing @code{TO} and @code{S"}. I
4387: recommend that you do not define such words, as cute as they may be:
4388: they make it hard to get at both parts of the word in some contexts.
4389: E.g., assume you want to get an execution token for the compilation
4390: part. Instead, define two words, one that embodies the interpretation
4391: part, and one that embodies the compilation part. Once you have done
4392: that, you can define a combined word with @code{interpret/compile:} for
4393: the convenience of your users.
1.1 anton 4394:
1.26 crook 4395: You might try to use this feature to provide an optimizing
4396: implementation of the default compilation semantics of a word. For
4397: example, by defining:
1.1 anton 4398: @example
1.26 crook 4399: :noname
4400: foo bar ;
4401: :noname
4402: POSTPONE foo POSTPONE bar ;
1.29 ! crook 4403: interpret/compile: opti-foobar
1.1 anton 4404: @end example
1.26 crook 4405:
1.23 crook 4406: @noindent
1.26 crook 4407: as an optimizing version of:
4408:
1.1 anton 4409: @example
1.26 crook 4410: : foobar
4411: foo bar ;
1.1 anton 4412: @end example
4413:
1.26 crook 4414: Unfortunately, this does not work correctly with @code{[compile]},
4415: because @code{[compile]} assumes that the compilation semantics of all
4416: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
1.29 ! crook 4417: opti-foobar} would compile compilation semantics, whereas
! 4418: @code{[compile] foobar} would compile interpretation semantics.
1.1 anton 4419:
1.26 crook 4420: @cindex state-smart words (are a bad idea)
1.29 ! crook 4421: Some people try to use @dfn{state-smart} words to emulate the feature provided
1.26 crook 4422: by @code{interpret/compile:} (words are state-smart if they check
4423: @code{STATE} during execution). E.g., they would try to code
4424: @code{foobar} like this:
1.1 anton 4425:
1.26 crook 4426: @example
4427: : foobar
4428: STATE @@
4429: IF ( compilation state )
4430: POSTPONE foo POSTPONE bar
4431: ELSE
4432: foo bar
4433: ENDIF ; immediate
4434: @end example
1.1 anton 4435:
1.26 crook 4436: Although this works if @code{foobar} is only processed by the text
4437: interpreter, it does not work in other contexts (like @code{'} or
4438: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
4439: for a state-smart word, not for the interpretation semantics of the
4440: original @code{foobar}; when you execute this execution token (directly
4441: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
4442: state, the result will not be what you expected (i.e., it will not
4443: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
4444: write them@footnote{For a more detailed discussion of this topic, see
4445: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
4446: Ertl; presented at EuroForth '98 and available from
4447: @url{http://www.complang.tuwien.ac.at/papers/}}!
1.1 anton 4448:
1.26 crook 4449: @cindex defining words with arbitrary semantics combinations
4450: It is also possible to write defining words that define words with
4451: arbitrary combinations of interpretation and compilation semantics. In
4452: general, they look like this:
1.1 anton 4453:
1.26 crook 4454: @example
4455: : def-word
4456: create-interpret/compile
1.29 ! crook 4457: @i{code1}
1.26 crook 4458: interpretation>
1.29 ! crook 4459: @i{code2}
1.26 crook 4460: <interpretation
4461: compilation>
1.29 ! crook 4462: @i{code3}
1.26 crook 4463: <compilation ;
4464: @end example
1.1 anton 4465:
1.29 ! crook 4466: For a @i{word} defined with @code{def-word}, the interpretation
! 4467: semantics are to push the address of the body of @i{word} and perform
! 4468: @i{code2}, and the compilation semantics are to push the address of
! 4469: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
1.26 crook 4470: can also be defined like this (except that the defined constants don't
4471: behave correctly when @code{[compile]}d):
1.1 anton 4472:
1.26 crook 4473: @example
4474: : constant ( n "name" -- )
4475: create-interpret/compile
4476: ,
4477: interpretation> ( -- n )
4478: @@
4479: <interpretation
4480: compilation> ( compilation. -- ; run-time. -- n )
4481: @@ postpone literal
4482: <compilation ;
4483: @end example
1.1 anton 4484:
1.26 crook 4485: doc-create-interpret/compile
4486: doc-interpretation>
4487: doc-<interpretation
4488: doc-compilation>
4489: doc-<compilation
1.1 anton 4490:
1.29 ! crook 4491: Words defined with @code{interpret/compile:} and
1.26 crook 4492: @code{create-interpret/compile} have an extended header structure that
4493: differs from other words; however, unless you try to access them with
4494: plain address arithmetic, you should not notice this. Words for
4495: accessing the header structure usually know how to deal with this; e.g.,
1.29 ! crook 4496: @code{'} @i{word} @code{>body} also gives you the body of a word created
! 4497: with @code{create-interpret/compile}.
1.1 anton 4498:
1.27 crook 4499: doc-postpone
1.29 ! crook 4500: @comment TODO -- expand glossary text for POSTPONE
1.27 crook 4501:
1.26 crook 4502: @c ----------------------------------------------------------
4503: @node The Text Interpreter, Tokens for Words, Defining Words, Words
4504: @section The Text Interpreter
4505: @cindex interpreter - outer
4506: @cindex text interpreter
4507: @cindex outer interpreter
1.1 anton 4508:
1.29 ! crook 4509: The text interpreter@footnote{This is an expanded version of the
! 4510: material in @ref{Introducing the Text Interpreter}.} is an endless loop
! 4511: that processes input from the current input device. A popular
! 4512: implementation technique for Forth is to implement a @dfn{forth virtual
! 4513: machine} using a loop called the @dfn{inner interpreter}. Because of
! 4514: this naming, the text interpreter is also known as the @dfn{outer
1.27 crook 4515: interpreter}.
4516:
1.29 ! crook 4517: @cindex interpret state
! 4518: @cindex compile state
! 4519: The text interpreter operates in one of two states: @dfn{interpret
! 4520: state} and @dfn{compile state}. The current state is defined by the
! 4521: aptly-named variable, @code{state}.
! 4522:
! 4523: This section starts by describing how the text interpreter behaves when
! 4524: it is in interpret state, processing input from the user input device --
! 4525: the keyboard. This is the mode that a Forth system is in after it starts
! 4526: up.
! 4527:
! 4528: @cindex input buffer
! 4529: @cindex terminal input buffer
! 4530: The text interpreter works from an area of memory called the @dfn{input
! 4531: buffer}@footnote{When the text interpreter is processing input from the
! 4532: keyboard, this area of memory is called the @dfn{terminal input buffer}
! 4533: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
! 4534: @code{#TIB}.}, which stores your keyboard input when you press the
! 4535: <return> key. Starting at the beginning of the input buffer, it skips
! 4536: leading spaces (called @dfn{delimiters}) then parses a string (a
! 4537: sequence of non-space characters) until it reaches either a space
! 4538: character or the end of the buffer. Having parsed a string, it makes two
! 4539: attempts to process it:
1.27 crook 4540:
1.29 ! crook 4541: @cindex dictionary
1.27 crook 4542: @itemize @bullet
4543: @item
1.29 ! crook 4544: It looks for the string in a @dfn{dictionary} of definitions. If the
! 4545: string is found, the string names a @dfn{definition} (also known as a
! 4546: @dfn{word}) and the dictionary search returns information that allows
! 4547: the text interpreter to perform the word's @dfn{interpretation
! 4548: semantics}. In most cases, this simply means that the word will be
! 4549: executed.
1.27 crook 4550: @item
4551: If the string is not found in the dictionary, the text interpreter
1.29 ! crook 4552: attempts to treat it as a number, using the rules described in
! 4553: @ref{Number Conversion}. If the string represents a legal number in the
! 4554: current radix, the number is pushed onto a parameter stack (the data
! 4555: stack for integers, the floating-point stack for floating-point
! 4556: numbers).
! 4557: @end itemize
! 4558:
! 4559: If both attempts fail, or if the word is found in the dictionary but has
! 4560: no interpretation semantics@footnote{This happens if the word was
! 4561: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
! 4562: remainder of the input buffer, issues an error message and waits for
! 4563: more input. If one of the attempts succeeds, the text interpreter
! 4564: repeats the parsing process until the whole of the input buffer has been
! 4565: processed, at which point it prints the status message ``@code{ ok}''
! 4566: and waits for more input.
! 4567:
! 4568: @cindex parse area
! 4569: The text interpreter keeps track of its position in the input buffer by
! 4570: updating a variable called @code{>IN} (pronounced ``to-in''). The value
! 4571: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
! 4572: of the input buffer. The region from offset @code{>IN @@} to the end of
! 4573: the input buffer is called the @dfn{parse area}@footnote{In other words,
! 4574: the text interpreter processes the contents of the input buffer by
! 4575: parsing strings from the parse area until the parse area is empty.}.
! 4576: This example shows how @code{>IN} changes as the text interpreter parses
! 4577: the input buffer:
! 4578:
! 4579: @example
! 4580: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
! 4581: CR ." ->" TYPE ." <-" ; IMMEDIATE
! 4582:
! 4583: 1 2 3 remaining + remaining .
! 4584:
! 4585: : foo 1 2 3 remaining SWAP remaining ;
! 4586: @end example
! 4587:
! 4588: @noindent
! 4589: The result is:
! 4590:
! 4591: @example
! 4592: ->+ remaining .<-
! 4593: ->.<-5 ok
! 4594:
! 4595: ->SWAP remaining ;-<
! 4596: ->;<- ok
! 4597: @end example
! 4598:
! 4599: @cindex parsing words
! 4600: The value of @code{>IN} can also be modified by a word in the input
! 4601: buffer that is executed by the text interpreter. This means that a word
! 4602: can ``trick'' the text interpreter into either skipping a section of the
! 4603: input buffer@footnote{This is how parsing words work.} or into parsing a
! 4604: section twice. For example:
1.27 crook 4605:
1.29 ! crook 4606: @example
! 4607: : lat ." <<lat>>" ;
! 4608: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
! 4609: @end example
! 4610:
! 4611: @noindent
! 4612: When @code{flat} is executed, this output is produced@footnote{Exercise
! 4613: for the reader: what would happen if the @code{3} were replaced with
! 4614: @code{4}?}:
! 4615:
! 4616: @example
! 4617: <<flat>><<lat>>
! 4618: @end example
! 4619:
! 4620: @noindent
! 4621: Two important notes about the behaviour of the text interpreter:
1.27 crook 4622:
4623: @itemize @bullet
4624: @item
4625: It processes each input string to completion before parsing additional
1.29 ! crook 4626: characters from the input buffer.
! 4627: @item
! 4628: It treats the input buffer as a read-only region (and so must your code).
! 4629: @end itemize
! 4630:
! 4631: @noindent
! 4632: When the text interpreter is in compile state, its behaviour changes in
! 4633: these ways:
! 4634:
! 4635: @itemize @bullet
! 4636: @item
! 4637: If a parsed string is found in the dictionary, the text interpreter will
! 4638: perform the word's @dfn{compilation semantics}. In most cases, this
! 4639: simply means that the execution semantics of the word will be appended
! 4640: to the current definition.
1.27 crook 4641: @item
1.29 ! crook 4642: When a number is encountered, it is compiled into the current definition
! 4643: (as a literal) rather than being pushed onto a parameter stack.
! 4644: @item
! 4645: If an error occurs, @code{state} is modified to put the text interpreter
! 4646: back into interpret state.
! 4647: @item
! 4648: Each time a line is entered from the keyboard, Gforth prints
! 4649: ``@code{ compiled}'' rather than `` @code{ok}''.
! 4650: @end itemize
! 4651:
! 4652: @cindex text interpreter - input sources
! 4653: When the text interpreter is using an input device other than the
! 4654: keyboard, its behaviour changes in these ways:
! 4655:
! 4656: @itemize @bullet
! 4657: @item
! 4658: When the parse area is empty, the text interpreter attempts to refill
! 4659: the input buffer from the input source. When the input source is
! 4660: exhausted, the input source is set back to the user input device.
! 4661: @item
! 4662: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
! 4663: time the parse area is emptied.
! 4664: @item
! 4665: If an error occurs, the input source is set back to the user input
! 4666: device.
1.27 crook 4667: @end itemize
1.21 crook 4668:
1.29 ! crook 4669: @ref{Input Sources} describes this in more detail.
! 4670:
1.26 crook 4671: doc->in
1.27 crook 4672: doc-source
4673:
1.26 crook 4674: doc-tib
4675: doc-#tib
1.1 anton 4676:
1.26 crook 4677: @menu
1.29 ! crook 4678: * Input Sources::
1.26 crook 4679: * Number Conversion::
4680: * Interpret/Compile states::
4681: * Literals::
4682: * Interpreter Directives::
4683: @end menu
1.1 anton 4684:
1.29 ! crook 4685: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
! 4686: @subsection Input Sources
! 4687: @cindex input sources
! 4688: @cindex text interpreter - input sources
! 4689:
! 4690: By default, the text interpreter accepts input from the user input
! 4691: device (the keyboard) when Forth starts up. The text interpreter can
! 4692: process input from any of these sources:
! 4693:
! 4694: @itemize @bullet
! 4695: @item
! 4696: The user input device -- the keyboard.
! 4697: @item
! 4698: A file, using the words described in @ref{Forth source files}.
! 4699: @item
! 4700: A block, using the words described in @ref{Blocks}.
! 4701: @item
! 4702: A text string, using @code{evaluate}.
! 4703: @end itemize
! 4704:
! 4705: A program can identify the current input device from the values of
! 4706: @code{source-id} and @code{blk}.
! 4707:
! 4708: doc-source-id
! 4709: doc-blk
! 4710:
! 4711: doc-save-input
! 4712: doc-restore-input
! 4713:
! 4714: doc-evaluate
1.1 anton 4715:
1.29 ! crook 4716:
! 4717: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
1.26 crook 4718: @subsection Number Conversion
4719: @cindex number conversion
4720: @cindex double-cell numbers, input format
4721: @cindex input format for double-cell numbers
4722: @cindex single-cell numbers, input format
4723: @cindex input format for single-cell numbers
4724: @cindex floating-point numbers, input format
4725: @cindex input format for floating-point numbers
1.1 anton 4726:
1.29 ! crook 4727: This section describes the rules that the text interpreter uses when it
! 4728: tries to convert a string into a number.
1.1 anton 4729:
1.26 crook 4730: Let <digit> represent any character that is a legal digit in the current
1.29 ! crook 4731: number base@footnote{For example, 0-9 when the number base is decimal or
! 4732: 0-9, A-F when the number base is hexadecimal.}.
1.1 anton 4733:
1.26 crook 4734: Let <decimal digit> represent any character in the range 0-9.
1.1 anton 4735:
1.29 ! crook 4736: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
! 4737: in the braces (@i{a} or @i{b} or neither).
1.1 anton 4738:
1.26 crook 4739: Let * represent any number of instances of the previous character
4740: (including none).
1.1 anton 4741:
1.26 crook 4742: Let any other character represent itself.
1.1 anton 4743:
1.29 ! crook 4744: @noindent
1.26 crook 4745: Now, the conversion rules are:
1.21 crook 4746:
1.26 crook 4747: @itemize @bullet
4748: @item
4749: A string of the form <digit><digit>* is treated as a single-precision
1.29 ! crook 4750: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
1.26 crook 4751: @item
4752: A string of the form -<digit><digit>* is treated as a single-precision
1.29 ! crook 4753: (cell-sized) negative integer, and is represented using 2's-complement
1.26 crook 4754: arithmetic. Examples are -45 -5681 -0
4755: @item
4756: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
1.29 ! crook 4757: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
! 4758: (all three of these represent the same number).
1.26 crook 4759: @item
4760: A string of the form -<digit><digit>*.<digit>* is treated as a
1.29 ! crook 4761: double-precision (double-cell-sized) negative integer, and is
1.26 crook 4762: represented using 2's-complement arithmetic. Examples are -3465. -3.465
1.29 ! crook 4763: -34.65 (all three of these represent the same number).
1.26 crook 4764: @item
1.29 ! crook 4765: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
! 4766: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
1.26 crook 4767: number. Examples are 1e0 1.e 1.e0 +1e+0 (which all represent the same
1.29 ! crook 4768: number) +12.E-4
1.26 crook 4769: @end itemize
1.1 anton 4770:
1.26 crook 4771: By default, the number base used for integer number conversion is given
1.29 ! crook 4772: by the contents of the variable @code{BASE}. Base 10 (decimal) is
1.26 crook 4773: always used for floating-point number conversion.
1.1 anton 4774:
1.29 ! crook 4775: doc-dpl
1.26 crook 4776: doc-base
4777: doc-hex
4778: doc-decimal
1.1 anton 4779:
1.26 crook 4780: @cindex '-prefix for character strings
4781: @cindex &-prefix for decimal numbers
4782: @cindex %-prefix for binary numbers
4783: @cindex $-prefix for hexadecimal numbers
1.29 ! crook 4784: Gforth allows you to override the value of @code{BASE} by using a
! 4785: prefix@footnote{Some Forth implementations provide a similar scheme by
! 4786: implementing @code{$} etc. as parsing words that process the subsequent
! 4787: number in the input stream and push it onto the stack. For example, see
! 4788: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
! 4789: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
! 4790: is required between the prefix and the number.} before the first digit
! 4791: of an (integer) number. Four prefixes are supported:
1.1 anton 4792:
1.26 crook 4793: @itemize @bullet
4794: @item
4795: @code{&} -- decimal number
4796: @item
4797: @code{%} -- binary number
4798: @item
4799: @code{$} -- hexadecimal number
4800: @item
4801: @code{'} -- base 256 number
4802: @end itemize
1.1 anton 4803:
1.26 crook 4804: Here are some examples, with the equivalent decimal number shown after
4805: in braces:
1.1 anton 4806:
1.26 crook 4807: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
4808: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
4809: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
4810: &905 (905), $abc (2478), $ABC (2478).
1.1 anton 4811:
1.26 crook 4812: @cindex number conversion - traps for the unwary
1.29 ! crook 4813: @noindent
1.26 crook 4814: Number conversion has a number of traps for the unwary:
1.1 anton 4815:
1.26 crook 4816: @itemize @bullet
4817: @item
4818: You cannot determine the current number base using the code sequence
4819: @code{BASE @@ .} -- the number base is always 10 in the current number
4820: base. Instead, use something like @code{BASE @@ DECIMAL DUP . BASE !}
4821: @item
4822: If the number base is set to a value greater than 14 (for example,
4823: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
4824: it to be intepreted as either a single-precision integer or a
4825: floating-point number (Gforth treats it as an integer). The ambiguity
4826: can be resolved by explicitly stating the sign of the mantissa and/or
4827: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
4828: ambiguity arises; either representation will be treated as a
4829: floating-point number.
4830: @item
1.29 ! crook 4831: There is a word @code{bin} but it does @i{not} set the number base!
1.26 crook 4832: It is used to specify file types.
4833: @item
4834: ANS Forth requires the @code{.} of a double-precision number to
4835: be the final character in the string. Allowing the @code{.} to be
4836: anywhere after the first digit is a Gforth extension.
4837: @item
4838: The number conversion process does not check for overflow.
4839: @item
4840: In Gforth, number conversion to floating-point numbers always use base
4841: 10, irrespective of the value of @code{BASE}. In ANS Forth,
4842: conversion to floating-point numbers whilst the value of
4843: @code{BASE} is not 10 is an ambiguous condition.
4844: @end itemize
1.1 anton 4845:
1.29 ! crook 4846: @ref{Input} describes words that you can use to read numbers into your
! 4847: programs.
1.1 anton 4848:
1.26 crook 4849: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
4850: @subsection Interpret/Compile states
4851: @cindex Interpret/Compile states
1.1 anton 4852:
1.29 ! crook 4853: A standard program is not permitted to change @code{state}
! 4854: explicitly. However, it can change @code{state} implicitly, using the
! 4855: words @code{[} and @code{]}. When @code{[} is executed it switches
! 4856: @code{state} to interpret state, and therefore the text interpreter
! 4857: starts interpreting. When @code{]} is executed it switches @code{state}
! 4858: to compile state and therefore the text interpreter starts
! 4859: compiling. The most common usage for these words is to compile literals,
! 4860: as shown in @ref{Literals}. However, they give you the freedom to switch
! 4861: modes at will. Here is an example of building a jump-table of execution
! 4862: tokens:
! 4863:
! 4864: @example
! 4865: : AA ." this is A" ;
! 4866: : BB ." this is B" ;
! 4867: : CC ." this is C" ;
! 4868:
! 4869: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
! 4870: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
! 4871: cells table + @ execute ;
! 4872: @end example
! 4873:
! 4874: @noindent
! 4875: Now @code{0 go} will display ``@code{this is A}''. The table can be
! 4876: built far more neatly@footnote{The source code is neater.. what is
! 4877: compiled in memory in each case is identical.} like this:
! 4878:
! 4879: @example
! 4880: create table ] aa bb cc [
! 4881: @end example
! 4882:
! 4883: The problem with this code is that it is not portable; it will only work
! 4884: on systems where code space and data space co-incide. The reason is that
! 4885: both tables @i{compile} execution tokens -- into code space. The
! 4886: Standard only allows data space to be assigned for a @code{CREATE}d
! 4887: word. In addition, the Standard only allows @code{@@} to access data
! 4888: space, whilst this example is using it to access code space. The only
! 4889: portable, Standard way to build this table is to build it in data space,
! 4890: like this:
! 4891:
! 4892: @example
! 4893: create table ' aa , ' bb , ' cc ,
! 4894: @end example
! 4895:
! 4896: @noindent
! 4897: A similar technique can be used to build a table of constants:
! 4898:
! 4899: @example
! 4900: create primes 1 , 3 , 5 , 7 , 11 ,
! 4901: @end example
1.1 anton 4902:
1.26 crook 4903: doc-state
4904: doc-[
4905: doc-]
1.1 anton 4906:
1.26 crook 4907: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
4908: @subsection Literals
4909: @cindex Literals
1.21 crook 4910:
1.29 ! crook 4911: Often, you want to use a number within a colon definition. When you do
! 4912: this, the text interpreter automatically compiles the number as a
! 4913: @i{literal}. A literal is a number whose run-time effect is to be pushed
! 4914: onto the stack. If you had to do some maths to generate the number, you
! 4915: might write it like this:
! 4916:
! 4917: @example
! 4918: : HOUR-TO-SEC ( n1 -- n2 )
! 4919: 60 * \ to minutes
! 4920: 60 * ; \ to seconds
! 4921: @end example
! 4922:
! 4923: It is very clear what this definition is doing, but it's inefficient
! 4924: since it is performing 2 multiples at run-time. An alternative would be
! 4925: to write:
! 4926:
! 4927: @example
! 4928: : HOUR-TO-SEC ( n1 -- n2 )
! 4929: 3600 * ; \ to seconds
! 4930: @end example
! 4931:
! 4932: Which does the same thing, and has the advantage of using a single
! 4933: multiply. Ideally, we'd like the efficiency of the second with the
! 4934: readability of the first.
! 4935:
! 4936: @code{Literal} allows us to achieve that. It takes a number from the
! 4937: stack and lays it down in the current definition just as though the
! 4938: number had been typed directly into the definition. Our first attempt
! 4939: might look like this:
! 4940:
! 4941: @example
! 4942: 60 \ mins per hour
! 4943: 60 * \ seconds per minute
! 4944: : HOUR-TO-SEC ( n1 -- n2 )
! 4945: Literal * ; \ to seconds
! 4946: @end example
! 4947:
! 4948: But this produces the error message @code{unstructured}. What happened?
! 4949: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
! 4950: @i{colon-sys} is implementation-defined. In other words, once we start a
! 4951: colon definition we can't portably access anything that was on the stack
! 4952: before the definition began@footnote{@cite{Two Problems in ANS Forth},
! 4953: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
! 4954: some situations where you might want to access stack items above
! 4955: colon-sys, and provides a solution to the problem.}. The correct way of
! 4956: solving this problem in this instance is to use @code{[ ]} like this:
! 4957:
! 4958: @example
! 4959: : HOUR-TO-SEC ( n1 -- n2 )
! 4960: [ 60 \ minutes per hour
! 4961: 60 * ] \ seconds per minute
! 4962: LITERAL * ; \ to seconds
! 4963: @end example
1.23 crook 4964:
1.26 crook 4965: doc-literal
4966: doc-]L
4967: doc-2literal
4968: doc-fliteral
1.1 anton 4969:
1.29 ! crook 4970: @node Interpreter Directives, , Literals, The Text Interpreter
1.26 crook 4971: @subsection Interpreter Directives
4972: @cindex interpreter directives
1.1 anton 4973:
1.29 ! crook 4974: These words are usually used in interpret state; typically to control
! 4975: which parts of a source file are processed by the text
1.26 crook 4976: interpreter. There are only a few ANS Forth Standard words, but Gforth
4977: supplements these with a rich set of immediate control structure words
4978: to compensate for the fact that the non-immediate versions can only be
1.29 ! crook 4979: used in compile state (@pxref{Control Structures}). Typical usages:
! 4980:
! 4981: @example
! 4982: FALSE Constant ASSEMBLER
! 4983: .
! 4984: .
! 4985: ASSEMBLER [IF]
! 4986: : ASSEMBLER-FEATURE
! 4987: ...
! 4988: ;
! 4989: [ENDIF]
! 4990: .
! 4991: .
! 4992: : SEE
! 4993: ... \ general-purpose SEE code
! 4994: [ ASSEMBLER [IF] ]
! 4995: ... \ assembler-specific SEE code
! 4996: [ [ENDIF] ]
! 4997: ;
! 4998: @end example
1.1 anton 4999:
1.26 crook 5000: doc-[IF]
5001: doc-[ELSE]
5002: doc-[THEN]
5003: doc-[ENDIF]
1.1 anton 5004:
1.26 crook 5005: doc-[IFDEF]
5006: doc-[IFUNDEF]
1.1 anton 5007:
1.26 crook 5008: doc-[?DO]
5009: doc-[DO]
5010: doc-[FOR]
5011: doc-[LOOP]
5012: doc-[+LOOP]
5013: doc-[NEXT]
1.1 anton 5014:
1.26 crook 5015: doc-[BEGIN]
5016: doc-[UNTIL]
5017: doc-[AGAIN]
5018: doc-[WHILE]
5019: doc-[REPEAT]
1.1 anton 5020:
1.27 crook 5021:
5022:
1.26 crook 5023: @c -------------------------------------------------------------
5024: @node Tokens for Words, Word Lists, The Text Interpreter, Words
5025: @section Tokens for Words
5026: @cindex tokens for words
1.1 anton 5027:
1.28 crook 5028: This section describes the creation and use of tokens that represent
1.29 ! crook 5029: words.
! 5030:
! 5031: Named words have information stored in their name dictionary entries to
! 5032: indicate any non-default semantics (@pxref{Interpretation and
! 5033: Compilation Semantics}). The semantics can be modified, using
! 5034: @code{immediate} and/or @code{compile-only}, at the time that the words
! 5035: are defined. Unnamed words have (by definition) no name dictionary
! 5036: entry, and therefore must have default semantics.
1.21 crook 5037:
1.26 crook 5038: Named words have interpretation and compilation semantics. Unnamed words
5039: just have execution semantics.
1.21 crook 5040:
1.29 ! crook 5041: @cindex xt
! 5042: @cindex execution token
! 5043: The execution semantics of an unnamed word are represented by an
! 5044: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
! 5045: the execution token of the last word defined can be produced with
! 5046: @code{lastxt}.
! 5047:
! 5048: The interpretation semantics of a named word are also represented by an
! 5049: execution token. You can produce the execution token using @code{'} or
! 5050: @code{[']}. A simple example shows the difference between the two:
1.21 crook 5051:
1.29 ! crook 5052: @example
! 5053: : greet ( -- ) ." Hello" ;
! 5054: : foo ( -- xt ) ['] greet ; \ ['] parses greet at compile-time
! 5055: : bar ( -- ) ' EXECUTE ; \ ' parses at run-time
1.1 anton 5056:
1.29 ! crook 5057: \ the next four lines all do the same thing
! 5058: foo EXECUTE
! 5059: greet
! 5060: ' greet EXECUTE
! 5061: boo greet
! 5062: @end example
1.1 anton 5063:
1.29 ! crook 5064: An execution token occupies one cell.
1.26 crook 5065: @cindex code field address
5066: @cindex CFA
1.29 ! crook 5067: In Gforth, the abstract data type @i{execution token} is implemented
1.26 crook 5068: as a code field address (CFA).
5069: @comment TODO note that the standard does not say what it represents..
5070: @comment and you cannot necessarily compile it in all Forths (eg native
5071: @comment compilers?).
1.1 anton 5072:
1.29 ! crook 5073: For literals, use @code{'} in interpreted code and @code{[']} in
! 5074: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
! 5075: unusually by complaining about compile-only words. To get the execution
! 5076: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
! 5077: or @code{[COMP'] @i{name} DROP}.
1.1 anton 5078:
1.26 crook 5079: @cindex compilation token
1.29 ! crook 5080: The compilation semantics of a named word are represented by a
! 5081: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
! 5082: @i{xt} is an execution token. The compilation semantics represented by
! 5083: the compilation token can be performed with @code{execute}, which
! 5084: consumes the whole compilation token, with an additional stack effect
! 5085: determined by the represented compilation semantics.
! 5086:
! 5087: At present, the @i{w} part of a compilation token is an execution token,
! 5088: and the @i{xt} part represents either @code{execute} or
! 5089: @code{compile,}@footnote{Depending upon the compilation semantics of the
! 5090: word. If the word has default compilation semantics, the @i{xt} will
! 5091: represent @code{compile,}. If the word is @code{immediate}, the @i{xt}
! 5092: will represent @code{execute}.}. However, don't rely on that knowledge,
! 5093: unless necessary; future versions of Gforth may introduce unusual
! 5094: compilation tokens (e.g., a compilation token that represents the
! 5095: compilation semantics of a literal).
1.1 anton 5096:
1.26 crook 5097: You can compile the compilation semantics with @code{postpone,}. I.e.,
1.29 ! crook 5098: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
! 5099: @i{word}}.
1.21 crook 5100:
1.26 crook 5101: @cindex name token
5102: @cindex name field address
5103: @cindex NFA
1.29 ! crook 5104: Named words are also represented by the @dfn{name token}, (@i{nt}). In
! 5105: Gforth, the abstract data type @emph{name token} is implemented as a
! 5106: name field address (NFA).
! 5107:
! 5108: doc-execute
! 5109: doc-compile,
! 5110: doc-[']
! 5111: doc-'
! 5112: doc-[comp']
! 5113: doc-comp'
! 5114: doc-postpone,
1.1 anton 5115:
1.26 crook 5116: doc-find-name
5117: doc-name>int
5118: doc-name?int
5119: doc-name>comp
5120: doc-name>string
1.1 anton 5121:
1.26 crook 5122: @c -------------------------------------------------------------
5123: @node Word Lists, Environmental Queries, Tokens for Words, Words
5124: @section Word Lists
5125: @cindex word lists
5126: @cindex name dictionary
1.1 anton 5127:
1.26 crook 5128: @cindex wid
5129: All definitions other than those created by @code{:noname} have an entry
5130: in the name dictionary. The name dictionary is fragmented into a number
1.29 ! crook 5131: of parts, called @dfn{word lists}. A word list is identified by a
! 5132: cell-sized word list identifier (@i{wid}) in much the same way as a
1.26 crook 5133: file is identified by a file handle. The numerical value of the wid has
5134: no (portable) meaning, and might change from session to session.
1.1 anton 5135:
1.26 crook 5136: @cindex compilation word list
5137: At any one time, a single word list is defined as the word list to which
1.29 ! crook 5138: all new definitions will be added -- this is called the @dfn{compilation
1.26 crook 5139: word list}. When Gforth is started, the compilation word list is the
5140: word list called @code{FORTH-WORDLIST}.
1.1 anton 5141:
1.26 crook 5142: @cindex search order stack
1.29 ! crook 5143: Forth maintains a stack of word lists, representing the @dfn{search
1.26 crook 5144: order}. When the name dictionary is searched (for example, when
5145: attempting to find a word's execution token during compilation), only
5146: those word lists that are currently in the search order are
5147: searched. The most recently-defined word in the word list at the top of
5148: the word list stack is searched first, and the search proceeds until
5149: either the word is located or the oldest definition in the word list at
5150: the bottom of the stack is reached. Definitions of the word may exist in
5151: more than one word lists; the search order determines which version will
5152: be found.
1.1 anton 5153:
1.29 ! crook 5154: The ANS Forth ``Search order'' word set is intended to provide a set of
! 5155: low-level tools that allow various different schemes to be
1.26 crook 5156: implemented. Gforth provides @code{vocabulary}, a traditional Forth
5157: word. @file{compat/vocabulary.fs} provides an implementation in ANS
5158: Standard Forth.
1.1 anton 5159:
1.27 crook 5160: @comment TODO: locals section refers to here, saying that every word list (aka
5161: @comment vocabulary) has its own methods for searching etc. Need to document that.
1.1 anton 5162:
1.27 crook 5163: @comment the thisone- prefix is used to pick out the true definition of a
5164: @comment word from the source files, rather than some alias.
1.26 crook 5165: doc-forth-wordlist
5166: doc-definitions
5167: doc-get-current
5168: doc-set-current
5169: doc-get-order
1.27 crook 5170: doc---thisone-set-order
1.26 crook 5171: doc-wordlist
5172: doc-also
1.27 crook 5173: doc---thisone-forth
1.26 crook 5174: doc-only
1.27 crook 5175: doc---thisone-order
1.26 crook 5176: doc-previous
1.15 anton 5177:
1.26 crook 5178: doc-find
5179: doc-search-wordlist
1.15 anton 5180:
1.26 crook 5181: doc-words
5182: doc-vlist
1.1 anton 5183:
1.26 crook 5184: doc-mappedwordlist
5185: doc-root
5186: doc-vocabulary
5187: doc-seal
5188: doc-vocs
5189: doc-current
5190: doc-context
1.1 anton 5191:
1.26 crook 5192: @menu
5193: * Why use word lists?::
5194: * Word list examples::
5195: @end menu
5196:
5197: @node Why use word lists?, Word list examples, Word Lists, Word Lists
5198: @subsection Why use word lists?
5199: @cindex word lists - why use them?
5200:
1.29 ! crook 5201: Here are some reasons for using multiple word lists:
1.26 crook 5202:
5203: @itemize @bullet
5204: @item
5205: To improve compilation speed by reducing the number of name dictionary
5206: entries that must be searched. This is achieved by creating a new
5207: word list that contains all of the definitions that are used in the
5208: definition of a Forth system but which would not usually be used by
5209: programs running on that system. That word list would be on the search
5210: list when the Forth system was compiled but would be removed from the
5211: search list for normal operation. This can be a useful technique for
5212: low-performance systems (for example, 8-bit processors in embedded
5213: systems) but is unlikely to be necessary in high-performance desktop
5214: systems.
5215: @item
5216: To prevent a set of words from being used outside the context in which
5217: they are valid. Two classic examples of this are an integrated editor
5218: (all of the edit commands are defined in a separate word list; the
5219: search order is set to the editor word list when the editor is invoked;
5220: the old search order is restored when the editor is terminated) and an
5221: integrated assembler (the op-codes for the machine are defined in a
5222: separate word list which is used when a @code{CODE} word is defined).
5223: @item
5224: To prevent a name-space clash between multiple definitions with the same
5225: name. For example, when building a cross-compiler you might have a word
5226: @code{IF} that generates conditional code for your target system. By
5227: placing this definition in a different word list you can control whether
5228: the host system's @code{IF} or the target system's @code{IF} get used in
5229: any particular context by controlling the order of the word lists on the
5230: search order stack.
5231: @end itemize
1.1 anton 5232:
1.26 crook 5233: @node Word list examples, ,Why use word lists?, Word Lists
5234: @subsection Word list examples
5235: @cindex word lists - examples
1.1 anton 5236:
1.26 crook 5237: Here is an example of creating and using a new wordlist using ANS
5238: Forth Standard words:
1.1 anton 5239:
5240: @example
1.26 crook 5241: wordlist constant my-new-words-wordlist
5242: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
1.21 crook 5243:
1.26 crook 5244: \ add it to the search order
5245: also my-new-words
1.21 crook 5246:
1.26 crook 5247: \ alternatively, add it to the search order and make it
5248: \ the compilation word list
5249: also my-new-words definitions
5250: \ type "order" to see the problem
1.21 crook 5251: @end example
5252:
1.26 crook 5253: The problem with this example is that @code{order} has no way to
5254: associate the name @code{my-new-words} with the wid of the word list (in
5255: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
5256: that has no associated name). There is no Standard way of associating a
5257: name with a wid.
5258:
5259: In Gforth, this example can be re-coded using @code{vocabulary}, which
5260: associates a name with a wid:
1.21 crook 5261:
1.26 crook 5262: @example
5263: vocabulary my-new-words
1.21 crook 5264:
1.26 crook 5265: \ add it to the search order
5266: my-new-words
1.21 crook 5267:
1.26 crook 5268: \ alternatively, add it to the search order and make it
5269: \ the compilation word list
5270: my-new-words definitions
5271: \ type "order" to see that the problem is solved
5272: @end example
1.23 crook 5273:
1.26 crook 5274: @c -------------------------------------------------------------
5275: @node Environmental Queries, Files, Word Lists, Words
5276: @section Environmental Queries
5277: @cindex environmental queries
1.21 crook 5278:
1.26 crook 5279: ANS Forth introduced the idea of ``environmental queries'' as a way
5280: for a program running on a system to determine certain characteristics of the system.
5281: The Standard specifies a number of strings that might be recognised by a system.
1.21 crook 5282:
1.26 crook 5283: The Standard requires that the name space used for environmental queries
5284: be distinct from the name space used for definitions.
1.21 crook 5285:
1.26 crook 5286: Typically, environmental queries are supported by creating a set of
1.29 ! crook 5287: definitions in a word list that is @i{only} used during environmental
1.26 crook 5288: queries; that is what Gforth does. There is no Standard way of adding
5289: definitions to the set of recognised environmental queries, but any
5290: implementation that supports the loading of optional word sets must have
5291: some mechanism for doing this (after loading the word set, the
5292: associated environmental query string must return @code{true}). In
5293: Gforth, the word list used to honour environmental queries can be
5294: manipulated just like any other word list.
1.21 crook 5295:
1.26 crook 5296: doc-environment?
5297: doc-environment-wordlist
1.21 crook 5298:
1.26 crook 5299: doc-gforth
5300: doc-os-class
1.21 crook 5301:
1.26 crook 5302: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
5303: returning two items on the stack, querying it using @code{environment?}
5304: will return an additional item; the @code{true} flag that shows that the
5305: string was recognised.
1.21 crook 5306:
1.26 crook 5307: @comment TODO Document the standard strings or note where they are documented herein
1.21 crook 5308:
1.26 crook 5309: Here are some examples of using environmental queries:
1.21 crook 5310:
1.26 crook 5311: @example
5312: s" address-unit-bits" environment? 0=
5313: [IF]
5314: cr .( environmental attribute address-units-bits unknown... ) cr
5315: [THEN]
1.21 crook 5316:
1.26 crook 5317: s" block" environment? [IF] DROP include block.fs [THEN]
1.21 crook 5318:
1.26 crook 5319: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
1.21 crook 5320:
1.26 crook 5321: s" gforth" environment? [IF] .( Gforth version ) TYPE
5322: [ELSE] .( Not Gforth..) [THEN]
5323: @end example
1.21 crook 5324:
5325:
1.26 crook 5326: Here is an example of adding a definition to the environment word list:
1.21 crook 5327:
1.26 crook 5328: @example
5329: get-current environment-wordlist set-current
5330: true constant block
5331: true constant block-ext
5332: set-current
5333: @end example
1.21 crook 5334:
1.26 crook 5335: You can see what definitions are in the environment word list like this:
1.21 crook 5336:
1.26 crook 5337: @example
5338: get-order 1+ environment-wordlist swap set-order words previous
5339: @end example
1.21 crook 5340:
5341:
1.26 crook 5342: @c -------------------------------------------------------------
5343: @node Files, Blocks, Environmental Queries, Words
5344: @section Files
1.28 crook 5345: @cindex files
5346: @cindex I/O - file-handling
1.21 crook 5347:
1.26 crook 5348: Gforth provides facilities for accessing files that are stored in the
5349: host operating system's file-system. Files that are processed by Gforth
5350: can be divided into two categories:
1.21 crook 5351:
1.23 crook 5352: @itemize @bullet
5353: @item
1.29 ! crook 5354: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
1.23 crook 5355: @item
1.29 ! crook 5356: Files that are processed by some other program (@dfn{general files}).
1.26 crook 5357: @end itemize
5358:
5359: @menu
5360: * Forth source files::
5361: * General files::
5362: * Search Paths::
5363: * Forth Search Paths::
5364: * General Search Paths::
5365: @end menu
5366:
1.21 crook 5367:
1.26 crook 5368: @c -------------------------------------------------------------
5369: @node Forth source files, General files, Files, Files
5370: @subsection Forth source files
5371: @cindex including files
5372: @cindex Forth source files
1.21 crook 5373:
1.26 crook 5374: The simplest way to interpret the contents of a file is to use one of
5375: these two formats:
1.21 crook 5376:
1.26 crook 5377: @example
5378: include mysource.fs
5379: s" mysource.fs" included
5380: @end example
1.21 crook 5381:
1.26 crook 5382: Sometimes you want to include a file only if it is not included already
5383: (by, say, another source file). In that case, you can use one of these
5384: fomats:
1.21 crook 5385:
1.26 crook 5386: @example
5387: require mysource.fs
5388: needs mysource.fs
5389: s" mysource.fs" required
5390: @end example
1.21 crook 5391:
1.26 crook 5392: @cindex stack effect of included files
5393: @cindex including files, stack effect
5394: I recommend that you write your source files such that interpreting them
5395: does not change the stack. This allows using these files with
5396: @code{required} and friends without complications. For example:
1.21 crook 5397:
1.26 crook 5398: @example
5399: 1 require foo.fs drop
5400: @end example
1.21 crook 5401:
1.26 crook 5402: doc-include-file
5403: doc-included
1.28 crook 5404: doc-included?
1.26 crook 5405: doc-include
5406: doc-required
5407: doc-require
5408: doc-needs
1.28 crook 5409: doc-init-included-files
1.21 crook 5410:
1.26 crook 5411: A definition in ANS Forth for @code{required} is provided in
5412: @file{compat/required.fs}.
1.21 crook 5413:
1.26 crook 5414: @c -------------------------------------------------------------
5415: @node General files, Search Paths, Forth source files, Files
5416: @subsection General files
5417: @cindex general files
5418: @cindex file-handling
1.21 crook 5419:
1.26 crook 5420: Files are opened/created by name and type. The following types are
5421: recognised:
1.1 anton 5422:
1.26 crook 5423: doc-r/o
5424: doc-r/w
5425: doc-w/o
5426: doc-bin
1.1 anton 5427:
1.26 crook 5428: When a file is opened/created, it returns a file identifier,
1.29 ! crook 5429: @i{wfileid} that is used for all other file commands. All file
! 5430: commands also return a status value, @i{wior}, that is 0 for a
1.26 crook 5431: successful operation and an implementation-defined non-zero value in the
5432: case of an error.
1.21 crook 5433:
1.26 crook 5434: doc-open-file
5435: doc-create-file
1.21 crook 5436:
1.26 crook 5437: doc-close-file
5438: doc-delete-file
5439: doc-rename-file
5440: doc-read-file
5441: doc-read-line
5442: doc-write-file
5443: doc-write-line
5444: doc-emit-file
5445: doc-flush-file
1.21 crook 5446:
1.26 crook 5447: doc-file-status
5448: doc-file-position
5449: doc-reposition-file
5450: doc-file-size
5451: doc-resize-file
1.21 crook 5452:
1.26 crook 5453: @c ---------------------------------------------------------
5454: @node Search Paths, Forth Search Paths, General files, Files
5455: @subsection Search Paths
5456: @cindex path for @code{included}
5457: @cindex file search path
5458: @cindex @code{include} search path
5459: @cindex search path for files
1.21 crook 5460:
1.26 crook 5461: If you specify an absolute filename (i.e., a filename starting with
5462: @file{/} or @file{~}, or with @file{:} in the second position (as in
5463: @samp{C:...})) for @code{included} and friends, that file is included
5464: just as you would expect.
1.21 crook 5465:
1.26 crook 5466: For relative filenames, Gforth uses a search path similar to Forth's
5467: search order (@pxref{Word Lists}). It tries to find the given filename
5468: in the directories present in the path, and includes the first one it
5469: finds. There are separate search paths for Forth source files and
5470: general files.
1.21 crook 5471:
1.26 crook 5472: If the search path contains the directory @file{.} (as it should), this
5473: refers to the directory that the present file was @code{included}
5474: from. This allows files to include other files relative to their own
5475: position (irrespective of the current working directory or the absolute
5476: position). This feature is essential for libraries consisting of
5477: several files, where a file may include other files from the library.
5478: It corresponds to @code{#include "..."} in C. If the current input
5479: source is not a file, @file{.} refers to the directory of the innermost
5480: file being included, or, if there is no file being included, to the
5481: current working directory.
1.21 crook 5482:
1.26 crook 5483: Use @file{~+} to refer to the current working directory (as in the
5484: @code{bash}).
1.1 anton 5485:
1.26 crook 5486: If the filename starts with @file{./}, the search path is not searched
5487: (just as with absolute filenames), and the @file{.} has the same meaning
5488: as described above.
1.1 anton 5489:
1.26 crook 5490: @c ---------------------------------------------------------
5491: @node Forth Search Paths, General Search Paths, Search Paths, Files
5492: @subsubsection Forth Search Paths
1.28 crook 5493: @cindex search path control - Forth
1.5 anton 5494:
1.26 crook 5495: The search path is initialized when you start Gforth (@pxref{Invoking
5496: Gforth}). You can display it and change it using these words:
1.5 anton 5497:
1.26 crook 5498: doc-.fpath
5499: doc-fpath+
5500: doc-fpath=
5501: doc-open-fpath-file
1.5 anton 5502:
1.26 crook 5503: Here is an example of using @code{fpath} and @code{require}:
1.5 anton 5504:
1.26 crook 5505: @example
5506: fpath= /usr/lib/forth/|./
5507: require timer.fs
5508: @end example
1.5 anton 5509:
1.26 crook 5510: @c ---------------------------------------------------------
5511: @node General Search Paths, , Forth Search Paths, Files
5512: @subsubsection General Search Paths
5513: @cindex search path control - for user applications
1.5 anton 5514:
1.26 crook 5515: Your application may need to search files in several directories, like
5516: @code{included} does. To facilitate this, Gforth allows you to define
5517: and use your own search paths, by providing generic equivalents of the
5518: Forth search path words:
1.5 anton 5519:
1.26 crook 5520: doc-.path
5521: doc-path+
5522: doc-path=
5523: doc-open-path-file
1.5 anton 5524:
1.26 crook 5525: Here's an example of creating a search path:
1.5 anton 5526:
1.26 crook 5527: @example
5528: \ Make a buffer for the path:
5529: create mypath 100 chars , \ maximum length (is checked)
5530: 0 , \ real len
5531: 100 chars allot \ space for path
5532: @end example
1.5 anton 5533:
1.26 crook 5534: @c -------------------------------------------------------------
5535: @node Blocks, Other I/O, Files, Words
5536: @section Blocks
1.28 crook 5537: @cindex I/O - blocks
5538: @cindex blocks
5539:
5540: When you run Gforth on a modern desk-top computer, it runs under the
5541: control of an operating system which provides certain services. One of
5542: these services is @var{file services}, which allows Forth source code
5543: and data to be stored in files and read into Gforth (@pxref{Files}).
5544:
5545: Traditionally, Forth has been an important programming language on
5546: systems where it has interfaced directly to the underlying hardware with
5547: no intervening operating system. Forth provides a mechanism, called
1.29 ! crook 5548: @dfn{blocks}, for accessing mass storage on such systems.
1.28 crook 5549:
5550: A block is a 1024-byte data area, which can be used to hold data or
5551: Forth source code. No structure is imposed on the contents of the
5552: block. A block is identified by its number; blocks are numbered
5553: contiguously from 1 to an implementation-defined maximum.
5554:
5555: A typical system that used blocks but no operating system might use a
5556: single floppy-disk drive for mass storage, with the disks formatted to
5557: provide 256-byte sectors. Blocks would be implemented by assigning the
5558: first four sectors of the disk to block 1, the second four sectors to
5559: block 2 and so on, up to the limit of the capacity of the disk. The disk
5560: would not contain any file system information, just the set of blocks.
5561:
1.29 ! crook 5562: @cindex blocks file
1.28 crook 5563: On systems that do provide file services, blocks are typically
1.29 ! crook 5564: implemented by storing a sequence of blocks within a single @dfn{blocks
1.28 crook 5565: file}. The size of the blocks file will be an exact multiple of 1024
5566: bytes, corresponding to the number of blocks it contains. This is the
5567: mechanism that Gforth uses.
5568:
1.29 ! crook 5569: @cindex @file{blocks.fb}
1.28 crook 5570: Only 1 blocks file can be open at a time. If you use block words without
5571: having specified a blocks file, Gforth defaults to the blocks file
5572: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
5573: locate a blocks file (@pxref{Forth Search Paths}).
5574:
1.29 ! crook 5575: @cindex block buffers
1.28 crook 5576: When you read and write blocks under program control, Gforth uses a
1.29 ! crook 5577: number of @dfn{block buffers} as intermediate storage. These buffers are
1.28 crook 5578: not used when you use @code{load} to interpret the contents of a block.
5579:
5580: The behaviour of the block buffers is directly analagous to that of a
5581: cache. Each block buffer has three states:
5582:
5583: @itemize @bullet
5584: @item
5585: Unassigned
5586: @item
5587: Assigned-clean
5588: @item
5589: Assigned-dirty
5590: @end itemize
5591:
1.29 ! crook 5592: Initially, all block buffers are @i{unassigned}. In order to access a
1.28 crook 5593: block, the block (specified by its block number) must be assigned to a
5594: block buffer.
5595:
5596: The assignment of a block to a block buffer is performed by @code{block}
5597: or @code{buffer}. Use @code{block} when you wish to modify the existing
5598: contents of a block. Use @code{buffer} when you don't care about the
5599: existing contents of the block@footnote{The ANS Forth definition of
5600: @code{block} is intended not to cause disk I/O; if the data associated
5601: with the particular block is already stored in a block buffer due to an
5602: earlier @code{block} command, @code{buffer} will return that block
5603: buffer and the existing contents of the block will be
5604: available. Otherwise, @code{buffer} will simply assign a new, empty
1.29 ! crook 5605: block buffer for the block.}.
1.28 crook 5606:
5607: Once a block has been assigned to a block buffer, the block buffer state
1.29 ! crook 5608: becomes @i{assigned-clean}. Data can now be manipulated within the
1.28 crook 5609: block buffer.
5610:
5611: When the contents of a block buffer is changed it is necessary,
5612: @i{before calling} @code{block} @i{or} @code{buffer} @i{again}, to
5613: either abandon the changes (by doing nothing) or commit the changes,
5614: using @code{update}. Using @code{update} does not change the blocks
1.29 ! crook 5615: file; it simply changes a block buffer's state to @i{assigned-dirty}.
1.28 crook 5616:
1.29 ! crook 5617: The word @code{flush} causes all @i{assigned-dirty} blocks to be
1.28 crook 5618: written back to the blocks file on disk. Leaving Gforth using @code{bye}
5619: also causes a @code{flush} to be performed.
5620:
1.29 ! crook 5621: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
1.28 crook 5622: algorithm to assign a block buffer to a block. That means that any
5623: particular block can only be assigned to one specific block buffer,
1.29 ! crook 5624: called (for the particular operation) the @i{victim buffer}. If the
! 5625: victim buffer is @i{unassigned} or @i{assigned-clean} it can be
! 5626: allocated to the new block immediately. If it is @i{assigned-dirty}
1.28 crook 5627: its current contents must be written out to disk before it can be
5628: allocated to the new block.
5629:
5630: Although no structure is imposed on the contents of a block, it is
5631: traditional to display the contents as 16 lines each of 64 characters. A
5632: block provides a single, continuous stream of input (for example, it
5633: acts as a single parse area) -- there are no end-of-line characters
5634: within a block, and no end-of-file character at the end of a
5635: block. There are two consequences of this:
1.26 crook 5636:
1.28 crook 5637: @itemize @bullet
5638: @item
5639: The last character of one line wraps straight into the first character
5640: of the following line
5641: @item
5642: The word @code{\} -- comment to end of line -- requires special
5643: treatment; in the context of a block it causes all characters until the
5644: end of the current 64-character ``line'' to be ignored.
5645: @end itemize
5646:
5647: In Gforth, when you use @code{block} with a non-existent block number,
5648: the current block file will be extended to the appropriate size and the
5649: block buffer will be initialised with spaces.
5650:
1.29 ! crook 5651: Gforth doesn't encourage the use of blocks; the mechanism is only
! 5652: provided for backward compatibility -- ANS Forth requires blocks to be
! 5653: available when files are.
1.28 crook 5654:
5655: Common techniques that are used when working with blocks include:
5656:
5657: @itemize @bullet
5658: @item
5659: A screen editor that allows you to edit blocks without leaving the Forth
5660: environment.
5661: @item
5662: Shadow screens; where every code block has an associated block
5663: containing comments (for example: code in odd block numbers, comments in
5664: even block numbers). Typically, the block editor provides a convenient
5665: mechanism to toggle between code and comments.
5666: @item
5667: Load blocks; a single block (typically block 1) contains a number of
5668: @code{thru} commands which @code{load} the whole of the application.
5669: @item
5670: Chaining blocks; a block terminates with a @code{-->} so that a whole
5671: application can be @code{load}ed by @code{load}ing a single block.
5672: @end itemize
1.26 crook 5673:
1.29 ! crook 5674: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
! 5675: integrated into a Forth programming environment.
1.26 crook 5676:
5677: @comment TODO what about errors on open-blocks?
5678: doc-open-blocks
5679: doc-use
5680: doc-get-block-fid
5681: doc-block-position
1.28 crook 5682:
5683: doc-scr
5684: doc-list
5685:
5686: doc---block-block
5687: doc-buffer
5688:
1.26 crook 5689: doc-update
1.28 crook 5690: doc-updated?
1.26 crook 5691: doc-save-buffers
5692: doc-empty-buffers
5693: doc-empty-buffer
5694: doc-flush
1.28 crook 5695:
1.26 crook 5696: doc-load
5697: doc-thru
5698: doc-+load
5699: doc-+thru
5700: doc---block--->
5701: doc-block-included
5702:
5703: @c -------------------------------------------------------------
5704: @node Other I/O, Programming Tools, Blocks, Words
5705: @section Other I/O
1.28 crook 5706: @cindex I/O - keyboard and display
1.26 crook 5707:
5708: @menu
5709: * Simple numeric output:: Predefined formats
5710: * Formatted numeric output:: Formatted (pictured) output
5711: * String Formats:: How Forth stores strings in memory
5712: * Displaying characters and strings:: Other stuff
5713: * Input:: Input
5714: @end menu
5715:
5716: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
5717: @subsection Simple numeric output
1.28 crook 5718: @cindex numeric output - simple/free-format
1.5 anton 5719:
1.26 crook 5720: The simplest output functions are those that display numbers from the
5721: data or floating-point stacks. Floating-point output is always displayed
5722: using base 10. Numbers displayed from the data stack use the value stored
5723: in @code{base}.
1.5 anton 5724:
1.26 crook 5725: doc-.
5726: doc-dec.
5727: doc-hex.
5728: doc-u.
5729: doc-.r
5730: doc-u.r
5731: doc-d.
5732: doc-ud.
5733: doc-d.r
5734: doc-ud.r
5735: doc-f.
5736: doc-fe.
5737: doc-fs.
1.5 anton 5738:
1.26 crook 5739: Examples of printing the number 1234.5678E23 in the different floating-point output
5740: formats are shown below:
1.5 anton 5741:
5742: @example
1.26 crook 5743: f. 123456779999999000000000000.
5744: fe. 123.456779999999E24
5745: fs. 1.23456779999999E26
1.5 anton 5746: @end example
5747:
5748:
1.26 crook 5749: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
5750: @subsection Formatted numeric output
1.28 crook 5751: @cindex formatted numeric output
1.26 crook 5752: @cindex pictured numeric output
1.28 crook 5753: @cindex numeric output - formatted
1.26 crook 5754:
1.29 ! crook 5755: Forth traditionally uses a technique called @dfn{pictured numeric
1.26 crook 5756: output} for formatted printing of integers. In this technique, digits
5757: are extracted from the number (using the current output radix defined by
5758: @code{base}), converted to ASCII codes and appended to a string that is
5759: built in a scratch-pad area of memory (@pxref{core-idef,
5760: Implementation-defined options, Implementation-defined
5761: options}). Arbitrary characters can be appended to the string during the
5762: extraction process. The completed string is specified by an address
5763: and length and can be manipulated (@code{TYPE}ed, copied, modified)
5764: under program control.
1.5 anton 5765:
1.26 crook 5766: All of the words described in the previous section for simple numeric
5767: output are implemented in Gforth using pictured numeric output.
1.5 anton 5768:
1.26 crook 5769: Three important things to remember about Pictured Numeric Output:
1.5 anton 5770:
1.26 crook 5771: @itemize @bullet
5772: @item
1.28 crook 5773: It always operates on double-precision numbers; to display a
5774: single-precision number, convert it first (@pxref{Double precision} for
5775: ways of doing this).
1.26 crook 5776: @item
1.28 crook 5777: It always treats the double-precision number as though it were
5778: unsigned. The examples below show ways of printing signed numbers.
1.26 crook 5779: @item
5780: The string is built up from right to left; least significant digit first.
5781: @end itemize
1.5 anton 5782:
1.26 crook 5783: doc-<#
5784: doc-#
5785: doc-#s
5786: doc-hold
5787: doc-sign
5788: doc-#>
1.5 anton 5789:
1.26 crook 5790: doc-represent
1.5 anton 5791:
1.26 crook 5792: Here are some examples of using pictured numeric output:
1.5 anton 5793:
5794: @example
1.26 crook 5795: : my-u. ( u -- )
5796: \ Simplest use of pns.. behaves like Standard u.
5797: 0 \ convert to unsigned double
5798: <# \ start conversion
5799: #s \ convert all digits
5800: #> \ complete conversion
5801: TYPE SPACE ; \ display, with trailing space
1.5 anton 5802:
1.26 crook 5803: : cents-only ( u -- )
5804: 0 \ convert to unsigned double
5805: <# \ start conversion
5806: # # \ convert two least-significant digits
5807: #> \ complete conversion, discard other digits
5808: TYPE SPACE ; \ display, with trailing space
1.5 anton 5809:
1.26 crook 5810: : dollars-and-cents ( u -- )
5811: 0 \ convert to unsigned double
5812: <# \ start conversion
5813: # # \ convert two least-significant digits
5814: [char] . hold \ insert decimal point
5815: #s \ convert remaining digits
5816: [char] $ hold \ append currency symbol
5817: #> \ complete conversion
5818: TYPE SPACE ; \ display, with trailing space
1.5 anton 5819:
1.26 crook 5820: : my-. ( n -- )
5821: \ handling negatives.. behaves like Standard .
5822: s>d \ convert to signed double
5823: swap over dabs \ leave sign byte followed by unsigned double
5824: <# \ start conversion
5825: #s \ convert all digits
5826: rot sign \ get at sign byte, append "-" if needed
5827: #> \ complete conversion
5828: TYPE SPACE ; \ display, with trailing space
1.5 anton 5829:
1.26 crook 5830: : account. ( n -- )
5831: \ accountants don't like minus signs, they use braces
5832: \ for negative numbers
5833: s>d \ convert to signed double
5834: swap over dabs \ leave sign byte followed by unsigned double
5835: <# \ start conversion
5836: 2 pick \ get copy of sign byte
5837: 0< IF [char] ) hold THEN \ right-most character of output
5838: #s \ convert all digits
5839: rot \ get at sign byte
5840: 0< IF [char] ( hold THEN
5841: #> \ complete conversion
5842: TYPE SPACE ; \ display, with trailing space
1.5 anton 5843: @end example
5844:
1.26 crook 5845: Here are some examples of using these words:
1.5 anton 5846:
5847: @example
1.26 crook 5848: 1 my-u. 1
5849: hex -1 my-u. decimal FFFFFFFF
5850: 1 cents-only 01
5851: 1234 cents-only 34
5852: 2 dollars-and-cents $0.02
5853: 1234 dollars-and-cents $12.34
5854: 123 my-. 123
5855: -123 my. -123
5856: 123 account. 123
5857: -456 account. (456)
1.5 anton 5858: @end example
5859:
5860:
1.26 crook 5861: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
5862: @subsection String Formats
1.27 crook 5863: @cindex strings - see character strings
5864: @cindex character strings - formats
1.28 crook 5865: @cindex I/O - see character strings
1.26 crook 5866:
1.27 crook 5867: Forth commonly uses two different methods for representing character
5868: strings:
1.26 crook 5869:
5870: @itemize @bullet
5871: @item
5872: @cindex address of counted string
1.29 ! crook 5873: As a @dfn{counted string}, represented by a @i{c-addr}. The char
! 5874: addressed by @i{c-addr} contains a character-count, @i{n}, of the
! 5875: string and the string occupies the subsequent @i{n} char addresses in
1.26 crook 5876: memory.
5877: @item
1.29 ! crook 5878: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
! 5879: of the string in characters, and @i{c-addr} is the address of the
1.26 crook 5880: first byte of the string.
5881: @end itemize
5882:
5883: ANS Forth encourages the use of the second format when representing
5884: strings on the stack, whilst conceeding that the counted string format
5885: remains useful as a way of storing strings in memory.
5886:
5887: doc-count
5888:
5889: @xref{Memory Blocks} for words that move, copy and search
5890: for strings. @xref{Displaying characters and strings,} for words that
5891: display characters and strings.
5892:
5893:
5894: @node Displaying characters and strings, Input, String Formats, Other I/O
5895: @subsection Displaying characters and strings
1.27 crook 5896: @cindex characters - compiling and displaying
5897: @cindex character strings - compiling and displaying
1.26 crook 5898:
5899: This section starts with a glossary of Forth words and ends with a set
5900: of examples.
5901:
5902: doc-bl
5903: doc-space
5904: doc-spaces
5905: doc-emit
5906: doc-toupper
5907: doc-."
5908: doc-.(
5909: doc-type
5910: doc-cr
1.27 crook 5911: @cindex cursor control
1.26 crook 5912: doc-at-xy
5913: doc-page
5914: doc-s"
5915: doc-c"
5916: doc-char
5917: doc-[char]
5918: doc-sliteral
5919:
5920: As an example, consider the following text, stored in a file @file{test.fs}:
1.5 anton 5921:
5922: @example
1.26 crook 5923: .( text-1)
5924: : my-word
5925: ." text-2" cr
5926: .( text-3)
5927: ;
5928:
5929: ." text-4"
5930:
5931: : my-char
5932: [char] ALPHABET emit
5933: char emit
5934: ;
1.5 anton 5935: @end example
5936:
1.26 crook 5937: When you load this code into Gforth, the following output is generated:
1.5 anton 5938:
1.26 crook 5939: @example
5940: @kbd{include test.fs <return>} text-1text-3text-4 ok
5941: @end example
1.5 anton 5942:
1.26 crook 5943: @itemize @bullet
5944: @item
5945: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
5946: is an immediate word; it behaves in the same way whether it is used inside
5947: or outside a colon definition.
5948: @item
5949: Message @code{text-4} is displayed because of Gforth's added interpretation
5950: semantics for @code{."}.
5951: @item
1.29 ! crook 5952: Message @code{text-2} is @i{not} displayed, because the text interpreter
1.26 crook 5953: performs the compilation semantics for @code{."} within the definition of
5954: @code{my-word}.
5955: @end itemize
1.5 anton 5956:
1.26 crook 5957: Here are some examples of executing @code{my-word} and @code{my-char}:
1.5 anton 5958:
1.26 crook 5959: @example
5960: @kbd{my-word <return>} text-2
5961: ok
5962: @kbd{my-char fred <return>} Af ok
5963: @kbd{my-char jim <return>} Aj ok
5964: @end example
1.5 anton 5965:
5966: @itemize @bullet
5967: @item
1.26 crook 5968: Message @code{text-2} is displayed because of the run-time behaviour of
5969: @code{."}.
5970: @item
5971: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
5972: on the stack at run-time. @code{emit} always displays the character
5973: when @code{my-char} is executed.
5974: @item
5975: @code{char} parses a string at run-time and the second @code{emit} displays
5976: the first character of the string.
1.5 anton 5977: @item
1.26 crook 5978: If you type @code{see my-char} you can see that @code{[char]} discarded
5979: the text ``LPHABET'' and only compiled the display code for ``A'' into the
5980: definition of @code{my-char}.
1.5 anton 5981: @end itemize
5982:
5983:
5984:
1.26 crook 5985: @node Input, , Displaying characters and strings, Other I/O
5986: @subsection Input
5987: @cindex input
1.28 crook 5988: @cindex I/O - see input
5989: @cindex parsing a string
1.5 anton 5990:
1.27 crook 5991: @xref{String Formats} for ways of storing character strings in memory.
1.5 anton 5992:
1.27 crook 5993: @comment TODO examples for >number >float accept key key? pad parse word refill
1.29 ! crook 5994: @comment then index them
1.27 crook 5995:
5996: doc-key
5997: doc-key?
1.26 crook 5998: doc->number
5999: doc->float
6000: doc-accept
1.27 crook 6001: doc-pad
6002: doc-parse
6003: doc-word
6004: doc-sword
6005: doc-refill
6006: @comment obsolescent words..
6007: doc-convert
1.26 crook 6008: doc-query
6009: doc-expect
1.27 crook 6010: doc-span
1.5 anton 6011:
6012:
6013: @c -------------------------------------------------------------
1.26 crook 6014: @node Programming Tools, Assembler and Code Words, Other I/O, Words
6015: @section Programming Tools
6016: @cindex programming tools
1.12 anton 6017:
6018: @menu
1.26 crook 6019: * Debugging:: Simple and quick.
6020: * Assertions:: Making your programs self-checking.
6021: * Singlestep Debugger:: Executing your program word by word.
1.5 anton 6022: @end menu
6023:
1.26 crook 6024: @node Debugging, Assertions, Programming Tools, Programming Tools
6025: @subsection Debugging
6026: @cindex debugging
1.5 anton 6027:
1.26 crook 6028: Languages with a slow edit/compile/link/test development loop tend to
6029: require sophisticated tracing/stepping debuggers to facilate
6030: productive debugging.
1.5 anton 6031:
1.26 crook 6032: A much better (faster) way in fast-compiling languages is to add
6033: printing code at well-selected places, let the program run, look at
6034: the output, see where things went wrong, add more printing code, etc.,
6035: until the bug is found.
1.5 anton 6036:
1.26 crook 6037: The simple debugging aids provided in @file{debugs.fs}
6038: are meant to support this style of debugging. In addition, there are
6039: words for non-destructively inspecting the stack and memory:
1.5 anton 6040:
1.26 crook 6041: doc-.s
6042: doc-f.s
1.5 anton 6043:
1.29 ! crook 6044: There is a word @code{.r} but it does @i{not} display the return
1.26 crook 6045: stack! It is used for formatted numeric output.
1.5 anton 6046:
1.26 crook 6047: doc-depth
6048: doc-fdepth
6049: doc-clearstack
6050: doc-?
6051: doc-dump
1.5 anton 6052:
1.26 crook 6053: The word @code{~~} prints debugging information (by default the source
6054: location and the stack contents). It is easy to insert. If you use Emacs
6055: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
6056: query-replace them with nothing). The deferred words
6057: @code{printdebugdata} and @code{printdebugline} control the output of
6058: @code{~~}. The default source location output format works well with
6059: Emacs' compilation mode, so you can step through the program at the
6060: source level using @kbd{C-x `} (the advantage over a stepping debugger
6061: is that you can step in any direction and you know where the crash has
6062: happened or where the strange data has occurred).
1.5 anton 6063:
1.26 crook 6064: The default actions of @code{~~} clobber the contents of the pictured
6065: numeric output string, so you should not use @code{~~}, e.g., between
6066: @code{<#} and @code{#>}.
1.5 anton 6067:
1.26 crook 6068: doc-~~
6069: doc-printdebugdata
6070: doc-printdebugline
1.5 anton 6071:
1.26 crook 6072: doc-see
6073: doc-marker
1.5 anton 6074:
1.26 crook 6075: Here's an example of using @code{marker} at the start of a source file
6076: that you are debugging; it ensures that you only ever have one copy of
6077: the file's definitions compiled at any time:
1.5 anton 6078:
1.26 crook 6079: @example
6080: [IFDEF] my-code
6081: my-code
6082: [ENDIF]
1.5 anton 6083:
1.26 crook 6084: marker my-code
1.28 crook 6085: init-included-files
1.5 anton 6086:
1.26 crook 6087: \ .. definitions start here
6088: \ .
6089: \ .
6090: \ end
6091: @end example
1.5 anton 6092:
6093:
6094:
1.26 crook 6095: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
6096: @subsection Assertions
6097: @cindex assertions
1.5 anton 6098:
1.26 crook 6099: It is a good idea to make your programs self-checking, especially if you
6100: make an assumption that may become invalid during maintenance (for
6101: example, that a certain field of a data structure is never zero). Gforth
1.29 ! crook 6102: supports @dfn{assertions} for this purpose. They are used like this:
1.23 crook 6103:
1.26 crook 6104: @example
1.29 ! crook 6105: assert( @i{flag} )
1.26 crook 6106: @end example
1.23 crook 6107:
1.26 crook 6108: The code between @code{assert(} and @code{)} should compute a flag, that
6109: should be true if everything is alright and false otherwise. It should
6110: not change anything else on the stack. The overall stack effect of the
6111: assertion is @code{( -- )}. E.g.
1.23 crook 6112:
1.26 crook 6113: @example
6114: assert( 1 1 + 2 = ) \ what we learn in school
6115: assert( dup 0<> ) \ assert that the top of stack is not zero
6116: assert( false ) \ this code should not be reached
6117: @end example
1.23 crook 6118:
1.26 crook 6119: The need for assertions is different at different times. During
6120: debugging, we want more checking, in production we sometimes care more
6121: for speed. Therefore, assertions can be turned off, i.e., the assertion
6122: becomes a comment. Depending on the importance of an assertion and the
6123: time it takes to check it, you may want to turn off some assertions and
6124: keep others turned on. Gforth provides several levels of assertions for
6125: this purpose:
1.23 crook 6126:
1.26 crook 6127: doc-assert0(
6128: doc-assert1(
6129: doc-assert2(
6130: doc-assert3(
6131: doc-assert(
6132: doc-)
1.23 crook 6133:
1.26 crook 6134: The variable @code{assert-level} specifies the highest assertions that
6135: are turned on. I.e., at the default @code{assert-level} of one,
6136: @code{assert0(} and @code{assert1(} assertions perform checking, while
6137: @code{assert2(} and @code{assert3(} assertions are treated as comments.
6138:
6139: The value of @code{assert-level} is evaluated at compile-time, not at
6140: run-time. Therefore you cannot turn assertions on or off at run-time;
6141: you have to set the @code{assert-level} appropriately before compiling a
6142: piece of code. You can compile different pieces of code at different
6143: @code{assert-level}s (e.g., a trusted library at level 1 and
6144: newly-written code at level 3).
1.23 crook 6145:
1.26 crook 6146: doc-assert-level
1.23 crook 6147:
1.26 crook 6148: If an assertion fails, a message compatible with Emacs' compilation mode
6149: is produced and the execution is aborted (currently with @code{ABORT"}.
6150: If there is interest, we will introduce a special throw code. But if you
6151: intend to @code{catch} a specific condition, using @code{throw} is
6152: probably more appropriate than an assertion).
1.23 crook 6153:
1.26 crook 6154: Definitions in ANS Forth for these assertion words are provided
6155: in @file{compat/assert.fs}.
1.23 crook 6156:
6157:
1.26 crook 6158: @node Singlestep Debugger, , Assertions, Programming Tools
6159: @subsection Singlestep Debugger
6160: @cindex singlestep Debugger
6161: @cindex debugging Singlestep
6162: @cindex @code{dbg}
6163: @cindex @code{BREAK:}
6164: @cindex @code{BREAK"}
1.23 crook 6165:
1.26 crook 6166: When you create a new word there's often the need to check whether it
6167: behaves correctly or not. You can do this by typing @code{dbg
6168: badword}. A debug session might look like this:
1.23 crook 6169:
1.26 crook 6170: @example
6171: : badword 0 DO i . LOOP ; ok
6172: 2 dbg badword
6173: : badword
6174: Scanning code...
1.23 crook 6175:
1.26 crook 6176: Nesting debugger ready!
1.23 crook 6177:
1.26 crook 6178: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
6179: 400D4740 8049F68 DO -> [ 0 ]
6180: 400D4744 804A0C8 i -> [ 1 ] 00000
6181: 400D4748 400C5E60 . -> 0 [ 0 ]
6182: 400D474C 8049D0C LOOP -> [ 0 ]
6183: 400D4744 804A0C8 i -> [ 1 ] 00001
6184: 400D4748 400C5E60 . -> 1 [ 0 ]
6185: 400D474C 8049D0C LOOP -> [ 0 ]
6186: 400D4758 804B384 ; -> ok
6187: @end example
1.23 crook 6188:
1.26 crook 6189: Each line displayed is one step. You always have to hit return to
6190: execute the next word that is displayed. If you don't want to execute
6191: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
6192: an overview what keys are available:
1.23 crook 6193:
1.26 crook 6194: @table @i
1.23 crook 6195:
1.26 crook 6196: @item <return>
6197: Next; Execute the next word.
1.23 crook 6198:
1.26 crook 6199: @item n
6200: Nest; Single step through next word.
1.5 anton 6201:
1.26 crook 6202: @item u
6203: Unnest; Stop debugging and execute rest of word. If we got to this word
6204: with nest, continue debugging with the calling word.
1.5 anton 6205:
1.26 crook 6206: @item d
6207: Done; Stop debugging and execute rest.
1.5 anton 6208:
1.26 crook 6209: @item s
6210: Stop; Abort immediately.
1.5 anton 6211:
1.26 crook 6212: @end table
1.5 anton 6213:
1.26 crook 6214: Debugging large application with this mechanism is very difficult, because
6215: you have to nest very deeply into the program before the interesting part
6216: begins. This takes a lot of time.
1.5 anton 6217:
1.26 crook 6218: To do it more directly put a @code{BREAK:} command into your source code.
6219: When program execution reaches @code{BREAK:} the single step debugger is
6220: invoked and you have all the features described above.
1.23 crook 6221:
1.26 crook 6222: If you have more than one part to debug it is useful to know where the
6223: program has stopped at the moment. You can do this by the
6224: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
6225: string is typed out when the ``breakpoint'' is reached.
6226:
6227: doc-dbg
6228: doc-BREAK:
6229: doc-BREAK"
6230:
6231:
6232: @c -------------------------------------------------------------
6233: @node Assembler and Code Words, Threading Words, Programming Tools, Words
6234: @section Assembler and Code Words
6235: @cindex assembler
6236: @cindex code words
1.5 anton 6237:
1.26 crook 6238: Gforth provides some words for defining primitives (words written in
1.29 ! crook 6239: machine code), and for defining the machine-code equivalent of
1.26 crook 6240: @code{DOES>}-based defining words. However, the machine-independent
6241: nature of Gforth poses a few problems: First of all, Gforth runs on
6242: several architectures, so it can provide no standard assembler. What's
6243: worse is that the register allocation not only depends on the processor,
6244: but also on the @code{gcc} version and options used.
1.5 anton 6245:
1.29 ! crook 6246: The words that Gforth offers encapsulate some system dependences (e.g.,
! 6247: the header structure), so a system-independent assembler may be used in
1.26 crook 6248: Gforth. If you do not have an assembler, you can compile machine code
1.29 ! crook 6249: directly with @code{,} and @code{c,}@footnote{This isn't portable,
! 6250: because these words emit stuff in @i{data} space; it works because
! 6251: Gforth has unified code/data spaces. Assembler isn't likely to be
! 6252: portable anyway.}.
1.5 anton 6253:
1.26 crook 6254: doc-assembler
6255: doc-code
6256: doc-end-code
6257: doc-;code
6258: doc-flush-icache
1.5 anton 6259:
1.26 crook 6260: If @code{flush-icache} does not work correctly, @code{code} words
6261: etc. will not work (reliably), either.
1.5 anton 6262:
1.29 ! crook 6263: The typical usage of these @code{code} words can be shown most easily by
! 6264: analogy to the equivalent high-level defining words:
! 6265:
! 6266: @example
! 6267: : foo code foo
! 6268: <high-level Forth words> <assembler>
! 6269: ; end-code
! 6270:
! 6271: : bar : bar
! 6272: <high-level Forth words> <high-level Forth words>
! 6273: CREATE CREATE
! 6274: <high-level Forth words> <high-level Forth words>
! 6275: DOES> ;code
! 6276: <high-level Forth words> <assembler>
! 6277: ; end-code
! 6278: @end example
! 6279:
1.26 crook 6280: @code{flush-icache} is always present. The other words are rarely used
6281: and reside in @code{code.fs}, which is usually not loaded. You can load
6282: it with @code{require code.fs}.
1.5 anton 6283:
1.26 crook 6284: @cindex registers of the inner interpreter
6285: In the assembly code you will want to refer to the inner interpreter's
6286: registers (e.g., the data stack pointer) and you may want to use other
6287: registers for temporary storage. Unfortunately, the register allocation
6288: is installation-dependent.
1.5 anton 6289:
1.26 crook 6290: The easiest solution is to use explicit register declarations
6291: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
6292: GNU C Manual}) for all of the inner interpreter's registers: You have to
6293: compile Gforth with @code{-DFORCE_REG} (configure option
6294: @code{--enable-force-reg}) and the appropriate declarations must be
6295: present in the @code{machine.h} file (see @code{mips.h} for an example;
6296: you can find a full list of all declarable register symbols with
6297: @code{grep register engine.c}). If you give explicit registers to all
6298: variables that are declared at the beginning of @code{engine()}, you
6299: should be able to use the other caller-saved registers for temporary
6300: storage. Alternatively, you can use the @code{gcc} option
6301: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
6302: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
6303: (however, this restriction on register allocation may slow Gforth
6304: significantly).
1.5 anton 6305:
1.26 crook 6306: If this solution is not viable (e.g., because @code{gcc} does not allow
6307: you to explicitly declare all the registers you need), you have to find
6308: out by looking at the code where the inner interpreter's registers
6309: reside and which registers can be used for temporary storage. You can
6310: get an assembly listing of the engine's code with @code{make engine.s}.
1.5 anton 6311:
1.26 crook 6312: In any case, it is good practice to abstract your assembly code from the
6313: actual register allocation. E.g., if the data stack pointer resides in
6314: register @code{$17}, create an alias for this register called @code{sp},
6315: and use that in your assembly code.
1.5 anton 6316:
1.26 crook 6317: @cindex code words, portable
6318: Another option for implementing normal and defining words efficiently
6319: is to add the desired functionality to the source of Gforth. For normal
6320: words you just have to edit @file{primitives} (@pxref{Automatic
6321: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
6322: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
6323: @file{prims2x.fs}, and possibly @file{cross.fs}.
1.5 anton 6324:
6325:
1.26 crook 6326: @c -------------------------------------------------------------
6327: @node Threading Words, Locals, Assembler and Code Words, Words
6328: @section Threading Words
6329: @cindex threading words
1.5 anton 6330:
1.26 crook 6331: @cindex code address
6332: These words provide access to code addresses and other threading stuff
6333: in Gforth (and, possibly, other interpretive Forths). It more or less
6334: abstracts away the differences between direct and indirect threading
6335: (and, for direct threading, the machine dependences). However, at
6336: present this wordset is still incomplete. It is also pretty low-level;
6337: some day it will hopefully be made unnecessary by an internals wordset
6338: that abstracts implementation details away completely.
1.5 anton 6339:
1.26 crook 6340: doc-threading-method
6341: doc->code-address
6342: doc->does-code
6343: doc-code-address!
6344: doc-does-code!
6345: doc-does-handler!
6346: doc-/does-handler
1.5 anton 6347:
1.26 crook 6348: The code addresses produced by various defining words are produced by
6349: the following words:
1.5 anton 6350:
1.26 crook 6351: doc-docol:
6352: doc-docon:
6353: doc-dovar:
6354: doc-douser:
6355: doc-dodefer:
6356: doc-dofield:
1.5 anton 6357:
1.26 crook 6358: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
6359: with @code{>does-code}. If the word was defined in that way, the value
6360: returned is non-zero and identifies the @code{DOES>} used by the
6361: defining word.
6362: @comment TODO should that be ``identifies the xt of the DOES> ??''
1.5 anton 6363:
1.26 crook 6364: @c -------------------------------------------------------------
6365: @node Locals, Structures, Threading Words, Words
6366: @section Locals
6367: @cindex locals
1.5 anton 6368:
1.26 crook 6369: Local variables can make Forth programming more enjoyable and Forth
6370: programs easier to read. Unfortunately, the locals of ANS Forth are
6371: laden with restrictions. Therefore, we provide not only the ANS Forth
6372: locals wordset, but also our own, more powerful locals wordset (we
6373: implemented the ANS Forth locals wordset through our locals wordset).
1.5 anton 6374:
1.26 crook 6375: The ideas in this section have also been published in the paper
6376: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
6377: at EuroForth '94; it is available at
6378: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1.5 anton 6379:
1.26 crook 6380: @menu
6381: * Gforth locals::
6382: * ANS Forth locals::
6383: @end menu
1.5 anton 6384:
1.26 crook 6385: @node Gforth locals, ANS Forth locals, Locals, Locals
6386: @subsection Gforth locals
6387: @cindex Gforth locals
6388: @cindex locals, Gforth style
1.5 anton 6389:
1.26 crook 6390: Locals can be defined with
1.5 anton 6391:
6392: @example
1.26 crook 6393: @{ local1 local2 ... -- comment @}
6394: @end example
6395: or
6396: @example
6397: @{ local1 local2 ... @}
1.5 anton 6398: @end example
6399:
1.26 crook 6400: E.g.,
1.5 anton 6401: @example
1.26 crook 6402: : max @{ n1 n2 -- n3 @}
6403: n1 n2 > if
6404: n1
6405: else
6406: n2
6407: endif ;
1.5 anton 6408: @end example
6409:
1.26 crook 6410: The similarity of locals definitions with stack comments is intended. A
6411: locals definition often replaces the stack comment of a word. The order
6412: of the locals corresponds to the order in a stack comment and everything
6413: after the @code{--} is really a comment.
1.5 anton 6414:
1.26 crook 6415: This similarity has one disadvantage: It is too easy to confuse locals
6416: declarations with stack comments, causing bugs and making them hard to
6417: find. However, this problem can be avoided by appropriate coding
6418: conventions: Do not use both notations in the same program. If you do,
6419: they should be distinguished using additional means, e.g. by position.
6420:
6421: @cindex types of locals
6422: @cindex locals types
6423: The name of the local may be preceded by a type specifier, e.g.,
6424: @code{F:} for a floating point value:
6425:
6426: @example
6427: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
6428: \ complex multiplication
6429: Ar Br f* Ai Bi f* f-
6430: Ar Bi f* Ai Br f* f+ ;
6431: @end example
6432:
6433: @cindex flavours of locals
6434: @cindex locals flavours
6435: @cindex value-flavoured locals
6436: @cindex variable-flavoured locals
6437: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
6438: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
6439: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
6440: with @code{W:}, @code{D:} etc.) produces its value and can be changed
6441: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
6442: produces its address (which becomes invalid when the variable's scope is
6443: left). E.g., the standard word @code{emit} can be defined in terms of
6444: @code{type} like this:
1.5 anton 6445:
6446: @example
1.26 crook 6447: : emit @{ C^ char* -- @}
6448: char* 1 type ;
1.5 anton 6449: @end example
6450:
1.26 crook 6451: @cindex default type of locals
6452: @cindex locals, default type
6453: A local without type specifier is a @code{W:} local. Both flavours of
6454: locals are initialized with values from the data or FP stack.
1.5 anton 6455:
1.26 crook 6456: Currently there is no way to define locals with user-defined data
6457: structures, but we are working on it.
1.5 anton 6458:
1.26 crook 6459: Gforth allows defining locals everywhere in a colon definition. This
6460: poses the following questions:
1.5 anton 6461:
1.26 crook 6462: @menu
6463: * Where are locals visible by name?::
6464: * How long do locals live?::
6465: * Programming Style::
6466: * Implementation::
6467: @end menu
1.5 anton 6468:
1.26 crook 6469: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
6470: @subsubsection Where are locals visible by name?
6471: @cindex locals visibility
6472: @cindex visibility of locals
6473: @cindex scope of locals
1.5 anton 6474:
1.26 crook 6475: Basically, the answer is that locals are visible where you would expect
6476: it in block-structured languages, and sometimes a little longer. If you
6477: want to restrict the scope of a local, enclose its definition in
6478: @code{SCOPE}...@code{ENDSCOPE}.
1.5 anton 6479:
1.26 crook 6480: doc-scope
6481: doc-endscope
1.5 anton 6482:
1.26 crook 6483: These words behave like control structure words, so you can use them
6484: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
6485: arbitrary ways.
1.5 anton 6486:
1.26 crook 6487: If you want a more exact answer to the visibility question, here's the
6488: basic principle: A local is visible in all places that can only be
6489: reached through the definition of the local@footnote{In compiler
6490: construction terminology, all places dominated by the definition of the
6491: local.}. In other words, it is not visible in places that can be reached
6492: without going through the definition of the local. E.g., locals defined
6493: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
6494: defined in @code{BEGIN}...@code{UNTIL} are visible after the
6495: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1.5 anton 6496:
1.26 crook 6497: The reasoning behind this solution is: We want to have the locals
6498: visible as long as it is meaningful. The user can always make the
6499: visibility shorter by using explicit scoping. In a place that can
6500: only be reached through the definition of a local, the meaning of a
6501: local name is clear. In other places it is not: How is the local
6502: initialized at the control flow path that does not contain the
6503: definition? Which local is meant, if the same name is defined twice in
6504: two independent control flow paths?
1.5 anton 6505:
1.26 crook 6506: This should be enough detail for nearly all users, so you can skip the
6507: rest of this section. If you really must know all the gory details and
6508: options, read on.
1.5 anton 6509:
1.26 crook 6510: In order to implement this rule, the compiler has to know which places
6511: are unreachable. It knows this automatically after @code{AHEAD},
6512: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
6513: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
6514: compiler that the control flow never reaches that place. If
6515: @code{UNREACHABLE} is not used where it could, the only consequence is
6516: that the visibility of some locals is more limited than the rule above
6517: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
6518: lie to the compiler), buggy code will be produced.
1.5 anton 6519:
1.26 crook 6520: doc-unreachable
1.5 anton 6521:
1.26 crook 6522: Another problem with this rule is that at @code{BEGIN}, the compiler
6523: does not know which locals will be visible on the incoming
6524: back-edge. All problems discussed in the following are due to this
6525: ignorance of the compiler (we discuss the problems using @code{BEGIN}
6526: loops as examples; the discussion also applies to @code{?DO} and other
6527: loops). Perhaps the most insidious example is:
1.5 anton 6528: @example
1.26 crook 6529: AHEAD
6530: BEGIN
6531: x
6532: [ 1 CS-ROLL ] THEN
6533: @{ x @}
6534: ...
6535: UNTIL
6536: @end example
1.5 anton 6537:
1.26 crook 6538: This should be legal according to the visibility rule. The use of
6539: @code{x} can only be reached through the definition; but that appears
6540: textually below the use.
1.5 anton 6541:
1.26 crook 6542: From this example it is clear that the visibility rules cannot be fully
6543: implemented without major headaches. Our implementation treats common
6544: cases as advertised and the exceptions are treated in a safe way: The
6545: compiler makes a reasonable guess about the locals visible after a
6546: @code{BEGIN}; if it is too pessimistic, the
6547: user will get a spurious error about the local not being defined; if the
6548: compiler is too optimistic, it will notice this later and issue a
6549: warning. In the case above the compiler would complain about @code{x}
6550: being undefined at its use. You can see from the obscure examples in
6551: this section that it takes quite unusual control structures to get the
6552: compiler into trouble, and even then it will often do fine.
1.5 anton 6553:
1.26 crook 6554: If the @code{BEGIN} is reachable from above, the most optimistic guess
6555: is that all locals visible before the @code{BEGIN} will also be
6556: visible after the @code{BEGIN}. This guess is valid for all loops that
6557: are entered only through the @code{BEGIN}, in particular, for normal
6558: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
6559: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
6560: compiler. When the branch to the @code{BEGIN} is finally generated by
6561: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
6562: warns the user if it was too optimistic:
6563: @example
6564: IF
6565: @{ x @}
6566: BEGIN
6567: \ x ?
6568: [ 1 cs-roll ] THEN
6569: ...
6570: UNTIL
1.5 anton 6571: @end example
6572:
1.26 crook 6573: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
6574: optimistically assumes that it lives until the @code{THEN}. It notices
6575: this difference when it compiles the @code{UNTIL} and issues a
6576: warning. The user can avoid the warning, and make sure that @code{x}
6577: is not used in the wrong area by using explicit scoping:
6578: @example
6579: IF
6580: SCOPE
6581: @{ x @}
6582: ENDSCOPE
6583: BEGIN
6584: [ 1 cs-roll ] THEN
6585: ...
6586: UNTIL
6587: @end example
1.5 anton 6588:
1.26 crook 6589: Since the guess is optimistic, there will be no spurious error messages
6590: about undefined locals.
1.5 anton 6591:
1.26 crook 6592: If the @code{BEGIN} is not reachable from above (e.g., after
6593: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
6594: optimistic guess, as the locals visible after the @code{BEGIN} may be
6595: defined later. Therefore, the compiler assumes that no locals are
6596: visible after the @code{BEGIN}. However, the user can use
6597: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
6598: visible at the BEGIN as at the point where the top control-flow stack
6599: item was created.
1.5 anton 6600:
1.26 crook 6601: doc-assume-live
1.5 anton 6602:
1.26 crook 6603: E.g.,
1.5 anton 6604: @example
1.26 crook 6605: @{ x @}
6606: AHEAD
6607: ASSUME-LIVE
6608: BEGIN
6609: x
6610: [ 1 CS-ROLL ] THEN
6611: ...
6612: UNTIL
1.5 anton 6613: @end example
6614:
1.26 crook 6615: Other cases where the locals are defined before the @code{BEGIN} can be
6616: handled by inserting an appropriate @code{CS-ROLL} before the
6617: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
6618: behind the @code{ASSUME-LIVE}).
1.5 anton 6619:
1.26 crook 6620: Cases where locals are defined after the @code{BEGIN} (but should be
6621: visible immediately after the @code{BEGIN}) can only be handled by
6622: rearranging the loop. E.g., the ``most insidious'' example above can be
6623: arranged into:
1.5 anton 6624: @example
1.26 crook 6625: BEGIN
6626: @{ x @}
6627: ... 0=
6628: WHILE
6629: x
6630: REPEAT
1.5 anton 6631: @end example
6632:
1.26 crook 6633: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
6634: @subsubsection How long do locals live?
6635: @cindex locals lifetime
6636: @cindex lifetime of locals
1.5 anton 6637:
1.26 crook 6638: The right answer for the lifetime question would be: A local lives at
6639: least as long as it can be accessed. For a value-flavoured local this
6640: means: until the end of its visibility. However, a variable-flavoured
6641: local could be accessed through its address far beyond its visibility
6642: scope. Ultimately, this would mean that such locals would have to be
6643: garbage collected. Since this entails un-Forth-like implementation
6644: complexities, I adopted the same cowardly solution as some other
6645: languages (e.g., C): The local lives only as long as it is visible;
6646: afterwards its address is invalid (and programs that access it
6647: afterwards are erroneous).
1.5 anton 6648:
1.26 crook 6649: @node Programming Style, Implementation, How long do locals live?, Gforth locals
6650: @subsubsection Programming Style
6651: @cindex locals programming style
6652: @cindex programming style, locals
1.5 anton 6653:
1.26 crook 6654: The freedom to define locals anywhere has the potential to change
6655: programming styles dramatically. In particular, the need to use the
6656: return stack for intermediate storage vanishes. Moreover, all stack
6657: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
6658: determined arguments) can be eliminated: If the stack items are in the
6659: wrong order, just write a locals definition for all of them; then
6660: write the items in the order you want.
1.5 anton 6661:
1.26 crook 6662: This seems a little far-fetched and eliminating stack manipulations is
6663: unlikely to become a conscious programming objective. Still, the number
6664: of stack manipulations will be reduced dramatically if local variables
6665: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
6666: a traditional implementation of @code{max}).
1.5 anton 6667:
1.26 crook 6668: This shows one potential benefit of locals: making Forth programs more
6669: readable. Of course, this benefit will only be realized if the
6670: programmers continue to honour the principle of factoring instead of
6671: using the added latitude to make the words longer.
1.5 anton 6672:
1.26 crook 6673: @cindex single-assignment style for locals
6674: Using @code{TO} can and should be avoided. Without @code{TO},
6675: every value-flavoured local has only a single assignment and many
6676: advantages of functional languages apply to Forth. I.e., programs are
6677: easier to analyse, to optimize and to read: It is clear from the
6678: definition what the local stands for, it does not turn into something
6679: different later.
1.5 anton 6680:
1.26 crook 6681: E.g., a definition using @code{TO} might look like this:
1.5 anton 6682: @example
1.26 crook 6683: : strcmp @{ addr1 u1 addr2 u2 -- n @}
6684: u1 u2 min 0
6685: ?do
6686: addr1 c@@ addr2 c@@ -
6687: ?dup-if
6688: unloop exit
6689: then
6690: addr1 char+ TO addr1
6691: addr2 char+ TO addr2
6692: loop
6693: u1 u2 - ;
1.5 anton 6694: @end example
1.26 crook 6695: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
6696: every loop iteration. @code{strcmp} is a typical example of the
6697: readability problems of using @code{TO}. When you start reading
6698: @code{strcmp}, you think that @code{addr1} refers to the start of the
6699: string. Only near the end of the loop you realize that it is something
6700: else.
1.5 anton 6701:
1.26 crook 6702: This can be avoided by defining two locals at the start of the loop that
6703: are initialized with the right value for the current iteration.
1.5 anton 6704: @example
1.26 crook 6705: : strcmp @{ addr1 u1 addr2 u2 -- n @}
6706: addr1 addr2
6707: u1 u2 min 0
6708: ?do @{ s1 s2 @}
6709: s1 c@@ s2 c@@ -
6710: ?dup-if
6711: unloop exit
6712: then
6713: s1 char+ s2 char+
6714: loop
6715: 2drop
6716: u1 u2 - ;
1.5 anton 6717: @end example
1.26 crook 6718: Here it is clear from the start that @code{s1} has a different value
6719: in every loop iteration.
1.5 anton 6720:
1.26 crook 6721: @node Implementation, , Programming Style, Gforth locals
6722: @subsubsection Implementation
6723: @cindex locals implementation
6724: @cindex implementation of locals
1.5 anton 6725:
1.26 crook 6726: @cindex locals stack
6727: Gforth uses an extra locals stack. The most compelling reason for
6728: this is that the return stack is not float-aligned; using an extra stack
6729: also eliminates the problems and restrictions of using the return stack
6730: as locals stack. Like the other stacks, the locals stack grows toward
6731: lower addresses. A few primitives allow an efficient implementation:
1.5 anton 6732:
1.26 crook 6733: doc-@local#
6734: doc-f@local#
6735: doc-laddr#
6736: doc-lp+!#
6737: doc-lp!
6738: doc->l
6739: doc-f>l
1.5 anton 6740:
1.26 crook 6741: In addition to these primitives, some specializations of these
6742: primitives for commonly occurring inline arguments are provided for
6743: efficiency reasons, e.g., @code{@@local0} as specialization of
6744: @code{@@local#} for the inline argument 0. The following compiling words
6745: compile the right specialized version, or the general version, as
6746: appropriate:
1.6 pazsan 6747:
1.26 crook 6748: doc-compile-@local
6749: doc-compile-f@local
6750: doc-compile-lp+!
1.12 anton 6751:
1.26 crook 6752: Combinations of conditional branches and @code{lp+!#} like
6753: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
6754: is taken) are provided for efficiency and correctness in loops.
1.6 pazsan 6755:
1.26 crook 6756: A special area in the dictionary space is reserved for keeping the
6757: local variable names. @code{@{} switches the dictionary pointer to this
6758: area and @code{@}} switches it back and generates the locals
6759: initializing code. @code{W:} etc.@ are normal defining words. This
6760: special area is cleared at the start of every colon definition.
1.6 pazsan 6761:
1.26 crook 6762: @cindex word list for defining locals
6763: A special feature of Gforth's dictionary is used to implement the
6764: definition of locals without type specifiers: every word list (aka
6765: vocabulary) has its own methods for searching
6766: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
6767: with a special search method: When it is searched for a word, it
6768: actually creates that word using @code{W:}. @code{@{} changes the search
6769: order to first search the word list containing @code{@}}, @code{W:} etc.,
6770: and then the word list for defining locals without type specifiers.
1.12 anton 6771:
1.26 crook 6772: The lifetime rules support a stack discipline within a colon
6773: definition: The lifetime of a local is either nested with other locals
6774: lifetimes or it does not overlap them.
1.6 pazsan 6775:
1.26 crook 6776: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
6777: pointer manipulation is generated. Between control structure words
6778: locals definitions can push locals onto the locals stack. @code{AGAIN}
6779: is the simplest of the other three control flow words. It has to
6780: restore the locals stack depth of the corresponding @code{BEGIN}
6781: before branching. The code looks like this:
6782: @format
6783: @code{lp+!#} current-locals-size @minus{} dest-locals-size
6784: @code{branch} <begin>
6785: @end format
1.6 pazsan 6786:
1.26 crook 6787: @code{UNTIL} is a little more complicated: If it branches back, it
6788: must adjust the stack just like @code{AGAIN}. But if it falls through,
6789: the locals stack must not be changed. The compiler generates the
6790: following code:
6791: @format
6792: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
6793: @end format
6794: The locals stack pointer is only adjusted if the branch is taken.
1.6 pazsan 6795:
1.26 crook 6796: @code{THEN} can produce somewhat inefficient code:
6797: @format
6798: @code{lp+!#} current-locals-size @minus{} orig-locals-size
6799: <orig target>:
6800: @code{lp+!#} orig-locals-size @minus{} new-locals-size
6801: @end format
6802: The second @code{lp+!#} adjusts the locals stack pointer from the
1.29 ! crook 6803: level at the @i{orig} point to the level after the @code{THEN}. The
1.26 crook 6804: first @code{lp+!#} adjusts the locals stack pointer from the current
6805: level to the level at the orig point, so the complete effect is an
6806: adjustment from the current level to the right level after the
6807: @code{THEN}.
1.6 pazsan 6808:
1.26 crook 6809: @cindex locals information on the control-flow stack
6810: @cindex control-flow stack items, locals information
6811: In a conventional Forth implementation a dest control-flow stack entry
6812: is just the target address and an orig entry is just the address to be
6813: patched. Our locals implementation adds a word list to every orig or dest
6814: item. It is the list of locals visible (or assumed visible) at the point
6815: described by the entry. Our implementation also adds a tag to identify
6816: the kind of entry, in particular to differentiate between live and dead
6817: (reachable and unreachable) orig entries.
1.6 pazsan 6818:
1.26 crook 6819: A few unusual operations have to be performed on locals word lists:
1.6 pazsan 6820:
1.26 crook 6821: doc-common-list
6822: doc-sub-list?
6823: doc-list-size
1.6 pazsan 6824:
1.26 crook 6825: Several features of our locals word list implementation make these
6826: operations easy to implement: The locals word lists are organised as
6827: linked lists; the tails of these lists are shared, if the lists
6828: contain some of the same locals; and the address of a name is greater
6829: than the address of the names behind it in the list.
1.6 pazsan 6830:
1.26 crook 6831: Another important implementation detail is the variable
6832: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
6833: determine if they can be reached directly or only through the branch
6834: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
6835: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
6836: definition, by @code{BEGIN} and usually by @code{THEN}.
1.6 pazsan 6837:
1.26 crook 6838: Counted loops are similar to other loops in most respects, but
6839: @code{LEAVE} requires special attention: It performs basically the same
6840: service as @code{AHEAD}, but it does not create a control-flow stack
6841: entry. Therefore the information has to be stored elsewhere;
6842: traditionally, the information was stored in the target fields of the
6843: branches created by the @code{LEAVE}s, by organizing these fields into a
6844: linked list. Unfortunately, this clever trick does not provide enough
6845: space for storing our extended control flow information. Therefore, we
6846: introduce another stack, the leave stack. It contains the control-flow
6847: stack entries for all unresolved @code{LEAVE}s.
1.6 pazsan 6848:
1.26 crook 6849: Local names are kept until the end of the colon definition, even if
6850: they are no longer visible in any control-flow path. In a few cases
6851: this may lead to increased space needs for the locals name area, but
6852: usually less than reclaiming this space would cost in code size.
1.6 pazsan 6853:
6854:
1.26 crook 6855: @node ANS Forth locals, , Gforth locals, Locals
6856: @subsection ANS Forth locals
6857: @cindex locals, ANS Forth style
1.6 pazsan 6858:
1.26 crook 6859: The ANS Forth locals wordset does not define a syntax for locals, but
6860: words that make it possible to define various syntaxes. One of the
6861: possible syntaxes is a subset of the syntax we used in the Gforth locals
6862: wordset, i.e.:
1.6 pazsan 6863:
6864: @example
1.26 crook 6865: @{ local1 local2 ... -- comment @}
1.6 pazsan 6866: @end example
1.23 crook 6867: @noindent
1.26 crook 6868: or
1.6 pazsan 6869: @example
1.26 crook 6870: @{ local1 local2 ... @}
1.6 pazsan 6871: @end example
6872:
1.26 crook 6873: The order of the locals corresponds to the order in a stack comment. The
6874: restrictions are:
1.6 pazsan 6875:
6876: @itemize @bullet
6877: @item
1.26 crook 6878: Locals can only be cell-sized values (no type specifiers are allowed).
1.6 pazsan 6879: @item
1.26 crook 6880: Locals can be defined only outside control structures.
1.6 pazsan 6881: @item
1.26 crook 6882: Locals can interfere with explicit usage of the return stack. For the
6883: exact (and long) rules, see the standard. If you don't use return stack
6884: accessing words in a definition using locals, you will be all right. The
6885: purpose of this rule is to make locals implementation on the return
6886: stack easier.
1.6 pazsan 6887: @item
1.26 crook 6888: The whole definition must be in one line.
6889: @end itemize
1.6 pazsan 6890:
1.26 crook 6891: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
6892: Defining Words}). I.e., they are initialized from the stack. Using their
6893: name produces their value. Their value can be changed using @code{TO}.
1.6 pazsan 6894:
1.26 crook 6895: Since this syntax is supported by Gforth directly, you need not do
6896: anything to use it. If you want to port a program using this syntax to
6897: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
6898: syntax on the other system.
1.6 pazsan 6899:
1.26 crook 6900: Note that a syntax shown in the standard, section A.13 looks
6901: similar, but is quite different in having the order of locals
6902: reversed. Beware!
1.6 pazsan 6903:
1.26 crook 6904: The ANS Forth locals wordset itself consists of a word:
1.6 pazsan 6905:
1.26 crook 6906: doc-(local)
1.6 pazsan 6907:
1.26 crook 6908: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
6909: awful that we strongly recommend not to use it. We have implemented this
6910: syntax to make porting to Gforth easy, but do not document it here. The
6911: problem with this syntax is that the locals are defined in an order
6912: reversed with respect to the standard stack comment notation, making
6913: programs harder to read, and easier to misread and miswrite. The only
6914: merit of this syntax is that it is easy to implement using the ANS Forth
6915: locals wordset.
1.7 pazsan 6916:
6917:
1.26 crook 6918: @c ----------------------------------------------------------
6919: @node Structures, Object-oriented Forth, Locals, Words
6920: @section Structures
6921: @cindex structures
6922: @cindex records
1.7 pazsan 6923:
1.26 crook 6924: This section presents the structure package that comes with Gforth. A
6925: version of the package implemented in ANS Forth is available in
6926: @file{compat/struct.fs}. This package was inspired by a posting on
6927: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
6928: possibly John Hayes). A version of this section has been published in
6929: ???. Marcel Hendrix provided helpful comments.
1.7 pazsan 6930:
1.26 crook 6931: @menu
6932: * Why explicit structure support?::
6933: * Structure Usage::
6934: * Structure Naming Convention::
6935: * Structure Implementation::
6936: * Structure Glossary::
6937: @end menu
1.7 pazsan 6938:
1.26 crook 6939: @node Why explicit structure support?, Structure Usage, Structures, Structures
6940: @subsection Why explicit structure support?
1.7 pazsan 6941:
1.26 crook 6942: @cindex address arithmetic for structures
6943: @cindex structures using address arithmetic
6944: If we want to use a structure containing several fields, we could simply
6945: reserve memory for it, and access the fields using address arithmetic
1.27 crook 6946: (@pxref{Address Arithmetic}). As an example, consider a structure with
1.26 crook 6947: the following fields
1.7 pazsan 6948:
1.26 crook 6949: @table @code
6950: @item a
6951: is a float
6952: @item b
6953: is a cell
6954: @item c
6955: is a float
6956: @end table
1.7 pazsan 6957:
1.26 crook 6958: Given the (float-aligned) base address of the structure we get the
6959: address of the field
1.13 pazsan 6960:
1.26 crook 6961: @table @code
6962: @item a
6963: without doing anything further.
6964: @item b
6965: with @code{float+}
6966: @item c
6967: with @code{float+ cell+ faligned}
6968: @end table
1.13 pazsan 6969:
1.26 crook 6970: It is easy to see that this can become quite tiring.
1.13 pazsan 6971:
1.26 crook 6972: Moreover, it is not very readable, because seeing a
6973: @code{cell+} tells us neither which kind of structure is
6974: accessed nor what field is accessed; we have to somehow infer the kind
6975: of structure, and then look up in the documentation, which field of
6976: that structure corresponds to that offset.
1.13 pazsan 6977:
1.26 crook 6978: Finally, this kind of address arithmetic also causes maintenance
6979: troubles: If you add or delete a field somewhere in the middle of the
6980: structure, you have to find and change all computations for the fields
6981: afterwards.
1.13 pazsan 6982:
1.26 crook 6983: So, instead of using @code{cell+} and friends directly, how
6984: about storing the offsets in constants:
1.13 pazsan 6985:
6986: @example
1.26 crook 6987: 0 constant a-offset
6988: 0 float+ constant b-offset
6989: 0 float+ cell+ faligned c-offset
1.13 pazsan 6990: @end example
6991:
1.26 crook 6992: Now we can get the address of field @code{x} with @code{x-offset
6993: +}. This is much better in all respects. Of course, you still
6994: have to change all later offset definitions if you add a field. You can
6995: fix this by declaring the offsets in the following way:
1.13 pazsan 6996:
6997: @example
1.26 crook 6998: 0 constant a-offset
6999: a-offset float+ constant b-offset
7000: b-offset cell+ faligned constant c-offset
1.13 pazsan 7001: @end example
7002:
1.26 crook 7003: Since we always use the offsets with @code{+}, we could use a defining
7004: word @code{cfield} that includes the @code{+} in the action of the
7005: defined word:
1.8 pazsan 7006:
7007: @example
1.26 crook 7008: : cfield ( n "name" -- )
7009: create ,
7010: does> ( name execution: addr1 -- addr2 )
7011: @@ + ;
1.13 pazsan 7012:
1.26 crook 7013: 0 cfield a
7014: 0 a float+ cfield b
7015: 0 b cell+ faligned cfield c
1.13 pazsan 7016: @end example
7017:
1.26 crook 7018: Instead of @code{x-offset +}, we now simply write @code{x}.
7019:
7020: The structure field words now can be used quite nicely. However,
7021: their definition is still a bit cumbersome: We have to repeat the
7022: name, the information about size and alignment is distributed before
7023: and after the field definitions etc. The structure package presented
7024: here addresses these problems.
7025:
7026: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
7027: @subsection Structure Usage
7028: @cindex structure usage
1.13 pazsan 7029:
1.26 crook 7030: @cindex @code{field} usage
7031: @cindex @code{struct} usage
7032: @cindex @code{end-struct} usage
7033: You can define a structure for a (data-less) linked list with:
1.13 pazsan 7034: @example
1.26 crook 7035: struct
7036: cell% field list-next
7037: end-struct list%
1.13 pazsan 7038: @end example
7039:
1.26 crook 7040: With the address of the list node on the stack, you can compute the
7041: address of the field that contains the address of the next node with
7042: @code{list-next}. E.g., you can determine the length of a list
7043: with:
1.13 pazsan 7044:
7045: @example
1.26 crook 7046: : list-length ( list -- n )
7047: \ "list" is a pointer to the first element of a linked list
7048: \ "n" is the length of the list
7049: 0 BEGIN ( list1 n1 )
7050: over
7051: WHILE ( list1 n1 )
7052: 1+ swap list-next @@ swap
7053: REPEAT
7054: nip ;
1.13 pazsan 7055: @end example
7056:
1.26 crook 7057: You can reserve memory for a list node in the dictionary with
7058: @code{list% %allot}, which leaves the address of the list node on the
7059: stack. For the equivalent allocation on the heap you can use @code{list%
7060: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
7061: use @code{list% %allocate}). You can get the the size of a list
7062: node with @code{list% %size} and its alignment with @code{list%
7063: %alignment}.
1.13 pazsan 7064:
1.26 crook 7065: Note that in ANS Forth the body of a @code{create}d word is
7066: @code{aligned} but not necessarily @code{faligned};
7067: therefore, if you do a:
1.13 pazsan 7068: @example
1.26 crook 7069: create @emph{name} foo% %allot
1.8 pazsan 7070: @end example
7071:
1.26 crook 7072: @noindent
7073: then the memory alloted for @code{foo%} is
7074: guaranteed to start at the body of @code{@emph{name}} only if
7075: @code{foo%} contains only character, cell and double fields.
1.20 pazsan 7076:
1.26 crook 7077: @cindex strcutures containing structures
7078: You can include a structure @code{foo%} as a field of
7079: another structure, like this:
1.20 pazsan 7080: @example
1.26 crook 7081: struct
7082: ...
7083: foo% field ...
7084: ...
7085: end-struct ...
1.20 pazsan 7086: @end example
7087:
1.26 crook 7088: @cindex structure extension
7089: @cindex extended records
7090: Instead of starting with an empty structure, you can extend an
7091: existing structure. E.g., a plain linked list without data, as defined
7092: above, is hardly useful; You can extend it to a linked list of integers,
7093: like this:@footnote{This feature is also known as @emph{extended
7094: records}. It is the main innovation in the Oberon language; in other
7095: words, adding this feature to Modula-2 led Wirth to create a new
7096: language, write a new compiler etc. Adding this feature to Forth just
7097: required a few lines of code.}
1.20 pazsan 7098:
7099: @example
1.26 crook 7100: list%
7101: cell% field intlist-int
7102: end-struct intlist%
1.20 pazsan 7103: @end example
7104:
1.26 crook 7105: @code{intlist%} is a structure with two fields:
7106: @code{list-next} and @code{intlist-int}.
1.20 pazsan 7107:
1.26 crook 7108: @cindex structures containing arrays
7109: You can specify an array type containing @emph{n} elements of
7110: type @code{foo%} like this:
1.20 pazsan 7111:
7112: @example
1.26 crook 7113: foo% @emph{n} *
1.20 pazsan 7114: @end example
7115:
1.26 crook 7116: You can use this array type in any place where you can use a normal
7117: type, e.g., when defining a @code{field}, or with
7118: @code{%allot}.
1.20 pazsan 7119:
1.26 crook 7120: @cindex first field optimization
7121: The first field is at the base address of a structure and the word
7122: for this field (e.g., @code{list-next}) actually does not change
7123: the address on the stack. You may be tempted to leave it away in the
7124: interest of run-time and space efficiency. This is not necessary,
7125: because the structure package optimizes this case and compiling such
7126: words does not generate any code. So, in the interest of readability
7127: and maintainability you should include the word for the field when
7128: accessing the field.
1.20 pazsan 7129:
1.26 crook 7130: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
7131: @subsection Structure Naming Convention
7132: @cindex structure naming convention
1.20 pazsan 7133:
1.26 crook 7134: The field names that come to (my) mind are often quite generic, and,
7135: if used, would cause frequent name clashes. E.g., many structures
7136: probably contain a @code{counter} field. The structure names
7137: that come to (my) mind are often also the logical choice for the names
7138: of words that create such a structure.
1.20 pazsan 7139:
1.26 crook 7140: Therefore, I have adopted the following naming conventions:
1.20 pazsan 7141:
1.26 crook 7142: @itemize @bullet
7143: @cindex field naming convention
7144: @item
7145: The names of fields are of the form
7146: @code{@emph{struct}-@emph{field}}, where
7147: @code{@emph{struct}} is the basic name of the structure, and
7148: @code{@emph{field}} is the basic name of the field. You can
7149: think of field words as converting the (address of the)
7150: structure into the (address of the) field.
1.20 pazsan 7151:
1.26 crook 7152: @cindex structure naming convention
7153: @item
7154: The names of structures are of the form
7155: @code{@emph{struct}%}, where
7156: @code{@emph{struct}} is the basic name of the structure.
7157: @end itemize
1.20 pazsan 7158:
1.26 crook 7159: This naming convention does not work that well for fields of extended
7160: structures; e.g., the integer list structure has a field
7161: @code{intlist-int}, but has @code{list-next}, not
7162: @code{intlist-next}.
1.20 pazsan 7163:
1.26 crook 7164: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
7165: @subsection Structure Implementation
7166: @cindex structure implementation
7167: @cindex implementation of structures
1.20 pazsan 7168:
1.26 crook 7169: The central idea in the implementation is to pass the data about the
7170: structure being built on the stack, not in some global
7171: variable. Everything else falls into place naturally once this design
7172: decision is made.
1.20 pazsan 7173:
1.26 crook 7174: The type description on the stack is of the form @emph{align
7175: size}. Keeping the size on the top-of-stack makes dealing with arrays
7176: very simple.
1.20 pazsan 7177:
1.26 crook 7178: @code{field} is a defining word that uses @code{Create}
7179: and @code{DOES>}. The body of the field contains the offset
7180: of the field, and the normal @code{DOES>} action is simply:
1.20 pazsan 7181:
7182: @example
1.26 crook 7183: @ +
1.20 pazsan 7184: @end example
7185:
1.23 crook 7186: @noindent
1.26 crook 7187: i.e., add the offset to the address, giving the stack effect
1.29 ! crook 7188: @i{addr1 -- addr2} for a field.
1.20 pazsan 7189:
1.26 crook 7190: @cindex first field optimization, implementation
7191: This simple structure is slightly complicated by the optimization
7192: for fields with offset 0, which requires a different
7193: @code{DOES>}-part (because we cannot rely on there being
7194: something on the stack if such a field is invoked during
7195: compilation). Therefore, we put the different @code{DOES>}-parts
7196: in separate words, and decide which one to invoke based on the
7197: offset. For a zero offset, the field is basically a noop; it is
7198: immediate, and therefore no code is generated when it is compiled.
1.20 pazsan 7199:
1.26 crook 7200: @node Structure Glossary, , Structure Implementation, Structures
7201: @subsection Structure Glossary
7202: @cindex structure glossary
1.20 pazsan 7203:
1.26 crook 7204: doc-%align
7205: doc-%alignment
7206: doc-%alloc
7207: doc-%allocate
7208: doc-%allot
7209: doc-cell%
7210: doc-char%
7211: doc-dfloat%
7212: doc-double%
7213: doc-end-struct
7214: doc-field
7215: doc-float%
7216: doc-naligned
7217: doc-sfloat%
7218: doc-%size
7219: doc-struct
1.23 crook 7220:
1.26 crook 7221: @c -------------------------------------------------------------
7222: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
7223: @section Object-oriented Forth
1.20 pazsan 7224:
1.26 crook 7225: Gforth comes with three packages for object-oriented programming:
7226: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
7227: is preloaded, so you have to @code{include} them before use. The most
7228: important differences between these packages (and others) are discussed
7229: in @ref{Comparison with other object models}. All packages are written
7230: in ANS Forth and can be used with any other ANS Forth.
1.20 pazsan 7231:
1.26 crook 7232: @menu
7233: * Why object-oriented programming?::
7234: * Object-Oriented Terminology::
7235: * Objects::
7236: * OOF::
7237: * Mini-OOF::
7238: * Comparison with other object models::
7239: @end menu
1.20 pazsan 7240:
1.23 crook 7241:
1.26 crook 7242: @node Why object-oriented programming?, Object-Oriented Terminology, , Object-oriented Forth
7243: @subsubsection Why object-oriented programming?
7244: @cindex object-oriented programming motivation
7245: @cindex motivation for object-oriented programming
1.23 crook 7246:
1.26 crook 7247: Often we have to deal with several data structures (@emph{objects}),
7248: that have to be treated similarly in some respects, but differently in
7249: others. Graphical objects are the textbook example: circles, triangles,
7250: dinosaurs, icons, and others, and we may want to add more during program
7251: development. We want to apply some operations to any graphical object,
7252: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
7253: has to do something different for every kind of object.
7254: @comment TODO add some other operations eg perimeter, area
7255: @comment and tie in to concrete examples later..
1.23 crook 7256:
1.26 crook 7257: We could implement @code{draw} as a big @code{CASE}
7258: control structure that executes the appropriate code depending on the
7259: kind of object to be drawn. This would be not be very elegant, and,
7260: moreover, we would have to change @code{draw} every time we add
7261: a new kind of graphical object (say, a spaceship).
1.23 crook 7262:
1.26 crook 7263: What we would rather do is: When defining spaceships, we would tell
7264: the system: ``Here's how you @code{draw} a spaceship; you figure
7265: out the rest''.
1.23 crook 7266:
1.26 crook 7267: This is the problem that all systems solve that (rightfully) call
7268: themselves object-oriented; the object-oriented packages presented here
7269: solve this problem (and not much else).
7270: @comment TODO ?list properties of oo systems.. oo vs o-based?
1.23 crook 7271:
1.26 crook 7272: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
7273: @subsubsection Object-Oriented Terminology
7274: @cindex object-oriented terminology
7275: @cindex terminology for object-oriented programming
1.23 crook 7276:
1.26 crook 7277: This section is mainly for reference, so you don't have to understand
7278: all of it right away. The terminology is mainly Smalltalk-inspired. In
7279: short:
1.23 crook 7280:
1.26 crook 7281: @table @emph
7282: @cindex class
7283: @item class
7284: a data structure definition with some extras.
1.23 crook 7285:
1.26 crook 7286: @cindex object
7287: @item object
7288: an instance of the data structure described by the class definition.
1.23 crook 7289:
1.26 crook 7290: @cindex instance variables
7291: @item instance variables
7292: fields of the data structure.
1.23 crook 7293:
1.26 crook 7294: @cindex selector
7295: @cindex method selector
7296: @cindex virtual function
7297: @item selector
7298: (or @emph{method selector}) a word (e.g.,
7299: @code{draw}) that performs an operation on a variety of data
7300: structures (classes). A selector describes @emph{what} operation to
7301: perform. In C++ terminology: a (pure) virtual function.
1.23 crook 7302:
1.26 crook 7303: @cindex method
7304: @item method
7305: the concrete definition that performs the operation
7306: described by the selector for a specific class. A method specifies
7307: @emph{how} the operation is performed for a specific class.
1.23 crook 7308:
1.26 crook 7309: @cindex selector invocation
7310: @cindex message send
7311: @cindex invoking a selector
7312: @item selector invocation
7313: a call of a selector. One argument of the call (the TOS (top-of-stack))
7314: is used for determining which method is used. In Smalltalk terminology:
7315: a message (consisting of the selector and the other arguments) is sent
7316: to the object.
1.1 anton 7317:
1.26 crook 7318: @cindex receiving object
7319: @item receiving object
7320: the object used for determining the method executed by a selector
7321: invocation. In the @file{objects.fs} model, it is the object that is on
7322: the TOS when the selector is invoked. (@emph{Receiving} comes from
7323: the Smalltalk @emph{message} terminology.)
1.1 anton 7324:
1.26 crook 7325: @cindex child class
7326: @cindex parent class
7327: @cindex inheritance
7328: @item child class
7329: a class that has (@emph{inherits}) all properties (instance variables,
7330: selectors, methods) from a @emph{parent class}. In Smalltalk
7331: terminology: The subclass inherits from the superclass. In C++
7332: terminology: The derived class inherits from the base class.
1.1 anton 7333:
1.26 crook 7334: @end table
1.21 crook 7335:
1.26 crook 7336: @c If you wonder about the message sending terminology, it comes from
7337: @c a time when each object had it's own task and objects communicated via
7338: @c message passing; eventually the Smalltalk developers realized that
7339: @c they can do most things through simple (indirect) calls. They kept the
7340: @c terminology.
1.1 anton 7341:
7342:
1.26 crook 7343: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
7344: @subsection The @file{objects.fs} model
7345: @cindex objects
7346: @cindex object-oriented programming
1.1 anton 7347:
1.26 crook 7348: @cindex @file{objects.fs}
7349: @cindex @file{oof.fs}
1.1 anton 7350:
1.26 crook 7351: This section describes the @file{objects.fs} package. This material also has been published in @cite{Yet Another Forth Objects Package} by Anton Ertl and appeared in Forth Dimensions 19(2), pages 37--43 (@url{http://www.complang.tuwien.ac.at/forth/objects/objects.html}).
7352: @c McKewan's and Zsoter's packages
1.1 anton 7353:
1.26 crook 7354: This section assumes that you have read @ref{Structures}.
1.1 anton 7355:
1.26 crook 7356: The techniques on which this model is based have been used to implement
7357: the parser generator, Gray, and have also been used in Gforth for
7358: implementing the various flavours of word lists (hashed or not,
7359: case-sensitive or not, special-purpose word lists for locals etc.).
1.1 anton 7360:
7361:
1.26 crook 7362: @menu
7363: * Properties of the Objects model::
7364: * Basic Objects Usage::
7365: * The Objects base class::
7366: * Creating objects::
7367: * Object-Oriented Programming Style::
7368: * Class Binding::
7369: * Method conveniences::
7370: * Classes and Scoping::
7371: * Object Interfaces::
7372: * Objects Implementation::
7373: * Objects Glossary::
7374: @end menu
1.1 anton 7375:
1.26 crook 7376: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
7377: and Bernd Paysan helped me with the related works section.
1.1 anton 7378:
1.26 crook 7379: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
7380: @subsubsection Properties of the @file{objects.fs} model
7381: @cindex @file{objects.fs} properties
1.1 anton 7382:
1.26 crook 7383: @itemize @bullet
7384: @item
7385: It is straightforward to pass objects on the stack. Passing
7386: selectors on the stack is a little less convenient, but possible.
1.1 anton 7387:
1.26 crook 7388: @item
7389: Objects are just data structures in memory, and are referenced by their
7390: address. You can create words for objects with normal defining words
7391: like @code{constant}. Likewise, there is no difference between instance
7392: variables that contain objects and those that contain other data.
1.1 anton 7393:
1.26 crook 7394: @item
7395: Late binding is efficient and easy to use.
1.21 crook 7396:
1.26 crook 7397: @item
7398: It avoids parsing, and thus avoids problems with state-smartness
7399: and reduced extensibility; for convenience there are a few parsing
7400: words, but they have non-parsing counterparts. There are also a few
7401: defining words that parse. This is hard to avoid, because all standard
7402: defining words parse (except @code{:noname}); however, such
7403: words are not as bad as many other parsing words, because they are not
7404: state-smart.
1.21 crook 7405:
1.26 crook 7406: @item
7407: It does not try to incorporate everything. It does a few things and does
7408: them well (IMO). In particular, this model was not designed to support
7409: information hiding (although it has features that may help); you can use
7410: a separate package for achieving this.
1.21 crook 7411:
1.26 crook 7412: @item
7413: It is layered; you don't have to learn and use all features to use this
7414: model. Only a few features are necessary (@xref{Basic Objects Usage},
7415: @xref{The Objects base class}, @xref{Creating objects}.), the others
7416: are optional and independent of each other.
1.21 crook 7417:
1.26 crook 7418: @item
7419: An implementation in ANS Forth is available.
1.21 crook 7420:
1.26 crook 7421: @end itemize
1.21 crook 7422:
7423:
1.26 crook 7424: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
7425: @subsubsection Basic @file{objects.fs} Usage
7426: @cindex basic objects usage
7427: @cindex objects, basic usage
1.21 crook 7428:
1.26 crook 7429: You can define a class for graphical objects like this:
1.21 crook 7430:
1.26 crook 7431: @cindex @code{class} usage
7432: @cindex @code{end-class} usage
7433: @cindex @code{selector} usage
7434: @example
7435: object class \ "object" is the parent class
7436: selector draw ( x y graphical -- )
7437: end-class graphical
7438: @end example
1.21 crook 7439:
1.26 crook 7440: This code defines a class @code{graphical} with an
7441: operation @code{draw}. We can perform the operation
7442: @code{draw} on any @code{graphical} object, e.g.:
1.21 crook 7443:
1.26 crook 7444: @example
7445: 100 100 t-rex draw
7446: @end example
1.21 crook 7447:
1.26 crook 7448: @noindent
7449: where @code{t-rex} is a word (say, a constant) that produces a
7450: graphical object.
1.21 crook 7451:
1.29 ! crook 7452: @comment TODO add a 2nd operation eg perimeter.. and use for
1.26 crook 7453: @comment a concrete example
1.21 crook 7454:
1.26 crook 7455: @cindex abstract class
7456: How do we create a graphical object? With the present definitions,
7457: we cannot create a useful graphical object. The class
7458: @code{graphical} describes graphical objects in general, but not
7459: any concrete graphical object type (C++ users would call it an
7460: @emph{abstract class}); e.g., there is no method for the selector
7461: @code{draw} in the class @code{graphical}.
1.21 crook 7462:
1.26 crook 7463: For concrete graphical objects, we define child classes of the
7464: class @code{graphical}, e.g.:
1.21 crook 7465:
1.26 crook 7466: @cindex @code{overrides} usage
7467: @cindex @code{field} usage in class definition
7468: @example
7469: graphical class \ "graphical" is the parent class
7470: cell% field circle-radius
1.21 crook 7471:
1.26 crook 7472: :noname ( x y circle -- )
7473: circle-radius @@ draw-circle ;
7474: overrides draw
1.21 crook 7475:
1.26 crook 7476: :noname ( n-radius circle -- )
7477: circle-radius ! ;
7478: overrides construct
1.21 crook 7479:
1.26 crook 7480: end-class circle
1.21 crook 7481: @end example
7482:
1.26 crook 7483: Here we define a class @code{circle} as a child of @code{graphical},
7484: with field @code{circle-radius} (which behaves just like a field
7485: (@pxref{Structures}); it defines (using @code{overrides}) new methods
7486: for the selectors @code{draw} and @code{construct} (@code{construct} is
7487: defined in @code{object}, the parent class of @code{graphical}).
1.21 crook 7488:
1.26 crook 7489: Now we can create a circle on the heap (i.e.,
7490: @code{allocate}d memory) with:
1.21 crook 7491:
1.26 crook 7492: @cindex @code{heap-new} usage
1.21 crook 7493: @example
1.26 crook 7494: 50 circle heap-new constant my-circle
7495: @end example
1.21 crook 7496:
1.26 crook 7497: @noindent
7498: @code{heap-new} invokes @code{construct}, thus
7499: initializing the field @code{circle-radius} with 50. We can draw
7500: this new circle at (100,100) with:
1.21 crook 7501:
1.26 crook 7502: @example
7503: 100 100 my-circle draw
1.21 crook 7504: @end example
7505:
1.26 crook 7506: @cindex selector invocation, restrictions
7507: @cindex class definition, restrictions
7508: Note: You can only invoke a selector if the object on the TOS
7509: (the receiving object) belongs to the class where the selector was
7510: defined or one of its descendents; e.g., you can invoke
7511: @code{draw} only for objects belonging to @code{graphical}
7512: or its descendents (e.g., @code{circle}). Immediately before
7513: @code{end-class}, the search order has to be the same as
7514: immediately after @code{class}.
1.21 crook 7515:
1.26 crook 7516: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
7517: @subsubsection The @file{object.fs} base class
7518: @cindex @code{object} class
1.21 crook 7519:
1.26 crook 7520: When you define a class, you have to specify a parent class. So how do
7521: you start defining classes? There is one class available from the start:
7522: @code{object}. It is ancestor for all classes and so is the
7523: only class that has no parent. It has two selectors: @code{construct}
7524: and @code{print}.
1.21 crook 7525:
1.26 crook 7526: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
7527: @subsubsection Creating objects
7528: @cindex creating objects
7529: @cindex object creation
7530: @cindex object allocation options
1.21 crook 7531:
1.26 crook 7532: @cindex @code{heap-new} discussion
7533: @cindex @code{dict-new} discussion
7534: @cindex @code{construct} discussion
7535: You can create and initialize an object of a class on the heap with
7536: @code{heap-new} ( ... class -- object ) and in the dictionary
7537: (allocation with @code{allot}) with @code{dict-new} (
7538: ... class -- object ). Both words invoke @code{construct}, which
7539: consumes the stack items indicated by "..." above.
1.21 crook 7540:
1.26 crook 7541: @cindex @code{init-object} discussion
7542: @cindex @code{class-inst-size} discussion
7543: If you want to allocate memory for an object yourself, you can get its
7544: alignment and size with @code{class-inst-size 2@@} ( class --
7545: align size ). Once you have memory for an object, you can initialize
7546: it with @code{init-object} ( ... class object -- );
7547: @code{construct} does only a part of the necessary work.
1.21 crook 7548:
1.26 crook 7549: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
7550: @subsubsection Object-Oriented Programming Style
7551: @cindex object-oriented programming style
1.21 crook 7552:
1.26 crook 7553: This section is not exhaustive.
1.1 anton 7554:
1.26 crook 7555: @cindex stack effects of selectors
7556: @cindex selectors and stack effects
7557: In general, it is a good idea to ensure that all methods for the
7558: same selector have the same stack effect: when you invoke a selector,
7559: you often have no idea which method will be invoked, so, unless all
7560: methods have the same stack effect, you will not know the stack effect
7561: of the selector invocation.
1.21 crook 7562:
1.26 crook 7563: One exception to this rule is methods for the selector
7564: @code{construct}. We know which method is invoked, because we
7565: specify the class to be constructed at the same place. Actually, I
7566: defined @code{construct} as a selector only to give the users a
7567: convenient way to specify initialization. The way it is used, a
7568: mechanism different from selector invocation would be more natural
7569: (but probably would take more code and more space to explain).
1.21 crook 7570:
1.26 crook 7571: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
7572: @subsubsection Class Binding
7573: @cindex class binding
7574: @cindex early binding
1.21 crook 7575:
1.26 crook 7576: @cindex late binding
7577: Normal selector invocations determine the method at run-time depending
7578: on the class of the receiving object. This run-time selection is called
1.29 ! crook 7579: @i{late binding}.
1.21 crook 7580:
1.26 crook 7581: Sometimes it's preferable to invoke a different method. For example,
7582: you might want to use the simple method for @code{print}ing
7583: @code{object}s instead of the possibly long-winded @code{print} method
7584: of the receiver class. You can achieve this by replacing the invocation
7585: of @code{print} with:
1.21 crook 7586:
1.26 crook 7587: @cindex @code{[bind]} usage
7588: @example
7589: [bind] object print
1.21 crook 7590: @end example
7591:
1.26 crook 7592: @noindent
7593: in compiled code or:
1.21 crook 7594:
1.26 crook 7595: @cindex @code{bind} usage
1.21 crook 7596: @example
1.26 crook 7597: bind object print
1.21 crook 7598: @end example
7599:
1.26 crook 7600: @cindex class binding, alternative to
7601: @noindent
7602: in interpreted code. Alternatively, you can define the method with a
7603: name (e.g., @code{print-object}), and then invoke it through the
7604: name. Class binding is just a (often more convenient) way to achieve
7605: the same effect; it avoids name clutter and allows you to invoke
7606: methods directly without naming them first.
7607:
7608: @cindex superclass binding
7609: @cindex parent class binding
7610: A frequent use of class binding is this: When we define a method
7611: for a selector, we often want the method to do what the selector does
7612: in the parent class, and a little more. There is a special word for
7613: this purpose: @code{[parent]}; @code{[parent]
7614: @emph{selector}} is equivalent to @code{[bind] @emph{parent
7615: selector}}, where @code{@emph{parent}} is the parent
7616: class of the current class. E.g., a method definition might look like:
1.21 crook 7617:
1.26 crook 7618: @cindex @code{[parent]} usage
1.21 crook 7619: @example
1.26 crook 7620: :noname
7621: dup [parent] foo \ do parent's foo on the receiving object
7622: ... \ do some more
7623: ; overrides foo
1.21 crook 7624: @end example
7625:
1.26 crook 7626: @cindex class binding as optimization
7627: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
7628: March 1997), Andrew McKewan presents class binding as an optimization
7629: technique. I recommend not using it for this purpose unless you are in
7630: an emergency. Late binding is pretty fast with this model anyway, so the
7631: benefit of using class binding is small; the cost of using class binding
7632: where it is not appropriate is reduced maintainability.
1.21 crook 7633:
1.26 crook 7634: While we are at programming style questions: You should bind
7635: selectors only to ancestor classes of the receiving object. E.g., say,
7636: you know that the receiving object is of class @code{foo} or its
7637: descendents; then you should bind only to @code{foo} and its
7638: ancestors.
1.21 crook 7639:
1.26 crook 7640: @node Method conveniences, Classes and Scoping, Class Binding, Objects
7641: @subsubsection Method conveniences
7642: @cindex method conveniences
1.1 anton 7643:
1.26 crook 7644: In a method you usually access the receiving object pretty often. If
7645: you define the method as a plain colon definition (e.g., with
7646: @code{:noname}), you may have to do a lot of stack
7647: gymnastics. To avoid this, you can define the method with @code{m:
7648: ... ;m}. E.g., you could define the method for
7649: @code{draw}ing a @code{circle} with
1.20 pazsan 7650:
1.26 crook 7651: @cindex @code{this} usage
7652: @cindex @code{m:} usage
7653: @cindex @code{;m} usage
7654: @example
7655: m: ( x y circle -- )
7656: ( x y ) this circle-radius @@ draw-circle ;m
7657: @end example
1.20 pazsan 7658:
1.26 crook 7659: @cindex @code{exit} in @code{m: ... ;m}
7660: @cindex @code{exitm} discussion
7661: @cindex @code{catch} in @code{m: ... ;m}
7662: When this method is executed, the receiver object is removed from the
7663: stack; you can access it with @code{this} (admittedly, in this
7664: example the use of @code{m: ... ;m} offers no advantage). Note
7665: that I specify the stack effect for the whole method (i.e. including
7666: the receiver object), not just for the code between @code{m:}
7667: and @code{;m}. You cannot use @code{exit} in
7668: @code{m:...;m}; instead, use
7669: @code{exitm}.@footnote{Moreover, for any word that calls
7670: @code{catch} and was defined before loading
7671: @code{objects.fs}, you have to redefine it like I redefined
7672: @code{catch}: @code{: catch this >r catch r> to-this ;}}
1.20 pazsan 7673:
1.26 crook 7674: @cindex @code{inst-var} usage
7675: You will frequently use sequences of the form @code{this
7676: @emph{field}} (in the example above: @code{this
7677: circle-radius}). If you use the field only in this way, you can
7678: define it with @code{inst-var} and eliminate the
7679: @code{this} before the field name. E.g., the @code{circle}
7680: class above could also be defined with:
1.20 pazsan 7681:
1.26 crook 7682: @example
7683: graphical class
7684: cell% inst-var radius
1.20 pazsan 7685:
1.26 crook 7686: m: ( x y circle -- )
7687: radius @@ draw-circle ;m
7688: overrides draw
1.20 pazsan 7689:
1.26 crook 7690: m: ( n-radius circle -- )
7691: radius ! ;m
7692: overrides construct
1.12 anton 7693:
1.26 crook 7694: end-class circle
7695: @end example
1.12 anton 7696:
1.26 crook 7697: @code{radius} can only be used in @code{circle} and its
7698: descendent classes and inside @code{m:...;m}.
1.12 anton 7699:
1.26 crook 7700: @cindex @code{inst-value} usage
7701: You can also define fields with @code{inst-value}, which is
7702: to @code{inst-var} what @code{value} is to
7703: @code{variable}. You can change the value of such a field with
7704: @code{[to-inst]}. E.g., we could also define the class
7705: @code{circle} like this:
1.12 anton 7706:
1.26 crook 7707: @example
7708: graphical class
7709: inst-value radius
1.12 anton 7710:
1.26 crook 7711: m: ( x y circle -- )
7712: radius draw-circle ;m
7713: overrides draw
1.12 anton 7714:
1.26 crook 7715: m: ( n-radius circle -- )
7716: [to-inst] radius ;m
7717: overrides construct
1.21 crook 7718:
1.26 crook 7719: end-class circle
1.12 anton 7720: @end example
7721:
7722:
1.26 crook 7723: @node Classes and Scoping, Object Interfaces, Method conveniences, Objects
7724: @subsubsection Classes and Scoping
7725: @cindex classes and scoping
7726: @cindex scoping and classes
1.12 anton 7727:
1.26 crook 7728: Inheritance is frequent, unlike structure extension. This exacerbates
7729: the problem with the field name convention (@pxref{Structure Naming
7730: Convention}): One always has to remember in which class the field was
7731: originally defined; changing a part of the class structure would require
7732: changes for renaming in otherwise unaffected code.
1.12 anton 7733:
1.26 crook 7734: @cindex @code{inst-var} visibility
7735: @cindex @code{inst-value} visibility
7736: To solve this problem, I added a scoping mechanism (which was not in my
7737: original charter): A field defined with @code{inst-var} (or
7738: @code{inst-value}) is visible only in the class where it is defined and in
7739: the descendent classes of this class. Using such fields only makes
7740: sense in @code{m:}-defined methods in these classes anyway.
1.12 anton 7741:
1.26 crook 7742: This scoping mechanism allows us to use the unadorned field name,
7743: because name clashes with unrelated words become much less likely.
1.12 anton 7744:
1.26 crook 7745: @cindex @code{protected} discussion
7746: @cindex @code{private} discussion
7747: Once we have this mechanism, we can also use it for controlling the
7748: visibility of other words: All words defined after
7749: @code{protected} are visible only in the current class and its
7750: descendents. @code{public} restores the compilation
7751: (i.e. @code{current}) word list that was in effect before. If you
7752: have several @code{protected}s without an intervening
7753: @code{public} or @code{set-current}, @code{public}
7754: will restore the compilation word list in effect before the first of
7755: these @code{protected}s.
1.12 anton 7756:
1.26 crook 7757: @node Object Interfaces, Objects Implementation, Classes and Scoping, Objects
7758: @subsubsection Object Interfaces
7759: @cindex object interfaces
7760: @cindex interfaces for objects
1.12 anton 7761:
1.26 crook 7762: In this model you can only call selectors defined in the class of the
7763: receiving objects or in one of its ancestors. If you call a selector
7764: with a receiving object that is not in one of these classes, the
7765: result is undefined; if you are lucky, the program crashes
7766: immediately.
1.12 anton 7767:
1.26 crook 7768: @cindex selectors common to hardly-related classes
7769: Now consider the case when you want to have a selector (or several)
7770: available in two classes: You would have to add the selector to a
7771: common ancestor class, in the worst case to @code{object}. You
7772: may not want to do this, e.g., because someone else is responsible for
7773: this ancestor class.
1.12 anton 7774:
1.26 crook 7775: The solution for this problem is interfaces. An interface is a
7776: collection of selectors. If a class implements an interface, the
7777: selectors become available to the class and its descendents. A class
7778: can implement an unlimited number of interfaces. For the problem
7779: discussed above, we would define an interface for the selector(s), and
7780: both classes would implement the interface.
1.12 anton 7781:
1.26 crook 7782: As an example, consider an interface @code{storage} for
7783: writing objects to disk and getting them back, and a class
7784: @code{foo} that implements it. The code would look like this:
1.12 anton 7785:
1.26 crook 7786: @cindex @code{interface} usage
7787: @cindex @code{end-interface} usage
7788: @cindex @code{implementation} usage
7789: @example
7790: interface
7791: selector write ( file object -- )
7792: selector read1 ( file object -- )
7793: end-interface storage
1.12 anton 7794:
1.26 crook 7795: bar class
7796: storage implementation
1.12 anton 7797:
1.26 crook 7798: ... overrides write
7799: ... overrides read
7800: ...
7801: end-class foo
1.12 anton 7802: @end example
7803:
1.26 crook 7804: @noindent
1.29 ! crook 7805: (I would add a word @code{read} @i{( file -- object )} that uses
1.26 crook 7806: @code{read1} internally, but that's beyond the point illustrated
7807: here.)
1.12 anton 7808:
1.26 crook 7809: Note that you cannot use @code{protected} in an interface; and
7810: of course you cannot define fields.
1.12 anton 7811:
1.26 crook 7812: In the Neon model, all selectors are available for all classes;
7813: therefore it does not need interfaces. The price you pay in this model
7814: is slower late binding, and therefore, added complexity to avoid late
7815: binding.
1.12 anton 7816:
1.26 crook 7817: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
7818: @subsubsection @file{objects.fs} Implementation
7819: @cindex @file{objects.fs} implementation
1.12 anton 7820:
1.26 crook 7821: @cindex @code{object-map} discussion
7822: An object is a piece of memory, like one of the data structures
7823: described with @code{struct...end-struct}. It has a field
7824: @code{object-map} that points to the method map for the object's
7825: class.
1.12 anton 7826:
1.26 crook 7827: @cindex method map
7828: @cindex virtual function table
7829: The @emph{method map}@footnote{This is Self terminology; in C++
7830: terminology: virtual function table.} is an array that contains the
1.29 ! crook 7831: execution tokens (@i{xt}s) of the methods for the object's class. Each
1.26 crook 7832: selector contains an offset into a method map.
1.12 anton 7833:
1.26 crook 7834: @cindex @code{selector} implementation, class
7835: @code{selector} is a defining word that uses
7836: @code{CREATE} and @code{DOES>}. The body of the
7837: selector contains the offset; the @code{does>} action for a
7838: class selector is, basically:
1.21 crook 7839:
1.26 crook 7840: @example
7841: ( object addr ) @@ over object-map @@ + @@ execute
7842: @end example
1.12 anton 7843:
1.26 crook 7844: Since @code{object-map} is the first field of the object, it
7845: does not generate any code. As you can see, calling a selector has a
7846: small, constant cost.
1.12 anton 7847:
1.26 crook 7848: @cindex @code{current-interface} discussion
7849: @cindex class implementation and representation
7850: A class is basically a @code{struct} combined with a method
7851: map. During the class definition the alignment and size of the class
7852: are passed on the stack, just as with @code{struct}s, so
7853: @code{field} can also be used for defining class
7854: fields. However, passing more items on the stack would be
7855: inconvenient, so @code{class} builds a data structure in memory,
7856: which is accessed through the variable
7857: @code{current-interface}. After its definition is complete, the
7858: class is represented on the stack by a pointer (e.g., as parameter for
7859: a child class definition).
1.1 anton 7860:
1.26 crook 7861: A new class starts off with the alignment and size of its parent,
7862: and a copy of the parent's method map. Defining new fields extends the
7863: size and alignment; likewise, defining new selectors extends the
1.29 ! crook 7864: method map. @code{overrides} just stores a new @i{xt} in the method
1.26 crook 7865: map at the offset given by the selector.
1.20 pazsan 7866:
1.26 crook 7867: @cindex class binding, implementation
1.29 ! crook 7868: Class binding just gets the @i{xt} at the offset given by the selector
1.26 crook 7869: from the class's method map and @code{compile,}s (in the case of
7870: @code{[bind]}) it.
1.21 crook 7871:
1.26 crook 7872: @cindex @code{this} implementation
7873: @cindex @code{catch} and @code{this}
7874: @cindex @code{this} and @code{catch}
7875: I implemented @code{this} as a @code{value}. At the
7876: start of an @code{m:...;m} method the old @code{this} is
7877: stored to the return stack and restored at the end; and the object on
7878: the TOS is stored @code{TO this}. This technique has one
7879: disadvantage: If the user does not leave the method via
7880: @code{;m}, but via @code{throw} or @code{exit},
7881: @code{this} is not restored (and @code{exit} may
7882: crash). To deal with the @code{throw} problem, I have redefined
7883: @code{catch} to save and restore @code{this}; the same
7884: should be done with any word that can catch an exception. As for
7885: @code{exit}, I simply forbid it (as a replacement, there is
7886: @code{exitm}).
1.21 crook 7887:
1.26 crook 7888: @cindex @code{inst-var} implementation
7889: @code{inst-var} is just the same as @code{field}, with
7890: a different @code{DOES>} action:
7891: @example
7892: @@ this +
7893: @end example
7894: Similar for @code{inst-value}.
1.21 crook 7895:
1.26 crook 7896: @cindex class scoping implementation
7897: Each class also has a word list that contains the words defined with
7898: @code{inst-var} and @code{inst-value}, and its protected
7899: words. It also has a pointer to its parent. @code{class} pushes
7900: the word lists of the class and all its ancestors onto the search order stack,
7901: and @code{end-class} drops them.
1.21 crook 7902:
1.26 crook 7903: @cindex interface implementation
7904: An interface is like a class without fields, parent and protected
7905: words; i.e., it just has a method map. If a class implements an
7906: interface, its method map contains a pointer to the method map of the
7907: interface. The positive offsets in the map are reserved for class
7908: methods, therefore interface map pointers have negative
7909: offsets. Interfaces have offsets that are unique throughout the
7910: system, unlike class selectors, whose offsets are only unique for the
7911: classes where the selector is available (invokable).
1.21 crook 7912:
1.26 crook 7913: This structure means that interface selectors have to perform one
7914: indirection more than class selectors to find their method. Their body
7915: contains the interface map pointer offset in the class method map, and
7916: the method offset in the interface method map. The
7917: @code{does>} action for an interface selector is, basically:
1.21 crook 7918:
7919: @example
1.26 crook 7920: ( object selector-body )
7921: 2dup selector-interface @@ ( object selector-body object interface-offset )
7922: swap object-map @@ + @@ ( object selector-body map )
7923: swap selector-offset @@ + @@ execute
1.21 crook 7924: @end example
7925:
1.26 crook 7926: where @code{object-map} and @code{selector-offset} are
7927: first fields and generate no code.
7928:
7929: As a concrete example, consider the following code:
1.21 crook 7930:
1.26 crook 7931: @example
7932: interface
7933: selector if1sel1
7934: selector if1sel2
7935: end-interface if1
1.21 crook 7936:
1.26 crook 7937: object class
7938: if1 implementation
7939: selector cl1sel1
7940: cell% inst-var cl1iv1
1.21 crook 7941:
1.26 crook 7942: ' m1 overrides construct
7943: ' m2 overrides if1sel1
7944: ' m3 overrides if1sel2
7945: ' m4 overrides cl1sel2
7946: end-class cl1
1.21 crook 7947:
1.26 crook 7948: create obj1 object dict-new drop
7949: create obj2 cl1 dict-new drop
7950: @end example
1.21 crook 7951:
1.26 crook 7952: The data structure created by this code (including the data structure
7953: for @code{object}) is shown in the <a
7954: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
1.29 ! crook 7955: @comment TODO add this diagram..
1.21 crook 7956:
1.26 crook 7957: @node Objects Glossary, , Objects Implementation, Objects
7958: @subsubsection @file{objects.fs} Glossary
7959: @cindex @file{objects.fs} Glossary
1.21 crook 7960:
1.26 crook 7961: doc---objects-bind
7962: doc---objects-<bind>
7963: doc---objects-bind'
7964: doc---objects-[bind]
7965: doc---objects-class
7966: doc---objects-class->map
7967: doc---objects-class-inst-size
7968: doc---objects-class-override!
7969: doc---objects-construct
7970: doc---objects-current'
7971: doc---objects-[current]
7972: doc---objects-current-interface
7973: doc---objects-dict-new
7974: doc---objects-drop-order
7975: doc---objects-end-class
7976: doc---objects-end-class-noname
7977: doc---objects-end-interface
7978: doc---objects-end-interface-noname
7979: doc---objects-exitm
7980: doc---objects-heap-new
7981: doc---objects-implementation
7982: doc---objects-init-object
7983: doc---objects-inst-value
7984: doc---objects-inst-var
7985: doc---objects-interface
7986: doc---objects-;m
7987: doc---objects-m:
7988: doc---objects-method
7989: doc---objects-object
7990: doc---objects-overrides
7991: doc---objects-[parent]
7992: doc---objects-print
7993: doc---objects-protected
7994: doc---objects-public
7995: doc---objects-push-order
7996: doc---objects-selector
7997: doc---objects-this
7998: doc---objects-<to-inst>
7999: doc---objects-[to-inst]
8000: doc---objects-to-this
8001: doc---objects-xt-new
1.21 crook 8002:
1.26 crook 8003: @c -------------------------------------------------------------
8004: @node OOF, Mini-OOF, Objects, Object-oriented Forth
8005: @subsection The @file{oof.fs} model
8006: @cindex oof
8007: @cindex object-oriented programming
1.21 crook 8008:
1.26 crook 8009: @cindex @file{objects.fs}
8010: @cindex @file{oof.fs}
1.21 crook 8011:
1.26 crook 8012: This section describes the @file{oof.fs} package.
1.21 crook 8013:
1.26 crook 8014: The package described in this section has been used in bigFORTH since 1991, and
8015: used for two large applications: a chromatographic system used to
8016: create new medicaments, and a graphic user interface library (MINOS).
1.21 crook 8017:
1.26 crook 8018: You can find a description (in German) of @file{oof.fs} in @cite{Object
8019: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
8020: 10(2), 1994.
1.21 crook 8021:
1.26 crook 8022: @menu
8023: * Properties of the OOF model::
8024: * Basic OOF Usage::
8025: * The OOF base class::
8026: * Class Declaration::
8027: * Class Implementation::
8028: @end menu
1.21 crook 8029:
1.26 crook 8030: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
8031: @subsubsection Properties of the @file{oof.fs} model
8032: @cindex @file{oof.fs} properties
1.21 crook 8033:
1.26 crook 8034: @itemize @bullet
8035: @item
8036: This model combines object oriented programming with information
8037: hiding. It helps you writing large application, where scoping is
8038: necessary, because it provides class-oriented scoping.
1.21 crook 8039:
1.26 crook 8040: @item
8041: Named objects, object pointers, and object arrays can be created,
8042: selector invocation uses the ``object selector'' syntax. Selector invocation
8043: to objects and/or selectors on the stack is a bit less convenient, but
8044: possible.
1.21 crook 8045:
1.26 crook 8046: @item
8047: Selector invocation and instance variable usage of the active object is
8048: straightforward, since both make use of the active object.
1.21 crook 8049:
1.26 crook 8050: @item
8051: Late binding is efficient and easy to use.
1.21 crook 8052:
1.26 crook 8053: @item
8054: State-smart objects parse selectors. However, extensibility is provided
8055: using a (parsing) selector @code{postpone} and a selector @code{'}.
1.21 crook 8056:
8057: @item
1.26 crook 8058: An implementation in ANS Forth is available.
8059:
1.21 crook 8060: @end itemize
8061:
8062:
1.26 crook 8063: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
8064: @subsubsection Basic @file{oof.fs} Usage
8065: @cindex @file{oof.fs} usage
8066:
8067: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
1.21 crook 8068:
1.26 crook 8069: You can define a class for graphical objects like this:
1.21 crook 8070:
1.26 crook 8071: @cindex @code{class} usage
8072: @cindex @code{class;} usage
8073: @cindex @code{method} usage
8074: @example
8075: object class graphical \ "object" is the parent class
8076: method draw ( x y graphical -- )
8077: class;
8078: @end example
1.21 crook 8079:
1.26 crook 8080: This code defines a class @code{graphical} with an
8081: operation @code{draw}. We can perform the operation
8082: @code{draw} on any @code{graphical} object, e.g.:
1.21 crook 8083:
1.26 crook 8084: @example
8085: 100 100 t-rex draw
8086: @end example
1.21 crook 8087:
1.26 crook 8088: @noindent
8089: where @code{t-rex} is an object or object pointer, created with e.g.
8090: @code{graphical : t-rex}.
1.21 crook 8091:
1.26 crook 8092: @cindex abstract class
8093: How do we create a graphical object? With the present definitions,
8094: we cannot create a useful graphical object. The class
8095: @code{graphical} describes graphical objects in general, but not
8096: any concrete graphical object type (C++ users would call it an
8097: @emph{abstract class}); e.g., there is no method for the selector
8098: @code{draw} in the class @code{graphical}.
1.21 crook 8099:
1.26 crook 8100: For concrete graphical objects, we define child classes of the
8101: class @code{graphical}, e.g.:
1.21 crook 8102:
8103: @example
1.26 crook 8104: graphical class circle \ "graphical" is the parent class
8105: cell var circle-radius
8106: how:
8107: : draw ( x y -- )
8108: circle-radius @@ draw-circle ;
8109:
8110: : init ( n-radius -- (
8111: circle-radius ! ;
8112: class;
8113: @end example
8114:
8115: Here we define a class @code{circle} as a child of @code{graphical},
8116: with a field @code{circle-radius}; it defines new methods for the
8117: selectors @code{draw} and @code{init} (@code{init} is defined in
8118: @code{object}, the parent class of @code{graphical}).
1.21 crook 8119:
1.26 crook 8120: Now we can create a circle in the dictionary with:
1.21 crook 8121:
1.26 crook 8122: @example
8123: 50 circle : my-circle
1.21 crook 8124: @end example
8125:
1.26 crook 8126: @noindent
8127: @code{:} invokes @code{init}, thus initializing the field
8128: @code{circle-radius} with 50. We can draw this new circle at (100,100)
8129: with:
1.21 crook 8130:
8131: @example
1.26 crook 8132: 100 100 my-circle draw
1.21 crook 8133: @end example
8134:
1.26 crook 8135: @cindex selector invocation, restrictions
8136: @cindex class definition, restrictions
8137: Note: You can only invoke a selector if the receiving object belongs to
8138: the class where the selector was defined or one of its descendents;
8139: e.g., you can invoke @code{draw} only for objects belonging to
8140: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
8141: mechanism will check if you try to invoke a selector that is not
8142: defined in this class hierarchy, so you'll get an error at compilation
8143: time.
8144:
8145:
8146: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
8147: @subsubsection The @file{oof.fs} base class
8148: @cindex @file{oof.fs} base class
8149:
8150: When you define a class, you have to specify a parent class. So how do
8151: you start defining classes? There is one class available from the start:
8152: @code{object}. You have to use it as ancestor for all classes. It is the
8153: only class that has no parent. Classes are also objects, except that
8154: they don't have instance variables; class manipulation such as
8155: inheritance or changing definitions of a class is handled through
8156: selectors of the class @code{object}.
8157:
8158: @code{object} provides a number of selectors:
8159:
1.21 crook 8160: @itemize @bullet
8161: @item
1.26 crook 8162: @code{class} for subclassing, @code{definitions} to add definitions
8163: later on, and @code{class?} to get type informations (is the class a
8164: subclass of the class passed on the stack?).
8165: doc---object-class
8166: doc---object-definitions
8167: doc---object-class?
8168:
1.21 crook 8169: @item
1.26 crook 8170: @code{init} and @code{dispose} as constructor and destructor of the
8171: object. @code{init} is invocated after the object's memory is allocated,
8172: while @code{dispose} also handles deallocation. Thus if you redefine
8173: @code{dispose}, you have to call the parent's dispose with @code{super
8174: dispose}, too.
8175: doc---object-init
8176: doc---object-dispose
8177:
1.21 crook 8178: @item
1.26 crook 8179: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
8180: @code{[]} to create named and unnamed objects and object arrays or
8181: object pointers.
8182: doc---object-new
8183: doc---object-new[]
8184: doc---object-:
8185: doc---object-ptr
8186: doc---object-asptr
8187: doc---object-[]
1.21 crook 8188:
1.26 crook 8189: @item
8190: @code{::} and @code{super} for explicit scoping. You should use explicit
8191: scoping only for super classes or classes with the same set of instance
8192: variables. Explicitly-scoped selectors use early binding.
8193: doc---object-::
8194: doc---object-super
1.21 crook 8195:
1.26 crook 8196: @item
8197: @code{self} to get the address of the object
8198: doc---object-self
1.21 crook 8199:
8200: @item
1.26 crook 8201: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
8202: pointers and instance defers.
8203: doc---object-bind
8204: doc---object-bound
8205: doc---object-link
8206: doc---object-is
8207:
1.21 crook 8208: @item
1.26 crook 8209: @code{'} to obtain selector tokens, @code{send} to invocate selectors
8210: form the stack, and @code{postpone} to generate selector invocation code.
8211: doc---object-'
8212: doc---object-postpone
8213:
1.21 crook 8214: @item
1.26 crook 8215: @code{with} and @code{endwith} to select the active object from the
8216: stack, and enable its scope. Using @code{with} and @code{endwith}
8217: also allows you to create code using selector @code{postpone} without being
8218: trapped by the state-smart objects.
8219: doc---object-with
8220: doc---object-endwith
8221:
1.21 crook 8222: @end itemize
8223:
1.26 crook 8224: @node Class Declaration, Class Implementation, The OOF base class, OOF
8225: @subsubsection Class Declaration
8226: @cindex class declaration
8227:
8228: @itemize @bullet
8229: @item
8230: Instance variables
8231: doc---oof-var
1.21 crook 8232:
1.26 crook 8233: @item
8234: Object pointers
8235: doc---oof-ptr
8236: doc---oof-asptr
1.21 crook 8237:
1.26 crook 8238: @item
8239: Instance defers
8240: doc---oof-defer
1.21 crook 8241:
1.26 crook 8242: @item
8243: Method selectors
8244: doc---oof-early
8245: doc---oof-method
1.21 crook 8246:
1.26 crook 8247: @item
8248: Class-wide variables
8249: doc---oof-static
1.21 crook 8250:
1.26 crook 8251: @item
8252: End declaration
8253: doc---oof-how:
8254: doc---oof-class;
1.21 crook 8255:
1.26 crook 8256: @end itemize
1.21 crook 8257:
1.26 crook 8258: @c -------------------------------------------------------------
8259: @node Class Implementation, , Class Declaration, OOF
8260: @subsubsection Class Implementation
8261: @cindex class implementation
1.21 crook 8262:
1.26 crook 8263: @c -------------------------------------------------------------
8264: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
8265: @subsection The @file{mini-oof.fs} model
8266: @cindex mini-oof
1.1 anton 8267:
1.26 crook 8268: Gforth's third object oriented Forth package is a 12-liner. It uses a
8269: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
8270: and reduces to the bare minimum of features. This is based on a posting
8271: of Bernd Paysan in comp.arch.
1.1 anton 8272:
8273: @menu
1.26 crook 8274: * Basic Mini-OOF Usage::
8275: * Mini-OOF Example::
8276: * Mini-OOF Implementation::
1.1 anton 8277: @end menu
8278:
1.26 crook 8279: @c -------------------------------------------------------------
8280: @node Basic Mini-OOF Usage, Mini-OOF Example, , Mini-OOF
8281: @subsubsection Basic @file{mini-oof.fs} Usage
8282: @cindex mini-oof usage
1.1 anton 8283:
1.28 crook 8284: There is a base class (@code{class}, which allocates one cell for the
8285: object pointer) plus seven other words: to define a method, a variable,
8286: a class; to end a class, to resolve binding, to allocate an object and
8287: to compile a class method.
1.26 crook 8288: @comment TODO better description of the last one
1.1 anton 8289:
1.26 crook 8290: doc-object
8291: doc-method
8292: doc-var
8293: doc-class
8294: doc-end-class
8295: doc-defines
8296: doc-new
8297: doc-::
1.1 anton 8298:
1.21 crook 8299:
1.26 crook 8300: @c -------------------------------------------------------------
8301: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
8302: @subsubsection Mini-OOF Example
8303: @cindex mini-oof example
1.21 crook 8304:
1.26 crook 8305: A short example shows how to use this package. This example, in slightly
8306: extended form, is supplied as @file{moof-exm.fs}
1.29 ! crook 8307: @comment TODO could flesh this out with some comments from the Forthwrite article
1.21 crook 8308:
1.26 crook 8309: @example
8310: object class
8311: method init
8312: method draw
8313: end-class graphical
8314: @end example
1.21 crook 8315:
1.26 crook 8316: This code defines a class @code{graphical} with an
8317: operation @code{draw}. We can perform the operation
8318: @code{draw} on any @code{graphical} object, e.g.:
1.1 anton 8319:
1.26 crook 8320: @example
8321: 100 100 t-rex draw
8322: @end example
1.1 anton 8323:
1.26 crook 8324: where @code{t-rex} is an object or object pointer, created with e.g.
8325: @code{graphical new Constant t-rex}.
1.1 anton 8326:
1.26 crook 8327: For concrete graphical objects, we define child classes of the
8328: class @code{graphical}, e.g.:
1.21 crook 8329:
8330: @example
1.26 crook 8331: graphical class
8332: cell var circle-radius
8333: end-class circle \ "graphical" is the parent class
1.21 crook 8334:
1.26 crook 8335: :noname ( x y -- )
8336: circle-radius @@ draw-circle ; circle defines draw
8337: :noname ( r -- )
8338: circle-radius ! ; circle defines init
1.21 crook 8339: @end example
8340:
1.26 crook 8341: There is no implicit init method, so we have to define one. The creation
8342: code of the object now has to call init explicitely.
1.21 crook 8343:
1.26 crook 8344: @example
8345: circle new Constant my-circle
8346: 50 my-circle init
8347: @end example
1.21 crook 8348:
1.26 crook 8349: It is also possible to add a function to create named objects with
8350: automatic call of @code{init}, given that all objects have @code{init}
8351: on the same place:
1.1 anton 8352:
8353: @example
1.26 crook 8354: : new: ( .. o "name" -- )
8355: new dup Constant init ;
8356: 80 circle new: large-circle
1.1 anton 8357: @end example
8358:
1.26 crook 8359: We can draw this new circle at (100,100) with:
1.1 anton 8360:
8361: @example
1.26 crook 8362: 100 100 my-circle draw
1.1 anton 8363: @end example
8364:
1.26 crook 8365: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
8366: @subsubsection @file{mini-oof.fs} Implementation
1.1 anton 8367:
1.26 crook 8368: Object-oriented systems with late binding typically use a
8369: ``vtable''-approach: the first variable in each object is a pointer to a
8370: table, which contains the methods as function pointers. The vtable
8371: may also contain other information.
1.1 anton 8372:
1.26 crook 8373: So first, let's declare methods:
1.1 anton 8374:
1.26 crook 8375: @example
8376: : method ( m v -- m' v ) Create over , swap cell+ swap
8377: DOES> ( ... o -- ... ) @ over @ + @ execute ;
8378: @end example
1.1 anton 8379:
1.26 crook 8380: During method declaration, the number of methods and instance
8381: variables is on the stack (in address units). @code{method} creates
8382: one method and increments the method number. To execute a method, it
8383: takes the object, fetches the vtable pointer, adds the offset, and
1.29 ! crook 8384: executes the @i{xt} stored there. Each method takes the object it is
1.26 crook 8385: invoked from as top of stack parameter. The method itself should
8386: consume that object.
1.1 anton 8387:
1.26 crook 8388: Now, we also have to declare instance variables
1.21 crook 8389:
1.26 crook 8390: @example
8391: : var ( m v size -- m v' ) Create over , +
8392: DOES> ( o -- addr ) @ + ;
8393: @end example
1.21 crook 8394:
1.26 crook 8395: As before, a word is created with the current offset. Instance
8396: variables can have different sizes (cells, floats, doubles, chars), so
8397: all we do is take the size and add it to the offset. If your machine
8398: has alignment restrictions, put the proper @code{aligned} or
8399: @code{faligned} before the variable, to adjust the variable
8400: offset. That's why it is on the top of stack.
1.2 jwilke 8401:
1.26 crook 8402: We need a starting point (the base object) and some syntactic sugar:
1.21 crook 8403:
1.26 crook 8404: @example
8405: Create object 1 cells , 2 cells ,
8406: : class ( class -- class methods vars ) dup 2@ ;
8407: @end example
1.21 crook 8408:
1.26 crook 8409: For inheritance, the vtable of the parent object has to be
8410: copied when a new, derived class is declared. This gives all the
8411: methods of the parent class, which can be overridden, though.
1.21 crook 8412:
1.2 jwilke 8413: @example
1.26 crook 8414: : end-class ( class methods vars -- )
8415: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
8416: cell+ dup cell+ r> rot @ 2 cells /string move ;
8417: @end example
8418:
8419: The first line creates the vtable, initialized with
8420: @code{noop}s. The second line is the inheritance mechanism, it
8421: copies the xts from the parent vtable.
1.2 jwilke 8422:
1.26 crook 8423: We still have no way to define new methods, let's do that now:
1.2 jwilke 8424:
1.26 crook 8425: @example
8426: : defines ( xt class -- ) ' >body @ + ! ;
1.2 jwilke 8427: @end example
8428:
1.26 crook 8429: To allocate a new object, we need a word, too:
1.2 jwilke 8430:
1.26 crook 8431: @example
8432: : new ( class -- o ) here over @ allot swap over ! ;
8433: @end example
1.2 jwilke 8434:
1.26 crook 8435: Sometimes derived classes want to access the method of the
8436: parent object. There are two ways to achieve this with Mini-OOF:
8437: first, you could use named words, and second, you could look up the
8438: vtable of the parent object.
1.2 jwilke 8439:
1.26 crook 8440: @example
8441: : :: ( class "name" -- ) ' >body @ + @ compile, ;
8442: @end example
1.2 jwilke 8443:
8444:
1.26 crook 8445: Nothing can be more confusing than a good example, so here is
8446: one. First let's declare a text object (called
8447: @code{button}), that stores text and position:
1.2 jwilke 8448:
1.26 crook 8449: @example
8450: object class
8451: cell var text
8452: cell var len
8453: cell var x
8454: cell var y
8455: method init
8456: method draw
8457: end-class button
8458: @end example
1.2 jwilke 8459:
1.26 crook 8460: @noindent
8461: Now, implement the two methods, @code{draw} and @code{init}:
1.2 jwilke 8462:
1.26 crook 8463: @example
8464: :noname ( o -- )
8465: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
8466: button defines draw
8467: :noname ( addr u o -- )
8468: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
8469: button defines init
8470: @end example
1.2 jwilke 8471:
1.26 crook 8472: @noindent
8473: To demonstrate inheritance, we define a class @code{bold-button}, with no
8474: new data and no new methods:
1.2 jwilke 8475:
1.26 crook 8476: @example
8477: button class
8478: end-class bold-button
1.1 anton 8479:
1.26 crook 8480: : bold 27 emit ." [1m" ;
8481: : normal 27 emit ." [0m" ;
8482: @end example
1.1 anton 8483:
1.26 crook 8484: @noindent
8485: The class @code{bold-button} has a different draw method to
8486: @code{button}, but the new method is defined in terms of the draw method
8487: for @code{button}:
1.1 anton 8488:
1.26 crook 8489: @example
8490: :noname bold [ button :: draw ] normal ; bold-button defines draw
8491: @end example
1.1 anton 8492:
1.26 crook 8493: @noindent
8494: Finally, create two objects and apply methods:
1.1 anton 8495:
1.26 crook 8496: @example
8497: button new Constant foo
8498: s" thin foo" foo init
8499: page
8500: foo draw
8501: bold-button new Constant bar
8502: s" fat bar" bar init
8503: 1 bar y !
8504: bar draw
8505: @end example
1.1 anton 8506:
8507:
1.26 crook 8508: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
8509: @subsubsection Comparison with other object models
8510: @cindex comparison of object models
8511: @cindex object models, comparison
1.1 anton 8512:
1.26 crook 8513: Many object-oriented Forth extensions have been proposed (@cite{A survey
8514: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
8515: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
8516: relation of the object models described here to two well-known and two
8517: closely-related (by the use of method maps) models.
1.1 anton 8518:
1.26 crook 8519: @cindex Neon model
8520: The most popular model currently seems to be the Neon model (see
8521: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
8522: 1997) by Andrew McKewan) but this model has a number of limitations
8523: @footnote{A longer version of this critique can be
8524: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
8525: Dimensions, May 1997) by Anton Ertl.}:
1.1 anton 8526:
1.26 crook 8527: @itemize @bullet
8528: @item
8529: It uses a @code{@emph{selector
8530: object}} syntax, which makes it unnatural to pass objects on the
8531: stack.
1.1 anton 8532:
1.26 crook 8533: @item
8534: It requires that the selector parses the input stream (at
8535: compile time); this leads to reduced extensibility and to bugs that are+
8536: hard to find.
1.1 anton 8537:
1.26 crook 8538: @item
8539: It allows using every selector to every object;
8540: this eliminates the need for classes, but makes it harder to create
8541: efficient implementations.
8542: @end itemize
1.1 anton 8543:
1.26 crook 8544: @cindex Pountain's object-oriented model
8545: Another well-known publication is @cite{Object-Oriented Forth} (Academic
8546: Press, London, 1987) by Dick Pountain. However, it is not really about
8547: object-oriented programming, because it hardly deals with late
8548: binding. Instead, it focuses on features like information hiding and
8549: overloading that are characteristic of modular languages like Ada (83).
1.1 anton 8550:
1.26 crook 8551: @cindex Zsoter's object-oriented model
8552: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1) 1996, pages 31-35)
8553: Andras Zsoter describes a model that makes heavy use of an active object
8554: (like @code{this} in @file{objects.fs}): The active object is not only
8555: used for accessing all fields, but also specifies the receiving object
8556: of every selector invocation; you have to change the active object
8557: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
8558: changes more or less implicitly at @code{m: ... ;m}. Such a change at
8559: the method entry point is unnecessary with the Zsoter's model, because
8560: the receiving object is the active object already. On the other hand, the explicit
8561: change is absolutely necessary in that model, because otherwise no one
8562: could ever change the active object. An ANS Forth implementation of this
8563: model is available at @url{http://www.forth.org/fig/oopf.html}.
1.1 anton 8564:
1.26 crook 8565: @cindex @file{oof.fs}, differences to other models
8566: The @file{oof.fs} model combines information hiding and overloading
8567: resolution (by keeping names in various word lists) with object-oriented
8568: programming. It sets the active object implicitly on method entry, but
8569: also allows explicit changing (with @code{>o...o>} or with
8570: @code{with...endwith}). It uses parsing and state-smart objects and
8571: classes for resolving overloading and for early binding: the object or
8572: class parses the selector and determines the method from this. If the
8573: selector is not parsed by an object or class, it performs a call to the
8574: selector for the active object (late binding), like Zsoter's model.
8575: Fields are always accessed through the active object. The big
8576: disadvantage of this model is the parsing and the state-smartness, which
8577: reduces extensibility and increases the opportunities for subtle bugs;
8578: essentially, you are only safe if you never tick or @code{postpone} an
8579: object or class (Bernd disagrees, but I (Anton) am not convinced).
1.1 anton 8580:
1.26 crook 8581: @cindex @file{mini-oof.fs}, differences to other models
8582: The @file{mini-oof.fs} model is quite similar to a very stripped-down version of
8583: the @file{objects.fs} model, but syntactically it is a mixture of the @file{objects.fs} and
8584: @file{oof.fs} models.
1.1 anton 8585:
1.26 crook 8586: @c -------------------------------------------------------------
8587: @node Passing Commands to the OS, Miscellaneous Words, Object-oriented Forth, Words
1.21 crook 8588: @section Passing Commands to the Operating System
8589: @cindex operating system - passing commands
8590: @cindex shell commands
8591:
8592: Gforth allows you to pass an arbitrary string to the host operating
8593: system shell (if such a thing exists) for execution.
8594:
8595: doc-sh
8596: doc-system
8597: doc-$?
1.23 crook 8598: doc-getenv
1.21 crook 8599:
1.26 crook 8600: @c -------------------------------------------------------------
1.21 crook 8601: @node Miscellaneous Words, , Passing Commands to the OS, Words
8602: @section Miscellaneous Words
8603: @cindex miscellaneous words
8604:
1.29 ! crook 8605: @comment TODO find homes for these
! 8606:
1.26 crook 8607: These section lists the ANS Forth words that are not documented
1.21 crook 8608: elsewhere in this manual. Ultimately, they all need proper homes.
8609:
8610: doc-ms
8611: doc-time&date
1.27 crook 8612:
1.21 crook 8613: doc-[compile]
8614:
1.26 crook 8615: The following ANS Forth words are not currently supported by Gforth
1.27 crook 8616: (@pxref{ANS conformance}):
1.21 crook 8617:
8618: @code{EDITOR}
8619: @code{EKEY}
8620: @code{EKEY>CHAR}
8621: @code{EKEY?}
8622: @code{EMIT?}
8623: @code{FORGET}
8624:
1.24 anton 8625: @c ******************************************************************
8626: @node Error messages, Tools, Words, Top
8627: @chapter Error messages
8628: @cindex error messages
8629: @cindex backtrace
8630:
8631: A typical Gforth error message looks like this:
8632:
8633: @example
8634: in file included from :-1
8635: in file included from ./yyy.fs:1
8636: ./xxx.fs:4: Invalid memory address
8637: bar
8638: ^^^
1.25 anton 8639: $400E664C @@
8640: $400E6664 foo
1.24 anton 8641: @end example
8642:
8643: The message identifying the error is @code{Invalid memory address}. The
8644: error happened when text-interpreting line 4 of the file
8645: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
8646: word on the line where the error happened, is pointed out (with
8647: @code{^^^}).
8648:
8649: The file containing the error was included in line 1 of @file{./yyy.fs},
8650: and @file{yyy.fs} was included from a non-file (in this case, by giving
8651: @file{yyy.fs} as command-line parameter to Gforth).
8652:
8653: At the end of the error message you find a return stack dump that can be
8654: interpreted as a backtrace (possibly empty). On top you find the top of
8655: the return stack when the @code{throw} happened, and at the bottom you
8656: find the return stack entry just above the return stack of the topmost
8657: text interpreter.
8658:
8659: To the right of most return stack entries you see a guess for the word
8660: that pushed that return stack entry as its return address. This gives a
8661: backtrace. In our case we see that @code{bar} called @code{foo}, and
8662: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
8663: address} exception).
8664:
8665: Note that the backtrace is not perfect: We don't know which return stack
8666: entries are return addresses (so we may get false positives); and in
8667: some cases (e.g., for @code{abort"}) we cannot determine from the return
8668: address the word that pushed the return address, so for some return
8669: addresses you see no names in the return stack dump.
1.25 anton 8670:
8671: @cindex @code{catch} and backtraces
8672: The return stack dump represents the return stack at the time when a
8673: specific @code{throw} was executed. In programs that make use of
8674: @code{catch}, it is not necessarily clear which @code{throw} should be
8675: used for the return stack dump (e.g., consider one @code{throw} that
8676: indicates an error, which is caught, and during recovery another error
8677: happens; which @code{throw} should be used for the stack dump). Gforth
8678: presents the return stack dump for the first @code{throw} after the last
8679: executed (not returned-to) @code{catch}; this works well in the usual
8680: case.
8681:
8682: @cindex @code{gforth-fast} and backtraces
8683: @cindex @code{gforth-fast}, difference from @code{gforth}
8684: @cindex backtraces with @code{gforth-fast}
8685: @cindex return stack dump with @code{gforth-fast}
8686: @code{gforth} is able to do a return stack dump for throws generated
8687: from primitives (e.g., invalid memory address, stack empty etc.);
8688: @code{gforth-fast} is only able to do a return stack dump from a
8689: directly called @code{throw} (including @code{abort} etc.). This is the
8690: only difference (apart from a speed difference of about 30%) between
8691: @code{gforth} and @code{gforth-fast}. Given an exception caused by a
8692: primitive in @code{gforth-fast}, you will typically see no return stack
8693: dump at all; however, if the exception is caught by @code{catch} (e.g.,
8694: for restoring some state), and then @code{throw}n again, the return
8695: stack dump will be for the first such @code{throw}.
1.2 jwilke 8696:
1.5 anton 8697: @c ******************************************************************
1.24 anton 8698: @node Tools, ANS conformance, Error messages, Top
1.1 anton 8699: @chapter Tools
8700:
8701: @menu
8702: * ANS Report:: Report the words used, sorted by wordset.
8703: @end menu
8704:
8705: See also @ref{Emacs and Gforth}.
8706:
8707: @node ANS Report, , Tools, Tools
8708: @section @file{ans-report.fs}: Report the words used, sorted by wordset
8709: @cindex @file{ans-report.fs}
8710: @cindex report the words used in your program
8711: @cindex words used in your program
8712:
8713: If you want to label a Forth program as ANS Forth Program, you must
8714: document which wordsets the program uses; for extension wordsets, it is
8715: helpful to list the words the program requires from these wordsets
8716: (because Forth systems are allowed to provide only some words of them).
8717:
8718: The @file{ans-report.fs} tool makes it easy for you to determine which
8719: words from which wordset and which non-ANS words your application
8720: uses. You simply have to include @file{ans-report.fs} before loading the
8721: program you want to check. After loading your program, you can get the
8722: report with @code{print-ans-report}. A typical use is to run this as
8723: batch job like this:
8724: @example
8725: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
8726: @end example
8727:
8728: The output looks like this (for @file{compat/control.fs}):
8729: @example
8730: The program uses the following words
8731: from CORE :
8732: : POSTPONE THEN ; immediate ?dup IF 0=
8733: from BLOCK-EXT :
8734: \
8735: from FILE :
8736: (
8737: @end example
8738:
8739: @subsection Caveats
8740:
8741: Note that @file{ans-report.fs} just checks which words are used, not whether
8742: they are used in an ANS Forth conforming way!
8743:
8744: Some words are defined in several wordsets in the
8745: standard. @file{ans-report.fs} reports them for only one of the
8746: wordsets, and not necessarily the one you expect. It depends on usage
8747: which wordset is the right one to specify. E.g., if you only use the
8748: compilation semantics of @code{S"}, it is a Core word; if you also use
8749: its interpretation semantics, it is a File word.
8750:
8751: @c ******************************************************************
8752: @node ANS conformance, Model, Tools, Top
8753: @chapter ANS conformance
8754: @cindex ANS conformance of Gforth
8755:
8756: To the best of our knowledge, Gforth is an
8757:
8758: ANS Forth System
8759: @itemize @bullet
8760: @item providing the Core Extensions word set
8761: @item providing the Block word set
8762: @item providing the Block Extensions word set
8763: @item providing the Double-Number word set
8764: @item providing the Double-Number Extensions word set
8765: @item providing the Exception word set
8766: @item providing the Exception Extensions word set
8767: @item providing the Facility word set
8768: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
8769: @item providing the File Access word set
8770: @item providing the File Access Extensions word set
8771: @item providing the Floating-Point word set
8772: @item providing the Floating-Point Extensions word set
8773: @item providing the Locals word set
8774: @item providing the Locals Extensions word set
8775: @item providing the Memory-Allocation word set
8776: @item providing the Memory-Allocation Extensions word set (that one's easy)
8777: @item providing the Programming-Tools word set
8778: @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
8779: @item providing the Search-Order word set
8780: @item providing the Search-Order Extensions word set
8781: @item providing the String word set
8782: @item providing the String Extensions word set (another easy one)
8783: @end itemize
8784:
8785: @cindex system documentation
8786: In addition, ANS Forth systems are required to document certain
8787: implementation choices. This chapter tries to meet these
8788: requirements. In many cases it gives a way to ask the system for the
8789: information instead of providing the information directly, in
8790: particular, if the information depends on the processor, the operating
8791: system or the installation options chosen, or if they are likely to
8792: change during the maintenance of Gforth.
8793:
8794: @comment The framework for the rest has been taken from pfe.
8795:
8796: @menu
8797: * The Core Words::
8798: * The optional Block word set::
8799: * The optional Double Number word set::
8800: * The optional Exception word set::
8801: * The optional Facility word set::
8802: * The optional File-Access word set::
8803: * The optional Floating-Point word set::
8804: * The optional Locals word set::
8805: * The optional Memory-Allocation word set::
8806: * The optional Programming-Tools word set::
8807: * The optional Search-Order word set::
8808: @end menu
8809:
8810:
8811: @c =====================================================================
8812: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
8813: @comment node-name, next, previous, up
8814: @section The Core Words
8815: @c =====================================================================
8816: @cindex core words, system documentation
8817: @cindex system documentation, core words
8818:
8819: @menu
8820: * core-idef:: Implementation Defined Options
8821: * core-ambcond:: Ambiguous Conditions
8822: * core-other:: Other System Documentation
8823: @end menu
8824:
8825: @c ---------------------------------------------------------------------
8826: @node core-idef, core-ambcond, The Core Words, The Core Words
8827: @subsection Implementation Defined Options
8828: @c ---------------------------------------------------------------------
8829: @cindex core words, implementation-defined options
8830: @cindex implementation-defined options, core words
8831:
8832:
8833: @table @i
8834: @item (Cell) aligned addresses:
8835: @cindex cell-aligned addresses
8836: @cindex aligned addresses
8837: processor-dependent. Gforth's alignment words perform natural alignment
8838: (e.g., an address aligned for a datum of size 8 is divisible by
8839: 8). Unaligned accesses usually result in a @code{-23 THROW}.
8840:
8841: @item @code{EMIT} and non-graphic characters:
8842: @cindex @code{EMIT} and non-graphic characters
8843: @cindex non-graphic characters and @code{EMIT}
8844: The character is output using the C library function (actually, macro)
8845: @code{putc}.
8846:
8847: @item character editing of @code{ACCEPT} and @code{EXPECT}:
8848: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
8849: @cindex editing in @code{ACCEPT} and @code{EXPECT}
8850: @cindex @code{ACCEPT}, editing
8851: @cindex @code{EXPECT}, editing
8852: This is modeled on the GNU readline library (@pxref{Readline
8853: Interaction, , Command Line Editing, readline, The GNU Readline
8854: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
8855: producing a full word completion every time you type it (instead of
1.28 crook 8856: producing the common prefix of all completions). @xref{Command-line editing}.
1.1 anton 8857:
8858: @item character set:
8859: @cindex character set
8860: The character set of your computer and display device. Gforth is
8861: 8-bit-clean (but some other component in your system may make trouble).
8862:
8863: @item Character-aligned address requirements:
8864: @cindex character-aligned address requirements
8865: installation-dependent. Currently a character is represented by a C
8866: @code{unsigned char}; in the future we might switch to @code{wchar_t}
8867: (Comments on that requested).
8868:
8869: @item character-set extensions and matching of names:
8870: @cindex character-set extensions and matching of names
1.26 crook 8871: @cindex case-sensitivity for name lookup
8872: @cindex name lookup, case-sensitivity
8873: @cindex locale and case-sensitivity
1.21 crook 8874: Any character except the ASCII NUL character can be used in a
1.1 anton 8875: name. Matching is case-insensitive (except in @code{TABLE}s). The
8876: matching is performed using the C function @code{strncasecmp}, whose
8877: function is probably influenced by the locale. E.g., the @code{C} locale
8878: does not know about accents and umlauts, so they are matched
8879: case-sensitively in that locale. For portability reasons it is best to
8880: write programs such that they work in the @code{C} locale. Then one can
8881: use libraries written by a Polish programmer (who might use words
8882: containing ISO Latin-2 encoded characters) and by a French programmer
8883: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
8884: funny results for some of the words (which ones, depends on the font you
8885: are using)). Also, the locale you prefer may not be available in other
8886: operating systems. Hopefully, Unicode will solve these problems one day.
8887:
8888: @item conditions under which control characters match a space delimiter:
8889: @cindex space delimiters
8890: @cindex control characters as delimiters
8891: If @code{WORD} is called with the space character as a delimiter, all
8892: white-space characters (as identified by the C macro @code{isspace()})
8893: are delimiters. @code{PARSE}, on the other hand, treats space like other
8894: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
8895: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
8896: interpreter (aka text interpreter) by default, treats all white-space
8897: characters as delimiters.
8898:
1.26 crook 8899: @item format of the control-flow stack:
8900: @cindex control-flow stack, format
8901: The data stack is used as control-flow stack. The size of a control-flow
1.1 anton 8902: stack item in cells is given by the constant @code{cs-item-size}. At the
8903: time of this writing, an item consists of a (pointer to a) locals list
8904: (third), an address in the code (second), and a tag for identifying the
8905: item (TOS). The following tags are used: @code{defstart},
8906: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
8907: @code{scopestart}.
8908:
8909: @item conversion of digits > 35
8910: @cindex digits > 35
8911: The characters @code{[\]^_'} are the digits with the decimal value
8912: 36@minus{}41. There is no way to input many of the larger digits.
8913:
8914: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
8915: @cindex @code{EXPECT}, display after end of input
8916: @cindex @code{ACCEPT}, display after end of input
8917: The cursor is moved to the end of the entered string. If the input is
8918: terminated using the @kbd{Return} key, a space is typed.
8919:
8920: @item exception abort sequence of @code{ABORT"}:
8921: @cindex exception abort sequence of @code{ABORT"}
8922: @cindex @code{ABORT"}, exception abort sequence
8923: The error string is stored into the variable @code{"error} and a
8924: @code{-2 throw} is performed.
8925:
8926: @item input line terminator:
8927: @cindex input line terminator
8928: @cindex line terminator on input
1.26 crook 8929: @cindex newline character on input
1.1 anton 8930: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
8931: lines. One of these characters is typically produced when you type the
8932: @kbd{Enter} or @kbd{Return} key.
8933:
8934: @item maximum size of a counted string:
8935: @cindex maximum size of a counted string
8936: @cindex counted string, maximum size
8937: @code{s" /counted-string" environment? drop .}. Currently 255 characters
8938: on all ports, but this may change.
8939:
8940: @item maximum size of a parsed string:
8941: @cindex maximum size of a parsed string
8942: @cindex parsed string, maximum size
8943: Given by the constant @code{/line}. Currently 255 characters.
8944:
8945: @item maximum size of a definition name, in characters:
8946: @cindex maximum size of a definition name, in characters
8947: @cindex name, maximum length
8948: 31
8949:
8950: @item maximum string length for @code{ENVIRONMENT?}, in characters:
8951: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
8952: @cindex @code{ENVIRONMENT?} string length, maximum
8953: 31
8954:
8955: @item method of selecting the user input device:
8956: @cindex user input device, method of selecting
8957: The user input device is the standard input. There is currently no way to
8958: change it from within Gforth. However, the input can typically be
8959: redirected in the command line that starts Gforth.
8960:
8961: @item method of selecting the user output device:
8962: @cindex user output device, method of selecting
8963: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
1.10 anton 8964: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
8965: output when the user output device is a terminal, otherwise the output
8966: is buffered.
1.1 anton 8967:
8968: @item methods of dictionary compilation:
8969: What are we expected to document here?
8970:
8971: @item number of bits in one address unit:
8972: @cindex number of bits in one address unit
8973: @cindex address unit, size in bits
8974: @code{s" address-units-bits" environment? drop .}. 8 in all current
8975: ports.
8976:
8977: @item number representation and arithmetic:
8978: @cindex number representation and arithmetic
8979: Processor-dependent. Binary two's complement on all current ports.
8980:
8981: @item ranges for integer types:
8982: @cindex ranges for integer types
8983: @cindex integer types, ranges
8984: Installation-dependent. Make environmental queries for @code{MAX-N},
8985: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
8986: unsigned (and positive) types is 0. The lower bound for signed types on
8987: two's complement and one's complement machines machines can be computed
8988: by adding 1 to the upper bound.
8989:
8990: @item read-only data space regions:
8991: @cindex read-only data space regions
8992: @cindex data-space, read-only regions
8993: The whole Forth data space is writable.
8994:
8995: @item size of buffer at @code{WORD}:
8996: @cindex size of buffer at @code{WORD}
8997: @cindex @code{WORD} buffer size
8998: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
8999: shared with the pictured numeric output string. If overwriting
9000: @code{PAD} is acceptable, it is as large as the remaining dictionary
9001: space, although only as much can be sensibly used as fits in a counted
9002: string.
9003:
9004: @item size of one cell in address units:
9005: @cindex cell size
9006: @code{1 cells .}.
9007:
9008: @item size of one character in address units:
9009: @cindex char size
9010: @code{1 chars .}. 1 on all current ports.
9011:
9012: @item size of the keyboard terminal buffer:
9013: @cindex size of the keyboard terminal buffer
9014: @cindex terminal buffer, size
9015: Varies. You can determine the size at a specific time using @code{lp@@
9016: tib - .}. It is shared with the locals stack and TIBs of files that
9017: include the current file. You can change the amount of space for TIBs
9018: and locals stack at Gforth startup with the command line option
9019: @code{-l}.
9020:
9021: @item size of the pictured numeric output buffer:
9022: @cindex size of the pictured numeric output buffer
9023: @cindex pictured numeric output buffer, size
9024: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9025: shared with @code{WORD}.
9026:
9027: @item size of the scratch area returned by @code{PAD}:
9028: @cindex size of the scratch area returned by @code{PAD}
9029: @cindex @code{PAD} size
9030: The remainder of dictionary space. @code{unused pad here - - .}.
9031:
9032: @item system case-sensitivity characteristics:
9033: @cindex case-sensitivity characteristics
1.26 crook 9034: Dictionary searches are case-insensitive (except in
1.1 anton 9035: @code{TABLE}s). However, as explained above under @i{character-set
9036: extensions}, the matching for non-ASCII characters is determined by the
9037: locale you are using. In the default @code{C} locale all non-ASCII
9038: characters are matched case-sensitively.
9039:
9040: @item system prompt:
9041: @cindex system prompt
9042: @cindex prompt
9043: @code{ ok} in interpret state, @code{ compiled} in compile state.
9044:
9045: @item division rounding:
9046: @cindex division rounding
9047: installation dependent. @code{s" floored" environment? drop .}. We leave
9048: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
9049: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
9050:
9051: @item values of @code{STATE} when true:
9052: @cindex @code{STATE} values
9053: -1.
9054:
9055: @item values returned after arithmetic overflow:
9056: On two's complement machines, arithmetic is performed modulo
9057: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9058: arithmetic (with appropriate mapping for signed types). Division by zero
9059: typically results in a @code{-55 throw} (Floating-point unidentified
9060: fault), although a @code{-10 throw} (divide by zero) would be more
9061: appropriate.
9062:
9063: @item whether the current definition can be found after @t{DOES>}:
9064: @cindex @t{DOES>}, visibility of current definition
9065: No.
9066:
9067: @end table
9068:
9069: @c ---------------------------------------------------------------------
9070: @node core-ambcond, core-other, core-idef, The Core Words
9071: @subsection Ambiguous conditions
9072: @c ---------------------------------------------------------------------
9073: @cindex core words, ambiguous conditions
9074: @cindex ambiguous conditions, core words
9075:
9076: @table @i
9077:
9078: @item a name is neither a word nor a number:
9079: @cindex name not found
1.26 crook 9080: @cindex undefined word
1.1 anton 9081: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
9082: preserves the data and FP stack, so you don't lose more work than
9083: necessary.
9084:
9085: @item a definition name exceeds the maximum length allowed:
1.26 crook 9086: @cindex word name too long
1.1 anton 9087: @code{-19 throw} (Word name too long)
9088:
9089: @item addressing a region not inside the various data spaces of the forth system:
9090: @cindex Invalid memory address
9091: The stacks, code space and name space are accessible. Machine code space is
9092: typically readable. Accessing other addresses gives results dependent on
9093: the operating system. On decent systems: @code{-9 throw} (Invalid memory
9094: address).
9095:
9096: @item argument type incompatible with parameter:
1.26 crook 9097: @cindex argument type mismatch
1.1 anton 9098: This is usually not caught. Some words perform checks, e.g., the control
9099: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
9100: mismatch).
9101:
9102: @item attempting to obtain the execution token of a word with undefined execution semantics:
9103: @cindex Interpreting a compile-only word, for @code{'} etc.
9104: @cindex execution token of words with undefined execution semantics
9105: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
9106: get an execution token for @code{compile-only-error} (which performs a
9107: @code{-14 throw} when executed).
9108:
9109: @item dividing by zero:
9110: @cindex dividing by zero
9111: @cindex floating point unidentified fault, integer division
1.24 anton 9112: On better platforms, this produces a @code{-10 throw} (Division by
9113: zero); on other systems, this typically results in a @code{-55 throw}
9114: (Floating-point unidentified fault).
1.1 anton 9115:
9116: @item insufficient data stack or return stack space:
9117: @cindex insufficient data stack or return stack space
9118: @cindex stack overflow
1.26 crook 9119: @cindex address alignment exception, stack overflow
1.1 anton 9120: @cindex Invalid memory address, stack overflow
9121: Depending on the operating system, the installation, and the invocation
9122: of Gforth, this is either checked by the memory management hardware, or
1.24 anton 9123: it is not checked. If it is checked, you typically get a @code{-3 throw}
9124: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
9125: throw} (Invalid memory address) (depending on the platform and how you
9126: achieved the overflow) as soon as the overflow happens. If it is not
9127: checked, overflows typically result in mysterious illegal memory
9128: accesses, producing @code{-9 throw} (Invalid memory address) or
9129: @code{-23 throw} (Address alignment exception); they might also destroy
9130: the internal data structure of @code{ALLOCATE} and friends, resulting in
9131: various errors in these words.
1.1 anton 9132:
9133: @item insufficient space for loop control parameters:
9134: @cindex insufficient space for loop control parameters
9135: like other return stack overflows.
9136:
9137: @item insufficient space in the dictionary:
9138: @cindex insufficient space in the dictionary
9139: @cindex dictionary overflow
1.12 anton 9140: If you try to allot (either directly with @code{allot}, or indirectly
9141: with @code{,}, @code{create} etc.) more memory than available in the
9142: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
9143: to access memory beyond the end of the dictionary, the results are
9144: similar to stack overflows.
1.1 anton 9145:
9146: @item interpreting a word with undefined interpretation semantics:
9147: @cindex interpreting a word with undefined interpretation semantics
9148: @cindex Interpreting a compile-only word
9149: For some words, we have defined interpretation semantics. For the
9150: others: @code{-14 throw} (Interpreting a compile-only word).
9151:
9152: @item modifying the contents of the input buffer or a string literal:
9153: @cindex modifying the contents of the input buffer or a string literal
9154: These are located in writable memory and can be modified.
9155:
9156: @item overflow of the pictured numeric output string:
9157: @cindex overflow of the pictured numeric output string
9158: @cindex pictured numeric output string, overflow
1.24 anton 9159: @code{-17 throw} (Pictured numeric ouput string overflow).
1.1 anton 9160:
9161: @item parsed string overflow:
9162: @cindex parsed string overflow
9163: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
9164:
9165: @item producing a result out of range:
9166: @cindex result out of range
9167: On two's complement machines, arithmetic is performed modulo
9168: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9169: arithmetic (with appropriate mapping for signed types). Division by zero
1.24 anton 9170: typically results in a @code{-10 throw} (divide by zero) or @code{-55
9171: throw} (floating point unidentified fault). @code{convert} and
9172: @code{>number} currently overflow silently.
1.1 anton 9173:
9174: @item reading from an empty data or return stack:
9175: @cindex stack empty
9176: @cindex stack underflow
1.24 anton 9177: @cindex return stack underflow
1.1 anton 9178: The data stack is checked by the outer (aka text) interpreter after
9179: every word executed. If it has underflowed, a @code{-4 throw} (Stack
9180: underflow) is performed. Apart from that, stacks may be checked or not,
1.24 anton 9181: depending on operating system, installation, and invocation. If they are
9182: caught by a check, they typically result in @code{-4 throw} (Stack
9183: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
9184: (Invalid memory address), depending on the platform and which stack
9185: underflows and by how much. Note that even if the system uses checking
9186: (through the MMU), your program may have to underflow by a significant
9187: number of stack items to trigger the reaction (the reason for this is
9188: that the MMU, and therefore the checking, works with a page-size
9189: granularity). If there is no checking, the symptoms resulting from an
9190: underflow are similar to those from an overflow. Unbalanced return
9191: stack errors result in a variaty of symptoms, including @code{-9 throw}
9192: (Invalid memory address) and Illegal Instruction (typically @code{-260
9193: throw}).
1.1 anton 9194:
9195: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
9196: @cindex unexpected end of the input buffer
9197: @cindex zero-length string as a name
9198: @cindex Attempt to use zero-length string as a name
9199: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
9200: use zero-length string as a name). Words like @code{'} probably will not
9201: find what they search. Note that it is possible to create zero-length
9202: names with @code{nextname} (should it not?).
9203:
9204: @item @code{>IN} greater than input buffer:
9205: @cindex @code{>IN} greater than input buffer
9206: The next invocation of a parsing word returns a string with length 0.
9207:
9208: @item @code{RECURSE} appears after @code{DOES>}:
9209: @cindex @code{RECURSE} appears after @code{DOES>}
9210: Compiles a recursive call to the defining word, not to the defined word.
9211:
9212: @item argument input source different than current input source for @code{RESTORE-INPUT}:
9213: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
1.26 crook 9214: @cindex argument type mismatch, @code{RESTORE-INPUT}
1.1 anton 9215: @cindex @code{RESTORE-INPUT}, Argument type mismatch
9216: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
9217: the end of the file was reached), its source-id may be
9218: reused. Therefore, restoring an input source specification referencing a
9219: closed file may lead to unpredictable results instead of a @code{-12
9220: THROW}.
9221:
9222: In the future, Gforth may be able to restore input source specifications
9223: from other than the current input source.
9224:
9225: @item data space containing definitions gets de-allocated:
9226: @cindex data space containing definitions gets de-allocated
9227: Deallocation with @code{allot} is not checked. This typically results in
9228: memory access faults or execution of illegal instructions.
9229:
9230: @item data space read/write with incorrect alignment:
9231: @cindex data space read/write with incorrect alignment
9232: @cindex alignment faults
1.26 crook 9233: @cindex address alignment exception
1.1 anton 9234: Processor-dependent. Typically results in a @code{-23 throw} (Address
1.12 anton 9235: alignment exception). Under Linux-Intel on a 486 or later processor with
1.1 anton 9236: alignment turned on, incorrect alignment results in a @code{-9 throw}
9237: (Invalid memory address). There are reportedly some processors with
1.12 anton 9238: alignment restrictions that do not report violations.
1.1 anton 9239:
9240: @item data space pointer not properly aligned, @code{,}, @code{C,}:
9241: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
9242: Like other alignment errors.
9243:
9244: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
9245: Like other stack underflows.
9246:
9247: @item loop control parameters not available:
9248: @cindex loop control parameters not available
9249: Not checked. The counted loop words simply assume that the top of return
9250: stack items are loop control parameters and behave accordingly.
9251:
9252: @item most recent definition does not have a name (@code{IMMEDIATE}):
9253: @cindex most recent definition does not have a name (@code{IMMEDIATE})
9254: @cindex last word was headerless
9255: @code{abort" last word was headerless"}.
9256:
9257: @item name not defined by @code{VALUE} used by @code{TO}:
9258: @cindex name not defined by @code{VALUE} used by @code{TO}
9259: @cindex @code{TO} on non-@code{VALUE}s
9260: @cindex Invalid name argument, @code{TO}
9261: @code{-32 throw} (Invalid name argument) (unless name is a local or was
9262: defined by @code{CONSTANT}; in the latter case it just changes the constant).
9263:
9264: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
9265: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
1.26 crook 9266: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
1.1 anton 9267: @code{-13 throw} (Undefined word)
9268:
9269: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
9270: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
9271: Gforth behaves as if they were of the same type. I.e., you can predict
9272: the behaviour by interpreting all parameters as, e.g., signed.
9273:
9274: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
9275: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
9276: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
9277: compilation semantics of @code{TO}.
9278:
9279: @item String longer than a counted string returned by @code{WORD}:
1.26 crook 9280: @cindex string longer than a counted string returned by @code{WORD}
1.1 anton 9281: @cindex @code{WORD}, string overflow
9282: Not checked. The string will be ok, but the count will, of course,
9283: contain only the least significant bits of the length.
9284:
9285: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
9286: @cindex @code{LSHIFT}, large shift counts
9287: @cindex @code{RSHIFT}, large shift counts
9288: Processor-dependent. Typical behaviours are returning 0 and using only
9289: the low bits of the shift count.
9290:
9291: @item word not defined via @code{CREATE}:
9292: @cindex @code{>BODY} of non-@code{CREATE}d words
9293: @code{>BODY} produces the PFA of the word no matter how it was defined.
9294:
9295: @cindex @code{DOES>} of non-@code{CREATE}d words
9296: @code{DOES>} changes the execution semantics of the last defined word no
9297: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
9298: @code{CREATE , DOES>}.
9299:
9300: @item words improperly used outside @code{<#} and @code{#>}:
9301: Not checked. As usual, you can expect memory faults.
9302:
9303: @end table
9304:
9305:
9306: @c ---------------------------------------------------------------------
9307: @node core-other, , core-ambcond, The Core Words
9308: @subsection Other system documentation
9309: @c ---------------------------------------------------------------------
9310: @cindex other system documentation, core words
9311: @cindex core words, other system documentation
9312:
9313: @table @i
9314: @item nonstandard words using @code{PAD}:
9315: @cindex @code{PAD} use by nonstandard words
9316: None.
9317:
9318: @item operator's terminal facilities available:
9319: @cindex operator's terminal facilities available
9320: After processing the command line, Gforth goes into interactive mode,
9321: and you can give commands to Gforth interactively. The actual facilities
9322: available depend on how you invoke Gforth.
9323:
9324: @item program data space available:
9325: @cindex program data space available
9326: @cindex data space available
9327: @code{UNUSED .} gives the remaining dictionary space. The total
9328: dictionary space can be specified with the @code{-m} switch
9329: (@pxref{Invoking Gforth}) when Gforth starts up.
9330:
9331: @item return stack space available:
9332: @cindex return stack space available
9333: You can compute the total return stack space in cells with
9334: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
9335: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
9336:
9337: @item stack space available:
9338: @cindex stack space available
9339: You can compute the total data stack space in cells with
9340: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
9341: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
9342:
9343: @item system dictionary space required, in address units:
9344: @cindex system dictionary space required, in address units
9345: Type @code{here forthstart - .} after startup. At the time of this
9346: writing, this gives 80080 (bytes) on a 32-bit system.
9347: @end table
9348:
9349:
9350: @c =====================================================================
9351: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
9352: @section The optional Block word set
9353: @c =====================================================================
9354: @cindex system documentation, block words
9355: @cindex block words, system documentation
9356:
9357: @menu
9358: * block-idef:: Implementation Defined Options
9359: * block-ambcond:: Ambiguous Conditions
9360: * block-other:: Other System Documentation
9361: @end menu
9362:
9363:
9364: @c ---------------------------------------------------------------------
9365: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
9366: @subsection Implementation Defined Options
9367: @c ---------------------------------------------------------------------
9368: @cindex implementation-defined options, block words
9369: @cindex block words, implementation-defined options
9370:
9371: @table @i
9372: @item the format for display by @code{LIST}:
9373: @cindex @code{LIST} display format
9374: First the screen number is displayed, then 16 lines of 64 characters,
9375: each line preceded by the line number.
9376:
9377: @item the length of a line affected by @code{\}:
9378: @cindex length of a line affected by @code{\}
9379: @cindex @code{\}, line length in blocks
9380: 64 characters.
9381: @end table
9382:
9383:
9384: @c ---------------------------------------------------------------------
9385: @node block-ambcond, block-other, block-idef, The optional Block word set
9386: @subsection Ambiguous conditions
9387: @c ---------------------------------------------------------------------
9388: @cindex block words, ambiguous conditions
9389: @cindex ambiguous conditions, block words
9390:
9391: @table @i
9392: @item correct block read was not possible:
9393: @cindex block read not possible
9394: Typically results in a @code{throw} of some OS-derived value (between
9395: -512 and -2048). If the blocks file was just not long enough, blanks are
9396: supplied for the missing portion.
9397:
9398: @item I/O exception in block transfer:
9399: @cindex I/O exception in block transfer
9400: @cindex block transfer, I/O exception
9401: Typically results in a @code{throw} of some OS-derived value (between
9402: -512 and -2048).
9403:
9404: @item invalid block number:
9405: @cindex invalid block number
9406: @cindex block number invalid
9407: @code{-35 throw} (Invalid block number)
9408:
9409: @item a program directly alters the contents of @code{BLK}:
9410: @cindex @code{BLK}, altering @code{BLK}
9411: The input stream is switched to that other block, at the same
9412: position. If the storing to @code{BLK} happens when interpreting
9413: non-block input, the system will get quite confused when the block ends.
9414:
9415: @item no current block buffer for @code{UPDATE}:
9416: @cindex @code{UPDATE}, no current block buffer
9417: @code{UPDATE} has no effect.
9418:
9419: @end table
9420:
9421: @c ---------------------------------------------------------------------
9422: @node block-other, , block-ambcond, The optional Block word set
9423: @subsection Other system documentation
9424: @c ---------------------------------------------------------------------
9425: @cindex other system documentation, block words
9426: @cindex block words, other system documentation
9427:
9428: @table @i
9429: @item any restrictions a multiprogramming system places on the use of buffer addresses:
9430: No restrictions (yet).
9431:
9432: @item the number of blocks available for source and data:
9433: depends on your disk space.
9434:
9435: @end table
9436:
9437:
9438: @c =====================================================================
9439: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
9440: @section The optional Double Number word set
9441: @c =====================================================================
9442: @cindex system documentation, double words
9443: @cindex double words, system documentation
9444:
9445: @menu
9446: * double-ambcond:: Ambiguous Conditions
9447: @end menu
9448:
9449:
9450: @c ---------------------------------------------------------------------
9451: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
9452: @subsection Ambiguous conditions
9453: @c ---------------------------------------------------------------------
9454: @cindex double words, ambiguous conditions
9455: @cindex ambiguous conditions, double words
9456:
9457: @table @i
1.29 ! crook 9458: @item @i{d} outside of range of @i{n} in @code{D>S}:
! 9459: @cindex @code{D>S}, @i{d} out of range of @i{n}
! 9460: The least significant cell of @i{d} is produced.
1.1 anton 9461:
9462: @end table
9463:
9464:
9465: @c =====================================================================
9466: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
9467: @section The optional Exception word set
9468: @c =====================================================================
9469: @cindex system documentation, exception words
9470: @cindex exception words, system documentation
9471:
9472: @menu
9473: * exception-idef:: Implementation Defined Options
9474: @end menu
9475:
9476:
9477: @c ---------------------------------------------------------------------
9478: @node exception-idef, , The optional Exception word set, The optional Exception word set
9479: @subsection Implementation Defined Options
9480: @c ---------------------------------------------------------------------
9481: @cindex implementation-defined options, exception words
9482: @cindex exception words, implementation-defined options
9483:
9484: @table @i
9485: @item @code{THROW}-codes used in the system:
9486: @cindex @code{THROW}-codes used in the system
9487: The codes -256@minus{}-511 are used for reporting signals. The mapping
1.29 ! crook 9488: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
1.1 anton 9489: codes -512@minus{}-2047 are used for OS errors (for file and memory
9490: allocation operations). The mapping from OS error numbers to throw codes
9491: is -512@minus{}@code{errno}. One side effect of this mapping is that
9492: undefined OS errors produce a message with a strange number; e.g.,
9493: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
9494: @end table
9495:
9496: @c =====================================================================
9497: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
9498: @section The optional Facility word set
9499: @c =====================================================================
9500: @cindex system documentation, facility words
9501: @cindex facility words, system documentation
9502:
9503: @menu
9504: * facility-idef:: Implementation Defined Options
9505: * facility-ambcond:: Ambiguous Conditions
9506: @end menu
9507:
9508:
9509: @c ---------------------------------------------------------------------
9510: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
9511: @subsection Implementation Defined Options
9512: @c ---------------------------------------------------------------------
9513: @cindex implementation-defined options, facility words
9514: @cindex facility words, implementation-defined options
9515:
9516: @table @i
9517: @item encoding of keyboard events (@code{EKEY}):
9518: @cindex keyboard events, encoding in @code{EKEY}
9519: @cindex @code{EKEY}, encoding of keyboard events
9520: Not yet implemented.
9521:
9522: @item duration of a system clock tick:
9523: @cindex duration of a system clock tick
9524: @cindex clock tick duration
9525: System dependent. With respect to @code{MS}, the time is specified in
9526: microseconds. How well the OS and the hardware implement this, is
9527: another question.
9528:
9529: @item repeatability to be expected from the execution of @code{MS}:
9530: @cindex repeatability to be expected from the execution of @code{MS}
9531: @cindex @code{MS}, repeatability to be expected
9532: System dependent. On Unix, a lot depends on load. If the system is
9533: lightly loaded, and the delay is short enough that Gforth does not get
9534: swapped out, the performance should be acceptable. Under MS-DOS and
9535: other single-tasking systems, it should be good.
9536:
9537: @end table
9538:
9539:
9540: @c ---------------------------------------------------------------------
9541: @node facility-ambcond, , facility-idef, The optional Facility word set
9542: @subsection Ambiguous conditions
9543: @c ---------------------------------------------------------------------
9544: @cindex facility words, ambiguous conditions
9545: @cindex ambiguous conditions, facility words
9546:
9547: @table @i
9548: @item @code{AT-XY} can't be performed on user output device:
9549: @cindex @code{AT-XY} can't be performed on user output device
9550: Largely terminal dependent. No range checks are done on the arguments.
9551: No errors are reported. You may see some garbage appearing, you may see
9552: simply nothing happen.
9553:
9554: @end table
9555:
9556:
9557: @c =====================================================================
9558: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
9559: @section The optional File-Access word set
9560: @c =====================================================================
9561: @cindex system documentation, file words
9562: @cindex file words, system documentation
9563:
9564: @menu
9565: * file-idef:: Implementation Defined Options
9566: * file-ambcond:: Ambiguous Conditions
9567: @end menu
9568:
9569: @c ---------------------------------------------------------------------
9570: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
9571: @subsection Implementation Defined Options
9572: @c ---------------------------------------------------------------------
9573: @cindex implementation-defined options, file words
9574: @cindex file words, implementation-defined options
9575:
9576: @table @i
9577: @item file access methods used:
9578: @cindex file access methods used
9579: @code{R/O}, @code{R/W} and @code{BIN} work as you would
9580: expect. @code{W/O} translates into the C file opening mode @code{w} (or
9581: @code{wb}): The file is cleared, if it exists, and created, if it does
9582: not (with both @code{open-file} and @code{create-file}). Under Unix
9583: @code{create-file} creates a file with 666 permissions modified by your
9584: umask.
9585:
9586: @item file exceptions:
9587: @cindex file exceptions
9588: The file words do not raise exceptions (except, perhaps, memory access
9589: faults when you pass illegal addresses or file-ids).
9590:
9591: @item file line terminator:
9592: @cindex file line terminator
9593: System-dependent. Gforth uses C's newline character as line
9594: terminator. What the actual character code(s) of this are is
9595: system-dependent.
9596:
9597: @item file name format:
9598: @cindex file name format
9599: System dependent. Gforth just uses the file name format of your OS.
9600:
9601: @item information returned by @code{FILE-STATUS}:
9602: @cindex @code{FILE-STATUS}, returned information
9603: @code{FILE-STATUS} returns the most powerful file access mode allowed
9604: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
9605: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
9606: along with the returned mode.
9607:
9608: @item input file state after an exception when including source:
9609: @cindex exception when including source
9610: All files that are left via the exception are closed.
9611:
1.29 ! crook 9612: @item @i{ior} values and meaning:
! 9613: @cindex @i{ior} values and meaning
! 9614: The @i{ior}s returned by the file and memory allocation words are
1.1 anton 9615: intended as throw codes. They typically are in the range
9616: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
1.29 ! crook 9617: @i{ior}s is -512@minus{}@i{errno}.
1.1 anton 9618:
9619: @item maximum depth of file input nesting:
9620: @cindex maximum depth of file input nesting
9621: @cindex file input nesting, maximum depth
9622: limited by the amount of return stack, locals/TIB stack, and the number
9623: of open files available. This should not give you troubles.
9624:
9625: @item maximum size of input line:
9626: @cindex maximum size of input line
9627: @cindex input line size, maximum
9628: @code{/line}. Currently 255.
9629:
9630: @item methods of mapping block ranges to files:
9631: @cindex mapping block ranges to files
9632: @cindex files containing blocks
9633: @cindex blocks in files
9634: By default, blocks are accessed in the file @file{blocks.fb} in the
9635: current working directory. The file can be switched with @code{USE}.
9636:
9637: @item number of string buffers provided by @code{S"}:
9638: @cindex @code{S"}, number of string buffers
9639: 1
9640:
9641: @item size of string buffer used by @code{S"}:
9642: @cindex @code{S"}, size of string buffer
9643: @code{/line}. currently 255.
9644:
9645: @end table
9646:
9647: @c ---------------------------------------------------------------------
9648: @node file-ambcond, , file-idef, The optional File-Access word set
9649: @subsection Ambiguous conditions
9650: @c ---------------------------------------------------------------------
9651: @cindex file words, ambiguous conditions
9652: @cindex ambiguous conditions, file words
9653:
9654: @table @i
9655: @item attempting to position a file outside its boundaries:
9656: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
9657: @code{REPOSITION-FILE} is performed as usual: Afterwards,
9658: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
9659:
9660: @item attempting to read from file positions not yet written:
9661: @cindex reading from file positions not yet written
9662: End-of-file, i.e., zero characters are read and no error is reported.
9663:
1.29 ! crook 9664: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
! 9665: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
1.1 anton 9666: An appropriate exception may be thrown, but a memory fault or other
9667: problem is more probable.
9668:
1.29 ! crook 9669: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
! 9670: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
! 9671: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
! 9672: The @i{ior} produced by the operation, that discovered the problem, is
1.1 anton 9673: thrown.
9674:
9675: @item named file cannot be opened (@code{INCLUDED}):
9676: @cindex @code{INCLUDED}, named file cannot be opened
1.29 ! crook 9677: The @i{ior} produced by @code{open-file} is thrown.
1.1 anton 9678:
9679: @item requesting an unmapped block number:
9680: @cindex unmapped block numbers
9681: There are no unmapped legal block numbers. On some operating systems,
9682: writing a block with a large number may overflow the file system and
9683: have an error message as consequence.
9684:
9685: @item using @code{source-id} when @code{blk} is non-zero:
9686: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
9687: @code{source-id} performs its function. Typically it will give the id of
9688: the source which loaded the block. (Better ideas?)
9689:
9690: @end table
9691:
9692:
9693: @c =====================================================================
9694: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
9695: @section The optional Floating-Point word set
9696: @c =====================================================================
9697: @cindex system documentation, floating-point words
9698: @cindex floating-point words, system documentation
9699:
9700: @menu
9701: * floating-idef:: Implementation Defined Options
9702: * floating-ambcond:: Ambiguous Conditions
9703: @end menu
9704:
9705:
9706: @c ---------------------------------------------------------------------
9707: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
9708: @subsection Implementation Defined Options
9709: @c ---------------------------------------------------------------------
9710: @cindex implementation-defined options, floating-point words
9711: @cindex floating-point words, implementation-defined options
9712:
9713: @table @i
9714: @item format and range of floating point numbers:
9715: @cindex format and range of floating point numbers
9716: @cindex floating point numbers, format and range
9717: System-dependent; the @code{double} type of C.
9718:
1.29 ! crook 9719: @item results of @code{REPRESENT} when @i{float} is out of range:
! 9720: @cindex @code{REPRESENT}, results when @i{float} is out of range
1.1 anton 9721: System dependent; @code{REPRESENT} is implemented using the C library
9722: function @code{ecvt()} and inherits its behaviour in this respect.
9723:
9724: @item rounding or truncation of floating-point numbers:
9725: @cindex rounding of floating-point numbers
9726: @cindex truncation of floating-point numbers
9727: @cindex floating-point numbers, rounding or truncation
9728: System dependent; the rounding behaviour is inherited from the hosting C
9729: compiler. IEEE-FP-based (i.e., most) systems by default round to
9730: nearest, and break ties by rounding to even (i.e., such that the last
9731: bit of the mantissa is 0).
9732:
9733: @item size of floating-point stack:
9734: @cindex floating-point stack size
9735: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
9736: the floating-point stack (in floats). You can specify this on startup
9737: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
9738:
9739: @item width of floating-point stack:
9740: @cindex floating-point stack width
9741: @code{1 floats}.
9742:
9743: @end table
9744:
9745:
9746: @c ---------------------------------------------------------------------
9747: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
9748: @subsection Ambiguous conditions
9749: @c ---------------------------------------------------------------------
9750: @cindex floating-point words, ambiguous conditions
9751: @cindex ambiguous conditions, floating-point words
9752:
9753: @table @i
9754: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
9755: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
9756: System-dependent. Typically results in a @code{-23 THROW} like other
9757: alignment violations.
9758:
9759: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
9760: @cindex @code{f@@} used with an address that is not float aligned
9761: @cindex @code{f!} used with an address that is not float aligned
9762: System-dependent. Typically results in a @code{-23 THROW} like other
9763: alignment violations.
9764:
9765: @item floating-point result out of range:
9766: @cindex floating-point result out of range
9767: System-dependent. Can result in a @code{-55 THROW} (Floating-point
9768: unidentified fault), or can produce a special value representing, e.g.,
9769: Infinity.
9770:
9771: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
9772: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
9773: System-dependent. Typically results in an alignment fault like other
9774: alignment violations.
9775:
9776: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
9777: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
9778: The floating-point number is converted into decimal nonetheless.
9779:
9780: @item Both arguments are equal to zero (@code{FATAN2}):
9781: @cindex @code{FATAN2}, both arguments are equal to zero
9782: System-dependent. @code{FATAN2} is implemented using the C library
9783: function @code{atan2()}.
9784:
1.29 ! crook 9785: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
! 9786: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
! 9787: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
1.1 anton 9788: because of small errors and the tan will be a very large (or very small)
9789: but finite number.
9790:
1.29 ! crook 9791: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
! 9792: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
1.1 anton 9793: The result is rounded to the nearest float.
9794:
9795: @item dividing by zero:
9796: @cindex dividing by zero, floating-point
9797: @cindex floating-point dividing by zero
9798: @cindex floating-point unidentified fault, FP divide-by-zero
9799: @code{-55 throw} (Floating-point unidentified fault)
9800:
9801: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
9802: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
9803: System dependent. On IEEE-FP based systems the number is converted into
9804: an infinity.
9805:
1.29 ! crook 9806: @item @i{float}<1 (@code{FACOSH}):
! 9807: @cindex @code{FACOSH}, @i{float}<1
1.1 anton 9808: @cindex floating-point unidentified fault, @code{FACOSH}
9809: @code{-55 throw} (Floating-point unidentified fault)
9810:
1.29 ! crook 9811: @item @i{float}=<-1 (@code{FLNP1}):
! 9812: @cindex @code{FLNP1}, @i{float}=<-1
1.1 anton 9813: @cindex floating-point unidentified fault, @code{FLNP1}
9814: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
1.29 ! crook 9815: negative infinity is typically produced for @i{float}=-1.
1.1 anton 9816:
1.29 ! crook 9817: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
! 9818: @cindex @code{FLN}, @i{float}=<0
! 9819: @cindex @code{FLOG}, @i{float}=<0
1.1 anton 9820: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
9821: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
1.29 ! crook 9822: negative infinity is typically produced for @i{float}=0.
1.1 anton 9823:
1.29 ! crook 9824: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
! 9825: @cindex @code{FASINH}, @i{float}<0
! 9826: @cindex @code{FSQRT}, @i{float}<0
1.1 anton 9827: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
9828: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
9829: produces values for these inputs on my Linux box (Bug in the C library?)
9830:
1.29 ! crook 9831: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
! 9832: @cindex @code{FACOS}, |@i{float}|>1
! 9833: @cindex @code{FASIN}, |@i{float}|>1
! 9834: @cindex @code{FATANH}, |@i{float}|>1
1.1 anton 9835: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
9836: @code{-55 throw} (Floating-point unidentified fault).
9837:
1.29 ! crook 9838: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
! 9839: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
1.1 anton 9840: @cindex floating-point unidentified fault, @code{F>D}
9841: @code{-55 throw} (Floating-point unidentified fault).
9842:
9843: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
9844: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
9845: This does not happen.
9846: @end table
9847:
9848: @c =====================================================================
9849: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
9850: @section The optional Locals word set
9851: @c =====================================================================
9852: @cindex system documentation, locals words
9853: @cindex locals words, system documentation
9854:
9855: @menu
9856: * locals-idef:: Implementation Defined Options
9857: * locals-ambcond:: Ambiguous Conditions
9858: @end menu
9859:
9860:
9861: @c ---------------------------------------------------------------------
9862: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
9863: @subsection Implementation Defined Options
9864: @c ---------------------------------------------------------------------
9865: @cindex implementation-defined options, locals words
9866: @cindex locals words, implementation-defined options
9867:
9868: @table @i
9869: @item maximum number of locals in a definition:
9870: @cindex maximum number of locals in a definition
9871: @cindex locals, maximum number in a definition
9872: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
9873: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
9874: characters. The number of locals in a definition is bounded by the size
9875: of locals-buffer, which contains the names of the locals.
9876:
9877: @end table
9878:
9879:
9880: @c ---------------------------------------------------------------------
9881: @node locals-ambcond, , locals-idef, The optional Locals word set
9882: @subsection Ambiguous conditions
9883: @c ---------------------------------------------------------------------
9884: @cindex locals words, ambiguous conditions
9885: @cindex ambiguous conditions, locals words
9886:
9887: @table @i
9888: @item executing a named local in interpretation state:
9889: @cindex local in interpretation state
9890: @cindex Interpreting a compile-only word, for a local
9891: Locals have no interpretation semantics. If you try to perform the
9892: interpretation semantics, you will get a @code{-14 throw} somewhere
9893: (Interpreting a compile-only word). If you perform the compilation
9894: semantics, the locals access will be compiled (irrespective of state).
9895:
1.29 ! crook 9896: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
1.1 anton 9897: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
9898: @cindex @code{TO} on non-@code{VALUE}s and non-locals
9899: @cindex Invalid name argument, @code{TO}
9900: @code{-32 throw} (Invalid name argument)
9901:
9902: @end table
9903:
9904:
9905: @c =====================================================================
9906: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
9907: @section The optional Memory-Allocation word set
9908: @c =====================================================================
9909: @cindex system documentation, memory-allocation words
9910: @cindex memory-allocation words, system documentation
9911:
9912: @menu
9913: * memory-idef:: Implementation Defined Options
9914: @end menu
9915:
9916:
9917: @c ---------------------------------------------------------------------
9918: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
9919: @subsection Implementation Defined Options
9920: @c ---------------------------------------------------------------------
9921: @cindex implementation-defined options, memory-allocation words
9922: @cindex memory-allocation words, implementation-defined options
9923:
9924: @table @i
1.29 ! crook 9925: @item values and meaning of @i{ior}:
! 9926: @cindex @i{ior} values and meaning
! 9927: The @i{ior}s returned by the file and memory allocation words are
1.1 anton 9928: intended as throw codes. They typically are in the range
9929: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
1.29 ! crook 9930: @i{ior}s is -512@minus{}@i{errno}.
1.1 anton 9931:
9932: @end table
9933:
9934: @c =====================================================================
9935: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
9936: @section The optional Programming-Tools word set
9937: @c =====================================================================
9938: @cindex system documentation, programming-tools words
9939: @cindex programming-tools words, system documentation
9940:
9941: @menu
9942: * programming-idef:: Implementation Defined Options
9943: * programming-ambcond:: Ambiguous Conditions
9944: @end menu
9945:
9946:
9947: @c ---------------------------------------------------------------------
9948: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
9949: @subsection Implementation Defined Options
9950: @c ---------------------------------------------------------------------
9951: @cindex implementation-defined options, programming-tools words
9952: @cindex programming-tools words, implementation-defined options
9953:
9954: @table @i
9955: @item ending sequence for input following @code{;CODE} and @code{CODE}:
9956: @cindex @code{;CODE} ending sequence
9957: @cindex @code{CODE} ending sequence
9958: @code{END-CODE}
9959:
9960: @item manner of processing input following @code{;CODE} and @code{CODE}:
9961: @cindex @code{;CODE}, processing input
9962: @cindex @code{CODE}, processing input
9963: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
9964: the input is processed by the text interpreter, (starting) in interpret
9965: state.
9966:
9967: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
9968: @cindex @code{ASSEMBLER}, search order capability
9969: The ANS Forth search order word set.
9970:
9971: @item source and format of display by @code{SEE}:
9972: @cindex @code{SEE}, source and format of output
9973: The source for @code{see} is the intermediate code used by the inner
9974: interpreter. The current @code{see} tries to output Forth source code
9975: as well as possible.
9976:
9977: @end table
9978:
9979: @c ---------------------------------------------------------------------
9980: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
9981: @subsection Ambiguous conditions
9982: @c ---------------------------------------------------------------------
9983: @cindex programming-tools words, ambiguous conditions
9984: @cindex ambiguous conditions, programming-tools words
9985:
9986: @table @i
9987:
1.21 crook 9988: @item deleting the compilation word list (@code{FORGET}):
9989: @cindex @code{FORGET}, deleting the compilation word list
1.1 anton 9990: Not implemented (yet).
9991:
1.29 ! crook 9992: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
! 9993: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
! 9994: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
1.1 anton 9995: @cindex control-flow stack underflow
9996: This typically results in an @code{abort"} with a descriptive error
9997: message (may change into a @code{-22 throw} (Control structure mismatch)
9998: in the future). You may also get a memory access error. If you are
9999: unlucky, this ambiguous condition is not caught.
10000:
1.29 ! crook 10001: @item @i{name} can't be found (@code{FORGET}):
! 10002: @cindex @code{FORGET}, @i{name} can't be found
1.1 anton 10003: Not implemented (yet).
10004:
1.29 ! crook 10005: @item @i{name} not defined via @code{CREATE}:
! 10006: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
1.1 anton 10007: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
10008: the execution semantics of the last defined word no matter how it was
10009: defined.
10010:
10011: @item @code{POSTPONE} applied to @code{[IF]}:
10012: @cindex @code{POSTPONE} applied to @code{[IF]}
10013: @cindex @code{[IF]} and @code{POSTPONE}
10014: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
10015: equivalent to @code{[IF]}.
10016:
10017: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
10018: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
10019: Continue in the same state of conditional compilation in the next outer
10020: input source. Currently there is no warning to the user about this.
10021:
10022: @item removing a needed definition (@code{FORGET}):
10023: @cindex @code{FORGET}, removing a needed definition
10024: Not implemented (yet).
10025:
10026: @end table
10027:
10028:
10029: @c =====================================================================
10030: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
10031: @section The optional Search-Order word set
10032: @c =====================================================================
10033: @cindex system documentation, search-order words
10034: @cindex search-order words, system documentation
10035:
10036: @menu
10037: * search-idef:: Implementation Defined Options
10038: * search-ambcond:: Ambiguous Conditions
10039: @end menu
10040:
10041:
10042: @c ---------------------------------------------------------------------
10043: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
10044: @subsection Implementation Defined Options
10045: @c ---------------------------------------------------------------------
10046: @cindex implementation-defined options, search-order words
10047: @cindex search-order words, implementation-defined options
10048:
10049: @table @i
10050: @item maximum number of word lists in search order:
10051: @cindex maximum number of word lists in search order
10052: @cindex search order, maximum depth
10053: @code{s" wordlists" environment? drop .}. Currently 16.
10054:
10055: @item minimum search order:
10056: @cindex minimum search order
10057: @cindex search order, minimum
10058: @code{root root}.
10059:
10060: @end table
10061:
10062: @c ---------------------------------------------------------------------
10063: @node search-ambcond, , search-idef, The optional Search-Order word set
10064: @subsection Ambiguous conditions
10065: @c ---------------------------------------------------------------------
10066: @cindex search-order words, ambiguous conditions
10067: @cindex ambiguous conditions, search-order words
10068:
10069: @table @i
1.21 crook 10070: @item changing the compilation word list (during compilation):
10071: @cindex changing the compilation word list (during compilation)
10072: @cindex compilation word list, change before definition ends
10073: The word is entered into the word list that was the compilation word list
1.1 anton 10074: at the start of the definition. Any changes to the name field (e.g.,
10075: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
10076: are applied to the latest defined word (as reported by @code{last} or
1.21 crook 10077: @code{lastxt}), if possible, irrespective of the compilation word list.
1.1 anton 10078:
10079: @item search order empty (@code{previous}):
10080: @cindex @code{previous}, search order empty
1.26 crook 10081: @cindex vocstack empty, @code{previous}
1.1 anton 10082: @code{abort" Vocstack empty"}.
10083:
10084: @item too many word lists in search order (@code{also}):
10085: @cindex @code{also}, too many word lists in search order
1.26 crook 10086: @cindex vocstack full, @code{also}
1.1 anton 10087: @code{abort" Vocstack full"}.
10088:
10089: @end table
10090:
10091: @c ***************************************************************
10092: @node Model, Integrating Gforth, ANS conformance, Top
10093: @chapter Model
10094:
10095: This chapter has yet to be written. It will contain information, on
10096: which internal structures you can rely.
10097:
10098: @c ***************************************************************
10099: @node Integrating Gforth, Emacs and Gforth, Model, Top
10100: @chapter Integrating Gforth into C programs
10101:
10102: This is not yet implemented.
10103:
10104: Several people like to use Forth as scripting language for applications
10105: that are otherwise written in C, C++, or some other language.
10106:
10107: The Forth system ATLAST provides facilities for embedding it into
10108: applications; unfortunately it has several disadvantages: most
10109: importantly, it is not based on ANS Forth, and it is apparently dead
10110: (i.e., not developed further and not supported). The facilities
1.21 crook 10111: provided by Gforth in this area are inspired by ATLAST's facilities, so
1.1 anton 10112: making the switch should not be hard.
10113:
10114: We also tried to design the interface such that it can easily be
10115: implemented by other Forth systems, so that we may one day arrive at a
10116: standardized interface. Such a standard interface would allow you to
10117: replace the Forth system without having to rewrite C code.
10118:
10119: You embed the Gforth interpreter by linking with the library
10120: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
10121: global symbols in this library that belong to the interface, have the
10122: prefix @code{forth_}. (Global symbols that are used internally have the
10123: prefix @code{gforth_}).
10124:
10125: You can include the declarations of Forth types and the functions and
10126: variables of the interface with @code{#include <forth.h>}.
10127:
10128: Types.
10129:
10130: Variables.
10131:
10132: Data and FP Stack pointer. Area sizes.
10133:
10134: functions.
10135:
10136: forth_init(imagefile)
10137: forth_evaluate(string) exceptions?
10138: forth_goto(address) (or forth_execute(xt)?)
10139: forth_continue() (a corountining mechanism)
10140:
10141: Adding primitives.
10142:
10143: No checking.
10144:
10145: Signals?
10146:
10147: Accessing the Stacks
10148:
1.26 crook 10149: @c ******************************************************************
1.1 anton 10150: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
10151: @chapter Emacs and Gforth
10152: @cindex Emacs and Gforth
10153:
10154: @cindex @file{gforth.el}
10155: @cindex @file{forth.el}
10156: @cindex Rydqvist, Goran
10157: @cindex comment editing commands
10158: @cindex @code{\}, editing with Emacs
10159: @cindex debug tracer editing commands
10160: @cindex @code{~~}, removal with Emacs
10161: @cindex Forth mode in Emacs
10162: Gforth comes with @file{gforth.el}, an improved version of
10163: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
1.26 crook 10164: improvements are:
10165:
10166: @itemize @bullet
10167: @item
10168: A better (but still not perfect) handling of indentation.
10169: @item
10170: Comment paragraph filling (@kbd{M-q})
10171: @item
10172: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
10173: @item
10174: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
10175: @end itemize
10176:
10177: I left the stuff I do not use alone, even though some of it only makes
10178: sense for TILE. To get a description of these features, enter Forth mode
10179: and type @kbd{C-h m}.
1.1 anton 10180:
10181: @cindex source location of error or debugging output in Emacs
10182: @cindex error output, finding the source location in Emacs
10183: @cindex debugging output, finding the source location in Emacs
10184: In addition, Gforth supports Emacs quite well: The source code locations
10185: given in error messages, debugging output (from @code{~~}) and failed
10186: assertion messages are in the right format for Emacs' compilation mode
10187: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
10188: Manual}) so the source location corresponding to an error or other
10189: message is only a few keystrokes away (@kbd{C-x `} for the next error,
10190: @kbd{C-c C-c} for the error under the cursor).
10191:
10192: @cindex @file{TAGS} file
10193: @cindex @file{etags.fs}
10194: @cindex viewing the source of a word in Emacs
1.26 crook 10195: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file will
10196: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
1.1 anton 10197: contains the definitions of all words defined afterwards. You can then
10198: find the source for a word using @kbd{M-.}. Note that emacs can use
10199: several tags files at the same time (e.g., one for the Gforth sources
10200: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
10201: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
10202: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
10203: @file{/usr/local/share/gforth/0.2.0/TAGS}).
10204:
10205: @cindex @file{.emacs}
10206: To get all these benefits, add the following lines to your @file{.emacs}
10207: file:
10208:
10209: @example
10210: (autoload 'forth-mode "gforth.el")
10211: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
10212: @end example
10213:
1.26 crook 10214: @c ******************************************************************
1.1 anton 10215: @node Image Files, Engine, Emacs and Gforth, Top
10216: @chapter Image Files
1.26 crook 10217: @cindex image file
10218: @cindex @file{.fi} files
1.1 anton 10219: @cindex precompiled Forth code
10220: @cindex dictionary in persistent form
10221: @cindex persistent form of dictionary
10222:
10223: An image file is a file containing an image of the Forth dictionary,
10224: i.e., compiled Forth code and data residing in the dictionary. By
10225: convention, we use the extension @code{.fi} for image files.
10226:
10227: @menu
1.18 anton 10228: * Image Licensing Issues:: Distribution terms for images.
10229: * Image File Background:: Why have image files?
1.29 ! crook 10230: * Non-Relocatable Image Files:: don't always work.
1.18 anton 10231: * Data-Relocatable Image Files:: are better.
1.29 ! crook 10232: * Fully Relocatable Image Files:: better yet.
1.18 anton 10233: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
1.29 ! crook 10234: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
1.18 anton 10235: * Modifying the Startup Sequence:: and turnkey applications.
1.1 anton 10236: @end menu
10237:
1.18 anton 10238: @node Image Licensing Issues, Image File Background, Image Files, Image Files
10239: @section Image Licensing Issues
10240: @cindex license for images
10241: @cindex image license
10242:
10243: An image created with @code{gforthmi} (@pxref{gforthmi}) or
10244: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
10245: original image; i.e., according to copyright law it is a derived work of
10246: the original image.
10247:
10248: Since Gforth is distributed under the GNU GPL, the newly created image
10249: falls under the GNU GPL, too. In particular, this means that if you
10250: distribute the image, you have to make all of the sources for the image
10251: available, including those you wrote. For details see @ref{License, ,
10252: GNU General Public License (Section 3)}.
10253:
10254: If you create an image with @code{cross} (@pxref{cross.fs}), the image
10255: contains only code compiled from the sources you gave it; if none of
10256: these sources is under the GPL, the terms discussed above do not apply
10257: to the image. However, if your image needs an engine (a gforth binary)
10258: that is under the GPL, you should make sure that you distribute both in
10259: a way that is at most a @emph{mere aggregation}, if you don't want the
10260: terms of the GPL to apply to the image.
10261:
10262: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
1.1 anton 10263: @section Image File Background
10264: @cindex image file background
10265:
10266: Our Forth system consists not only of primitives, but also of
10267: definitions written in Forth. Since the Forth compiler itself belongs to
10268: those definitions, it is not possible to start the system with the
10269: primitives and the Forth source alone. Therefore we provide the Forth
1.26 crook 10270: code as an image file in nearly executable form. When Gforth starts up,
10271: a C routine loads the image file into memory, optionally relocates the
10272: addresses, then sets up the memory (stacks etc.) according to
10273: information in the image file, and (finally) starts executing Forth
10274: code.
1.1 anton 10275:
10276: The image file variants represent different compromises between the
10277: goals of making it easy to generate image files and making them
10278: portable.
10279:
10280: @cindex relocation at run-time
1.26 crook 10281: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
1.1 anton 10282: run-time. This avoids many of the complications discussed below (image
10283: files are data relocatable without further ado), but costs performance
10284: (one addition per memory access).
10285:
10286: @cindex relocation at load-time
1.26 crook 10287: By contrast, the Gforth loader performs relocation at image load time. The
10288: loader also has to replace tokens that represent primitive calls with the
1.1 anton 10289: appropriate code-field addresses (or code addresses in the case of
10290: direct threading).
10291:
10292: There are three kinds of image files, with different degrees of
10293: relocatability: non-relocatable, data-relocatable, and fully relocatable
10294: image files.
10295:
10296: @cindex image file loader
10297: @cindex relocating loader
10298: @cindex loader for image files
10299: These image file variants have several restrictions in common; they are
10300: caused by the design of the image file loader:
10301:
10302: @itemize @bullet
10303: @item
10304: There is only one segment; in particular, this means, that an image file
10305: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
1.26 crook 10306: them). The contents of the stacks are not represented, either.
1.1 anton 10307:
10308: @item
10309: The only kinds of relocation supported are: adding the same offset to
10310: all cells that represent data addresses; and replacing special tokens
10311: with code addresses or with pieces of machine code.
10312:
10313: If any complex computations involving addresses are performed, the
10314: results cannot be represented in the image file. Several applications that
10315: use such computations come to mind:
10316: @itemize @minus
10317: @item
10318: Hashing addresses (or data structures which contain addresses) for table
10319: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
10320: purpose, you will have no problem, because the hash tables are
10321: recomputed automatically when the system is started. If you use your own
10322: hash tables, you will have to do something similar.
10323:
10324: @item
10325: There's a cute implementation of doubly-linked lists that uses
10326: @code{XOR}ed addresses. You could represent such lists as singly-linked
10327: in the image file, and restore the doubly-linked representation on
10328: startup.@footnote{In my opinion, though, you should think thrice before
10329: using a doubly-linked list (whatever implementation).}
10330:
10331: @item
10332: The code addresses of run-time routines like @code{docol:} cannot be
10333: represented in the image file (because their tokens would be replaced by
10334: machine code in direct threaded implementations). As a workaround,
10335: compute these addresses at run-time with @code{>code-address} from the
10336: executions tokens of appropriate words (see the definitions of
10337: @code{docol:} and friends in @file{kernel.fs}).
10338:
10339: @item
10340: On many architectures addresses are represented in machine code in some
10341: shifted or mangled form. You cannot put @code{CODE} words that contain
10342: absolute addresses in this form in a relocatable image file. Workarounds
10343: are representing the address in some relative form (e.g., relative to
10344: the CFA, which is present in some register), or loading the address from
10345: a place where it is stored in a non-mangled form.
10346: @end itemize
10347: @end itemize
10348:
10349: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
10350: @section Non-Relocatable Image Files
10351: @cindex non-relocatable image files
1.26 crook 10352: @cindex image file, non-relocatable
1.1 anton 10353:
10354: These files are simple memory dumps of the dictionary. They are specific
10355: to the executable (i.e., @file{gforth} file) they were created
10356: with. What's worse, they are specific to the place on which the
10357: dictionary resided when the image was created. Now, there is no
10358: guarantee that the dictionary will reside at the same place the next
10359: time you start Gforth, so there's no guarantee that a non-relocatable
10360: image will work the next time (Gforth will complain instead of crashing,
10361: though).
10362:
10363: You can create a non-relocatable image file with
10364:
10365: doc-savesystem
10366:
10367: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
10368: @section Data-Relocatable Image Files
10369: @cindex data-relocatable image files
1.26 crook 10370: @cindex image file, data-relocatable
1.1 anton 10371:
10372: These files contain relocatable data addresses, but fixed code addresses
10373: (instead of tokens). They are specific to the executable (i.e.,
10374: @file{gforth} file) they were created with. For direct threading on some
10375: architectures (e.g., the i386), data-relocatable images do not work. You
10376: get a data-relocatable image, if you use @file{gforthmi} with a
10377: Gforth binary that is not doubly indirect threaded (@pxref{Fully
10378: Relocatable Image Files}).
10379:
10380: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
10381: @section Fully Relocatable Image Files
10382: @cindex fully relocatable image files
1.26 crook 10383: @cindex image file, fully relocatable
1.1 anton 10384:
10385: @cindex @file{kern*.fi}, relocatability
10386: @cindex @file{gforth.fi}, relocatability
10387: These image files have relocatable data addresses, and tokens for code
10388: addresses. They can be used with different binaries (e.g., with and
10389: without debugging) on the same machine, and even across machines with
10390: the same data formats (byte order, cell size, floating point
10391: format). However, they are usually specific to the version of Gforth
10392: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
10393: are fully relocatable.
10394:
10395: There are two ways to create a fully relocatable image file:
10396:
10397: @menu
1.29 ! crook 10398: * gforthmi:: The normal way
1.1 anton 10399: * cross.fs:: The hard way
10400: @end menu
10401:
10402: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
10403: @subsection @file{gforthmi}
10404: @cindex @file{comp-i.fs}
10405: @cindex @file{gforthmi}
10406:
10407: You will usually use @file{gforthmi}. If you want to create an
1.29 ! crook 10408: image @i{file} that contains everything you would load by invoking
! 10409: Gforth with @code{gforth @i{options}}, you simply say:
1.1 anton 10410: @example
1.29 ! crook 10411: gforthmi @i{file} @i{options}
1.1 anton 10412: @end example
10413:
10414: E.g., if you want to create an image @file{asm.fi} that has the file
10415: @file{asm.fs} loaded in addition to the usual stuff, you could do it
10416: like this:
10417:
10418: @example
10419: gforthmi asm.fi asm.fs
10420: @end example
10421:
1.27 crook 10422: @file{gforthmi} is implemented as a sh script and works like this: It
10423: produces two non-relocatable images for different addresses and then
10424: compares them. Its output reflects this: first you see the output (if
10425: any) of the two Gforth invocations that produce the nonrelocatable image
10426: files, then you see the output of the comparing program: It displays the
10427: offset used for data addresses and the offset used for code addresses;
1.1 anton 10428: moreover, for each cell that cannot be represented correctly in the
10429: image files, it displays a line like the following one:
10430:
10431: @example
10432: 78DC BFFFFA50 BFFFFA40
10433: @end example
10434:
10435: This means that at offset $78dc from @code{forthstart}, one input image
10436: contains $bffffa50, and the other contains $bffffa40. Since these cells
10437: cannot be represented correctly in the output image, you should examine
10438: these places in the dictionary and verify that these cells are dead
10439: (i.e., not read before they are written).
10440:
1.27 crook 10441: If you type @file{gforthmi} with no arguments, it prints some usage
10442: instructions.
10443:
1.1 anton 10444: @cindex @code{savesystem} during @file{gforthmi}
10445: @cindex @code{bye} during @file{gforthmi}
10446: @cindex doubly indirect threaded code
10447: @cindex environment variable @code{GFORTHD}
10448: @cindex @code{GFORTHD} environment variable
10449: @cindex @code{gforth-ditc}
1.29 ! crook 10450: There are a few wrinkles: After processing the passed @i{options}, the
1.1 anton 10451: words @code{savesystem} and @code{bye} must be visible. A special doubly
10452: indirect threaded version of the @file{gforth} executable is used for
10453: creating the nonrelocatable images; you can pass the exact filename of
10454: this executable through the environment variable @code{GFORTHD}
10455: (default: @file{gforth-ditc}); if you pass a version that is not doubly
10456: indirect threaded, you will not get a fully relocatable image, but a
1.27 crook 10457: data-relocatable image (because there is no code address offset). The
10458: normal @file{gforth} executable is used for creating the relocatable
10459: image; you can pass the exact filename of this executable through the
10460: environment variable @code{GFORTH}.
1.1 anton 10461:
10462: @node cross.fs, , gforthmi, Fully Relocatable Image Files
10463: @subsection @file{cross.fs}
10464: @cindex @file{cross.fs}
10465: @cindex cross-compiler
10466: @cindex metacompiler
10467:
10468: You can also use @code{cross}, a batch compiler that accepts a Forth-like
10469: programming language. This @code{cross} language has to be documented
10470: yet.
10471:
10472: @cindex target compiler
10473: @code{cross} also allows you to create image files for machines with
10474: different data sizes and data formats than the one used for generating
10475: the image file. You can also use it to create an application image that
10476: does not contain a Forth compiler. These features are bought with
10477: restrictions and inconveniences in programming. E.g., addresses have to
10478: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
10479: order to make the code relocatable.
10480:
10481:
10482: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
10483: @section Stack and Dictionary Sizes
10484: @cindex image file, stack and dictionary sizes
10485: @cindex dictionary size default
10486: @cindex stack size default
10487:
10488: If you invoke Gforth with a command line flag for the size
10489: (@pxref{Invoking Gforth}), the size you specify is stored in the
10490: dictionary. If you save the dictionary with @code{savesystem} or create
10491: an image with @file{gforthmi}, this size will become the default
10492: for the resulting image file. E.g., the following will create a
1.21 crook 10493: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
1.1 anton 10494:
10495: @example
10496: gforthmi gforth.fi -m 1M
10497: @end example
10498:
10499: In other words, if you want to set the default size for the dictionary
10500: and the stacks of an image, just invoke @file{gforthmi} with the
10501: appropriate options when creating the image.
10502:
10503: @cindex stack size, cache-friendly
10504: Note: For cache-friendly behaviour (i.e., good performance), you should
10505: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
10506: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
10507: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
10508:
10509: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
10510: @section Running Image Files
10511: @cindex running image files
10512: @cindex invoking image files
10513: @cindex image file invocation
10514:
10515: @cindex -i, invoke image file
10516: @cindex --image file, invoke image file
1.29 ! crook 10517: You can invoke Gforth with an image file @i{image} instead of the
1.1 anton 10518: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
10519: @example
1.29 ! crook 10520: gforth -i @i{image}
1.1 anton 10521: @end example
10522:
10523: @cindex executable image file
1.26 crook 10524: @cindex image file, executable
1.1 anton 10525: If your operating system supports starting scripts with a line of the
10526: form @code{#! ...}, you just have to type the image file name to start
10527: Gforth with this image file (note that the file extension @code{.fi} is
1.29 ! crook 10528: just a convention). I.e., to run Gforth with the image file @i{image},
! 10529: you can just type @i{image} instead of @code{gforth -i @i{image}}.
1.27 crook 10530: This works because every @code{.fi} file starts with a line of this
10531: format:
10532:
10533: @example
10534: #! /usr/local/bin/gforth-0.4.0 -i
10535: @end example
10536:
10537: The file and pathname for the Gforth engine specified on this line is
10538: the specific Gforth executable that it was built against; i.e. the value
10539: of the environment variable @code{GFORTH} at the time that
10540: @file{gforthmi} was executed.
1.1 anton 10541:
1.27 crook 10542: You can make use of the same shell capability to make a Forth source
10543: file into an executable. For example, if you place this text in a file:
1.26 crook 10544:
10545: @example
10546: #! /usr/local/bin/gforth
10547:
10548: ." Hello, world" CR
10549: bye
10550: @end example
10551:
10552: @noindent
1.27 crook 10553: and then make the file executable (chmod +x in Unix), you can run it
1.26 crook 10554: directly from the command line. The sequence @code{#!} is used in two
10555: ways; firstly, it is recognised as a ``magic sequence'' by the operating
1.29 ! crook 10556: system@footnote{The Unix kernel actually recognises two types of files:
! 10557: executable files and files of data, where the data is processed by an
! 10558: interpreter that is specified on the ``interpreter line'' -- the first
! 10559: line of the file, starting with the sequence #!. There may be a small
! 10560: limit (e.g., 32) on the number of characters that may be specified on
! 10561: the interpreter line.} secondly it is treated as a comment character by
! 10562: Gforth. Because of the second usage, a space is required between
! 10563: @code{#!} and the path to the executable.
1.27 crook 10564:
10565: The disadvantage of this latter technique, compared with using
10566: @file{gforthmi}, is that it is slower; the Forth source code is compiled
10567: on-the-fly, each time the program is invoked.
10568:
1.26 crook 10569: @comment TODO describe the #! magic with reference to the Power Tools book.
10570:
1.1 anton 10571: doc-#!
10572:
10573: @node Modifying the Startup Sequence, , Running Image Files, Image Files
10574: @section Modifying the Startup Sequence
10575: @cindex startup sequence for image file
10576: @cindex image file initialization sequence
10577: @cindex initialization sequence of image file
10578:
10579: You can add your own initialization to the startup sequence through the
1.26 crook 10580: deferred word @code{'cold}. @code{'cold} is invoked just before the
10581: image-specific command line processing (by default, loading files and
10582: evaluating (@code{-e}) strings) starts.
1.1 anton 10583:
10584: A sequence for adding your initialization usually looks like this:
10585:
10586: @example
10587: :noname
10588: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
10589: ... \ your stuff
10590: ; IS 'cold
10591: @end example
10592:
10593: @cindex turnkey image files
1.26 crook 10594: @cindex image file, turnkey applications
1.1 anton 10595: You can make a turnkey image by letting @code{'cold} execute a word
10596: (your turnkey application) that never returns; instead, it exits Gforth
10597: via @code{bye} or @code{throw}.
10598:
10599: @cindex command-line arguments, access
10600: @cindex arguments on the command line, access
10601: You can access the (image-specific) command-line arguments through the
1.26 crook 10602: variables @code{argc} and @code{argv}. @code{arg} provides convenient
1.1 anton 10603: access to @code{argv}.
10604:
1.26 crook 10605: If @code{'cold} exits normally, Gforth processes the command-line
10606: arguments as files to be loaded and strings to be evaluated. Therefore,
10607: @code{'cold} should remove the arguments it has used in this case.
10608:
10609: doc-'cold
1.1 anton 10610: doc-argc
10611: doc-argv
10612: doc-arg
10613:
10614:
10615: @c ******************************************************************
1.13 pazsan 10616: @node Engine, Binding to System Library, Image Files, Top
1.1 anton 10617: @chapter Engine
10618: @cindex engine
10619: @cindex virtual machine
10620:
1.26 crook 10621: Reading this chapter is not necessary for programming with Gforth. It
1.1 anton 10622: may be helpful for finding your way in the Gforth sources.
10623:
10624: The ideas in this section have also been published in the papers
10625: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
10626: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
10627: Ertl, presented at EuroForth '93; the latter is available at
10628: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
10629:
10630: @menu
10631: * Portability::
10632: * Threading::
10633: * Primitives::
10634: * Performance::
10635: @end menu
10636:
10637: @node Portability, Threading, Engine, Engine
10638: @section Portability
10639: @cindex engine portability
10640:
1.26 crook 10641: An important goal of the Gforth Project is availability across a wide
10642: range of personal machines. fig-Forth, and, to a lesser extent, F83,
10643: achieved this goal by manually coding the engine in assembly language
10644: for several then-popular processors. This approach is very
10645: labor-intensive and the results are short-lived due to progress in
10646: computer architecture.
1.1 anton 10647:
10648: @cindex C, using C for the engine
10649: Others have avoided this problem by coding in C, e.g., Mitch Bradley
10650: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
10651: particularly popular for UNIX-based Forths due to the large variety of
10652: architectures of UNIX machines. Unfortunately an implementation in C
10653: does not mix well with the goals of efficiency and with using
10654: traditional techniques: Indirect or direct threading cannot be expressed
10655: in C, and switch threading, the fastest technique available in C, is
10656: significantly slower. Another problem with C is that it is very
10657: cumbersome to express double integer arithmetic.
10658:
10659: @cindex GNU C for the engine
10660: @cindex long long
10661: Fortunately, there is a portable language that does not have these
10662: limitations: GNU C, the version of C processed by the GNU C compiler
10663: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
10664: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
10665: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
10666: threading possible, its @code{long long} type (@pxref{Long Long, ,
10667: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
10668: double numbers@footnote{Unfortunately, long longs are not implemented
10669: properly on all machines (e.g., on alpha-osf1, long longs are only 64
10670: bits, the same size as longs (and pointers), but they should be twice as
1.4 anton 10671: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
1.1 anton 10672: C Manual}). So, we had to implement doubles in C after all. Still, on
10673: most machines we can use long longs and achieve better performance than
10674: with the emulation package.}. GNU C is available for free on all
10675: important (and many unimportant) UNIX machines, VMS, 80386s running
10676: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
10677: on all these machines.
10678:
10679: Writing in a portable language has the reputation of producing code that
10680: is slower than assembly. For our Forth engine we repeatedly looked at
10681: the code produced by the compiler and eliminated most compiler-induced
10682: inefficiencies by appropriate changes in the source code.
10683:
10684: @cindex explicit register declarations
10685: @cindex --enable-force-reg, configuration flag
10686: @cindex -DFORCE_REG
10687: However, register allocation cannot be portably influenced by the
10688: programmer, leading to some inefficiencies on register-starved
10689: machines. We use explicit register declarations (@pxref{Explicit Reg
10690: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
10691: improve the speed on some machines. They are turned on by using the
10692: configuration flag @code{--enable-force-reg} (@code{gcc} switch
10693: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
10694: machine, but also on the compiler version: On some machines some
10695: compiler versions produce incorrect code when certain explicit register
10696: declarations are used. So by default @code{-DFORCE_REG} is not used.
10697:
10698: @node Threading, Primitives, Portability, Engine
10699: @section Threading
10700: @cindex inner interpreter implementation
10701: @cindex threaded code implementation
10702:
10703: @cindex labels as values
10704: GNU C's labels as values extension (available since @code{gcc-2.0},
10705: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
1.29 ! crook 10706: makes it possible to take the address of @i{label} by writing
! 10707: @code{&&@i{label}}. This address can then be used in a statement like
! 10708: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
1.1 anton 10709: @code{goto x}.
10710:
1.26 crook 10711: @cindex @code{NEXT}, indirect threaded
1.1 anton 10712: @cindex indirect threaded inner interpreter
10713: @cindex inner interpreter, indirect threaded
1.26 crook 10714: With this feature an indirect threaded @code{NEXT} looks like:
1.1 anton 10715: @example
10716: cfa = *ip++;
10717: ca = *cfa;
10718: goto *ca;
10719: @end example
10720: @cindex instruction pointer
10721: For those unfamiliar with the names: @code{ip} is the Forth instruction
10722: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
10723: execution token and points to the code field of the next word to be
10724: executed; The @code{ca} (code address) fetched from there points to some
10725: executable code, e.g., a primitive or the colon definition handler
10726: @code{docol}.
10727:
1.26 crook 10728: @cindex @code{NEXT}, direct threaded
1.1 anton 10729: @cindex direct threaded inner interpreter
10730: @cindex inner interpreter, direct threaded
10731: Direct threading is even simpler:
10732: @example
10733: ca = *ip++;
10734: goto *ca;
10735: @end example
10736:
10737: Of course we have packaged the whole thing neatly in macros called
1.26 crook 10738: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
1.1 anton 10739:
10740: @menu
10741: * Scheduling::
10742: * Direct or Indirect Threaded?::
10743: * DOES>::
10744: @end menu
10745:
10746: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
10747: @subsection Scheduling
10748: @cindex inner interpreter optimization
10749:
10750: There is a little complication: Pipelined and superscalar processors,
10751: i.e., RISC and some modern CISC machines can process independent
10752: instructions while waiting for the results of an instruction. The
10753: compiler usually reorders (schedules) the instructions in a way that
10754: achieves good usage of these delay slots. However, on our first tries
10755: the compiler did not do well on scheduling primitives. E.g., for
10756: @code{+} implemented as
10757: @example
10758: n=sp[0]+sp[1];
10759: sp++;
10760: sp[0]=n;
10761: NEXT;
10762: @end example
1.26 crook 10763: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
1.1 anton 10764: scheduling. After a little thought the problem becomes clear: The
1.21 crook 10765: compiler cannot know that @code{sp} and @code{ip} point to different
10766: addresses (and the version of @code{gcc} we used would not know it even
10767: if it was possible), so it could not move the load of the cfa above the
10768: store to the TOS. Indeed the pointers could be the same, if code on or
10769: very near the top of stack were executed. In the interest of speed we
10770: chose to forbid this probably unused ``feature'' and helped the compiler
1.26 crook 10771: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
1.21 crook 10772: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.1 anton 10773: @example
10774: n=sp[0]+sp[1];
10775: sp++;
10776: NEXT_P1;
10777: sp[0]=n;
10778: NEXT_P2;
10779: @end example
10780: This can be scheduled optimally by the compiler.
10781:
10782: This division can be turned off with the switch @code{-DCISC_NEXT}. This
10783: switch is on by default on machines that do not profit from scheduling
10784: (e.g., the 80386), in order to preserve registers.
10785:
10786: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
10787: @subsection Direct or Indirect Threaded?
10788: @cindex threading, direct or indirect?
10789:
10790: @cindex -DDIRECT_THREADED
10791: Both! After packaging the nasty details in macro definitions we
10792: realized that we could switch between direct and indirect threading by
10793: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
10794: defining a few machine-specific macros for the direct-threading case.
10795: On the Forth level we also offer access words that hide the
10796: differences between the threading methods (@pxref{Threading Words}).
10797:
10798: Indirect threading is implemented completely machine-independently.
10799: Direct threading needs routines for creating jumps to the executable
1.21 crook 10800: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
10801: machine-dependent, but they do not amount to many source lines. Therefore,
10802: even porting direct threading to a new machine requires little effort.
1.1 anton 10803:
10804: @cindex --enable-indirect-threaded, configuration flag
10805: @cindex --enable-direct-threaded, configuration flag
10806: The default threading method is machine-dependent. You can enforce a
10807: specific threading method when building Gforth with the configuration
10808: flag @code{--enable-direct-threaded} or
10809: @code{--enable-indirect-threaded}. Note that direct threading is not
10810: supported on all machines.
10811:
10812: @node DOES>, , Direct or Indirect Threaded?, Threading
10813: @subsection DOES>
10814: @cindex @code{DOES>} implementation
10815:
1.26 crook 10816: @cindex @code{dodoes} routine
10817: @cindex @code{DOES>}-code
1.1 anton 10818: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
10819: the chunk of code executed by every word defined by a
10820: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
10821: the Forth code to be executed, i.e. the code after the
1.26 crook 10822: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
1.1 anton 10823:
1.21 crook 10824: In fig-Forth the code field points directly to the @code{dodoes} and the
1.26 crook 10825: @code{DOES>}code address is stored in the cell after the code address (i.e. at
1.29 ! crook 10826: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
1.1 anton 10827: the Forth-79 and all later standards, because in fig-Forth this address
10828: lies in the body (which is illegal in these standards). However, by
10829: making the code field larger for all words this solution becomes legal
10830: again. We use this approach for the indirect threaded version and for
10831: direct threading on some machines. Leaving a cell unused in most words
10832: is a bit wasteful, but on the machines we are targeting this is hardly a
10833: problem. The other reason for having a code field size of two cells is
10834: to avoid having different image files for direct and indirect threaded
10835: systems (direct threaded systems require two-cell code fields on many
10836: machines).
10837:
1.26 crook 10838: @cindex @code{DOES>}-handler
1.1 anton 10839: The other approach is that the code field points or jumps to the cell
1.26 crook 10840: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
10841: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
10842: @code{DOES>}-code address by computing the code address, i.e., the address of
1.1 anton 10843: the jump to dodoes, and add the length of that jump field. A variant of
10844: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
10845: return address (which can be found in the return register on RISCs) is
1.26 crook 10846: the @code{DOES>}-code address. Since the two cells available in the code field
1.1 anton 10847: are used up by the jump to the code address in direct threading on many
10848: architectures, we use this approach for direct threading on these
10849: architectures. We did not want to add another cell to the code field.
10850:
10851: @node Primitives, Performance, Threading, Engine
10852: @section Primitives
10853: @cindex primitives, implementation
10854: @cindex virtual machine instructions, implementation
10855:
10856: @menu
10857: * Automatic Generation::
10858: * TOS Optimization::
10859: * Produced code::
10860: @end menu
10861:
10862: @node Automatic Generation, TOS Optimization, Primitives, Primitives
10863: @subsection Automatic Generation
10864: @cindex primitives, automatic generation
10865:
10866: @cindex @file{prims2x.fs}
10867: Since the primitives are implemented in a portable language, there is no
10868: longer any need to minimize the number of primitives. On the contrary,
10869: having many primitives has an advantage: speed. In order to reduce the
10870: number of errors in primitives and to make programming them easier, we
10871: provide a tool, the primitive generator (@file{prims2x.fs}), that
10872: automatically generates most (and sometimes all) of the C code for a
10873: primitive from the stack effect notation. The source for a primitive
10874: has the following form:
10875:
10876: @cindex primitive source format
10877: @format
1.29 ! crook 10878: @i{Forth-name} @i{stack-effect} @i{category} [@i{pronounc.}]
! 10879: [@code{""}@i{glossary entry}@code{""}]
! 10880: @i{C code}
1.1 anton 10881: [@code{:}
1.29 ! crook 10882: @i{Forth code}]
1.1 anton 10883: @end format
10884:
10885: The items in brackets are optional. The category and glossary fields
10886: are there for generating the documentation, the Forth code is there
10887: for manual implementations on machines without GNU C. E.g., the source
10888: for the primitive @code{+} is:
10889: @example
10890: + n1 n2 -- n core plus
10891: n = n1+n2;
10892: @end example
10893:
10894: This looks like a specification, but in fact @code{n = n1+n2} is C
10895: code. Our primitive generation tool extracts a lot of information from
10896: the stack effect notations@footnote{We use a one-stack notation, even
10897: though we have separate data and floating-point stacks; The separate
10898: notation can be generated easily from the unified notation.}: The number
10899: of items popped from and pushed on the stack, their type, and by what
10900: name they are referred to in the C code. It then generates a C code
10901: prelude and postlude for each primitive. The final C code for @code{+}
10902: looks like this:
10903:
10904: @example
10905: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
10906: /* */ /* documentation */
10907: @{
10908: DEF_CA /* definition of variable ca (indirect threading) */
10909: Cell n1; /* definitions of variables */
10910: Cell n2;
10911: Cell n;
10912: n1 = (Cell) sp[1]; /* input */
10913: n2 = (Cell) TOS;
10914: sp += 1; /* stack adjustment */
10915: NAME("+") /* debugging output (with -DDEBUG) */
10916: @{
10917: n = n1+n2; /* C code taken from the source */
10918: @}
10919: NEXT_P1; /* NEXT part 1 */
10920: TOS = (Cell)n; /* output */
10921: NEXT_P2; /* NEXT part 2 */
10922: @}
10923: @end example
10924:
10925: This looks long and inefficient, but the GNU C compiler optimizes quite
10926: well and produces optimal code for @code{+} on, e.g., the R3000 and the
10927: HP RISC machines: Defining the @code{n}s does not produce any code, and
10928: using them as intermediate storage also adds no cost.
10929:
1.26 crook 10930: There are also other optimizations that are not illustrated by this
10931: example: assignments between simple variables are usually for free (copy
1.1 anton 10932: propagation). If one of the stack items is not used by the primitive
10933: (e.g. in @code{drop}), the compiler eliminates the load from the stack
10934: (dead code elimination). On the other hand, there are some things that
10935: the compiler does not do, therefore they are performed by
10936: @file{prims2x.fs}: The compiler does not optimize code away that stores
10937: a stack item to the place where it just came from (e.g., @code{over}).
10938:
10939: While programming a primitive is usually easy, there are a few cases
10940: where the programmer has to take the actions of the generator into
10941: account, most notably @code{?dup}, but also words that do not (always)
1.26 crook 10942: fall through to @code{NEXT}.
1.1 anton 10943:
10944: @node TOS Optimization, Produced code, Automatic Generation, Primitives
10945: @subsection TOS Optimization
10946: @cindex TOS optimization for primitives
10947: @cindex primitives, keeping the TOS in a register
10948:
10949: An important optimization for stack machine emulators, e.g., Forth
10950: engines, is keeping one or more of the top stack items in
1.29 ! crook 10951: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
! 10952: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
1.1 anton 10953: @itemize @bullet
10954: @item
1.29 ! crook 10955: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
1.1 anton 10956: due to fewer loads from and stores to the stack.
1.29 ! crook 10957: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
! 10958: @i{y<n}, due to additional moves between registers.
1.1 anton 10959: @end itemize
10960:
10961: @cindex -DUSE_TOS
10962: @cindex -DUSE_NO_TOS
10963: In particular, keeping one item in a register is never a disadvantage,
10964: if there are enough registers. Keeping two items in registers is a
10965: disadvantage for frequent words like @code{?branch}, constants,
10966: variables, literals and @code{i}. Therefore our generator only produces
10967: code that keeps zero or one items in registers. The generated C code
10968: covers both cases; the selection between these alternatives is made at
10969: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
10970: code for @code{+} is just a simple variable name in the one-item case,
10971: otherwise it is a macro that expands into @code{sp[0]}. Note that the
10972: GNU C compiler tries to keep simple variables like @code{TOS} in
10973: registers, and it usually succeeds, if there are enough registers.
10974:
10975: @cindex -DUSE_FTOS
10976: @cindex -DUSE_NO_FTOS
10977: The primitive generator performs the TOS optimization for the
10978: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
10979: operations the benefit of this optimization is even larger:
10980: floating-point operations take quite long on most processors, but can be
10981: performed in parallel with other operations as long as their results are
10982: not used. If the FP-TOS is kept in a register, this works. If
10983: it is kept on the stack, i.e., in memory, the store into memory has to
10984: wait for the result of the floating-point operation, lengthening the
10985: execution time of the primitive considerably.
10986:
10987: The TOS optimization makes the automatic generation of primitives a
10988: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
10989: @code{TOS} is not sufficient. There are some special cases to
10990: consider:
10991: @itemize @bullet
10992: @item In the case of @code{dup ( w -- w w )} the generator must not
10993: eliminate the store to the original location of the item on the stack,
10994: if the TOS optimization is turned on.
10995: @item Primitives with stack effects of the form @code{--}
1.29 ! crook 10996: @i{out1}...@i{outy} must store the TOS to the stack at the start.
! 10997: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
1.1 anton 10998: must load the TOS from the stack at the end. But for the null stack
10999: effect @code{--} no stores or loads should be generated.
11000: @end itemize
11001:
11002: @node Produced code, , TOS Optimization, Primitives
11003: @subsection Produced code
11004: @cindex primitives, assembly code listing
11005:
11006: @cindex @file{engine.s}
11007: To see what assembly code is produced for the primitives on your machine
11008: with your compiler and your flag settings, type @code{make engine.s} and
11009: look at the resulting file @file{engine.s}.
11010:
11011: @node Performance, , Primitives, Engine
11012: @section Performance
11013: @cindex performance of some Forth interpreters
11014: @cindex engine performance
11015: @cindex benchmarking Forth systems
11016: @cindex Gforth performance
11017:
11018: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
11019: impossible to write a significantly faster engine.
11020:
11021: On register-starved machines like the 386 architecture processors
11022: improvements are possible, because @code{gcc} does not utilize the
11023: registers as well as a human, even with explicit register declarations;
11024: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
11025: and hand-tuned it for the 486; this system is 1.19 times faster on the
11026: Sieve benchmark on a 486DX2/66 than Gforth compiled with
11027: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
11028:
11029: @cindex Win32Forth performance
11030: @cindex NT Forth performance
11031: @cindex eforth performance
11032: @cindex ThisForth performance
11033: @cindex PFE performance
11034: @cindex TILE performance
11035: However, this potential advantage of assembly language implementations
11036: is not necessarily realized in complete Forth systems: We compared
11037: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
11038: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
11039: 1994) and Eforth (with and without peephole (aka pinhole) optimization
11040: of the threaded code); all these systems were written in assembly
11041: language. We also compared Gforth with three systems written in C:
11042: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
11043: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
1.21 crook 11044: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
11045: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
1.1 anton 11046: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
11047: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
11048: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
11049: 486DX2/66 with similar memory performance under Windows NT. Marcel
11050: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
11051: added the peephole optimizer, ran the benchmarks and reported the
11052: results.
11053:
11054: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
11055: matrix multiplication come from the Stanford integer benchmarks and have
11056: been translated into Forth by Martin Fraeman; we used the versions
11057: included in the TILE Forth package, but with bigger data set sizes; and
11058: a recursive Fibonacci number computation for benchmarking calling
11059: performance. The following table shows the time taken for the benchmarks
11060: scaled by the time taken by Gforth (in other words, it shows the speedup
11061: factor that Gforth achieved over the other systems).
11062:
11063: @example
11064: relative Win32- NT eforth This-
11065: time Gforth Forth Forth eforth +opt PFE Forth TILE
11066: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
11067: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
11068: matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
11069: fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
11070: @end example
11071:
1.26 crook 11072: You may be quite surprised by the good performance of Gforth when
11073: compared with systems written in assembly language. One important reason
11074: for the disappointing performance of these other systems is probably
11075: that they are not written optimally for the 486 (e.g., they use the
11076: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
11077: but costly method for relocating the Forth image: like @code{cforth}, it
11078: computes the actual addresses at run time, resulting in two address
11079: computations per @code{NEXT} (@pxref{Image File Background}).
11080:
11081: Only Eforth with the peephole optimizer has a performance that is
11082: comparable to Gforth. The speedups achieved with peephole optimization
11083: of threaded code are quite remarkable. Adding a peephole optimizer to
11084: Gforth should cause similar speedups.
1.1 anton 11085:
11086: The speedup of Gforth over PFE, ThisForth and TILE can be easily
11087: explained with the self-imposed restriction of the latter systems to
11088: standard C, which makes efficient threading impossible (however, the
1.4 anton 11089: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
1.1 anton 11090: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
11091: Moreover, current C compilers have a hard time optimizing other aspects
11092: of the ThisForth and the TILE source.
11093:
1.26 crook 11094: The performance of Gforth on 386 architecture processors varies widely
11095: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
11096: allocate any of the virtual machine registers into real machine
11097: registers by itself and would not work correctly with explicit register
11098: declarations, giving a 1.3 times slower engine (on a 486DX2/66 running
11099: the Sieve) than the one measured above.
1.1 anton 11100:
1.26 crook 11101: Note that there have been several releases of Win32Forth since the
11102: release presented here, so the results presented above may have little
1.1 anton 11103: predictive value for the performance of Win32Forth today.
11104:
11105: @cindex @file{Benchres}
11106: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
11107: Maierhofer (presented at EuroForth '95), an indirect threaded version of
11108: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
11109: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
11110: threaded version used here. The paper available at
11111: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
11112: it also contains numbers for some native code systems. You can find a
11113: newer version of these measurements at
11114: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
11115: find numbers for Gforth on various machines in @file{Benchres}.
11116:
1.26 crook 11117: @c ******************************************************************
1.13 pazsan 11118: @node Binding to System Library, Cross Compiler, Engine, Top
1.14 pazsan 11119: @chapter Binding to System Library
1.13 pazsan 11120:
11121: @node Cross Compiler, Bugs, Binding to System Library, Top
1.14 pazsan 11122: @chapter Cross Compiler
1.13 pazsan 11123:
11124: Cross Compiler
11125:
11126: @menu
11127: * Using the Cross Compiler::
11128: * How the Cross Compiler Works::
11129: @end menu
11130:
1.21 crook 11131: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
1.14 pazsan 11132: @section Using the Cross Compiler
1.13 pazsan 11133:
1.21 crook 11134: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
1.14 pazsan 11135: @section How the Cross Compiler Works
1.13 pazsan 11136:
11137: @node Bugs, Origin, Cross Compiler, Top
1.21 crook 11138: @appendix Bugs
1.1 anton 11139: @cindex bug reporting
11140:
1.21 crook 11141: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
1.1 anton 11142:
11143: If you find a bug, please send a bug report to
1.21 crook 11144: @email{bug-gforth@@gnu.ai.mit.edu}. A bug report should include this
11145: information:
11146:
11147: @itemize @bullet
11148: @item
11149: The Gforth version used (it is announced at the start of an
11150: interactive Gforth session).
11151: @item
11152: The machine and operating system (on Unix
11153: systems @code{uname -a} will report this information).
11154: @item
11155: The installation options (send the file @file{config.status}).
11156: @item
11157: A complete list of changes (if any) you (or your installer) have made to the
11158: Gforth sources.
11159: @item
11160: A program (or a sequence of keyboard commands) that reproduces the bug.
11161: @item
11162: A description of what you think constitutes the buggy behaviour.
11163: @end itemize
1.1 anton 11164:
11165: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
11166: to Report Bugs, gcc.info, GNU C Manual}.
11167:
11168:
1.21 crook 11169: @node Origin, Forth-related information, Bugs, Top
11170: @appendix Authors and Ancestors of Gforth
1.1 anton 11171:
11172: @section Authors and Contributors
11173: @cindex authors of Gforth
11174: @cindex contributors to Gforth
11175:
11176: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
11177: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
11178: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
11179: with their continuous feedback. Lennart Benshop contributed
11180: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
11181: support for calling C libraries. Helpful comments also came from Paul
11182: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
1.12 anton 11183: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
11184: release of Gforth-0.2.1 there were also helpful comments from many
11185: others; thank you all, sorry for not listing you here (but digging
1.23 crook 11186: through my mailbox to extract your names is on my to-do list). Since the
11187: release of Gforth-0.4.0 Neal Crook worked on the manual.
1.1 anton 11188:
11189: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
11190: and autoconf, among others), and to the creators of the Internet: Gforth
1.21 crook 11191: was developed across the Internet, and its authors did not meet
1.20 pazsan 11192: physically for the first 4 years of development.
1.1 anton 11193:
11194: @section Pedigree
1.26 crook 11195: @cindex pedigree of Gforth
1.1 anton 11196:
1.20 pazsan 11197: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
1.1 anton 11198: Dirk Zoller) will cross-fertilize each other. Of course, a significant
11199: part of the design of Gforth was prescribed by ANS Forth.
11200:
1.20 pazsan 11201: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
1.1 anton 11202: 32 bit native code version of VolksForth for the Atari ST, written
11203: mostly by Dietrich Weineck.
11204:
11205: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
11206: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
11207: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
11208:
11209: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
11210: Forth-83 standard. !! Pedigree? When?
11211:
11212: A team led by Bill Ragsdale implemented fig-Forth on many processors in
11213: 1979. Robert Selzer and Bill Ragsdale developed the original
11214: implementation of fig-Forth for the 6502 based on microForth.
11215:
11216: The principal architect of microForth was Dean Sanderson. microForth was
11217: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
11218: the 1802, and subsequently implemented on the 8080, the 6800 and the
11219: Z80.
11220:
11221: All earlier Forth systems were custom-made, usually by Charles Moore,
11222: who discovered (as he puts it) Forth during the late 60s. The first full
11223: Forth existed in 1971.
11224:
11225: A part of the information in this section comes from @cite{The Evolution
11226: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
11227: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
11228: Notices 28(3), 1993. You can find more historical and genealogical
11229: information about Forth there.
11230:
1.21 crook 11231: @node Forth-related information, Word Index, Origin, Top
11232: @appendix Other Forth-related information
11233: @cindex Forth-related information
11234:
11235: @menu
11236: * Internet resources::
11237: * Books::
11238: * The Forth Interest Group::
11239: * Conferences::
11240: @end menu
11241:
11242:
11243: @node Internet resources, Books, Forth-related information, Forth-related information
11244: @section Internet resources
1.26 crook 11245: @cindex internet resources
1.21 crook 11246:
11247: @cindex comp.lang.forth
11248: @cindex frequently asked questions
11249: There is an active newsgroup (comp.lang.forth) discussing Forth and
11250: Forth-related issues. A frequently-asked-questions (FAQ) list
11251: is posted to the newsgroup regulary, and archived at these sites:
11252:
11253: @itemize @bullet
11254: @item
11255: @url{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
11256: @item
11257: @url{ftp://ftp.forth.org/pub/Forth/FAQ/}
11258: @end itemize
11259:
11260: The FAQ list should be considered mandatory reading before posting to
11261: the newsgroup.
11262:
11263: Here are some other web sites holding Forth-related material:
11264:
11265: @itemize @bullet
11266: @item
11267: @url{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
11268: @item
11269: @url{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
11270: @item
11271: @url{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
11272: @item
11273: @url{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
11274: Research page, including links to the Journal of Forth Application and
11275: Research (JFAR) and a searchable Forth bibliography.
11276: @end itemize
11277:
11278:
11279: @node Books, The Forth Interest Group, Internet resources, Forth-related information
11280: @section Books
1.26 crook 11281: @cindex books on Forth
1.21 crook 11282:
11283: As the Standard is relatively new, there are not many books out yet. It
11284: is not recommended to learn Forth by using Gforth and a book that is not
11285: written for ANS Forth, as you will not know your mistakes from the
11286: deviations of the book. However, books based on the Forth-83 standard
11287: should be ok, because ANS Forth is primarily an extension of Forth-83.
11288:
11289: @cindex standard document for ANS Forth
11290: @cindex ANS Forth document
11291: The definite reference if you want to write ANS Forth programs is, of
1.26 crook 11292: course, the ANS Forth document. It is available in printed form from the
1.21 crook 11293: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
11294: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
11295: $200. You can also get it from Global Engineering Documents (Tel.: USA
11296: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
11297:
11298: @cite{dpANS6}, the last draft of the standard, which was then submitted
11299: to ANSI for publication is available electronically and for free in some
11300: MS Word format, and it has been converted to HTML
11301: (@url{http://www.taygeta.com/forth/dpans.html}; this is my favourite
11302: format); this HTML version also includes the answers to Requests for
11303: Interpretation (RFIs). Some pointers to these versions can be found
11304: through @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
11305:
1.26 crook 11306: @cindex introductory book on Forth
11307: @cindex book on Forth, introductory
1.21 crook 11308: @cindex Woehr, Jack: @cite{Forth: The New Model}
11309: @cindex @cite{Forth: The new model} (book)
11310: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
11311: introductory book based on a draft version of the standard. It does not
11312: cover the whole standard. It also contains interesting background
11313: information (Jack Woehr was in the ANS Forth Technical Committee). It is
11314: not appropriate for complete newbies, but programmers experienced in
11315: other languages should find it ok.
11316:
11317: @cindex Conklin, Edward K., and Elizabeth Rather: @cite{Forth Programmer's Handbook}
11318: @cindex Rather, Elizabeth and Edward K. Conklin: @cite{Forth Programmer's Handbook}
11319: @cindex @cite{Forth Programmer's Handbook} (book)
11320: @cite{Forth Programmer's Handbook} by Edward K. Conklin, Elizabeth
11321: D. Rather and the technical staff of Forth, Inc. (Forth, Inc., 1997;
11322: ISBN 0-9662156-0-5) contains little introductory material. The majority
11323: of the book is similar to @ref{Words}, but the book covers most of the
11324: standard words and some non-standard words (whereas this manual is
11325: quite incomplete). In addition, the book contains a chapter on
11326: programming style. The major drawback of this book is that it usually
11327: does not identify what is standard and what is specific to the Forth
11328: system described in the book (probably one of Forth, Inc.'s systems).
11329: Fortunately, many of the non-standard programming practices described in
11330: the book work in Gforth, too. Still, this drawback makes the book
11331: hardly more useful than a pre-ANS book.
11332:
11333: @node The Forth Interest Group, Conferences, Books, Forth-related information
11334: @section The Forth Interest Group
11335: @cindex Forth interest group (FIG)
11336:
11337: The Forth Interest Group (FIG) is a world-wide, non-profit,
1.26 crook 11338: member-supported organisation. It publishes a regular magazine,
11339: @var{FORTH Dimensions}, and offers other benefits of membership. You can
11340: contact the FIG through their office email address:
11341: @email{office@@forth.org} or by visiting their web site at
11342: @url{http://www.forth.org/}. This web site also includes links to FIG
11343: chapters in other countries and American cities
1.21 crook 11344: (@url{http://www.forth.org/chapters.html}).
11345:
11346: @node Conferences, , The Forth Interest Group, Forth-related information
11347: @section Conferences
11348: @cindex Conferences
11349:
11350: There are several regular conferences related to Forth. They are all
1.26 crook 11351: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
11352: news group:
1.21 crook 11353:
11354: @itemize @bullet
11355: @item
11356: FORML -- the Forth modification laboratory convenes every year near
11357: Monterey, California.
11358: @item
11359: The Rochester Forth Conference -- an annual conference traditionally
11360: held in Rochester, New York.
11361: @item
11362: EuroForth -- this European conference takes place annually.
11363: @end itemize
11364:
11365:
11366: @node Word Index, Concept Index, Forth-related information, Top
1.1 anton 11367: @unnumbered Word Index
11368:
1.26 crook 11369: This index is a list of Forth words that have ``glossary'' entries
11370: within this manual. Each word is listed with its stack effect and
11371: wordset.
1.1 anton 11372:
11373: @printindex fn
11374:
11375: @node Concept Index, , Word Index, Top
11376: @unnumbered Concept and Word Index
11377:
1.26 crook 11378: Not all entries listed in this index are present verbatim in the
11379: text. This index also duplicates, in abbreviated form, all of the words
11380: listed in the Word Index (only the names are listed for the words here).
1.1 anton 11381:
11382: @printindex cp
11383:
11384: @contents
11385: @bye
11386:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>