1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
3:
4: @comment TODO: nac29jan99 - a list of things to add in the next edit:
5: @comment 1. x-ref all ambiguous or implementation-defined features?
6: @comment 2. Describe the use of Auser Avariable AConstant A, etc.
7: @comment 3. words in miscellaneous section need a home.
8: @comment 4. search for TODO for other minor and major works required.
9: @comment 5. [rats] change all @var to @i in Forth source so that info
10: @comment file looks decent.
11: @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:
14:
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
23: @comment TODO this gets left in by HTML converter
24: @macro progstyle {}
25: Programming style note:
26: @end macro
27: @comment %**end of header (This is for running Texinfo on a region.)
28:
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:
51: @include version.texi
52:
53: @ifinfo
54: This file documents Gforth @value{VERSION}
55:
56: Copyright @copyright{} 1995-1999 Free Software Foundation, Inc.
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
88: @center for version @value{VERSION}
89: @sp 2
90: @center Anton Ertl
91: @center Bernd Paysan
92: @center Jens Wilke
93: @center Neal Crook
94: @sp 3
95: @center This manual is permanently under construction and was last updated on 04-May-1999
96:
97: @comment The following two commands start the copyright page.
98: @page
99: @vskip 0pt plus 1filll
100: Copyright @copyright{} 1995--1999 Free Software Foundation, Inc.
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
126: personal machines. This manual corresponds to version @value{VERSION}.
127: @end ifinfo
128:
129: @menu
130: * License:: The GPL
131: * Goals:: About the Gforth Project
132: * Gforth Environment:: Starting (and exiting) Gforth
133: * Introduction:: An introduction to ANS Forth
134: * Words:: Forth words available in Gforth
135: * Error messages:: How to interpret them
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
143: * Binding to System Library::
144: * Cross Compiler:: The Cross Compiler
145: * Bugs:: How to report them
146: * Origin:: Authors and ancestors of Gforth
147: * Forth-related information:: Books and places to look on the WWW
148: * Word Index:: An item for each Forth word
149: * Concept Index:: A menu covering many topics
150:
151: @detailmenu --- The Detailed Node Listing ---
152:
153: Goals of Gforth
154:
155: * Gforth Extensions Sinful?::
156:
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:
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::
174: * Where to go next::
175: * Exercises::
176:
177: Forth Words
178:
179: * Notation::
180: * Comments::
181: * Boolean Flags::
182: * Arithmetic::
183: * Stack Manipulation::
184: * Memory::
185: * Control Structures::
186: * Defining Words::
187: * The Text Interpreter::
188: * Tokens for Words::
189: * Word Lists::
190: * Environmental Queries::
191: * Files::
192: * Blocks::
193: * Other I/O::
194: * Programming Tools::
195: * Assembler and Code Words::
196: * Threading Words::
197: * Locals::
198: * Structures::
199: * Object-oriented Forth::
200: * Passing Commands to the OS::
201: * Miscellaneous Words::
202:
203: Arithmetic
204:
205: * Single precision::
206: * Bitwise operations::
207: * Double precision:: Double-cell integer arithmetic
208: * Numeric comparison::
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:
222: * Reserving Data Space::
223: * Memory Access::
224: * Address Arithmetic::
225: * Memory Blocks::
226: * Dynamic Allocation::
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:
245: The Text Interpreter
246:
247: * Input Sources::
248: * Number Conversion::
249: * Interpret/Compile states::
250: * Literals::
251: * Interpreter Directives::
252:
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:
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:
302: * Why object-oriented programming?::
303: * Object-Oriented Terminology::
304: * Objects::
305: * OOF::
306: * Mini-OOF::
307: * Comparison with other object models::
308:
309: The @file{objects.fs} model
310:
311: * Properties of the Objects model::
312: * Basic Objects Usage::
313: * The Objects base class::
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:
323: The @file{oof.fs} model
324:
325: * Properties of the OOF model::
326: * Basic OOF Usage::
327: * The OOF base class::
328: * Class Declaration::
329: * Class Implementation::
330:
331: The @file{mini-oof.fs} model
332:
333: * Basic Mini-OOF Usage::
334: * Mini-OOF Example::
335: * Mini-OOF Implementation::
336:
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:
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.
415: * Fully Relocatable Image Files:: better yet.
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.
419:
420: Fully Relocatable Image Files
421:
422: * gforthmi:: The normal way
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::
443:
444: Cross Compiler
445:
446: * Using the Cross Compiler::
447: * How the Cross Compiler Works::
448:
449: Other Forth-related information
450:
451: * Internet resources::
452: * Books::
453: * The Forth Interest Group::
454: * Conferences::
455:
456: @end detailmenu
457: @end menu
458:
459: @node License, Goals, Top, Top
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
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.
858: @end iftex
859:
860: @comment TODO much more blurb here.
861:
862: @c ******************************************************************
863: @node Goals, Gforth Environment, License, Top
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
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
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:
954:
955: @c ******************************************************************
956: @node Gforth Environment, Introduction, Goals, Top
957: @chapter Gforth Environment
958: @cindex Gforth environment
959:
960: Note: ultimately, the gforth man page will be auto-generated from the
961: material in this chapter.
962:
963: @menu
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
970: @end menu
971:
972: @xref{Image Files} for related information about the creation of images.
973:
974: @comment ----------------------------------------------
975: @node Invoking Gforth, Leaving Gforth, ,Gforth Environment
976: @section Invoking Gforth
977: @cindex invoking Gforth
978: @cindex running Gforth
979: @cindex command-line options
980: @cindex options on the command line
981: @cindex flags on the command line
982:
983: Gforth is made up of two parts; an executable ``engine'' (named
984: @file{gforth} or @file{gforth-fast}) and an image file. To start it, you
985: will usually just say @code{gforth} -- this automatically loads the
986: default image file @file{gforth.fi}. In many other cases the default
987: Gforth image will be invoked like this:
988: @example
989: gforth [file | -e forth-code] ...
990: @end example
991: @noindent
992: This interprets the contents of the files and the Forth code in the order they
993: are given.
994:
995: In addition to the @file{gforth} engine, there is also an engine called
996: @file{gforth-fast}, which is faster, but gives less informative error
997: messages (@pxref{Error messages}).
998:
999: In general, the command line looks like this:
1000:
1001: @example
1002: gforth[-fast] [engine options] [image options]
1003: @end example
1004:
1005: The engine options must come before the rest of the command
1006: line. They are:
1007:
1008: @table @code
1009: @cindex -i, command-line option
1010: @cindex --image-file, command-line option
1011: @item --image-file @i{file}
1012: @itemx -i @i{file}
1013: Loads the Forth image @i{file} instead of the default
1014: @file{gforth.fi} (@pxref{Image Files}).
1015:
1016: @cindex --path, command-line option
1017: @cindex -p, command-line option
1018: @item --path @i{path}
1019: @itemx -p @i{path}
1020: Uses @i{path} for searching the image file and Forth source code files
1021: instead of the default in the environment variable @code{GFORTHPATH} or
1022: the path specified at installation time (e.g.,
1023: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
1024: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
1025:
1026: @cindex --dictionary-size, command-line option
1027: @cindex -m, command-line option
1028: @cindex @i{size} parameters for command-line options
1029: @cindex size of the dictionary and the stacks
1030: @item --dictionary-size @i{size}
1031: @itemx -m @i{size}
1032: Allocate @i{size} space for the Forth dictionary space instead of
1033: using the default specified in the image (typically 256K). The
1034: @i{size} specification for this and subsequent options consists of
1035: an integer and a unit (e.g.,
1036: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
1037: size, in this case Cells), @code{k} (kilobytes), @code{M} (Megabytes),
1038: @code{G} (Gigabytes), and @code{T} (Terabytes). If no unit is specified,
1039: @code{e} is used.
1040:
1041: @cindex --data-stack-size, command-line option
1042: @cindex -d, command-line option
1043: @item --data-stack-size @i{size}
1044: @itemx -d @i{size}
1045: Allocate @i{size} space for the data stack instead of using the
1046: default specified in the image (typically 16K).
1047:
1048: @cindex --return-stack-size, command-line option
1049: @cindex -r, command-line option
1050: @item --return-stack-size @i{size}
1051: @itemx -r @i{size}
1052: Allocate @i{size} space for the return stack instead of using the
1053: default specified in the image (typically 15K).
1054:
1055: @cindex --fp-stack-size, command-line option
1056: @cindex -f, command-line option
1057: @item --fp-stack-size @i{size}
1058: @itemx -f @i{size}
1059: Allocate @i{size} space for the floating point stack instead of
1060: using the default specified in the image (typically 15.5K). In this case
1061: the unit specifier @code{e} refers to floating point numbers.
1062:
1063: @cindex --locals-stack-size, command-line option
1064: @cindex -l, command-line option
1065: @item --locals-stack-size @i{size}
1066: @itemx -l @i{size}
1067: Allocate @i{size} space for the locals stack instead of using the
1068: default specified in the image (typically 14.5K).
1069:
1070: @cindex -h, command-line option
1071: @cindex --help, command-line option
1072: @item --help
1073: @itemx -h
1074: Print a message about the command-line options
1075:
1076: @cindex -v, command-line option
1077: @cindex --version, command-line option
1078: @item --version
1079: @itemx -v
1080: Print version and exit
1081:
1082: @cindex --debug, command-line option
1083: @item --debug
1084: Print some information useful for debugging on startup.
1085:
1086: @cindex --offset-image, command-line option
1087: @item --offset-image
1088: Start the dictionary at a slightly different position than would be used
1089: otherwise (useful for creating data-relocatable images,
1090: @pxref{Data-Relocatable Image Files}).
1091:
1092: @cindex --no-offset-im, command-line option
1093: @item --no-offset-im
1094: Start the dictionary at the normal position.
1095:
1096: @cindex --clear-dictionary, command-line option
1097: @item --clear-dictionary
1098: Initialize all bytes in the dictionary to 0 before loading the image
1099: (@pxref{Data-Relocatable Image Files}).
1100:
1101: @cindex --die-on-signal, command-line-option
1102: @item --die-on-signal
1103: Normally Gforth handles most signals (e.g., the user interrupt SIGINT,
1104: or the segmentation violation SIGSEGV) by translating it into a Forth
1105: @code{THROW}. With this option, Gforth exits if it receives such a
1106: signal. This option is useful when the engine and/or the image might be
1107: severely broken (such that it causes another signal before recovering
1108: from the first); this option avoids endless loops in such cases.
1109: @end table
1110:
1111: @cindex loading files at startup
1112: @cindex executing code on startup
1113: @cindex batch processing with Gforth
1114: As explained above, the image-specific command-line arguments for the
1115: default image @file{gforth.fi} consist of a sequence of filenames and
1116: @code{-e @var{forth-code}} options that are interpreted in the sequence
1117: in which they are given. The @code{-e @var{forth-code}} or
1118: @code{--evaluate @var{forth-code}} option evaluates the Forth
1119: code. This option takes only one argument; if you want to evaluate more
1120: Forth words, you have to quote them or use @code{-e} several times. To exit
1121: after processing the command line (instead of entering interactive mode)
1122: append @code{-e bye} to the command line.
1123:
1124: @cindex versions, invoking other versions of Gforth
1125: If you have several versions of Gforth installed, @code{gforth} will
1126: invoke the version that was installed last. @code{gforth-@i{version}}
1127: invokes a specific version. You may want to use the option
1128: @code{--path}, if your environment contains the variable
1129: @code{GFORTHPATH}.
1130:
1131: Not yet implemented:
1132: On startup the system first executes the system initialization file
1133: (unless the option @code{--no-init-file} is given; note that the system
1134: resulting from using this option may not be ANS Forth conformant). Then
1135: the user initialization file @file{.gforth.fs} is executed, unless the
1136: option @code{--no-rc} is given; this file is first searched in @file{.},
1137: then in @file{~}, then in the normal path (see above).
1138:
1139:
1140:
1141: @comment ----------------------------------------------
1142: @node Leaving Gforth, Command-line editing, Invoking Gforth, Gforth Environment
1143: @section Leaving Gforth
1144: @cindex Gforth - leaving
1145: @cindex leaving Gforth
1146:
1147: You can leave Gforth by typing @code{bye} or @kbd{Ctrl-d} (at the start
1148: of a line) or (if you invoked Gforth with the @code{--die-on-signal}
1149: option) @kbd{Ctrl-c}. When you leave Gforth, all of your definitions and
1150: data are discarded. @xref{Image Files} for ways of saving the state of
1151: the system before leaving Gforth.
1152:
1153: doc-bye
1154:
1155: @comment ----------------------------------------------
1156: @node Command-line editing, Upper and lower case,Leaving Gforth,Gforth Environment
1157: @section Command-line editing
1158: @cindex command-line editing
1159:
1160: Gforth maintains a history file that records every line that you type to
1161: the text interpreter. This file is preserved between sessions, and is
1162: used to provide a command-line recall facility; if you type ctrl-P
1163: repeatedly you can recall successively older commands from this (or
1164: previous) session(s). The full list of command-line editing facilities is:
1165:
1166: @comment use @table? - anton
1167: @itemize @bullet
1168: @item
1169: @kbd{Ctrl-p} (``previous'') (or up-arrow) to recall successively older
1170: commands from the history buffer.
1171: @item
1172: @kbd{Ctrl-n} (``next'') (or down-arrow) to recall successively newer commands
1173: from the history buffer.
1174: @item
1175: @kbd{Ctrl-f} (or right-arrow) to move the cursor right, non-destructively.
1176: @item
1177: @kbd{Ctrl-b} (or left-arrow) to move the cursor left, non-destructively.
1178: @item
1179: @kbd{Ctrl-h} (backspace) to delete the character to the left of the cursor,
1180: closing up the line.
1181: @item
1182: @kbd{Ctrl-k} to delete (``kill'') from the cursor to the end of the line.
1183: @item
1184: @kbd{Ctrl-a} to move the cursor to the start of the line.
1185: @item
1186: @kbd{Ctrl-e} to move the cursor to the end of the line.
1187: @item
1188: @key{RET} (@kbd{Ctrl-m}) or @key{LFD} (@kbd{Ctrl-j}) to submit the current
1189: line.
1190: @item
1191: @key{TAB} to step through all possible full-word completions of the word
1192: currently being typed.
1193: @item
1194: @kbd{Ctrl-d} at the start of the line to terminate Gforth (gracefully,
1195: using @code{bye}).
1196: @end itemize
1197:
1198: When editing, displayable characters are inserted to the left of the
1199: cursor position; the line is always in ``insert'' (as opposed to
1200: ``overstrike'') mode.
1201:
1202: @cindex history file
1203: @cindex @file{.gforth-history}
1204: On Unix systems, the history file is @file{~/.gforth-history} by
1205: default@footnote{i.e. it is stored in the user's home directory.}. You
1206: can find out the name and location of your history file using:
1207:
1208: @example
1209: history-file type \ Unix-class systems
1210:
1211: history-file type \ Other systems
1212: history-dir type
1213: @end example
1214:
1215: If you enter long definitions by hand, you can use a text editor to
1216: paste them out of the history file into a Forth source file for reuse at
1217: a later time.
1218:
1219: Gforth never trims the size of the history file, so you should do this
1220: periodically, if necessary.
1221:
1222: @comment this is all defined in history.fs
1223:
1224:
1225:
1226: @comment ----------------------------------------------
1227: @node Upper and lower case, Environment variables,Command-line editing,Gforth Environment
1228: @section Upper and lower case
1229: @cindex case-sensitivity
1230: @cindex upper and lower case
1231:
1232: Gforth is case-insensitive, so you can enter definitions and invoke
1233: Standard words using upper, lower or mixed case (however,
1234: @pxref{core-idef, Implementation-defined options, Implementation-defined
1235: options}).
1236:
1237: ANS Forth only @i{requires} implementations to recognise Standard words
1238: when they are typed entirely in upper case. Therefore, a Standard
1239: program must use upper case for all Standard words. You can use whatever
1240: case you like for words that you define, but in a standard program you
1241: have to use the words in the same case that you defined them.
1242:
1243: Gforth supports case sensitivity through @code{table}s (case-sensitive
1244: wordlists, @pxref{Word Lists}).
1245:
1246: Two people have asked how to convert Gforth to case sensitivity; while
1247: we think this is a bad idea, you can change all wordlists into tables
1248: like this:
1249:
1250: @example
1251: ' table-find forth-wordlist wordlist-map @ !
1252: @end example
1253:
1254: Note that you now have to type the predefined words in the same case
1255: that we defined them, which are varying. You may want to convert them
1256: to your favourite case before doing this operation (I won't explain how,
1257: because if you are even contemplating to do this, you'd better have
1258: enough knowledge of Forth systems to know this already).
1259:
1260: @comment ----------------------------------------------
1261: @node Environment variables, Gforth Files, Upper and lower case,Gforth Environment
1262: @section Environment variables
1263: @cindex environment variables
1264:
1265: Gforth uses these environment variables:
1266:
1267: @itemize @bullet
1268: @item
1269: @cindex GFORTHHIST - environment variable
1270: GFORTHHIST - (Unix systems only) specifies the directory in which to
1271: open/create the history file, @file{.gforth-history}. Default:
1272: @code{$HOME}.
1273:
1274: @item
1275: @cindex GFORTHPATH - environment variable
1276: GFORTHPATH - specifies the path used when searching for the gforth image file and
1277: for Forth source-code files.
1278:
1279: @item
1280: @cindex GFORTH - environment variable
1281: GFORTH - used by @file{gforthmi} @xref{gforthmi}.
1282:
1283: @item
1284: @cindex GFORTHD - environment variable
1285: GFORTHD - used by @file{gforthmi} @xref{gforthmi}.
1286:
1287: @item
1288: @cindex TMP, TEMP - environment variable
1289: TMP, TEMP - (non-Unix systems only) used as a potential location for the
1290: history file.
1291: @end itemize
1292:
1293: @comment also POSIXELY_CORRECT LINES COLUMNS HOME but no interest in
1294: @comment mentioning these.
1295:
1296: All the Gforth environment variables default to sensible values if they
1297: are not set.
1298:
1299:
1300: @comment ----------------------------------------------
1301: @node Gforth Files, ,Environment variables,Gforth Environment
1302: @section Gforth files
1303: @cindex Gforth files
1304:
1305: When you Gforth on a Unix system in the default places, it installs
1306: files in these locations:
1307:
1308: @itemize @bullet
1309: @item
1310: @file{/usr/local/bin/gforth}
1311: @item
1312: @file{/usr/local/bin/gforthmi}
1313: @item
1314: @file{/usr/local/man/man1/gforth.1} - man page.
1315: @item
1316: @file{/usr/local/info} - the Info version of this manual.
1317: @item
1318: @file{/usr/local/lib/gforth/<version>/...} - Gforth @file{.fi} files.
1319: @item
1320: @file{/usr/local/share/gforth/<version>/TAGS} - Emacs TAGS file.
1321: @item
1322: @file{/usr/local/share/gforth/<version>/...} - Gforth source files.
1323: @item
1324: @file{.../emacs/site-lisp/gforth.el} - Emacs gforth mode.
1325: @end itemize
1326:
1327: You can select different places for installation by using
1328: @code{configure} options (listed with @code{configure --help}).
1329:
1330: @c ******************************************************************
1331: @node Introduction, Words, Gforth Environment, Top
1332: @comment node-name, next, previous, up
1333: @chapter An Introduction to ANS Forth
1334: @cindex Forth - an introduction
1335:
1336: The primary purpose of this manual is to document Gforth. However, since
1337: Forth is not a widely-known language and there is a lack of up-to-date
1338: teaching material, it seems worthwhile to provide some introductory
1339: material. @xref{Forth-related information} for other sources of Forth-related
1340: information.
1341:
1342: The examples in this section should work on any ANS Forth; the
1343: output shown was produced using Gforth. Each example attempts to
1344: reproduce the exact output that Gforth produces. If you try out the
1345: examples (and you should), what you should type is shown @kbd{like this}
1346: and Gforth's response is shown @code{like this}. The single exception is
1347: that, where the example shows @key{RET} it means that you should
1348: press the ``carriage return'' key. Unfortunately, some output formats for
1349: this manual cannot show the difference between @kbd{this} and
1350: @code{this} which will make trying out the examples harder (but not
1351: impossible).
1352:
1353: Forth is an unusual language. It provides an interactive development
1354: environment which includes both an interpreter and compiler. Forth
1355: programming style encourages you to break a problem down into many
1356: @cindex factoring
1357: small fragments (@dfn{factoring}), and then to develop and test each
1358: fragment interactively. Forth advocates assert that breaking the
1359: edit-compile-test cycle used by conventional programming languages can
1360: lead to great productivity improvements.
1361:
1362: @menu
1363: * Introducing the Text Interpreter::
1364: * Stacks and Postfix notation::
1365: * Your first definition::
1366: * How does that work?::
1367: * Forth is written in Forth::
1368: * Review - elements of a Forth system::
1369: * Where to go next::
1370: * Exercises::
1371: @end menu
1372:
1373: @comment ----------------------------------------------
1374: @node Introducing the Text Interpreter, Stacks and Postfix notation, Introduction, Introduction
1375: @section Introducing the Text Interpreter
1376: @cindex text interpreter
1377: @cindex outer interpreter
1378:
1379: @c IMO this is too detailed and the pace is too slow for
1380: @c an introduction. If you know German, take a look at
1381: @c http://www.complang.tuwien.ac.at/anton/lvas/skriptum-stack.html
1382: @c to see how I do it - anton
1383:
1384: When you invoke the Forth image, you will see a startup banner printed
1385: and nothing else (if you have Gforth installed on your system, try
1386: invoking it now, by typing @kbd{gforth@key{RET}}). Forth is now running
1387: its command line interpreter, which is called the @dfn{Text Interpreter}
1388: (also known as the @dfn{Outer Interpreter}). (You will learn a lot
1389: about the text interpreter as you read through this chapter, but
1390: @pxref{The Text Interpreter} for more detail).
1391:
1392: Although it's not obvious, Forth is actually waiting for your
1393: input. Type a number and press the @key{RET} key:
1394:
1395: @example
1396: @kbd{45@key{RET}} ok
1397: @end example
1398:
1399: Rather than give you a prompt to invite you to input something, the text
1400: interpreter prints a status message @i{after} it has processed a line
1401: of input. The status message in this case (``@code{ ok}'' followed by
1402: carriage-return) indicates that the text interpreter was able to process
1403: all of your input successfully. Now type something illegal:
1404:
1405: @example
1406: @kbd{qwer341@key{RET}}
1407: :1: Undefined word
1408: qwer341
1409: ^^^^^^^
1410: $400D2BA8 Bounce
1411: $400DBDA8 no.extensions
1412: @end example
1413:
1414: The exact text, other than the ``Undefined word'' may differ slightly on
1415: your system, but the effect is the same; when the text interpreter
1416: detects an error, it discards any remaining text on a line, resets
1417: certain internal state and prints an error message. @xref{Error
1418: messages} for a detailed description of error messages.
1419:
1420: The text interpreter waits for you to press carriage-return, and then
1421: processes your input line. Starting at the beginning of the line, it
1422: breaks the line into groups of characters separated by spaces. For each
1423: group of characters in turn, it makes two attempts to do something:
1424:
1425: @itemize @bullet
1426: @item
1427: It tries to treat it as a command. It does this by searching a @dfn{name
1428: dictionary}. If the group of characters matches an entry in the name
1429: dictionary, the name dictionary provides the text interpreter with
1430: information that allows the text interpreter perform some actions. In
1431: Forth jargon, we say that the group
1432: @cindex word
1433: @cindex definition
1434: @cindex execution token
1435: @cindex xt
1436: of characters names a @dfn{word}, that the dictionary search returns an
1437: @dfn{execution token (xt)} corresponding to the @dfn{definition} of the
1438: word, and that the text interpreter executes the xt. Often, the terms
1439: @dfn{word} and @dfn{definition} are used interchangeably.
1440: @item
1441: If the text interpreter fails to find a match in the name dictionary, it
1442: tries to treat the group of characters as a number in the current number
1443: base (when you start up Forth, the current number base is base 10). If
1444: the group of characters legitimately represents a number, the text
1445: interpreter pushes the number onto a stack (we'll learn more about that
1446: in the next section).
1447: @end itemize
1448:
1449: If the text interpreter is unable to do either of these things with any
1450: group of characters, it discards the group of characters and the rest of
1451: the line, then prints an error message. If the text interpreter reaches
1452: the end of the line without error, it prints the status message ``@code{ ok}''
1453: followed by carriage-return.
1454:
1455: This is the simplest command we can give to the text interpreter:
1456:
1457: @example
1458: @key{RET} ok
1459: @end example
1460:
1461: The text interpreter did everything we asked it to do (nothing) without
1462: an error, so it said that everything is ``@code{ ok}''. Try a slightly longer
1463: command:
1464:
1465: @example
1466: @kbd{12 dup fred dup@key{RET}}
1467: :1: Undefined word
1468: 12 dup fred dup
1469: ^^^^
1470: $400D2BA8 Bounce
1471: $400DBDA8 no.extensions
1472: @end example
1473:
1474: When you press the carriage-return key, the text interpreter starts to
1475: work its way along the line:
1476:
1477: @itemize @bullet
1478: @item
1479: When it gets to the space after the @code{2}, it takes the group of
1480: characters @code{12} and looks them up in the name
1481: dictionary@footnote{We can't tell if it found them or not, but assume
1482: for now that it did not}. There is no match for this group of characters
1483: in the name dictionary, so it tries to treat them as a number. It is
1484: able to do this successfully, so it puts the number, 12, ``on the stack''
1485: (whatever that means).
1486: @item
1487: The text interpreter resumes scanning the line and gets the next group
1488: of characters, @code{dup}. It looks it up in the name dictionary and
1489: (you'll have to take my word for this) finds it, and executes the word
1490: @code{dup} (whatever that means).
1491: @item
1492: Once again, the text interpreter resumes scanning the line and gets the
1493: group of characters @code{fred}. It looks them up in the name
1494: dictionary, but can't find them. It tries to treat them as a number, but
1495: they don't represent any legal number.
1496: @end itemize
1497:
1498: At this point, the text interpreter gives up and prints an error
1499: message. The error message shows exactly how far the text interpreter
1500: got in processing the line. In particular, it shows that the text
1501: interpreter made no attempt to do anything with the final character
1502: group, @code{dup}, even though we have good reason to believe that the
1503: text interpreter would have no problem looking that word up and
1504: executing it a second time.
1505:
1506:
1507: @comment ----------------------------------------------
1508: @node Stacks and Postfix notation, Your first definition, Introducing the Text Interpreter, Introduction
1509: @section Stacks, postfix notation and parameter passing
1510: @cindex text interpreter
1511: @cindex outer interpreter
1512:
1513: In procedural programming languages (like C and Pascal), the
1514: building-block of programs is the @dfn{function} or @dfn{procedure}. These
1515: functions or procedures are called with @dfn{explicit parameters}. For
1516: example, in C we might write:
1517:
1518: @example
1519: total = total + new_volume(length,height,depth);
1520: @end example
1521:
1522: @noindent
1523: where new_volume is a function-call to another piece of code, and total,
1524: length, height and depth are all variables. length, height and depth are
1525: parameters to the function-call.
1526:
1527: In Forth, the equivalent of the function or procedure is the
1528: @dfn{definition} and parameters are implicitly passed between
1529: definitions using a shared stack that is visible to the
1530: programmer. Although Forth does support variables, the existence of the
1531: stack means that they are used far less often than in most other
1532: programming languages. When the text interpreter encounters a number, it
1533: will place (@dfn{push}) it on the stack. There are several stacks (the
1534: actual number is implementation-dependent ...) and the particular stack
1535: used for any operation is implied unambiguously by the operation being
1536: performed. The stack used for all integer operations is called the @dfn{data
1537: stack} and, since this is the stack used most commonly, references to
1538: ``the data stack'' are often abbreviated to ``the stack''.
1539:
1540: The stacks have a last-in, first-out (LIFO) organisation. If you type:
1541:
1542: @example
1543: @kbd{1 2 3@key{RET}} ok
1544: @end example
1545:
1546: Then this instructs the text interpreter to placed three numbers on the
1547: (data) stack. An analogy for the behaviour of the stack is to take a
1548: pack of playing cards and deal out the ace (1), 2 and 3 into a pile on
1549: the table. The 3 was the last card onto the pile (``last-in'') and if
1550: you take a card off the pile then, unless you're prepared to fiddle a
1551: bit, the card that you take off will be the 3 (``first-out''). The
1552: number that will be first-out of the stack is called the @dfn{top of
1553: stack}, which
1554: @cindex TOS definition
1555: is often abbreviated to @dfn{TOS}.
1556:
1557: To understand how parameters are passed in Forth, consider the
1558: behaviour of the definition @code{+} (pronounced ``plus''). You will not
1559: be surprised to learn that this definition performs addition. More
1560: precisely, it adds two number together and produces a result. Where does
1561: it get the two numbers from? It takes the top two numbers off the
1562: stack. Where does it place the result? On the stack. You can act-out the
1563: behaviour of @code{+} with your playing cards like this:
1564:
1565: @itemize @bullet
1566: @item
1567: Pick up two cards from the stack on the table
1568: @item
1569: Stare at them intently and ask yourself ``what @i{is} the sum of these two
1570: numbers''
1571: @item
1572: Decide that the answer is 5
1573: @item
1574: Shuffle the two cards back into the pack and find a 5
1575: @item
1576: Put a 5 on the remaining ace that's on the table.
1577: @end itemize
1578:
1579: If you don't have a pack of cards handy but you do have Forth running,
1580: you can use the definition @code{.s} to show the current state of the stack,
1581: without affecting the stack. Type:
1582:
1583: @example
1584: @kbd{clearstack 1 2 3@key{RET}} ok
1585: @kbd{.s@key{RET}} <3> 1 2 3 ok
1586: @end example
1587:
1588: The text interpreter looks up the word @code{clearstack} and executes
1589: it; it tidies up the stack and removes any entries that may have been
1590: left on it by earlier examples. The text interpreter pushes each of the
1591: three numbers in turn onto the stack. Finally, the text interpreter
1592: looks up the word @code{.s} and executes it. The effect of executing
1593: @code{.s} is to print the ``<3>'' (the total number of items on the stack)
1594: followed by a list of all the items on the stack; the item on the far
1595: right-hand side is the TOS.
1596:
1597: You can now type:
1598:
1599: @example
1600: @kbd{+ .s@key{RET}} <2> 1 5 ok
1601: @end example
1602:
1603: @noindent
1604: which is correct; there are now 2 items on the stack and the result of
1605: the addition is 5.
1606:
1607: If you're playing with cards, try doing a second addition: pick up the
1608: two cards, work out that their sum is 6, shuffle them into the pack,
1609: look for a 6 and place that on the table. You now have just one item on
1610: the stack. What happens if you try to do a third addition? Pick up the
1611: first card, pick up the second card -- ah! There is no second card. This
1612: is called a @dfn{stack underflow} and consitutes an error. If you try to
1613: do the same thing with Forth it will report an error (probably a Stack
1614: Underflow or an Invalid Memory Address error).
1615:
1616: The opposite situation to a stack underflow is a @dfn{stack overflow},
1617: which simply accepts that there is a finite amount of storage space
1618: reserved for the stack. To stretch the playing card analogy, if you had
1619: enough packs of cards and you piled the cards up on the table, you would
1620: eventually be unable to add another card; you'd hit the ceiling. Gforth
1621: allows you to set the maximum size of the stacks. In general, the only
1622: time that you will get a stack overflow is because a definition has a
1623: bug in it and is generating data on the stack uncontrollably.
1624:
1625: There's one final use for the playing card analogy. If you model your
1626: stack using a pack of playing cards, the maximum number of items on
1627: your stack will be 52 (I assume you didn't use the Joker). The maximum
1628: @i{value} of any item on the stack is 13 (the King). In fact, the only
1629: possible numbers are positive integer numbers 1 through 13; you can't
1630: have (for example) 0 or 27 or 3.52 or -2. If you change the way you
1631: think about some of the cards, you can accommodate different
1632: numbers. For example, you could think of the Jack as representing 0,
1633: the Queen as representing -1 and the King as representing -2. Your
1634: *range* remains unchanged (you can still only represent a total of 13
1635: numbers) but the numbers that you can represent are -2 through 10.
1636:
1637: In that analogy, the limit was the amount of information that a single
1638: stack entry could hold, and Forth has a similar limit. In Forth, the
1639: size of a stack entry is called a @dfn{cell}. The actual size of a cell is
1640: implementation dependent and affects the maximum value that a stack
1641: entry can hold. A Standard Forth provides a cell size of at least
1642: 16-bits, and most desktop systems use a cell size of 32-bits.
1643:
1644: Forth does not do any type checking for you, so you are free to
1645: manipulate and combine stack items in any way you wish. A convenient way
1646: of treating stack items is as 2's complement signed integers, and that
1647: is what Standard words like @code{+} do. Therefore you can type:
1648:
1649: @example
1650: @kbd{-5 12 + .s@key{RET}} <1> 7 ok
1651: @end example
1652:
1653: If you use numbers and definitions like @code{+} in order to turn Forth
1654: into a great big pocket calculator, you will realise that it's rather
1655: different from a normal calculator. Rather than typing 2 + 3 = you had
1656: to type 2 3 + (ignore the fact that you had to use @code{.s} to see the
1657: result). The terminology used to describe this difference is to say that
1658: your calculator uses @dfn{Infix Notation} (parameters and operators are
1659: mixed) whilst Forth uses @dfn{Postfix Notation} (parameters and
1660: operators are separate), also called @dfn{Reverse Polish Notation}.
1661:
1662: Whilst postfix notation might look confusing to begin with, it has
1663: several important advantages:
1664:
1665: @itemize @bullet
1666: @item
1667: it is unambiguous
1668: @item
1669: it is more concise
1670: @item
1671: it fits naturally with a stack-based system
1672: @end itemize
1673:
1674: To examine these claims in more detail, consider these sums:
1675:
1676: @example
1677: 6 + 5 * 4 =
1678: 4 * 5 + 6 =
1679: @end example
1680:
1681: If you're just learning maths or your maths is very rusty, you will
1682: probably come up with the answer 44 for the first and 26 for the
1683: second. If you are a bit of a whizz at maths you will remember the
1684: @i{convention} that multiplication takes precendence over addition, and
1685: you'd come up with the answer 26 both times. To explain the answer 26
1686: to someone who got the answer 44, you'd probably rewrite the first sum
1687: like this:
1688:
1689: @example
1690: 6 + (5 * 4) =
1691: @end example
1692:
1693: If what you really wanted was to perform the addition before the
1694: multiplication, you would have to use parentheses to force it.
1695:
1696: If you did the first two sums on a pocket calculator you would probably
1697: get the right answers, unless you were very cautious and entered them using
1698: these keystroke sequences:
1699:
1700: 6 + 5 = * 4 =
1701: 4 * 5 = + 6 =
1702:
1703: Postfix notation is unambiguous because the order that the operators
1704: are applied is always explicit; that also means that parentheses are
1705: never required. The operators are @i{active} (the act of quoting the
1706: operator makes the operation occur) which removes the need for ``=''.
1707:
1708: The sum 6 + 5 * 4 can be written (in postfix notation) in two
1709: equivalent ways:
1710:
1711: @example
1712: 6 5 4 * + or:
1713: 5 4 * 6 +
1714: @end example
1715:
1716: An important thing that you should notice about this notation is that
1717: the @i{order} of the numbers does not change; if you want to subtract
1718: 2 from 10 you type @code{10 2 -}.
1719:
1720: The reason that Forth uses postfix notation is very simple to explain: it
1721: makes the implementation extremely simple, and it follows naturally from
1722: using the stack as a mechanism for passing parameters. Another way of
1723: thinking about this is to realise that all Forth definitions are
1724: @i{active}; they execute as they are encountered by the text
1725: interpreter. The result of this is that the syntax of Forth is trivially
1726: simple.
1727:
1728:
1729:
1730: @comment ----------------------------------------------
1731: @node Your first definition, How does that work?, Stacks and Postfix notation, Introduction
1732: @section Your first Forth definition
1733: @cindex first definition
1734:
1735: Until now, the examples we've seen have been trivial; we've just been
1736: using Forth as a bigger-than-pocket calculator. Also, each calculation
1737: we've shown has been a ``one-off'' -- to repeat it we'd need to type it in
1738: again@footnote{That's not quite true. If you press the up-arrow key on
1739: your keyboard you should be able to scroll back to any earlier command,
1740: edit it and re-enter it.} In this section we'll see how to add new
1741: words to Forth's vocabulary.
1742:
1743: The easiest way to create a new word is to use a @dfn{colon
1744: definition}. We'll define a few and try them out before worrying too
1745: much about how they work. Try typing in these examples; be careful to
1746: copy the spaces accurately:
1747:
1748: @example
1749: : add-two 2 + . ;
1750: : greet ." Hello and welcome" ;
1751: : demo 5 add-two ;
1752: @end example
1753:
1754: @noindent
1755: Now try them out:
1756:
1757: @example
1758: @kbd{greet@key{RET}} Hello and welcome ok
1759: @kbd{greet greet@key{RET}} Hello and welcomeHello and welcome ok
1760: @kbd{4 add-two@key{RET}} 6 ok
1761: @kbd{demo@key{RET}} 7 ok
1762: @kbd{9 greet demo add-two@key{RET}} Hello and welcome7 11 ok
1763: @end example
1764:
1765: The first new thing that we've introduced here is the pair of words
1766: @code{:} and @code{;}. These are used to start and terminate a new
1767: definition, respectively. The first word after the @code{:} is the name
1768: for the new definition.
1769:
1770: As you can see from the examples, a definition is built up of words that
1771: have already been defined; Forth makes no distinction between
1772: definitions that existed when you started the system up, and those that
1773: you define yourself.
1774:
1775: The examples also introduce the words @code{.} (dot), @code{."}
1776: (dot-quote) and @code{dup} (dewp). Dot takes the value from the top of
1777: the stack and displays it. It's like @code{.s} except that it only
1778: displays the top item of the stack and it is destructive; after it has
1779: executed, the number is no longer on the stack. There is always one
1780: space printed after the number, and no spaces before it. Dot-quote
1781: defines a string (a sequence of characters) that will be printed when
1782: the word is executed. The string can contain any printable characters
1783: except @code{"}. A @code{"} has a special function; it is not a Forth
1784: word but it acts as a delimiter (the way that delimiters work is
1785: described in the next section). Finally, @code{dup} duplicates the value
1786: at the top of the stack. Try typing @code{5 dup .s} to see what it does.
1787:
1788: We already know that the text interpreter searches through the
1789: dictionary to locate names. If you've followed the examples earlier, you
1790: will already have a definition called @code{add-two}. Lets try modifying
1791: it by typing in a new definition:
1792:
1793: @example
1794: @kbd{: add-two dup . ." + 2 =" 2 + . ;@key{RET}} redefined add-two ok
1795: @end example
1796:
1797: Forth recognised that we were defining a word that already exists, and
1798: printed a message to warn us of that fact. Let's try out the new
1799: definition:
1800:
1801: @example
1802: @kbd{9 add-two@key{RET}} 9 + 2 =11 ok
1803: @end example
1804:
1805: @noindent
1806: All that we've actually done here, though, is to create a new
1807: definition, with a particular name. The fact that there was already a
1808: definition with the same name did not make any difference to the way
1809: that the new definition was created (except that Forth printed a warning
1810: message). The old definition of add-two still exists (try @code{demo}
1811: again to see that this is true). Any new definition will use the new
1812: definition of @code{add-two}, but old definitions continue to use the
1813: version that already existed at the time that they were @code{compiled}.
1814:
1815: Before you go on to the next section, try defining and redefining some
1816: words of your own.
1817:
1818: @comment ----------------------------------------------
1819: @node How does that work?, Forth is written in Forth, Your first definition, Introduction
1820: @section How does that work?
1821: @cindex parsing words
1822:
1823: @c That's pretty deep (IMO way too deep) for an introduction. - anton
1824:
1825: @c Is it a good idea to talk about the interpretation semantics of a
1826: @c number? We don't have an xt to go along with it. - anton
1827:
1828: @c Now that I have eliminated execution semantics, I wonder if it would not
1829: @c be better to keep them (or add run-time semantics), to make it easier to
1830: @c explain what compilation semantics usually does. - anton
1831:
1832: Now we're going to take another look at the definition of @code{add-two}
1833: from the previous section. From our knowledge of the way that the text
1834: interpreter works, we would have expected this result when we tried to
1835: define @code{add-two}:
1836:
1837: @example
1838: @kbd{: add-two 2 + . " ;@key{RET}}
1839: ^^^^^^^
1840: Error: Undefined word
1841: @end example
1842:
1843: The reason that this didn't happen is bound up in the way that @code{:}
1844: works. The word @code{:} does two special things. The first special
1845: thing that it does prevents the text interpreter from ever seeing the
1846: characters @code{add-two}. The text interpreter uses a variable called
1847: @cindex modifying >IN
1848: @code{>IN} (pronounced ''to-in'') to keep track of where it is in the
1849: input line. When it encounters the word @code{:} it behaves in exactly
1850: the same way as it does for any other word; it looks it up in the name
1851: dictionary, finds its xt and executes it. When @code{:} executes, it
1852: looks at the input buffer, finds the word @code{add-two} and advances the
1853: value of @code{>IN} to point past it. It then does some other stuff
1854: associated with creating the new definition (including creating an entry
1855: for @code{add-two} in the name dictionary). When the execution of @code{:}
1856: completes, control returns to the text interpreter, which is oblivious
1857: to the fact that it has been tricked into ignoring part of the input
1858: line.
1859:
1860: @cindex parsing words
1861: Words like @code{:} -- words that advance the value of @code{>IN} and so
1862: prevent the text interpreter from acting on the whole of the input line
1863: -- are called @dfn{parsing words}.
1864:
1865: @cindex @code{state} - effect on the text interpreter
1866: @cindex text interpreter - effect of state
1867: The second special thing that @code{:} does is change the value of a
1868: variable called @code{state}, which affects the way that the text
1869: interpreter behaves. When Gforth starts up, @code{state} has the value
1870: 0, and the text interpreter is said to be @dfn{interpreting}. During a
1871: colon definition (started with @code{:}), @code{state} is set to -1 and
1872: the text interpreter is said to be @dfn{compiling}. The word @code{;}
1873: ends the definition -- one of the things that it does is to change the
1874: value of @code{state} back to 0.
1875:
1876: We have already seen how the text interpreter behaves when it is
1877: interpreting; it looks for each character sequence in the dictionary,
1878: finds its xt and executes it, or it converts it to a number and pushes
1879: it onto the stack, or it fails to do either and generates an error.
1880:
1881: When the text interpreter is compiling, its behaviour is slightly
1882: different; it still looks for each character sequence in the dictionary
1883: and finds it, or converts it to a number, or fails to do either and
1884: generates an error. But instead of the execution token of a word it
1885: finds and executes the compilation token. For most words executing the
1886: compilation token results in laying down (@dfn{compiling}) the execution
1887: token, i.e., some magic to make that xt or number get executed or pushed
1888: at a later time; at the time that @code{add-two} is
1889: @dfn{executed}. Therefore, when you execute @code{add-two} its
1890: @dfn{run-time effect} is exactly the same as if you had typed @code{2 +
1891: .} outside of a definition, and pressed carriage-return.
1892:
1893: In Forth, every word or number can be described in terms of two
1894: properties:
1895:
1896: @itemize @bullet
1897: @item
1898: Its @dfn{interpretation semantics}, represented by the execution token.
1899: @item
1900: Its @dfn{compilation semantics}, represented by the compilation token.
1901: @end itemize
1902:
1903: The value of @code{state} determines whether the text interpreter will
1904: use the compilation or interpretation semantics of a word or number that
1905: it encounters.
1906:
1907: @itemize @bullet
1908: @item
1909: @cindex interpretation semantics
1910: When the text interpreter encounters a word or number in @dfn{interpret}
1911: state, it performs the @dfn{interpretation semantics} of the word or
1912: number.
1913: @item
1914: @cindex compilation semantics
1915: When the text interpreter encounters a word or number in @dfn{compile}
1916: state, it performs the @dfn{compilation semantics} of the word or
1917: number.
1918: @end itemize
1919:
1920: @noindent
1921: Numbers are always treated in a fixed way:
1922:
1923: @itemize @bullet
1924: @item
1925: When the number is @dfn{interpreted}, its behaviour is to push the number onto the stack.
1926: @item
1927: When the number is @dfn{compiled}, a piece of code is appended to the
1928: current definition that pushes the number when it runs. (In other words,
1929: the compilation semantics of a number are to postpone its interpretation
1930: semantics until the run-time of the definition that it is being compiled
1931: into.)
1932: @end itemize
1933:
1934: The behaviour of a word is not so regular, but most have @i{default
1935: compilation semantics} which means that they behave like this:
1936:
1937: @itemize @bullet
1938: @item
1939: The @dfn{interpretation semantics} of the word are to do something useful.
1940: @item
1941: The @dfn{compilation semantics} of the word are to append its
1942: @dfn{interpretation semantics} to the current definition (so that its
1943: run-time behaviour is to do something useful).
1944: @end itemize
1945:
1946: @cindex immediate words
1947: The actual behaviour of any particular word depends upon the way in
1948: which it was defined. When the text interpreter finds the word in the
1949: name dictionary, it not only retrieves the xt for the word, it also
1950: retrieves some flags: the @dfn{compile-only} flag and the @dfn{immediate
1951: flag}. The compile-only flag indicates that the word has no
1952: interpretation semantics (the run-time behaviour for the default
1953: compilation semantics is not affected by this flag, however); any
1954: attempt to interpret a word that has the compile-only flag set will
1955: generate an error (for example, @code{IF} has no interpretation
1956: semantics). The immediate flag changes the compilation semantics of the
1957: word; if it is set, the compilation semantics are equal to the
1958: interpretation semantics (again ignoring the compile-only flag). it. In
1959: other words, these so-called @dfn{immediate} words behave like this:
1960:
1961: @itemize @bullet
1962: @item
1963: The @dfn{interpretation semantics} of the word are to do something useful.
1964: @item
1965: The @dfn{compilation semantics} of the word are to do something useful
1966: (and actually the same thing); i.e., it is executed during compilation.
1967: @end itemize
1968:
1969: This example shows the difference between an immediate and a
1970: non-immediate word:
1971:
1972: @example
1973: : show-state state @@ . ;
1974: : show-state-now show-state ; immediate
1975: : word1 show-state ;
1976: : word2 show-state-now ;
1977: @end example
1978:
1979: The word @code{immediate} after the definition of @code{show-state-now}
1980: makes that word an immediate word. These definitions introduce a new
1981: word: @code{@@} (pronounced ``fetch''). This word fetches the value of a
1982: variable, and leaves it on the stack. Therefore, the behaviour of
1983: @code{show-state} is to print a number that represents the current value
1984: of @code{state}.
1985:
1986: When you execute @code{word1}, it prints the number 0, indicating that
1987: the system is interpreting. When the text interpreter compiled the
1988: definition of @code{word1}, it encountered @code{show-state} whose
1989: compilation semantics are to append its interpretation semantics to the
1990: current definition. When you execute @code{word1}, it performs the
1991: interpretation semantics of @code{show-state}. At the time that @code{word1}
1992: (and therefore @code{show-state}) are executed, the system is
1993: interpreting.
1994:
1995: When you pressed @key{RET} after entering the definition of @code{word2},
1996: you should have seen the number -1 printed, followed by ``@code{
1997: ok}''. When the text interpreter compiled the definition of
1998: @code{word2}, it encountered @code{show-state-now}, an immediate word,
1999: whose compilation semantics are therefore to perform its interpretation
2000: semantics. It is executed straight away (even before the text
2001: interpreter has moved on to process another group of characters; the
2002: @code{;} in this example). The effect of executing it are to display the
2003: value of @code{state} @i{at the time that the definition of}
2004: @code{word2} @i{is being defined}. Printing -1 demonstrates that the
2005: system is compiling at this time. If you execute @code{word2} it does
2006: nothing at all.
2007:
2008: @cindex @code{."}, how it works
2009: Before leaving the subject of immediate words, consider the behaviour of
2010: @code{."} in the definition of @code{greet}, in the previous
2011: section. This word is both a parsing word and an immediate word. Notice
2012: that there is a space between @code{."} and the start of the text
2013: @code{Hello and welcome}, but that there is no space between the last
2014: letter of @code{welcome} and the @code{"} character. The reason for this
2015: is that @code{."} is a Forth word; it must have a space after it so that
2016: the text interpreter can identify it. The @code{"} is not a Forth word;
2017: it is a @dfn{delimiter}. The examples earlier show that, when the string
2018: is displayed, there is neither a space before the @code{H} nor after the
2019: @code{e}. Since @code{."} is an immediate word, it executes at the time
2020: that @code{greet} is defined. When it executes, its behaviour is to
2021: search forward in the input line looking for the delimiter. When it
2022: finds the delimiter, it updates @code{>IN} to point past the
2023: delimiter. It also compiles some magic code into the definition of
2024: @code{greet}; the xt of a run-time routine that prints a text string. It
2025: compiles the string @code{Hello and welcome} into memory so that it is
2026: available to be printed later. When the text interpreter gains control,
2027: the next word it finds in the input stream is @code{;} and so it
2028: terminates the definition of @code{greet}.
2029:
2030:
2031: @comment ----------------------------------------------
2032: @node Forth is written in Forth, Review - elements of a Forth system, How does that work?, Introduction
2033: @section Forth is written in Forth
2034: @cindex structure of Forth programs
2035:
2036: When you start up a Forth compiler, a large number of definitions
2037: already exist. In Forth, you develop a new application using bottom-up
2038: programming techniques to create new definitions that are defined in
2039: terms of existing definitions. As you create each definition you can
2040: test and debug it interactively.
2041:
2042: If you have tried out the examples in this section, you will probably
2043: have typed them in by hand; when you leave Gforth, your definitions will
2044: be lost. You can avoid this by using a text editor to enter Forth source
2045: code into a file, and then loading code from the file using
2046: @code{include} (@xref{Forth source files}). A Forth source file is
2047: processed by the text interpreter, just as though you had typed it in by
2048: hand@footnote{Actually, there are some subtle differences -- see
2049: @ref{The Text Interpreter}.}.
2050:
2051: Gforth also supports the traditional Forth alternative to using text
2052: files for program entry (@xref{Blocks}).
2053:
2054: In common with many, if not most, Forth compilers, most of Gforth is
2055: actually written in Forth. All of the @file{.fs} files in the
2056: installation directory@footnote{For example,
2057: @file{/usr/local/share/gforth...}} are Forth source files, which you can
2058: study to see examples of Forth programming.
2059:
2060: Gforth maintains a history file that records every line that you type to
2061: the text interpreter. This file is preserved between sessions, and is
2062: used to provide a command-line recall facility. If you enter long
2063: definitions by hand, you can use a text editor to paste them out of the
2064: history file into a Forth source file for reuse at a later time
2065: (@pxref{Command-line editing} for more information).
2066:
2067:
2068: @comment ----------------------------------------------
2069: @node Review - elements of a Forth system, Where to go next, Forth is written in Forth, Introduction
2070: @section Review - elements of a Forth system
2071: @cindex elements of a Forth system
2072:
2073: To summarise this chapter:
2074:
2075: @itemize @bullet
2076: @item
2077: Forth programs use @dfn{factoring} to break a problem down into small
2078: fragments called @dfn{words} or @dfn{definitions}.
2079: @item
2080: Forth program development is an interactive process.
2081: @item
2082: The main command loop that accepts input, and controls both
2083: interpretation and compilation, is called the @dfn{text interpreter}
2084: (also known as the @dfn{outer interpreter}).
2085: @item
2086: Forth has a very simple syntax, consisting of words and numbers
2087: separated by spaces or carriage-return characters. Any additional syntax
2088: is imposed by @dfn{parsing words}.
2089: @item
2090: Forth uses a stack to pass parameters between words. As a result, it
2091: uses postfix notation.
2092: @item
2093: To use a word that has previously been defined, the text interpreter
2094: searches for the word in the @dfn{name dictionary}.
2095: @item
2096: Words have @dfn{interpretation semantics} and @dfn{compilation semantics}.
2097: @item
2098: The text interpreter uses the value of @code{state} to select between
2099: the use of the @dfn{interpretation semantics} and the @dfn{compilation
2100: semantics} of a word that it encounters.
2101: @item
2102: The relationship between the @dfn{interpretation semantics} and
2103: @dfn{compilation semantics} for a word
2104: depend upon the way in which the word was defined (for example, whether
2105: it is an @dfn{immediate} word).
2106: @item
2107: Forth definitions can be implemented in Forth (called @dfn{high-level
2108: definitions}) or in some other way (usually a lower-level language and
2109: as a result often called @dfn{low-level definitions}, @dfn{code
2110: definitions} or @dfn{primitives}).
2111: @item
2112: Many Forth systems are implemented mainly in Forth.
2113: @end itemize
2114:
2115:
2116: @comment ----------------------------------------------
2117: @node Where to go next,Exercises,Review - elements of a Forth system, Introduction
2118: @section Where To Go Next
2119: @cindex where to go next
2120:
2121: Amazing as it may seem, if you have read (and understood) this far, you
2122: know almost all the fundamentals about the inner workings of a Forth
2123: system. You certainly know enough to be able to read and understand the
2124: rest of this manual and the ANS Forth document, to learn more about the
2125: facilities that Forth in general and Gforth in particular provide. Even
2126: scarier, you know almost enough to implement your own Forth system.
2127: However, that's not a good idea just yet... better to try writing some
2128: programs in Gforth.
2129:
2130: Forth has such a rich vocabulary that it can be hard to know where to
2131: start in learning it. This section suggests a few sets of words that are
2132: enough to write small but useful programs. Use the word index in this
2133: document to learn more about each word, then try it out and try to write
2134: small definitions using it. Start by experimenting with these words:
2135:
2136: @itemize @bullet
2137: @item
2138: Arithmetic: @code{+ - * / /MOD */ ABS INVERT}
2139: @item
2140: Comparison: @code{MIN MAX =}
2141: @item
2142: Logic: @code{AND OR XOR NOT}
2143: @item
2144: Stack manipulation: @code{DUP DROP SWAP OVER}
2145: @item
2146: Loops and decisions: @code{IF ELSE ENDIF ?DO I LOOP}
2147: @item
2148: Input/Output: @code{. ." EMIT CR KEY}
2149: @item
2150: Defining words: @code{: ; CREATE}
2151: @item
2152: Memory allocation words: @code{ALLOT ,}
2153: @item
2154: Tools: @code{SEE WORDS .S MARKER}
2155: @end itemize
2156:
2157: When you have mastered those, go on to:
2158:
2159: @itemize @bullet
2160: @item
2161: More defining words: @code{VARIABLE CONSTANT VALUE TO CREATE DOES>}
2162: @item
2163: Memory access: @code{@@ !}
2164: @end itemize
2165:
2166: When you have mastered these, there's nothing for it but to read through
2167: the whole of this manual and find out what you've missed.
2168:
2169: @comment ----------------------------------------------
2170: @node Exercises, ,Where to go next, Introduction
2171: @section Exercises
2172: @cindex exercises
2173:
2174: TODO: provide a set of programming excercises linked into the stuff done
2175: already and into other sections of the manual. Provide solutions to all
2176: the exercises in a .fs file in the distribution.
2177:
2178: @c Get some inspiration from Starting Forth and Kelly&Spies.
2179:
2180: @c excercises:
2181: @c 1. take inches and convert to feet and inches.
2182: @c 2. take temperature and convert from fahrenheight to celcius;
2183: @c may need to care about symmetric vs floored??
2184: @c 3. take input line and do character substitution
2185: @c to encipher or decipher
2186: @c 4. as above but work on a file for in and out
2187: @c 5. take input line and convert to pig-latin
2188: @c
2189: @c thing of sets of things to exercise then come up with
2190: @c problems that need those things.
2191:
2192:
2193: @c ******************************************************************
2194: @node Words, Error messages, Introduction, Top
2195: @chapter Forth Words
2196: @cindex words
2197:
2198: @menu
2199: * Notation::
2200: * Comments::
2201: * Boolean Flags::
2202: * Arithmetic::
2203: * Stack Manipulation::
2204: * Memory::
2205: * Control Structures::
2206: * Defining Words::
2207: * The Text Interpreter::
2208: * Tokens for Words::
2209: * Word Lists::
2210: * Environmental Queries::
2211: * Files::
2212: * Blocks::
2213: * Other I/O::
2214: * Programming Tools::
2215: * Assembler and Code Words::
2216: * Threading Words::
2217: * Locals::
2218: * Structures::
2219: * Object-oriented Forth::
2220: * Passing Commands to the OS::
2221: * Miscellaneous Words::
2222: @end menu
2223:
2224: @node Notation, Comments, Words, Words
2225: @section Notation
2226: @cindex notation of glossary entries
2227: @cindex format of glossary entries
2228: @cindex glossary notation format
2229: @cindex word glossary entry format
2230:
2231: The Forth words are described in this section in the glossary notation
2232: that has become a de-facto standard for Forth texts, i.e.,
2233:
2234: @format
2235: @i{word} @i{Stack effect} @i{wordset} @i{pronunciation}
2236: @end format
2237: @i{Description}
2238:
2239: @table @var
2240: @item word
2241: The name of the word.
2242:
2243: @item Stack effect
2244: @cindex stack effect
2245: The stack effect is written in the notation @code{@i{before} --
2246: @i{after}}, where @i{before} and @i{after} describe the top of
2247: stack entries before and after the execution of the word. The rest of
2248: the stack is not touched by the word. The top of stack is rightmost,
2249: i.e., a stack sequence is written as it is typed in. Note that Gforth
2250: uses a separate floating point stack, but a unified stack
2251: notation. Also, return stack effects are not shown in @i{stack
2252: effect}, but in @i{Description}. The name of a stack item describes
2253: the type and/or the function of the item. See below for a discussion of
2254: the types.
2255:
2256: All words have two stack effects: A compile-time stack effect and a
2257: run-time stack effect. The compile-time stack-effect of most words is
2258: @i{ -- }. If the compile-time stack-effect of a word deviates from
2259: this standard behaviour, or the word does other unusual things at
2260: compile time, both stack effects are shown; otherwise only the run-time
2261: stack effect is shown.
2262:
2263: @cindex pronounciation of words
2264: @item pronunciation
2265: How the word is pronounced.
2266:
2267: @cindex wordset
2268: @item wordset
2269: The ANS Forth standard is divided into several word sets. A standard
2270: system need not support all of them. Therefore, in theory, the fewer
2271: word sets your program uses the more portable it will be. However, we
2272: suspect that most ANS Forth systems on personal machines will feature
2273: all word sets. Words that are not defined in ANS Forth have
2274: @code{gforth} or @code{gforth-internal} as word set. @code{gforth}
2275: describes words that will work in future releases of Gforth;
2276: @code{gforth-internal} words are more volatile. Environmental query
2277: strings are also displayed like words; you can recognize them by the
2278: @code{environment} in the word set field.
2279:
2280: @item Description
2281: A description of the behaviour of the word.
2282: @end table
2283:
2284: @cindex types of stack items
2285: @cindex stack item types
2286: The type of a stack item is specified by the character(s) the name
2287: starts with:
2288:
2289: @table @code
2290: @item f
2291: @cindex @code{f}, stack item type
2292: Boolean flags, i.e. @code{false} or @code{true}.
2293: @item c
2294: @cindex @code{c}, stack item type
2295: Char
2296: @item w
2297: @cindex @code{w}, stack item type
2298: Cell, can contain an integer or an address
2299: @item n
2300: @cindex @code{n}, stack item type
2301: signed integer
2302: @item u
2303: @cindex @code{u}, stack item type
2304: unsigned integer
2305: @item d
2306: @cindex @code{d}, stack item type
2307: double sized signed integer
2308: @item ud
2309: @cindex @code{ud}, stack item type
2310: double sized unsigned integer
2311: @item r
2312: @cindex @code{r}, stack item type
2313: Float (on the FP stack)
2314: @item a-
2315: @cindex @code{a_}, stack item type
2316: Cell-aligned address
2317: @item c-
2318: @cindex @code{c_}, stack item type
2319: Char-aligned address (note that a Char may have two bytes in Windows NT)
2320: @item f-
2321: @cindex @code{f_}, stack item type
2322: Float-aligned address
2323: @item df-
2324: @cindex @code{df_}, stack item type
2325: Address aligned for IEEE double precision float
2326: @item sf-
2327: @cindex @code{sf_}, stack item type
2328: Address aligned for IEEE single precision float
2329: @item xt
2330: @cindex @code{xt}, stack item type
2331: Execution token, same size as Cell
2332: @item wid
2333: @cindex @code{wid}, stack item type
2334: Word list ID, same size as Cell
2335: @item f83name
2336: @cindex @code{f83name}, stack item type
2337: Pointer to a name structure
2338: @item "
2339: @cindex @code{"}, stack item type
2340: string in the input stream (not on the stack). The terminating character
2341: is a blank by default. If it is not a blank, it is shown in @code{<>}
2342: quotes.
2343: @end table
2344:
2345: @node Comments, Boolean Flags, Notation, Words
2346: @section Comments
2347: @cindex comments
2348:
2349: Forth supports two styles of comment; the traditional @i{in-line} comment,
2350: @code{(} and its modern cousin, the @i{comment to end of line}; @code{\}.
2351:
2352: doc-(
2353: doc-\
2354: doc-\G
2355:
2356: @node Boolean Flags, Arithmetic, Comments, Words
2357: @section Boolean Flags
2358: @cindex Boolean flags
2359:
2360: A Boolean flag is cell-sized. A cell with all bits clear represents the
2361: flag @code{false} and a flag with all bits set represents the flag
2362: @code{true}. Words that check a flag (for example, @code{IF}) will treat
2363: a cell that has @i{any} bit set as @code{true}.
2364:
2365: doc-true
2366: doc-false
2367: doc-on
2368: doc-off
2369:
2370: @node Arithmetic, Stack Manipulation, Boolean Flags, Words
2371: @section Arithmetic
2372: @cindex arithmetic words
2373:
2374: @cindex division with potentially negative operands
2375: Forth arithmetic is not checked, i.e., you will not hear about integer
2376: overflow on addition or multiplication, you may hear about division by
2377: zero if you are lucky. The operator is written after the operands, but
2378: the operands are still in the original order. I.e., the infix @code{2-1}
2379: corresponds to @code{2 1 -}. Forth offers a variety of division
2380: operators. If you perform division with potentially negative operands,
2381: you do not want to use @code{/} or @code{/mod} with its undefined
2382: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
2383: former, @pxref{Mixed precision}).
2384: @comment TODO discuss the different division forms and the std approach
2385:
2386: @menu
2387: * Single precision::
2388: * Bitwise operations::
2389: * Double precision:: Double-cell integer arithmetic
2390: * Numeric comparison::
2391: * Mixed precision:: Operations with single and double-cell integers
2392: * Floating Point::
2393: @end menu
2394:
2395: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
2396: @subsection Single precision
2397: @cindex single precision arithmetic words
2398:
2399: By default, numbers in Forth are single-precision integers that are 1
2400: cell in size. They can be signed or unsigned, depending upon how you
2401: treat them. @xref{Number Conversion} for the rules used by the text
2402: interpreter for recognising single-precision integers.
2403:
2404: doc-+
2405: doc-1+
2406: doc--
2407: doc-1-
2408: doc-*
2409: doc-/
2410: doc-mod
2411: doc-/mod
2412: doc-negate
2413: doc-abs
2414: doc-min
2415: doc-max
2416: doc-d>s
2417: doc-floored
2418:
2419: @node Bitwise operations, Double precision, Single precision, Arithmetic
2420: @subsection Bitwise operations
2421: @cindex bitwise operation words
2422:
2423: doc-and
2424: doc-or
2425: doc-xor
2426: doc-invert
2427: doc-lshift
2428: doc-rshift
2429: doc-2*
2430: doc-d2*
2431: doc-2/
2432: doc-d2/
2433:
2434: @node Double precision, Numeric comparison, Bitwise operations, Arithmetic
2435: @subsection Double precision
2436: @cindex double precision arithmetic words
2437:
2438: @xref{Number Conversion} for the rules used by the text interpreter for
2439: recognising double-precision integers.
2440:
2441: A double precision number is represented by a cell pair, with the most
2442: significant cell at the TOS. It is trivial to convert an unsigned
2443: single to an (unsigned) double; simply push a @code{0} onto the
2444: TOS. Since numbers are represented by Gforth using 2's complement
2445: arithmetic, converting a signed single to a (signed) double requires
2446: sign-extension across the most significant cell. This can be achieved
2447: using @code{s>d}. The moral of the story is that you cannot convert a
2448: number without knowing whether it represents an unsigned or a
2449: signed number.
2450:
2451: doc-s>d
2452: doc-d+
2453: doc-d-
2454: doc-dnegate
2455: doc-dabs
2456: doc-dmin
2457: doc-dmax
2458:
2459: @node Numeric comparison, Mixed precision, Double precision, Arithmetic
2460: @subsection Numeric comparison
2461: @cindex numeric comparison words
2462:
2463: doc-<
2464: doc-<=
2465: doc-<>
2466: doc-=
2467: doc->
2468: doc->=
2469:
2470: doc-0<
2471: doc-0<=
2472: doc-0<>
2473: doc-0=
2474: doc-0>
2475: doc-0>=
2476:
2477: doc-u<
2478: doc-u<=
2479: @c TODO why u<> and u= ... they are the same as <> and =
2480: @c commented them out because they are unnecessary
2481: @c doc-u<>
2482: @c doc-u=
2483: doc-u>
2484: doc-u>=
2485:
2486: doc-within
2487:
2488: doc-d<
2489: doc-d<=
2490: doc-d<>
2491: doc-d=
2492: doc-d>
2493: doc-d>=
2494:
2495: doc-d0<
2496: doc-d0<=
2497: doc-d0<>
2498: doc-d0=
2499: doc-d0>
2500: doc-d0>=
2501:
2502: doc-du<
2503: doc-du<=
2504: @c doc-du<>
2505: @c doc-du=
2506: doc-du>
2507: doc-du>=
2508:
2509: @node Mixed precision, Floating Point, Numeric comparison, Arithmetic
2510: @subsection Mixed precision
2511: @cindex mixed precision arithmetic words
2512:
2513: doc-m+
2514: doc-*/
2515: doc-*/mod
2516: doc-m*
2517: doc-um*
2518: doc-m*/
2519: doc-um/mod
2520: doc-fm/mod
2521: doc-sm/rem
2522:
2523: @node Floating Point, , Mixed precision, Arithmetic
2524: @subsection Floating Point
2525: @cindex floating point arithmetic words
2526:
2527: @xref{Number Conversion} for the rules used by the text interpreter for
2528: recognising floating-point numbers.
2529:
2530: @cindex angles in trigonometric operations
2531: @cindex trigonometric operations
2532: Angles in floating point operations are given in radians (a full circle
2533: has 2 pi radians). Gforth has a separate floating point
2534: stack, but the documentation uses the unified notation.
2535:
2536: @cindex floating-point arithmetic, pitfalls
2537: Floating point numbers have a number of unpleasant surprises for the
2538: unwary (e.g., floating point addition is not associative) and even a few
2539: for the wary. You should not use them unless you know what you are doing
2540: or you don't care that the results you get are totally bogus. If you
2541: want to learn about the problems of floating point numbers (and how to
2542: avoid them), you might start with @cite{David Goldberg, What Every
2543: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
2544: Computing Surveys 23(1):5@minus{}48, March 1991}
2545: (@url{http://www.validgh.com/goldberg/paper.ps}).
2546:
2547: doc-d>f
2548: doc-f>d
2549: doc-f+
2550: doc-f-
2551: doc-f*
2552: doc-f/
2553: doc-fnegate
2554: doc-fabs
2555: doc-fmax
2556: doc-fmin
2557: doc-floor
2558: doc-fround
2559: doc-f**
2560: doc-fsqrt
2561: doc-fexp
2562: doc-fexpm1
2563: doc-fln
2564: doc-flnp1
2565: doc-flog
2566: doc-falog
2567: doc-fsin
2568: doc-fcos
2569: doc-fsincos
2570: doc-ftan
2571: doc-fasin
2572: doc-facos
2573: doc-fatan
2574: doc-fatan2
2575: doc-fsinh
2576: doc-fcosh
2577: doc-ftanh
2578: doc-fasinh
2579: doc-facosh
2580: doc-fatanh
2581: doc-pi
2582:
2583: doc-f2*
2584: doc-f2/
2585: doc-1/f
2586: doc-precision
2587: doc-set-precision
2588:
2589: One particular problem with floating-point arithmetic is that comparison
2590: for equality often fails when you would expect it to succeed. For this
2591: reason approximate equality is often preferred (but you still have to
2592: know what you are doing). The comparison words are:
2593:
2594: doc-f~rel
2595: doc-f~abs
2596: doc-f=
2597: doc-f~
2598: doc-f<>
2599:
2600: doc-f<
2601: doc-f<=
2602: doc-f>
2603: doc-f>=
2604:
2605: doc-f0<
2606: doc-f0<=
2607: doc-f0<>
2608: doc-f0=
2609: doc-f0>
2610: doc-f0>=
2611:
2612:
2613: @node Stack Manipulation, Memory, Arithmetic, Words
2614: @section Stack Manipulation
2615: @cindex stack manipulation words
2616:
2617: @cindex floating-point stack in the standard
2618: Gforth maintains a number of separate stacks:
2619:
2620: @cindex data stack
2621: @cindex parameter stack
2622: @itemize @bullet
2623: @item
2624: A data stack (also known as the @dfn{parameter stack}) -- for
2625: characters, cells, addresses, and double cells.
2626:
2627: @cindex floating-point stack
2628: @item
2629: A floating point stack -- for floating point numbers.
2630:
2631: @cindex return stack
2632: @item
2633: A return stack -- for storing the return addresses of colon
2634: definitions and other data.
2635:
2636: @cindex locals stack
2637: @item
2638: A locals stack for storing local variables.
2639: @end itemize
2640:
2641: Whilst every sane Forth has a separate floating-point stack, it is not
2642: strictly required; an ANS Forth system could theoretically keep
2643: floating-point numbers on the data stack. As an additional difficulty,
2644: you don't know how many cells a floating-point number takes. It is
2645: reportedly possible to write words in a way that they work also for a
2646: unified stack model, but we do not recommend trying it. Instead, just
2647: say that your program has an environmental dependency on a separate
2648: floating-point stack.
2649:
2650: doc-floating-stack
2651:
2652: @cindex return stack and locals
2653: @cindex locals and return stack
2654: A Forth system is allowed to keep local variables on the
2655: return stack. This is reasonable, as local variables usually eliminate
2656: the need to use the return stack explicitly. So, if you want to produce
2657: a standard compliant program and you are using local variables in a
2658: word, forget about return stack manipulations in that word (refer to the
2659: standard document for the exact rules).
2660:
2661: @menu
2662: * Data stack::
2663: * Floating point stack::
2664: * Return stack::
2665: * Locals stack::
2666: * Stack pointer manipulation::
2667: @end menu
2668:
2669: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
2670: @subsection Data stack
2671: @cindex data stack manipulation words
2672: @cindex stack manipulations words, data stack
2673:
2674: doc-drop
2675: doc-nip
2676: doc-dup
2677: doc-over
2678: doc-tuck
2679: doc-swap
2680: doc-pick
2681: doc-rot
2682: doc--rot
2683: doc-?dup
2684: doc-roll
2685: doc-2drop
2686: doc-2nip
2687: doc-2dup
2688: doc-2over
2689: doc-2tuck
2690: doc-2swap
2691: doc-2rot
2692:
2693: @node Floating point stack, Return stack, Data stack, Stack Manipulation
2694: @subsection Floating point stack
2695: @cindex floating-point stack manipulation words
2696: @cindex stack manipulation words, floating-point stack
2697:
2698: doc-fdrop
2699: doc-fnip
2700: doc-fdup
2701: doc-fover
2702: doc-ftuck
2703: doc-fswap
2704: doc-fpick
2705: doc-frot
2706:
2707: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
2708: @subsection Return stack
2709: @cindex return stack manipulation words
2710: @cindex stack manipulation words, return stack
2711:
2712: doc->r
2713: doc-r>
2714: doc-r@
2715: doc-rdrop
2716: doc-2>r
2717: doc-2r>
2718: doc-2r@
2719: doc-2rdrop
2720:
2721: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
2722: @subsection Locals stack
2723:
2724: @comment TODO
2725:
2726: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
2727: @subsection Stack pointer manipulation
2728: @cindex stack pointer manipulation words
2729:
2730: doc-sp0
2731: doc-s0
2732: doc-sp@
2733: doc-sp!
2734: doc-fp0
2735: doc-fp@
2736: doc-fp!
2737: doc-rp0
2738: doc-r0
2739: doc-rp@
2740: doc-rp!
2741: doc-lp0
2742: doc-l0
2743: doc-lp@
2744: doc-lp!
2745:
2746: @node Memory, Control Structures, Stack Manipulation, Words
2747: @section Memory
2748: @cindex memory words
2749:
2750: @cindex dictionary
2751: Forth definitions are organised in memory structures that are
2752: collectively called the @dfn{dictionary}. The dictionary can be
2753: considered as three logical memory regions:
2754:
2755: @itemize @bullet
2756: @item
2757: @cindex code space
2758: @cindex code dictionary
2759: Code space, also known as the @dfn{code dictionary}.
2760: @item
2761: @cindex name space
2762: @cindex name dictionary
2763: Name space, also known as the @dfn{name dictionary}@footnote{Sometimes,
2764: the term @dfn{dictionary} is used simply to refer to the name
2765: dictionary, because it is the one region that is used for looking up
2766: names, just as you would in a conventional dictionary.}.
2767: @item
2768: @cindex data space
2769: Data space
2770: @end itemize
2771:
2772: When you create a colon definition, the text interpreter compiles the
2773: code for the definition into the code dictionary and compiles the name
2774: of the definition into the name dictionary, together with other
2775: information about the definition (such as its execution token).
2776:
2777: When you create a variable, the execution of @code{variable} will
2778: compile some code, assign once cell in data space, and compile the name
2779: of the variable into the name dictionary.
2780:
2781: @cindex memory regions - relationship between them
2782: ANS Forth does not specify the relationship between the three memory
2783: regions, and specifies that a Standard program must not access code or
2784: data space directly -- it may only access data space directly. In
2785: addition, the Standard defines what relationships you may and may not
2786: rely on when allocating regions in data space. These constraints are
2787: simply a reflection of the many diverse techniques that are used to
2788: implement Forth systems; understanding and following the requirements of
2789: the Standard allows you to write portable programs -- programs that run
2790: in the same way on any of these diverse systems. Another way of looking
2791: at this is to say that ANS Forth was designed to permit compliant Forth
2792: systems to be implemented in many diverse ways.
2793:
2794: @cindex memory regions - how they are assigned
2795: Here are some examples of ways in which name, code and data spaces
2796: might be assigned in different Forth implementations:
2797:
2798: @itemize @bullet
2799: @item
2800: For a Forth system that runs from RAM under a general-purpose operating
2801: system, it can be convenient to interleave name, code and data spaces in
2802: a single contiguous memory region. This organisation can be
2803: memory-efficient (for example, because the relationship between the name
2804: dictionary entry and the associated code dictionary entry can be
2805: implicit, rather than requiring an explicit memory pointer to reference
2806: from the name dictionary and the code dictionary). This is the
2807: organisation used by Gforth, as this example@footnote{The addresses
2808: in the example have been truncated to fit it onto the page, and the
2809: addresses and data shown will not match the output from your system} shows:
2810: @example
2811: hex
2812: variable fred 123456 fred !
2813: variable jim abcd jim !
2814: : foo + / - ;
2815: ' fred 10 - 50 dump
2816: ..80: 5C 46 0E 40 84 66 72 65 - 64 20 20 20 20 20 20 20 \F.@.fred
2817: ..90: D0 9B 04 08 00 00 00 00 - 56 34 12 00 80 46 0E 40 ........V4...F.@
2818: ..A0: 83 6A 69 6D 20 20 20 20 - D0 9B 04 08 00 00 00 00 .jim ........
2819: ..B0: CD AB 00 00 9C 46 0E 40 - 83 66 6F 6F 20 20 20 20 .....F.@.foo
2820: ..C0: 80 9B 04 08 00 00 00 00 - E4 2E 05 08 0C 2F 05 08 ............./..
2821: @end example
2822:
2823: @item
2824: For a high-performance system running on a modern RISC processor with a
2825: modified Harvard architecture (one that has a unified main memory but
2826: separate instruction and data caches), it is desirable to separate
2827: processor instructions from processor data. This encourages a high cache
2828: density and therefore a high cache hit rate. The Forth code dictionary
2829: is not necessarily made up entirely of processor instructions; its
2830: nature is dependent upon the Forth implementation.
2831:
2832: @item
2833: A Forth compiler that runs on a segmented 8086 processor could be
2834: designed to interleave the name, code and data spaces within a single
2835: 64Kbyte segment. A more common implementation choice is to use a
2836: separate 64Kbyte segment for each region, which provides more memory
2837: overall but provides an address map in which only the data space is
2838: accessible.
2839:
2840: @item
2841: Microprocessors exist that run Forth (or many of the primitives required
2842: to implement the Forth virtual machine efficiently) directly. On these
2843: processors, the relationship between name, code and data spaces may be
2844: imposed as a side-effect of the microarchitecture of the processor.
2845:
2846: @item
2847: A Forth compiler that executes from ROM on an embedded system needs its
2848: data space separated from the name and code spaces so that the data
2849: space can be mapped to a RAM area.
2850:
2851: @item
2852: A Forth compiler that runs on an embedded system may have a requirement
2853: for a small memory footprint. On such a system it can be useful to
2854: separate the name space from the data and code spaces; once the
2855: application has been compiled, the name dictionary is no longer
2856: required@footnote{more strictly speaking, most applications can be
2857: designed so that this is the case}. The name dictionary can be deleted
2858: entirely, or could be stored in memory on a remote @i{host} system for
2859: debug and development purposes. In the latter case, the compiler running
2860: on the @i{target} system could implement a protocol across a
2861: communication link that would allow it to interrogate the name dictionary.
2862: @end itemize
2863:
2864: @menu
2865: * Reserving Data Space::
2866: * Memory Access::
2867: * Address Arithmetic::
2868: * Memory Blocks::
2869: * Dynamic Allocation::
2870: @end menu
2871:
2872:
2873: @node Reserving Data Space, Memory Access, Memory, Memory
2874: @subsection Reserving Data Space
2875: @cindex reserving data space
2876: @cindex data space - reserving some
2877:
2878: @cindex data space pointer - contiguous regions
2879: Data space may be reserved as individual chars or cells or in contiguous
2880: regions. These are the rules for reserving contiguous regions in a
2881: Standard (i.e., portable) way:
2882: @itemize @bullet
2883: @item
2884: The value of the data-space pointer, @code{here}, always defines the
2885: beginning of a contiguous region of data space.
2886:
2887: @item
2888: @code{CREATE} establishes the beginning of a contiguous region of data
2889: space (the @code{CREATE}d definition returns the initial address of the
2890: region).
2891:
2892: @item
2893: @code{variable} does @i{not} establish the beginning of a contiguous
2894: region in data space; @code{variable} followed by @code{allot} is not
2895: guaranteed to allocate data space region that is contiguous with the
2896: storage allocated by @code{variable}. Instead, use @code{create} --
2897: @xref{Simple Defining Words} for examples.
2898:
2899: @item
2900: Successive calls to @code{allot}, @code{,} (comma), @code{2,} (2-comma),
2901: @code{c,} (c-comma) and @code{align} reserve a single contiguous region
2902: in data space. The contiguity of the region is interrupted by compiling
2903: (or removing) definitions from the dictionary.
2904:
2905: @item
2906: The most recently reserved contiguous region may be released by calling
2907: @code{allot} with a negative argument, provided that the region has not
2908: been interrupted by compiling (or removing) definitions from the
2909: dictionary.
2910: @end itemize
2911:
2912: @cindex data space pointer - alignment
2913: These factors affect the alignment of @code{here}, the data
2914: space pointer:
2915:
2916: @itemize @bullet
2917: @item
2918: If the data-space pointer is aligned@footnote{In ANS Forth-speak,
2919: @i{aligned} implictly means @code{CELL}-aligned.} before an
2920: @code{allot}, and a whole number of characters are reserved or released, it
2921: will remain aligned after the @code{allot}.
2922:
2923: @item
2924: If the data-space pointer is character-aligned before an @code{allot},
2925: and a whole number of cells are reserved or released, it will remain
2926: character-aligned after the @code{allot}.
2927:
2928: @item
2929: The initial contents of data space reserved using @code{allot} is
2930: undefined.
2931:
2932: @item
2933: Definitions created by @code{create}, @code{variable}, @code{2variable}
2934: return aligned addresses.
2935:
2936: @item
2937: After a definition is compiled or @code{align} is executed, the data
2938: space pointer is guaranteed to be aligned.
2939: @end itemize
2940:
2941: doc-here
2942: doc-unused
2943: doc-allot
2944: doc-c,
2945: doc-f,
2946: doc-,
2947: doc-2,
2948: @cindex user space
2949: doc-udp
2950: doc-uallot
2951:
2952:
2953: @node Memory Access, Address Arithmetic, Reserving Data Space, Memory
2954: @subsection Memory Access
2955: @cindex memory access words
2956:
2957: doc-@
2958: doc-!
2959: doc-+!
2960: doc-c@
2961: doc-c!
2962: doc-2@
2963: doc-2!
2964: doc-f@
2965: doc-f!
2966: doc-sf@
2967: doc-sf!
2968: doc-df@
2969: doc-df!
2970:
2971: @node Address Arithmetic, Memory Blocks, Memory Access, Memory
2972: @subsection Address Arithmetic
2973: @cindex address arithmetic words
2974:
2975: ANS Forth does not specify the sizes of the data types. Instead, it
2976: offers a number of words for computing sizes and doing address
2977: arithmetic. Address arithmetic is performed in terms of address units
2978: (aus); on most systems the address unit is one byte. Note that a
2979: character may have more than one au, so @code{chars} is no noop (on
2980: systems where it is a noop, it compiles to nothing).
2981:
2982: @cindex alignment of addresses for types
2983: ANS Forth also defines words for aligning addresses for specific
2984: types. Many computers require that accesses to specific data types
2985: must only occur at specific addresses; e.g., that cells may only be
2986: accessed at addresses divisible by 4. Even if a machine allows unaligned
2987: accesses, it can usually perform aligned accesses faster.
2988:
2989: For the performance-conscious: alignment operations are usually only
2990: necessary during the definition of a data structure, not during the
2991: (more frequent) accesses to it.
2992:
2993: ANS Forth defines no words for character-aligning addresses. This is not
2994: an oversight, but reflects the fact that addresses that are not
2995: char-aligned have no use in the standard and therefore will not be
2996: created.
2997:
2998: @cindex @code{CREATE} and alignment
2999: ANS Forth guarantees that addresses returned by @code{CREATE}d words
3000: are cell-aligned; in addition, Gforth guarantees that these addresses
3001: are aligned for all purposes.
3002:
3003: Note that the ANS Forth word @code{char} has nothing to do with address
3004: arithmetic.
3005:
3006: doc-chars
3007: doc-char+
3008: doc-cells
3009: doc-cell+
3010: doc-cell
3011: doc-align
3012: doc-aligned
3013: doc-floats
3014: doc-float+
3015: doc-float
3016: doc-falign
3017: doc-faligned
3018: doc-sfloats
3019: doc-sfloat+
3020: doc-sfalign
3021: doc-sfaligned
3022: doc-dfloats
3023: doc-dfloat+
3024: doc-dfalign
3025: doc-dfaligned
3026: doc-maxalign
3027: doc-maxaligned
3028: doc-cfalign
3029: doc-cfaligned
3030: doc-address-unit-bits
3031:
3032: @node Memory Blocks, Dynamic Allocation, Address Arithmetic, Memory
3033: @subsection Memory Blocks
3034: @cindex memory block words
3035: @cindex character strings - moving and copying
3036:
3037: Memory blocks often represent character strings; @xref{String Formats}
3038: for ways of storing character strings in memory. @xref{Displaying
3039: characters and strings} for other string-processing words.
3040:
3041: Some of these words work on address units (increments of @code{CELL}),
3042: and expect a @code{CELL}-aligned address. Others work on character units
3043: (increments of @code{CHAR}), and expect a @code{CHAR}-aligned
3044: address. Choose the correct operation depending upon your data type. If
3045: you are moving a block of memory (for example, a region reserved by
3046: @code{allot}) it is safe to use @code{move}, and it should be faster
3047: than using @code{cmove}. If you are moving (for example) a string
3048: compiled using @code{S"}, it is not portable to use @code{move}; the
3049: alignment of the string in memory could change, and the relationship
3050: between @code{CELL} and @code{CHAR} could change.
3051:
3052: When copying characters between overlapping memory regions, choose
3053: carefully between @code{cmove} and @code{cmove>}.
3054:
3055: You can only use any of these words @i{portably} to access data space.
3056:
3057: @comment TODO - think the naming of the arguments is wrong for move
3058: @comment well, really it seems to be the Standard that's wrong; it
3059: @comment describes MOVE as a word that requires a CELL-aligned source
3060: @comment and destination address but a xtranfer count that need not
3061: @comment be a multiple of CELL.
3062: doc-move
3063: doc-erase
3064: doc-cmove
3065: doc-cmove>
3066: doc-fill
3067: doc-blank
3068: doc-compare
3069: doc-search
3070: doc--trailing
3071: doc-/string
3072:
3073: @comment TODO examples
3074:
3075: @node Dynamic Allocation, ,Memory Blocks, Memory
3076: @subsection Dynamic Allocation of Memory
3077: @cindex dynamic allocation of memory
3078: @cindex memory-allocation word set
3079:
3080: The ANS Forth memory-allocation word set allows memory regions to be
3081: dynamically assigned, resized and released without affecting the data
3082: space pointer. In Gforth, these words are implemented using
3083: the standard C library calls malloc(), free() and resize().
3084:
3085: doc-allocate
3086: doc-free
3087: doc-resize
3088:
3089:
3090: @node Control Structures, Defining Words, Memory, Words
3091: @section Control Structures
3092: @cindex control structures
3093:
3094: Control structures in Forth cannot be used in interpret state, only in
3095: compile state@footnote{To be precise, they have no interpretation
3096: semantics (@pxref{Interpretation and Compilation Semantics}).}, i.e., in
3097: a colon definition. We do not like this limitation, but have not seen a
3098: satisfying way around it yet, although many schemes have been proposed.
3099:
3100: @menu
3101: * Selection:: IF.. ELSE.. ENDIF
3102: * Simple Loops:: BEGIN..
3103: * Counted Loops:: DO
3104: * Arbitrary control structures::
3105: * Calls and returns::
3106: * Exception Handling::
3107: @end menu
3108:
3109: @node Selection, Simple Loops, Control Structures, Control Structures
3110: @subsection Selection
3111: @cindex selection control structures
3112: @cindex control structures for selection
3113:
3114: @cindex @code{IF} control structure
3115: @example
3116: @i{flag}
3117: IF
3118: @i{code}
3119: ENDIF
3120: @end example
3121: @noindent
3122: or
3123: @example
3124: @i{flag}
3125: IF
3126: @i{code1}
3127: ELSE
3128: @i{code2}
3129: ENDIF
3130: @end example
3131:
3132: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
3133: standard, and @code{ENDIF} is not, although it is quite popular. We
3134: recommend using @code{ENDIF}, because it is less confusing for people
3135: who also know other languages (and is not prone to reinforcing negative
3136: prejudices against Forth in these people). Adding @code{ENDIF} to a
3137: system that only supplies @code{THEN} is simple:
3138: @example
3139: : ENDIF POSTPONE THEN ; immediate
3140: @end example
3141:
3142: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
3143: (adv.)} has the following meanings:
3144: @quotation
3145: ... 2b: following next after in order ... 3d: as a necessary consequence
3146: (if you were there, then you saw them).
3147: @end quotation
3148: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
3149: and many other programming languages has the meaning 3d.]
3150:
3151: Gforth also provides the words @code{?DUP-IF} and @code{?DUP-0=-IF}, so
3152: you can avoid using @code{?dup}. Using these alternatives is also more
3153: efficient than using @code{?dup}. Definitions in ANS Forth
3154: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
3155: @file{compat/control.fs}.
3156:
3157: @cindex @code{CASE} control structure
3158: @example
3159: @i{n}
3160: CASE
3161: @i{n1} OF @i{code1} ENDOF
3162: @i{n2} OF @i{code2} ENDOF
3163: @dots{}
3164: ENDCASE
3165: @end example
3166:
3167: Executes the first @i{codei}, where the @i{ni} is equal to
3168: @i{n}. A default case can be added by simply writing the code after
3169: the last @code{ENDOF}. It may use @i{n}, which is on top of the stack,
3170: but must not consume it.
3171:
3172: @node Simple Loops, Counted Loops, Selection, Control Structures
3173: @subsection Simple Loops
3174: @cindex simple loops
3175: @cindex loops without count
3176:
3177: @cindex @code{WHILE} loop
3178: @example
3179: BEGIN
3180: @i{code1}
3181: @i{flag}
3182: WHILE
3183: @i{code2}
3184: REPEAT
3185: @end example
3186:
3187: @i{code1} is executed and @i{flag} is computed. If it is true,
3188: @i{code2} is executed and the loop is restarted; If @i{flag} is
3189: false, execution continues after the @code{REPEAT}.
3190:
3191: @cindex @code{UNTIL} loop
3192: @example
3193: BEGIN
3194: @i{code}
3195: @i{flag}
3196: UNTIL
3197: @end example
3198:
3199: @i{code} is executed. The loop is restarted if @code{flag} is false.
3200:
3201: @cindex endless loop
3202: @cindex loops, endless
3203: @example
3204: BEGIN
3205: @i{code}
3206: AGAIN
3207: @end example
3208:
3209: This is an endless loop.
3210:
3211: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
3212: @subsection Counted Loops
3213: @cindex counted loops
3214: @cindex loops, counted
3215: @cindex @code{DO} loops
3216:
3217: The basic counted loop is:
3218: @example
3219: @i{limit} @i{start}
3220: ?DO
3221: @i{body}
3222: LOOP
3223: @end example
3224:
3225: This performs one iteration for every integer, starting from @i{start}
3226: and up to, but excluding @i{limit}. The counter, or @i{index}, can be
3227: accessed with @code{i}. For example, the loop:
3228: @example
3229: 10 0 ?DO
3230: i .
3231: LOOP
3232: @end example
3233: @noindent
3234: prints @code{0 1 2 3 4 5 6 7 8 9}
3235:
3236: The index of the innermost loop can be accessed with @code{i}, the index
3237: of the next loop with @code{j}, and the index of the third loop with
3238: @code{k}.
3239:
3240: doc-i
3241: doc-j
3242: doc-k
3243:
3244: The loop control data are kept on the return stack, so there are some
3245: restrictions on mixing return stack accesses and counted loop words. In
3246: particuler, if you put values on the return stack outside the loop, you
3247: cannot read them inside the loop@footnote{well, not in a way that is
3248: portable.}. If you put values on the return stack within a loop, you
3249: have to remove them before the end of the loop and before accessing the
3250: index of the loop.
3251:
3252: There are several variations on the counted loop:
3253:
3254: @itemize @bullet
3255: @item
3256: @code{LEAVE} leaves the innermost counted loop immediately; execution
3257: continues after the associated @code{LOOP} or @code{NEXT}. For example:
3258:
3259: @example
3260: 10 0 ?DO i DUP . 3 = IF LEAVE THEN LOOP
3261: @end example
3262: prints @code{0 1 2 3}
3263:
3264:
3265: @item
3266: @code{UNLOOP} prepares for an abnormal loop exit, e.g., via
3267: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
3268: return stack so @code{EXIT} can get to its return address. For example:
3269:
3270: @example
3271: : demo 10 0 ?DO i DUP . 3 = IF UNLOOP EXIT THEN LOOP ." Done" ;
3272: @end example
3273: prints @code{0 1 2 3}
3274:
3275:
3276: @item
3277: If @i{start} is greater than @i{limit}, a @code{?DO} loop is entered
3278: (and @code{LOOP} iterates until they become equal by wrap-around
3279: arithmetic). This behaviour is usually not what you want. Therefore,
3280: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
3281: @code{?DO}), which do not enter the loop if @i{start} is greater than
3282: @i{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
3283: unsigned loop parameters.
3284:
3285: @item
3286: @code{?DO} can be replaced by @code{DO}. @code{DO} always enters
3287: the loop, independent of the loop parameters. Do not use @code{DO}, even
3288: if you know that the loop is entered in any case. Such knowledge tends
3289: to become invalid during maintenance of a program, and then the
3290: @code{DO} will make trouble.
3291:
3292: @item
3293: @code{LOOP} can be replaced with @code{@i{n} +LOOP}; this updates the
3294: index by @i{n} instead of by 1. The loop is terminated when the border
3295: between @i{limit-1} and @i{limit} is crossed. E.g.:
3296:
3297: @example
3298: 4 0 +DO i . 2 +LOOP
3299: @end example
3300: @noindent
3301: prints @code{0 2}
3302:
3303: @example
3304: 4 1 +DO i . 2 +LOOP
3305: @end example
3306: @noindent
3307: prints @code{1 3}
3308:
3309:
3310: @cindex negative increment for counted loops
3311: @cindex counted loops with negative increment
3312: The behaviour of @code{@i{n} +LOOP} is peculiar when @i{n} is negative:
3313:
3314: @example
3315: -1 0 ?DO i . -1 +LOOP
3316: @end example
3317: @noindent
3318: prints @code{0 -1}
3319:
3320: @example
3321: 0 0 ?DO i . -1 +LOOP
3322: @end example
3323: prints nothing.
3324:
3325: Therefore we recommend avoiding @code{@i{n} +LOOP} with negative
3326: @i{n}. One alternative is @code{@i{u} -LOOP}, which reduces the
3327: index by @i{u} each iteration. The loop is terminated when the border
3328: between @i{limit+1} and @i{limit} is crossed. Gforth also provides
3329: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
3330:
3331: @example
3332: -2 0 -DO i . 1 -LOOP
3333: @end example
3334: @noindent
3335: prints @code{0 -1}
3336:
3337: @example
3338: -1 0 -DO i . 1 -LOOP
3339: @end example
3340: @noindent
3341: prints @code{0}
3342:
3343: @example
3344: 0 0 -DO i . 1 -LOOP
3345: @end example
3346: @noindent
3347: prints nothing.
3348:
3349: @end itemize
3350:
3351: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
3352: @code{-LOOP} are not defined in ANS Forth. However, an implementation
3353: for these words that uses only standard words is provided in
3354: @file{compat/loops.fs}.
3355:
3356:
3357: @cindex @code{FOR} loops
3358: Another counted loop is:
3359: @example
3360: @i{n}
3361: FOR
3362: @i{body}
3363: NEXT
3364: @end example
3365: This is the preferred loop of native code compiler writers who are too
3366: lazy to optimize @code{?DO} loops properly. This loop structure is not
3367: defined in ANS Forth. In Gforth, this loop iterates @i{n+1} times;
3368: @code{i} produces values starting with @i{n} and ending with 0. Other
3369: Forth systems may behave differently, even if they support @code{FOR}
3370: loops. To avoid problems, don't use @code{FOR} loops.
3371:
3372: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
3373: @subsection Arbitrary control structures
3374: @cindex control structures, user-defined
3375:
3376: @cindex control-flow stack
3377: ANS Forth permits and supports using control structures in a non-nested
3378: way. Information about incomplete control structures is stored on the
3379: control-flow stack. This stack may be implemented on the Forth data
3380: stack, and this is what we have done in Gforth.
3381:
3382: @cindex @code{orig}, control-flow stack item
3383: @cindex @code{dest}, control-flow stack item
3384: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
3385: entry represents a backward branch target. A few words are the basis for
3386: building any control structure possible (except control structures that
3387: need storage, like calls, coroutines, and backtracking).
3388:
3389: doc-if
3390: doc-ahead
3391: doc-then
3392: doc-begin
3393: doc-until
3394: doc-again
3395: doc-cs-pick
3396: doc-cs-roll
3397:
3398: The Standard words @code{CS-PICK} and @code{CS-ROLL} allow you to
3399: manipulate the control-flow stack in a portable way. Without them, you
3400: would need to know how many stack items are occupied by a control-flow
3401: entry (many systems use one cell. In Gforth they currently take three,
3402: but this may change in the future).
3403:
3404: Some standard control structure words are built from these words:
3405:
3406: doc-else
3407: doc-while
3408: doc-repeat
3409:
3410: Gforth adds some more control-structure words:
3411:
3412: doc-endif
3413: doc-?dup-if
3414: doc-?dup-0=-if
3415:
3416: Counted loop words constitute a separate group of words:
3417:
3418: doc-?do
3419: doc-+do
3420: doc-u+do
3421: doc--do
3422: doc-u-do
3423: doc-do
3424: doc-for
3425: doc-loop
3426: doc-+loop
3427: doc--loop
3428: doc-next
3429: doc-leave
3430: doc-?leave
3431: doc-unloop
3432: doc-done
3433:
3434: The standard does not allow using @code{CS-PICK} and @code{CS-ROLL} on
3435: @i{do-sys}. Gforth allows it, but it's your job to ensure that for
3436: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
3437: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
3438: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
3439: resolved (by using one of the loop-ending words or @code{DONE}).
3440:
3441: Another group of control structure words are:
3442:
3443: doc-case
3444: doc-endcase
3445: doc-of
3446: doc-endof
3447:
3448: @i{case-sys} and @i{of-sys} cannot be processed using @code{CS-PICK} and
3449: @code{CS-ROLL}.
3450:
3451: @subsubsection Programming Style
3452:
3453: In order to ensure readability we recommend that you do not create
3454: arbitrary control structures directly, but define new control structure
3455: words for the control structure you want and use these words in your
3456: program. For example, instead of writing:
3457:
3458: @example
3459: BEGIN
3460: ...
3461: IF [ 1 CS-ROLL ]
3462: ...
3463: AGAIN THEN
3464: @end example
3465:
3466: @noindent
3467: we recommend defining control structure words, e.g.,
3468:
3469: @example
3470: : WHILE ( DEST -- ORIG DEST )
3471: POSTPONE IF
3472: 1 CS-ROLL ; immediate
3473:
3474: : REPEAT ( orig dest -- )
3475: POSTPONE AGAIN
3476: POSTPONE THEN ; immediate
3477: @end example
3478:
3479: @noindent
3480: and then using these to create the control structure:
3481:
3482: @example
3483: BEGIN
3484: ...
3485: WHILE
3486: ...
3487: REPEAT
3488: @end example
3489:
3490: That's much easier to read, isn't it? Of course, @code{REPEAT} and
3491: @code{WHILE} are predefined, so in this example it would not be
3492: necessary to define them.
3493:
3494: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
3495: @subsection Calls and returns
3496: @cindex calling a definition
3497: @cindex returning from a definition
3498:
3499: @cindex recursive definitions
3500: A definition can be called simply be writing the name of the definition
3501: to be called. Normally a definition is invisible during its own
3502: definition. If you want to write a directly recursive definition, you
3503: can use @code{recursive} to make the current definition visible, or
3504: @code{recurse} to call the current definition directly.
3505:
3506: doc-recursive
3507: doc-recurse
3508:
3509: @comment TODO add example of the two recursion methods
3510: @quotation
3511: @progstyle
3512: I prefer using @code{recursive} to @code{recurse}, because calling the
3513: definition by name is more descriptive (if the name is well-chosen) than
3514: the somewhat cryptic @code{recurse}. E.g., in a quicksort
3515: implementation, it is much better to read (and think) ``now sort the
3516: partitions'' than to read ``now do a recursive call''.
3517: @end quotation
3518:
3519: For mutual recursion, use @code{Defer}red words, like this:
3520:
3521: @example
3522: Defer foo
3523:
3524: : bar ( ... -- ... )
3525: ... foo ... ;
3526:
3527: :noname ( ... -- ... )
3528: ... bar ... ;
3529: IS foo
3530: @end example
3531:
3532: The current definition returns control to the calling definition when
3533: the end of the definition is reached or @code{EXIT} is
3534: encountered. Deferred words are discussed in more detail in @ref{Simple
3535: Defining Words}.
3536:
3537: doc-exit
3538: doc-;s
3539:
3540: @node Exception Handling, , Calls and returns, Control Structures
3541: @subsection Exception Handling
3542: @cindex exceptions
3543:
3544: If your program detects a fatal error condition, the simplest action
3545: that it can take is to @code{quit}. This resets the return stack and
3546: restarts the text interpreter, but does not print any error message.
3547:
3548: The next stage in severity is to execute @code{abort}, which has the
3549: same effect as @code{quit}, with the addition that it resets the data
3550: stack.
3551:
3552: A slightly more sophisticated approach is use use @code{abort"}, which
3553: compiles a string to be used as an error message and does a conditional
3554: @code{abort} at run-time. For example:
3555:
3556: @example
3557: @kbd{: checker abort" That flag was true" ." A false flag" ;@key{RET}} ok
3558: @kbd{0 checker@key{RET}} A false flag ok
3559: @kbd{1 checker@key{RET}}
3560: :1: That flag was true
3561: 1 checker
3562: ^^^^^^^
3563: $400D1648 throw
3564: $400E4660
3565: @end example
3566:
3567: These simple techniques allow a program to react to a fatal error
3568: condition, but they are not exactly user-friendly. The ANS Forth
3569: Exception word set provides the pair of words @code{throw} and
3570: @code{catch}, which can be used to provide sophisticated error-handling.
3571:
3572: @code{catch} has a similar behaviour to @code{execute}, in that it takes
3573: an @i{xt} as a parameter and starts execution of the xt. However,
3574: before passing control to the xt, @code{catch} pushes an
3575: @dfn{exception frame} onto the @dfn{exception stack}. This exception
3576: frame is used to restore the system to a known state if a detected error
3577: occurs during the execution of the xt. A typical way to use @code{catch}
3578: would be:
3579:
3580: @example
3581: ... ['] foo catch IF ...
3582: @end example
3583:
3584: Whilst @code{foo} executes, it can call other words to any level of
3585: nesting, as usual. If @code{foo} (and all the words that it calls)
3586: execute successfully, control will ultimately passes to the word following
3587: the @code{catch}, and there will be a @code{true} flag (0) at
3588: TOS. However, if any word detects an error, it can terminate the
3589: execution of @code{foo} by pushing an error code onto the stack and then
3590: performing a @code{throw}. The execution of @code{throw} will pass
3591: control to the word following the @code{catch}, but this time the TOS
3592: will hold the error code. Therefore, the @code{IF} in the example
3593: can be used to determine whether @code{foo} executed successfully.
3594:
3595: This simple example shows how you can use @code{throw} and @code{catch}
3596: to ``take over'' exception handling from the system:
3597: @example
3598: : my-div ['] / catch if ." DIVIDE ERROR" else ." OK.. " . then ;
3599: @end example
3600:
3601: The next example is more sophisticated and shows a multi-level
3602: @code{throw} and @code{catch}. To understand this example, start at the
3603: definition of @code{top-level} and work backwards:
3604:
3605: @example
3606: : lowest-level ( -- c )
3607: key dup 27 = if
3608: 1 throw \ ESCAPE key pressed
3609: else
3610: ." lowest-level successfull" CR
3611: then
3612: ;
3613:
3614: : lower-level ( -- c )
3615: lowest-level
3616: \ at this level consider a CTRL-U to be a fatal error
3617: dup 21 = if \ CTRL-U
3618: 2 throw
3619: else
3620: ." lower-level successfull" CR
3621: then
3622: ;
3623:
3624: : low-level ( -- c )
3625: ['] lower-level catch
3626: ?dup if
3627: \ error occurred - do we recognise it?
3628: dup 1 = if
3629: \ ESCAPE key pressed.. pretend it was an E
3630: [char] E
3631: else throw \ propogate the error upwards
3632: then
3633: then
3634: ." low-level successfull" CR
3635: ;
3636:
3637: : top-level ( -- )
3638: CR ['] low-level catch \ CATCH is used like EXECUTE
3639: ?dup if \ error occurred..
3640: ." Error " . ." occurred - contact your supplier"
3641: else
3642: ." The '" emit ." ' key was pressed" CR
3643: then
3644: ;
3645: @end example
3646:
3647: The ANS Forth document assigns @code{throw} codes thus:
3648:
3649: @itemize @bullet
3650: @item
3651: codes in the range -1 -- -255 are reserved to be assigned by the
3652: Standard. Assignments for codes in the range -1 -- -58 are currently
3653: documented in the Standard. In particular, @code{-1 throw} is equivalent
3654: to @code{abort} and @code{-2 throw} is equivalent to @code{abort"}.
3655: @item
3656: codes in the range -256 -- -4095 are reserved to be assigned by the system.
3657: @item
3658: all other codes may be assigned by programs.
3659: @end itemize
3660:
3661: Gforth provides the word @code{exception} as a mechanism for assigning
3662: system throw codes to applications. This allows multiple applications to
3663: co-exist in memory without any clash of @code{throw} codes. A definition
3664: of @code{exception} in ANS Forth is provided in
3665: @file{compat/exception.fs}.
3666:
3667: doc-quit
3668: doc-abort
3669: doc-abort"
3670:
3671: doc-catch
3672: doc-throw
3673: doc---exception-exception
3674:
3675:
3676: @c -------------------------------------------------------------
3677: @node Defining Words, The Text Interpreter, Control Structures, Words
3678: @section Defining Words
3679: @cindex defining words
3680:
3681: @menu
3682: * Simple Defining Words:: Variables, values and constants
3683: * Colon Definitions::
3684: * User-defined Defining Words::
3685: * Supplying names::
3686: * Interpretation and Compilation Semantics::
3687: @end menu
3688:
3689: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
3690: @subsection Simple Defining Words
3691: @cindex simple defining words
3692: @cindex defining words, simple
3693:
3694: Defining words are used to create new entries in the dictionary. The
3695: simplest defining word is @code{CREATE}. @code{CREATE} is used like
3696: this:
3697:
3698: @example
3699: CREATE new-word1
3700: @end example
3701:
3702: @code{CREATE} is a parsing word that generates a dictionary entry for
3703: @code{new-word1}. When @code{new-word1} is executed, all that it does is
3704: leave an address on the stack. The address represents the value of
3705: the data space pointer (@code{HERE}) at the time that @code{new-word1}
3706: was defined. Therefore, @code{CREATE} is a way of associating a name
3707: with the address of a region of memory.
3708:
3709: By extending this example to reserve some memory in data space, we end
3710: up with a @i{variable}. Here are two different ways to do it:
3711:
3712: @example
3713: CREATE new-word2 1 cells allot \ reserve 1 cell - initial value undefined
3714: CREATE new-word3 4 , \ reserve 1 cell and initialise it (to 4)
3715: @end example
3716:
3717: The variable can be examined and modified using @code{@@} (``fetch'') and
3718: @code{!} (``store'') like this:
3719:
3720: @example
3721: new-word2 @@ . \ get address, fetch from it and display
3722: 1234 new-word2 ! \ new value, get address, store to it
3723: @end example
3724:
3725: As a final refinement, the whole code sequence can be wrapped up in a
3726: defining word (pre-empting the subject of the next section), making it
3727: easier to create new variables:
3728:
3729: @example
3730: : myvariable ( "name" -- a-addr ) CREATE 1 cells allot ;
3731:
3732: myvariable foo
3733: myvariable joe
3734:
3735: 45 3 * foo ! \ set foo to 135
3736: 1234 joe ! \ set joe to 1234
3737: 3 joe +! \ increment joe by 3.. to 1237
3738: @end example
3739:
3740: Not surprisingly, there is no need to define @code{myvariable}, since
3741: Forth already has a definition @code{Variable}. It behaves in exactly
3742: the same way as @code{myvariable} but it is implemented in an optimised
3743: way. Forth also provides @code{2Variable} and @code{fvariable} for
3744: double and floating-point variables, respectively.
3745:
3746: @cindex arrays
3747: A similar mechanism can be used to create arrays. For example, an
3748: 80-character text input buffer:
3749:
3750: @example
3751: CREATE text-buf 80 chars allot
3752:
3753: text-buf 0 chars c@@ \ the 1st character (offset 0)
3754: text-buf 3 chars c@@ \ the 4th character (offset 3)
3755: @end example
3756:
3757: You can build arbitrarily complex data structures by allocating
3758: appropriate areas of memory. @xref{Structures} for further discussions
3759: of this, and to learn about some Gforth tools that make it easier.
3760:
3761: @cindex user variables
3762: @cindex user space
3763: The defining word @code{User} behaves in the same way as @code{Variable}.
3764: The difference is that it reserves space in @i{user (data) space} rather
3765: than normal data space. In a Forth system that has a multi-tasker, each
3766: task has its own set of user variables.
3767:
3768: @comment TODO is that stuff about user variables strictly correct? Is it
3769: @comment just terminal tasks that have user variables?
3770: @comment should document tasker.fs (with some examples) elsewhere
3771: @comment in this manual, then expand on user space and user variables.
3772:
3773: After @code{CREATE} and @code{Variable}s, the next defining word to
3774: consider is @code{Constant}. @code{Constant} allows you to declare a
3775: fixed value and refer to it by name. For example:
3776:
3777: @example
3778: 12 Constant INCHES-PER-FOOT
3779: 3E+08 fconstant SPEED-O-LIGHT
3780: @end example
3781:
3782: A @code{Variable} can be both read and written, so its run-time
3783: behaviour is to supply an address through which its current value can be
3784: manipulated. In contrast, the value of a @code{Constant} cannot be
3785: changed once it has been declared@footnote{Well, often it can be -- but
3786: not in a Standard, portable way. It's safer to use a @code{Value} (read
3787: on).} so it's not necessary to supply the address -- it is more
3788: efficient to return the value of the constant directly. That's exactly
3789: what happens; the run-time effect of a constant is to put its value on
3790: the top of the stack (@ref{User-defined Defining Words} describes one
3791: way of implementing @code{Constant}).
3792:
3793: Gforth also provides @code{2Constant} and @code{fconstant} for defining
3794: double and floating-point constants, respectively.
3795:
3796: Constants in Forth behave differently from their equivalents in other
3797: programming languages. In other languages, a constant (such as an EQU in
3798: assembler or a #define in C) only exists at compile-time; in the
3799: executable program the constant has been translated into an absolute
3800: number and, unless you are using a symbolic debugger, it's impossible to
3801: know what abstract thing that number represents. In Forth a constant has
3802: an entry in the name dictionary and remains there after the code that
3803: uses it has been defined. In fact, it must remain in the dictionary
3804: since it has run-time duties to perform. For example:
3805:
3806: @example
3807: 12 Constant INCHES-PER-FOOT
3808: : FEET-TO-INCHES ( n1 -- n2 ) INCHES-PER-FOOT * ;
3809: @end example
3810:
3811: @cindex in-lining of constants
3812: When @code{FEET-TO-INCHES} is executed, it will in turn execute the xt
3813: associated with the constant @code{INCHES-PER-FOOT}. If you use
3814: @code{see} to decompile the definition of @code{FEET-TO-INCHES}, you can
3815: see that it makes a call to @code{INCHES-PER-FOOT}. Some Forth compilers
3816: attempt to optimise constants by in-lining them where they are used. You
3817: can force Gforth to in-line a constant like this:
3818:
3819: @example
3820: : FEET-TO-INCHES ( n1 -- n2 ) [ INCHES-PER-FOOT ] LITERAL * ;
3821: @end example
3822:
3823: If you use @code{see} to decompile @i{this} version of
3824: @code{FEET-TO-INCHES}, you can see that @code{INCHES-PER-FOOT} is no
3825: longer present. @xref{Interpret/Compile states} and @xref{Literals}
3826: explain to this works.
3827:
3828: In-lining constants in this way might improve execution time
3829: fractionally, and can ensure that a constant is now only referenced at
3830: compile-time. However, the definition of the constant still remains in
3831: the dictionary. Some Forth compilers provide a mechanism for controlling
3832: a second dictionary for holding transient words such that this second
3833: dictionary can be deleted later in order to recover memory
3834: space. However, there is no standard way of doing this.
3835:
3836: One aspect of constants and variables that can sometimes be confusing is
3837: that they have different stack effects; one returns its value whilst the
3838: other returns the address of its value. The defining word @code{Value}
3839: provides an alternative to @code{Variable}, and has the same stack
3840: effect as a constant. A @code{Value} needs an additional word, @code{TO}
3841: to allow its value to be changed. Here are some examples:
3842:
3843: @example
3844: 12 Value APPLES \ a Value is initialised when it is declared.. like a
3845: \ constant but unlike a variable
3846: 34 TO APPLES \ Change the value of APPLES. TO is a parsing word
3847: APPLES \ puts 34 on the top of the stack.
3848: @end example
3849:
3850: The defining word @code{Defer} allows you to define a word by name
3851: without defining its behaviour; the definition of its behaviour is
3852: deferred. Here are two situation where this can be useful:
3853:
3854: @itemize @bullet
3855: @item
3856: Where you want to allow the behaviour of a word to be altered later, and
3857: for all precompiled references to the word to change when its behaviour
3858: is changed.
3859: @item
3860: For mutual recursion; @xref{Calls and returns}.
3861: @end itemize
3862:
3863: In the following example, @code{foo} always invokes the version of
3864: @code{greet} that prints ``@code{Good morning}'' whilst @code{bar}
3865: always invokes the version that prints ``@code{Hello}''. There is no way
3866: of getting @code{foo} to use the later version without re-ordering the
3867: source code and recompilng it.
3868:
3869: @example
3870: : greet ." Good morning" ;
3871: : foo ... greet ... ;
3872: : greet ." Hello" ;
3873: : bar ... greet ... ;
3874: @end example
3875:
3876: This problem can be solved by defining @code{greet} as a @code{Defer}red
3877: word. The behaviour of a @code{Defer}red word can be defined and
3878: redefined at any time by using @code{IS} to associate the xt of a
3879: previously-defined word with it. The previous example becomes:
3880:
3881: @example
3882: Defer greet
3883: : foo ... greet ... ;
3884: : bar ... greet ... ;
3885: : greet1 ." Good morning" ;
3886: : greet2 ." Hello" ;
3887: ' greet2 IS greet \ make greet behave like greet2
3888: @end example
3889:
3890: A deferred word can only inherit default semantics from the xt (because
3891: that is all that an xt can represent -- @pxref{Tokens for Words} for
3892: more discussion of this). However, the semantics of the deferred word
3893: itself can be modified at the time that it is defined. For example:
3894:
3895: @example
3896: : bar .... ; compile-only
3897: Defer fred immediate
3898: Defer jim
3899:
3900: ' bar IS jim \ jim has default semantics
3901: ' bar IS fred \ fred is immediate
3902: @end example
3903:
3904: The defining word @code{Alias} allows you to define a word by name that
3905: has the same behaviour as some other word. Here are two situation where
3906: this can be useful:
3907:
3908: @itemize @bullet
3909: @item
3910: When you want access to a word's definition from a different word list
3911: (for an example of this, see the definition of the @code{Root} word list
3912: in the Gforth source).
3913: @item
3914: When you want to create a synonym; a definition that can be known by
3915: either of two names (for example, @code{THEN} and @code{ENDIF} are
3916: aliases).
3917: @end itemize
3918:
3919: The word whose behaviour the alias is to inherit is represented by an
3920: xt. Therefore, the alias can only inherits default semantics from its
3921: ancestor. The semantics of the alias itself can be modified at the time
3922: that it is defined. For example:
3923:
3924: @example
3925: : foo ... ; immediate
3926:
3927: ' foo Alias bar \ bar is not an immediate word
3928: ' foo Alias fooby immediate \ fooby is an immediate word
3929: @end example
3930:
3931: Words that are aliases have the same xt. Their semantics can differ
3932: because the rules about a word's semantics are stored in the name
3933: dictionary, and the aliases each have their own dictionary entry. It
3934: follows that words that are aliases have different name tokens and may
3935: have the same or different compilation tokens. Once again, see
3936: @ref{Tokens for Words} for more discussions of this.
3937:
3938: doc-create
3939: doc-variable
3940: doc-2variable
3941: doc-fvariable
3942: doc-user
3943: doc-constant
3944: doc-2constant
3945: doc-fconstant
3946: doc-value
3947: doc-to
3948: doc-defer
3949: doc-is
3950: doc-alias
3951: @comment TODO document these: what's defers <is> [is]
3952: doc-what's
3953: doc-defers
3954:
3955: Definitions in ANS Forth for @code{defer}, @code{<is>} and
3956: @code{[is]} are provided in @file{compat/defer.fs}.
3957:
3958:
3959: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
3960: @subsection Colon Definitions
3961: @cindex colon definitions
3962:
3963: @example
3964: : name ( ... -- ... )
3965: word1 word2 word3 ;
3966: @end example
3967:
3968: @noindent
3969: Creates a word called @code{name} that, upon execution, executes
3970: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
3971:
3972: The explanation above is somewhat superficial. @xref{Your first
3973: definition} for simple examples of colon definitions, then
3974: @xref{Interpretation and Compilation Semantics} for an in-depth
3975: discussion of some of the issues involved.
3976:
3977: doc-:
3978: doc-;
3979:
3980: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
3981: @subsection User-defined Defining Words
3982: @cindex user-defined defining words
3983: @cindex defining words, user-defined
3984:
3985: You can create a new defining word by wrapping defining-time code around
3986: an existing defining word and putting the sequence in a colon
3987: definition. For example, suppose that you have a word @code{stats} that
3988: gathers statistics about colon definitions given the @i{xt} of the
3989: definition, and you want every colon definition in your application to
3990: make a call to @code{stats}. You can define and use a new version of
3991: @code{:} like this:
3992:
3993: @example
3994: : stats ( xt -- ) DUP ." (Gathering statistics for " . ." )"
3995: ... ; \ other code
3996:
3997: : my: : lastxt postpone literal ['] stats compile, ;
3998:
3999: my: foo + - ;
4000: @end example
4001:
4002: When @code{foo} is defined using @code{my:} these steps occur:
4003:
4004: @itemize @bullet
4005: @item
4006: @code{my:} is executed.
4007: @item
4008: The @code{:} within the definition (the one between @code{my:} and
4009: @code{lastxt}) is executed, and does just what it always does; it parses
4010: the input stream for a name, builds a dictionary header for the name
4011: @code{foo} and switches @code{state} from interpret to compile.
4012: @item
4013: The word @code{lastxt} is executed. It puts the @i{xt} for the word that is
4014: being defined -- @code{foo} -- onto the stack.
4015: @item
4016: The code that was produced by @code{postpone literal} is executed; this
4017: causes the value on the stack to be compiled as a literal in the code
4018: area of @code{foo}.
4019: @item
4020: The code @code{['] stats} compiles a literal into the definition of
4021: @code{my:}. When @code{compile,} is executed, that literal -- the
4022: execution token for @code{stats} -- is layed down in the code area of
4023: @code{foo} , following the literal@footnote{Strictly speaking, the
4024: mechanism that @code{compile,} uses to convert an @i{xt} into something
4025: in the code area is implementation-dependent. A threaded implementation
4026: might spit out the execution token directly whilst another
4027: implementation might spit out a native code sequence.}.
4028: @item
4029: At this point, the execution of @code{my:} is complete, and control
4030: returns to the text interpreter. The text interpreter is in compile
4031: state, so subsequent text @code{+ -} is compiled into the definition of
4032: @code{foo} and the @code{;} terminates the definition as always.
4033: @end itemize
4034:
4035: You can use @code{see} to decompile a word that was defined using
4036: @code{my:} and see how it is different from a normal @code{:}
4037: definition. For example:
4038:
4039: @example
4040: : bar + - ; \ like foo but using : rather than my:
4041: see bar
4042: : bar
4043: + - ;
4044: see foo
4045: : foo
4046: 107645672 stats + - ;
4047:
4048: \ use ' stats . to show that 107645672 is the xt for stats
4049: @end example
4050:
4051:
4052: Rather than edit your application's source code to change every @code{:}
4053: to a @code{my:}, use a deferred word:
4054:
4055: @example
4056: : real: : ; \ retain access to the original
4057: defer : \ redefine as a deferred word
4058: ' my: IS : \ use special version of :
4059: \
4060: \ load application here
4061: \
4062: ' real: IS : \ go back to the original
4063: @end example
4064:
4065: You can use techniques like this to make new defining words in terms of
4066: @i{any} existing defining word.
4067:
4068:
4069: @cindex defining defining words
4070: @cindex @code{CREATE} ... @code{DOES>}
4071: If you want the words defined with your defining words to behave
4072: differently from words defined with standard defining words, you can
4073: write your defining word like this:
4074:
4075: @example
4076: : def-word ( "name" -- )
4077: CREATE @i{code1}
4078: DOES> ( ... -- ... )
4079: @i{code2} ;
4080:
4081: def-word name
4082: @end example
4083:
4084: @cindex child words
4085: This fragment defines a @dfn{defining word} @code{def-word} and then
4086: executes it. When @code{def-word} executes, it @code{CREATE}s a new
4087: word, @code{name}, and executes the code @i{code1}. The code @i{code2}
4088: is not executed at this time. The word @code{name} is sometimes called a
4089: @dfn{child} of @code{def-word}.
4090:
4091: When you execute @code{name}, the address of the body of @code{name} is
4092: put on the data stack and @i{code2} is executed (the address of the body
4093: of @code{name} is the address @code{HERE} returns immediately after the
4094: @code{CREATE}).
4095:
4096: @cindex atavism in child words
4097: You can use @code{def-word} to define a set of child word that behave
4098: differently, though atavistically; they all have a common run-time
4099: behaviour determined by @i{code2}. Typically, the @i{code1} sequence
4100: builds a data area in the body of the child word. The structure of the
4101: data is common to all children of @code{def-word}, but the data values
4102: are specific -- and private -- to each child word. When a child word is
4103: executed, the address of its private data area is passed as a parameter
4104: on TOS to be used and manipulated@footnote{It is legitimate both to read
4105: and write to this data area.} by @i{code2}.
4106:
4107: The two fragments of code that make up the defining words act (are
4108: executed) at two completely separate times:
4109:
4110: @itemize @bullet
4111: @item
4112: At @i{define time}, the defining word executes @i{code1} to generate a
4113: child word
4114: @item
4115: At @i{child execution time}, when a child word is invoked, @i{code2}
4116: is executed, using parameters (data) that are private and specific to
4117: the child word.
4118: @end itemize
4119:
4120: @c NAC I think this is a really bad example, because it diminishes
4121: @c rather than emphasising the fact that some important stuff happens
4122: @c at define time, and other important stuff happens at child-invocation
4123: @c time, and that those two times are potentially very different.
4124: @c
4125: @c In other words, if you make the following definitions:
4126: @c @example
4127: @c : def-word1 ( "name" -- )
4128: @c CREATE @i{code1} ;
4129: @c
4130: @c : action1 ( ... -- ... )
4131: @c @i{code2} ;
4132: @c
4133: @c def-word1 name1
4134: @c @end example
4135: @c
4136: @c Using @code{name1 action1} is equivalent to using @code{name}.
4137:
4138: The classic example is that you can define @code{CONSTANT} in this way:
4139:
4140: @example
4141: : CONSTANT ( w "name" -- )
4142: CREATE ,
4143: DOES> ( -- w )
4144: @@ ;
4145: @end example
4146:
4147: @comment There is a beautiful description of how this works and what
4148: @comment it does in the Forthwrite 100th edition.. as well as an elegant
4149: @comment commentary on the Counting Fruits problem.
4150:
4151: When you create a constant with @code{5 CONSTANT five}, a set of
4152: define-time actions take place; first a new word @code{five} is created,
4153: then the value 5 is laid down in the body of @code{five} with
4154: @code{,}. When @code{five} is invoked, the address of the body is put on
4155: the stack, and @code{@@} retrieves the value 5. The word @code{five} has
4156: no code of its own; it simply contains a data field and a pointer to the
4157: code that follows @code{DOES>} in its defining word. That makes words
4158: created in this way very compact.
4159:
4160: The final example in this section is intended to remind you that space
4161: reserved in @code{CREATE}d words is @i{data} space and therefore can be
4162: both read and written by a Standard program@footnote{Exercise: use this
4163: example as a starting point for your own implementation of @code{Value}
4164: and @code{TO} -- if you get stuck, investigate the behaviour of @code{'} and
4165: @code{[']}.}:
4166:
4167: @example
4168: : foo ( "name" -- )
4169: CREATE -1 ,
4170: DOES> ( -- )
4171: @@ .;
4172:
4173: foo first-word
4174: foo second-word
4175:
4176: 123 ' first-word >BODY !
4177: @end example
4178:
4179: If @code{first-word} had been a @code{CREATE}d word, we could simply
4180: have executed it to get the address of its data field. However, since it
4181: was defined to have @code{DOES>} actions, its execution semantics are to
4182: perform those @code{DOES>} actions. To get the address of its data field
4183: it's necessary to use @code{'} to get its xt, then @code{>BODY} to
4184: translate the xt into the address of the data field. When you execute
4185: @code{first-word}, it will display @code{123}. When you execute
4186: @code{second-word} it will display @code{-1}.
4187:
4188: @cindex stack effect of @code{DOES>}-parts
4189: @cindex @code{DOES>}-parts, stack effect
4190: In the examples above the stack comment after the @code{DOES>} specifies
4191: the stack effect of the defined words, not the stack effect of the
4192: following code (the following code expects the address of the body on
4193: the top of stack, which is not reflected in the stack comment). This is
4194: the convention that I use and recommend (it clashes a bit with using
4195: locals declarations for stack effect specification, though).
4196:
4197: @subsubsection Applications of @code{CREATE..DOES>}
4198: @cindex @code{CREATE} ... @code{DOES>}, applications
4199:
4200: You may wonder how to use this feature. Here are some usage patterns:
4201:
4202: @cindex factoring similar colon definitions
4203: When you see a sequence of code occurring several times, and you can
4204: identify a meaning, you will factor it out as a colon definition. When
4205: you see similar colon definitions, you can factor them using
4206: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
4207: that look very similar:
4208: @example
4209: : ori, ( reg-target reg-source n -- )
4210: 0 asm-reg-reg-imm ;
4211: : andi, ( reg-target reg-source n -- )
4212: 1 asm-reg-reg-imm ;
4213: @end example
4214:
4215: @noindent
4216: This could be factored with:
4217: @example
4218: : reg-reg-imm ( op-code -- )
4219: CREATE ,
4220: DOES> ( reg-target reg-source n -- )
4221: @@ asm-reg-reg-imm ;
4222:
4223: 0 reg-reg-imm ori,
4224: 1 reg-reg-imm andi,
4225: @end example
4226:
4227: @cindex currying
4228: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
4229: supply a part of the parameters for a word (known as @dfn{currying} in
4230: the functional language community). E.g., @code{+} needs two
4231: parameters. Creating versions of @code{+} with one parameter fixed can
4232: be done like this:
4233: @example
4234: : curry+ ( n1 -- )
4235: CREATE ,
4236: DOES> ( n2 -- n1+n2 )
4237: @@ + ;
4238:
4239: 3 curry+ 3+
4240: -2 curry+ 2-
4241: @end example
4242:
4243: @subsubsection The gory details of @code{CREATE..DOES>}
4244: @cindex @code{CREATE} ... @code{DOES>}, details
4245:
4246: doc-does>
4247:
4248: @cindex @code{DOES>} in a separate definition
4249: This means that you need not use @code{CREATE} and @code{DOES>} in the
4250: same definition; you can put the @code{DOES>}-part in a separate
4251: definition. This allows us to, e.g., select among different @code{DOES>}-parts:
4252: @example
4253: : does1
4254: DOES> ( ... -- ... )
4255: ... ;
4256:
4257: : does2
4258: DOES> ( ... -- ... )
4259: ... ;
4260:
4261: : def-word ( ... -- ... )
4262: create ...
4263: IF
4264: does1
4265: ELSE
4266: does2
4267: ENDIF ;
4268: @end example
4269:
4270: In this example, the selection of whether to use @code{does1} or
4271: @code{does2} is made at compile-time; at the time that the child word is
4272: @code{CREATE}d.
4273:
4274: @cindex @code{DOES>} in interpretation state
4275: In a standard program you can apply a @code{DOES>}-part only if the last
4276: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
4277: will override the behaviour of the last word defined in any case. In a
4278: standard program, you can use @code{DOES>} only in a colon
4279: definition. In Gforth, you can also use it in interpretation state, in a
4280: kind of one-shot mode; for example:
4281: @example
4282: CREATE name ( ... -- ... )
4283: @i{initialization}
4284: DOES>
4285: @i{code} ;
4286: @end example
4287:
4288: @noindent
4289: is equivalent to the standard:
4290: @example
4291: :noname
4292: DOES>
4293: @i{code} ;
4294: CREATE name EXECUTE ( ... -- ... )
4295: @i{initialization}
4296: @end example
4297:
4298: You can get the address of the body of a word with:
4299:
4300: doc->body
4301:
4302: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
4303: @subsection Supplying the name of a defined word
4304: @cindex names for defined words
4305: @cindex defining words, name parameter
4306:
4307: @cindex defining words, name given in a string
4308: By default, a defining word takes the name for the defined word from the
4309: input stream. Sometimes you want to supply the name from a string. You
4310: can do this with:
4311:
4312: doc-nextname
4313:
4314: For example:
4315:
4316: @example
4317: s" foo" nextname create
4318: @end example
4319: @noindent
4320: is equivalent to:
4321: @example
4322: create foo
4323: @end example
4324:
4325: @cindex defining words without name
4326: Sometimes you want to define an @dfn{anonymous word}; a word without a
4327: name. You can do this with:
4328:
4329: doc-:noname
4330:
4331: This leaves the execution token for the word on the stack after the
4332: closing @code{;}. Here's an example in which a deferred word is
4333: initialised with an @code{xt} from an anonymous colon definition:
4334: @example
4335: Defer deferred
4336: :noname ( ... -- ... )
4337: ... ;
4338: IS deferred
4339: @end example
4340:
4341: @noindent
4342: Gforth provides an alternative way of doing this, using two separate
4343: words:
4344:
4345: doc-noname
4346: @cindex execution token of last defined word
4347: doc-lastxt
4348:
4349: @noindent
4350: The previous example can be rewritten using @code{noname} and
4351: @code{lastxt}:
4352:
4353: @example
4354: Defer deferred
4355: noname : ( ... -- ... )
4356: ... ;
4357: lastxt IS deferred
4358: @end example
4359:
4360: @noindent
4361: @code{lastxt} also works when the last word was not defined as
4362: @code{noname}. It also has the useful property that is is valid as soon
4363: as the header for a definition has been build. Thus:
4364:
4365: @example
4366: lastxt . : foo [ lastxt . ] ; ' foo .
4367: @end example
4368:
4369: @noindent
4370: prints 3 numbers; the last two are the same.
4371:
4372:
4373: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
4374: @subsection Interpretation and Compilation Semantics
4375: @cindex semantics, interpretation and compilation
4376:
4377: @cindex interpretation semantics
4378: The @dfn{interpretation semantics} of a word are what the text
4379: interpreter does when it encounters the word in interpret state. It also
4380: appears in some other contexts, e.g., the execution token returned by
4381: @code{' @i{word}} identifies the interpretation semantics of
4382: @i{word} (in other words, @code{' @i{word} execute} is equivalent to
4383: interpret-state text interpretation of @code{@i{word}}).
4384:
4385: @cindex compilation semantics
4386: The @dfn{compilation semantics} of a word are what the text interpreter
4387: does when it encounters the word in compile state. It also appears in
4388: other contexts, e.g, @code{POSTPONE @i{word}} compiles@footnote{In
4389: standard terminology, ``appends to the current definition''.} the
4390: compilation semantics of @i{word}.
4391:
4392: @cindex execution semantics
4393: The standard also talks about @dfn{execution semantics}. They are used
4394: only for defining the interpretation and compilation semantics of many
4395: words. By default, the interpretation semantics of a word are to
4396: @code{execute} its execution semantics, and the compilation semantics of
4397: a word are to @code{compile,} its execution semantics.@footnote{In
4398: standard terminology: The default interpretation semantics are its
4399: execution semantics; the default compilation semantics are to append its
4400: execution semantics to the execution semantics of the current
4401: definition.}
4402:
4403: @comment TODO expand, make it co-operate with new sections on text interpreter.
4404:
4405: @cindex immediate words
4406: @cindex compile-only words
4407: You can change the semantics of the most-recently defined word:
4408:
4409: doc-immediate
4410: doc-compile-only
4411: doc-restrict
4412:
4413: Note that ticking (@code{'}) a compile-only word gives an error
4414: (``Interpreting a compile-only word'').
4415:
4416: Gforth also allows you to define words with arbitrary combinations of
4417: interpretation and compilation semantics.
4418:
4419: doc-interpret/compile:
4420:
4421: This feature was introduced for implementing @code{TO} and @code{S"}. I
4422: recommend that you do not define such words, as cute as they may be:
4423: they make it hard to get at both parts of the word in some contexts.
4424: E.g., assume you want to get an execution token for the compilation
4425: part. Instead, define two words, one that embodies the interpretation
4426: part, and one that embodies the compilation part. Once you have done
4427: that, you can define a combined word with @code{interpret/compile:} for
4428: the convenience of your users.
4429:
4430: You might try to use this feature to provide an optimizing
4431: implementation of the default compilation semantics of a word. For
4432: example, by defining:
4433: @example
4434: :noname
4435: foo bar ;
4436: :noname
4437: POSTPONE foo POSTPONE bar ;
4438: interpret/compile: opti-foobar
4439: @end example
4440:
4441: @noindent
4442: as an optimizing version of:
4443:
4444: @example
4445: : foobar
4446: foo bar ;
4447: @end example
4448:
4449: Unfortunately, this does not work correctly with @code{[compile]},
4450: because @code{[compile]} assumes that the compilation semantics of all
4451: @code{interpret/compile:} words are non-default. I.e., @code{[compile]
4452: opti-foobar} would compile compilation semantics, whereas
4453: @code{[compile] foobar} would compile interpretation semantics.
4454:
4455: @cindex state-smart words (are a bad idea)
4456: Some people try to use @dfn{state-smart} words to emulate the feature provided
4457: by @code{interpret/compile:} (words are state-smart if they check
4458: @code{STATE} during execution). E.g., they would try to code
4459: @code{foobar} like this:
4460:
4461: @example
4462: : foobar
4463: STATE @@
4464: IF ( compilation state )
4465: POSTPONE foo POSTPONE bar
4466: ELSE
4467: foo bar
4468: ENDIF ; immediate
4469: @end example
4470:
4471: Although this works if @code{foobar} is only processed by the text
4472: interpreter, it does not work in other contexts (like @code{'} or
4473: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
4474: for a state-smart word, not for the interpretation semantics of the
4475: original @code{foobar}; when you execute this execution token (directly
4476: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
4477: state, the result will not be what you expected (i.e., it will not
4478: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
4479: write them@footnote{For a more detailed discussion of this topic, see
4480: @cite{@code{State}-smartness -- Why it is Evil and How to Exorcise it} by Anton
4481: Ertl; presented at EuroForth '98 and available from
4482: @url{http://www.complang.tuwien.ac.at/papers/}}!
4483:
4484: @cindex defining words with arbitrary semantics combinations
4485: It is also possible to write defining words that define words with
4486: arbitrary combinations of interpretation and compilation semantics. In
4487: general, they look like this:
4488:
4489: @example
4490: : def-word
4491: create-interpret/compile
4492: @i{code1}
4493: interpretation>
4494: @i{code2}
4495: <interpretation
4496: compilation>
4497: @i{code3}
4498: <compilation ;
4499: @end example
4500:
4501: For a @i{word} defined with @code{def-word}, the interpretation
4502: semantics are to push the address of the body of @i{word} and perform
4503: @i{code2}, and the compilation semantics are to push the address of
4504: the body of @i{word} and perform @i{code3}. E.g., @code{constant}
4505: can also be defined like this (except that the defined constants don't
4506: behave correctly when @code{[compile]}d):
4507:
4508: @example
4509: : constant ( n "name" -- )
4510: create-interpret/compile
4511: ,
4512: interpretation> ( -- n )
4513: @@
4514: <interpretation
4515: compilation> ( compilation. -- ; run-time. -- n )
4516: @@ postpone literal
4517: <compilation ;
4518: @end example
4519:
4520: doc-create-interpret/compile
4521: doc-interpretation>
4522: doc-<interpretation
4523: doc-compilation>
4524: doc-<compilation
4525:
4526: Words defined with @code{interpret/compile:} and
4527: @code{create-interpret/compile} have an extended header structure that
4528: differs from other words; however, unless you try to access them with
4529: plain address arithmetic, you should not notice this. Words for
4530: accessing the header structure usually know how to deal with this; e.g.,
4531: @code{'} @i{word} @code{>body} also gives you the body of a word created
4532: with @code{create-interpret/compile}.
4533:
4534: doc-postpone
4535: @comment TODO -- expand glossary text for POSTPONE
4536:
4537: @c ----------------------------------------------------------
4538: @node The Text Interpreter, Tokens for Words, Defining Words, Words
4539: @section The Text Interpreter
4540: @cindex interpreter - outer
4541: @cindex text interpreter
4542: @cindex outer interpreter
4543:
4544: The text interpreter@footnote{This is an expanded version of the
4545: material in @ref{Introducing the Text Interpreter}.} is an endless loop
4546: that processes input from the current input device. A popular
4547: implementation technique for Forth is to implement a @dfn{forth virtual
4548: machine} using a loop called the @dfn{inner interpreter}. Because of
4549: this naming, the text interpreter is also known as the @dfn{outer
4550: interpreter}.
4551:
4552: @cindex interpret state
4553: @cindex compile state
4554: The text interpreter operates in one of two states: @dfn{interpret
4555: state} and @dfn{compile state}. The current state is defined by the
4556: aptly-named variable, @code{state}.
4557:
4558: This section starts by describing how the text interpreter behaves when
4559: it is in interpret state, processing input from the user input device --
4560: the keyboard. This is the mode that a Forth system is in after it starts
4561: up.
4562:
4563: @cindex input buffer
4564: @cindex terminal input buffer
4565: The text interpreter works from an area of memory called the @dfn{input
4566: buffer}@footnote{When the text interpreter is processing input from the
4567: keyboard, this area of memory is called the @dfn{terminal input buffer}
4568: (TIB) and is addressed by the (obsolescent) words @code{TIB} and
4569: @code{#TIB}.}, which stores your keyboard input when you press the
4570: @key{RET} key. Starting at the beginning of the input buffer, it skips
4571: leading spaces (called @dfn{delimiters}) then parses a string (a
4572: sequence of non-space characters) until it reaches either a space
4573: character or the end of the buffer. Having parsed a string, it makes two
4574: attempts to process it:
4575:
4576: @cindex dictionary
4577: @itemize @bullet
4578: @item
4579: It looks for the string in a @dfn{dictionary} of definitions. If the
4580: string is found, the string names a @dfn{definition} (also known as a
4581: @dfn{word}) and the dictionary search returns information that allows
4582: the text interpreter to perform the word's @dfn{interpretation
4583: semantics}. In most cases, this simply means that the word will be
4584: executed.
4585: @item
4586: If the string is not found in the dictionary, the text interpreter
4587: attempts to treat it as a number, using the rules described in
4588: @ref{Number Conversion}. If the string represents a legal number in the
4589: current radix, the number is pushed onto a parameter stack (the data
4590: stack for integers, the floating-point stack for floating-point
4591: numbers).
4592: @end itemize
4593:
4594: If both attempts fail, or if the word is found in the dictionary but has
4595: no interpretation semantics@footnote{This happens if the word was
4596: defined as @code{COMPILE-ONLY}.} the text interpreter discards the
4597: remainder of the input buffer, issues an error message and waits for
4598: more input. If one of the attempts succeeds, the text interpreter
4599: repeats the parsing process until the whole of the input buffer has been
4600: processed, at which point it prints the status message ``@code{ ok}''
4601: and waits for more input.
4602:
4603: @cindex parse area
4604: The text interpreter keeps track of its position in the input buffer by
4605: updating a variable called @code{>IN} (pronounced ``to-in''). The value
4606: of @code{>IN} starts out as 0, indicating an offset of 0 from the start
4607: of the input buffer. The region from offset @code{>IN @@} to the end of
4608: the input buffer is called the @dfn{parse area}@footnote{In other words,
4609: the text interpreter processes the contents of the input buffer by
4610: parsing strings from the parse area until the parse area is empty.}.
4611: This example shows how @code{>IN} changes as the text interpreter parses
4612: the input buffer:
4613:
4614: @example
4615: : remaining >IN @@ SOURCE 2 PICK - -ROT + SWAP
4616: CR ." ->" TYPE ." <-" ; IMMEDIATE
4617:
4618: 1 2 3 remaining + remaining .
4619:
4620: : foo 1 2 3 remaining SWAP remaining ;
4621: @end example
4622:
4623: @noindent
4624: The result is:
4625:
4626: @example
4627: ->+ remaining .<-
4628: ->.<-5 ok
4629:
4630: ->SWAP remaining ;-<
4631: ->;<- ok
4632: @end example
4633:
4634: @cindex parsing words
4635: The value of @code{>IN} can also be modified by a word in the input
4636: buffer that is executed by the text interpreter. This means that a word
4637: can ``trick'' the text interpreter into either skipping a section of the
4638: input buffer@footnote{This is how parsing words work.} or into parsing a
4639: section twice. For example:
4640:
4641: @example
4642: : lat ." <<lat>>" ;
4643: : flat ." <<flat>>" >IN DUP @@ 3 - SWAP ! ;
4644: @end example
4645:
4646: @noindent
4647: When @code{flat} is executed, this output is produced@footnote{Exercise
4648: for the reader: what would happen if the @code{3} were replaced with
4649: @code{4}?}:
4650:
4651: @example
4652: <<flat>><<lat>>
4653: @end example
4654:
4655: @noindent
4656: Two important notes about the behaviour of the text interpreter:
4657:
4658: @itemize @bullet
4659: @item
4660: It processes each input string to completion before parsing additional
4661: characters from the input buffer.
4662: @item
4663: It treats the input buffer as a read-only region (and so must your code).
4664: @end itemize
4665:
4666: @noindent
4667: When the text interpreter is in compile state, its behaviour changes in
4668: these ways:
4669:
4670: @itemize @bullet
4671: @item
4672: If a parsed string is found in the dictionary, the text interpreter will
4673: perform the word's @dfn{compilation semantics}. In most cases, this
4674: simply means that the execution semantics of the word will be appended
4675: to the current definition.
4676: @item
4677: When a number is encountered, it is compiled into the current definition
4678: (as a literal) rather than being pushed onto a parameter stack.
4679: @item
4680: If an error occurs, @code{state} is modified to put the text interpreter
4681: back into interpret state.
4682: @item
4683: Each time a line is entered from the keyboard, Gforth prints
4684: ``@code{ compiled}'' rather than `` @code{ok}''.
4685: @end itemize
4686:
4687: @cindex text interpreter - input sources
4688: When the text interpreter is using an input device other than the
4689: keyboard, its behaviour changes in these ways:
4690:
4691: @itemize @bullet
4692: @item
4693: When the parse area is empty, the text interpreter attempts to refill
4694: the input buffer from the input source. When the input source is
4695: exhausted, the input source is set back to the user input device.
4696: @item
4697: It doesn't print out ``@code{ ok}'' or ``@code{ compiled}'' messages each
4698: time the parse area is emptied.
4699: @item
4700: If an error occurs, the input source is set back to the user input
4701: device.
4702: @end itemize
4703:
4704: @ref{Input Sources} describes this in more detail.
4705:
4706: doc->in
4707: doc-source
4708:
4709: doc-tib
4710: doc-#tib
4711:
4712: @menu
4713: * Input Sources::
4714: * Number Conversion::
4715: * Interpret/Compile states::
4716: * Literals::
4717: * Interpreter Directives::
4718: @end menu
4719:
4720: @node Input Sources, Number Conversion, The Text Interpreter, The Text Interpreter
4721: @subsection Input Sources
4722: @cindex input sources
4723: @cindex text interpreter - input sources
4724:
4725: By default, the text interpreter accepts input from the user input
4726: device (the keyboard) when Forth starts up. The text interpreter can
4727: process input from any of these sources:
4728:
4729: @itemize @bullet
4730: @item
4731: The user input device -- the keyboard.
4732: @item
4733: A file, using the words described in @ref{Forth source files}.
4734: @item
4735: A block, using the words described in @ref{Blocks}.
4736: @item
4737: A text string, using @code{evaluate}.
4738: @end itemize
4739:
4740: A program can identify the current input device from the values of
4741: @code{source-id} and @code{blk}.
4742:
4743: doc-source-id
4744: doc-blk
4745:
4746: doc-save-input
4747: doc-restore-input
4748:
4749: doc-evaluate
4750:
4751:
4752: @node Number Conversion, Interpret/Compile states, Input Sources, The Text Interpreter
4753: @subsection Number Conversion
4754: @cindex number conversion
4755: @cindex double-cell numbers, input format
4756: @cindex input format for double-cell numbers
4757: @cindex single-cell numbers, input format
4758: @cindex input format for single-cell numbers
4759: @cindex floating-point numbers, input format
4760: @cindex input format for floating-point numbers
4761:
4762: This section describes the rules that the text interpreter uses when it
4763: tries to convert a string into a number.
4764:
4765: Let <digit> represent any character that is a legal digit in the current
4766: number base@footnote{For example, 0-9 when the number base is decimal or
4767: 0-9, A-F when the number base is hexadecimal.}.
4768:
4769: Let <decimal digit> represent any character in the range 0-9.
4770:
4771: Let @{@i{a b}@} represent the @i{optional} presence of any of the characters
4772: in the braces (@i{a} or @i{b} or neither).
4773:
4774: Let * represent any number of instances of the previous character
4775: (including none).
4776:
4777: Let any other character represent itself.
4778:
4779: @noindent
4780: Now, the conversion rules are:
4781:
4782: @itemize @bullet
4783: @item
4784: A string of the form <digit><digit>* is treated as a single-precision
4785: (cell-sized) positive integer. Examples are 0 123 6784532 32343212343456 42
4786: @item
4787: A string of the form -<digit><digit>* is treated as a single-precision
4788: (cell-sized) negative integer, and is represented using 2's-complement
4789: arithmetic. Examples are -45 -5681 -0
4790: @item
4791: A string of the form <digit><digit>*.<digit>* is treated as a double-precision
4792: (double-cell-sized) positive integer. Examples are 3465. 3.465 34.65
4793: (all three of these represent the same number).
4794: @item
4795: A string of the form -<digit><digit>*.<digit>* is treated as a
4796: double-precision (double-cell-sized) negative integer, and is
4797: represented using 2's-complement arithmetic. Examples are -3465. -3.465
4798: -34.65 (all three of these represent the same number).
4799: @item
4800: A string of the form @{+ -@}<decimal digit>@{.@}<decimal digit>*@{e
4801: E@}@{+ -@}<decimal digit><decimal digit>* is treated as a floating-point
4802: number. Examples are 1e0 1.e 1.e0 +1e+0 (which all represent the same
4803: number) +12.E-4
4804: @end itemize
4805:
4806: By default, the number base used for integer number conversion is given
4807: by the contents of the variable @code{BASE}. Base 10 (decimal) is
4808: always used for floating-point number conversion.
4809:
4810: doc-dpl
4811: doc-base
4812: doc-hex
4813: doc-decimal
4814:
4815: @cindex '-prefix for character strings
4816: @cindex &-prefix for decimal numbers
4817: @cindex %-prefix for binary numbers
4818: @cindex $-prefix for hexadecimal numbers
4819: Gforth allows you to override the value of @code{BASE} by using a
4820: prefix@footnote{Some Forth implementations provide a similar scheme by
4821: implementing @code{$} etc. as parsing words that process the subsequent
4822: number in the input stream and push it onto the stack. For example, see
4823: @cite{Number Conversion and Literals}, by Wil Baden; Forth Dimensions
4824: 20(3) pages 26--27. In such implementations, unlike in Gforth, a space
4825: is required between the prefix and the number.} before the first digit
4826: of an (integer) number. Four prefixes are supported:
4827:
4828: @itemize @bullet
4829: @item
4830: @code{&} -- decimal number
4831: @item
4832: @code{%} -- binary number
4833: @item
4834: @code{$} -- hexadecimal number
4835: @item
4836: @code{'} -- base 256 number
4837: @end itemize
4838:
4839: Here are some examples, with the equivalent decimal number shown after
4840: in braces:
4841:
4842: -$41 (-65), %1001101 (205), %1001.0001 (145 - a double-precision number),
4843: 'AB (16706; ascii A is 65, ascii B is 66, number is 65*256 + 66),
4844: 'ab (24930; ascii a is 97, ascii B is 98, number is 97*256 + 98),
4845: &905 (905), $abc (2478), $ABC (2478).
4846:
4847: @cindex number conversion - traps for the unwary
4848: @noindent
4849: Number conversion has a number of traps for the unwary:
4850:
4851: @itemize @bullet
4852: @item
4853: You cannot determine the current number base using the code sequence
4854: @code{BASE @@ .} -- the number base is always 10 in the current number
4855: base. Instead, use something like @code{BASE @@ DECIMAL DUP . BASE !}
4856: @item
4857: If the number base is set to a value greater than 14 (for example,
4858: hexadecimal), the number 123E4 is ambiguous; the conversion rules allow
4859: it to be intepreted as either a single-precision integer or a
4860: floating-point number (Gforth treats it as an integer). The ambiguity
4861: can be resolved by explicitly stating the sign of the mantissa and/or
4862: exponent: 123E+4 or +123E4 -- if the number base is decimal, no
4863: ambiguity arises; either representation will be treated as a
4864: floating-point number.
4865: @item
4866: There is a word @code{bin} but it does @i{not} set the number base!
4867: It is used to specify file types.
4868: @item
4869: ANS Forth requires the @code{.} of a double-precision number to
4870: be the final character in the string. Allowing the @code{.} to be
4871: anywhere after the first digit is a Gforth extension.
4872: @item
4873: The number conversion process does not check for overflow.
4874: @item
4875: In Gforth, number conversion to floating-point numbers always use base
4876: 10, irrespective of the value of @code{BASE}. In ANS Forth,
4877: conversion to floating-point numbers whilst the value of
4878: @code{BASE} is not 10 is an ambiguous condition.
4879: @end itemize
4880:
4881: @ref{Input} describes words that you can use to read numbers into your
4882: programs.
4883:
4884: @node Interpret/Compile states, Literals, Number Conversion, The Text Interpreter
4885: @subsection Interpret/Compile states
4886: @cindex Interpret/Compile states
4887:
4888: A standard program is not permitted to change @code{state}
4889: explicitly. However, it can change @code{state} implicitly, using the
4890: words @code{[} and @code{]}. When @code{[} is executed it switches
4891: @code{state} to interpret state, and therefore the text interpreter
4892: starts interpreting. When @code{]} is executed it switches @code{state}
4893: to compile state and therefore the text interpreter starts
4894: compiling. The most common usage for these words is to compile literals,
4895: as shown in @ref{Literals}. However, they give you the freedom to switch
4896: modes at will. Here is an example of building a jump-table of execution
4897: tokens:
4898:
4899: @example
4900: : AA ." this is A" ;
4901: : BB ." this is B" ;
4902: : CC ." this is C" ;
4903:
4904: create table ' aa COMPILE, ' bb COMPILE, ' cc COMPILE,
4905: : go ( n -- ) \ n is offset into table.. 0 for 1st entry
4906: cells table + @ execute ;
4907: @end example
4908:
4909: @noindent
4910: Now @code{0 go} will display ``@code{this is A}''. The table can be
4911: built far more neatly@footnote{The source code is neater.. what is
4912: compiled in memory in each case is identical.} like this:
4913:
4914: @example
4915: create table ] aa bb cc [
4916: @end example
4917:
4918: The problem with this code is that it is not portable; it will only work
4919: on systems where code space and data space co-incide. The reason is that
4920: both tables @i{compile} execution tokens -- into code space. The
4921: Standard only allows data space to be assigned for a @code{CREATE}d
4922: word. In addition, the Standard only allows @code{@@} to access data
4923: space, whilst this example is using it to access code space. The only
4924: portable, Standard way to build this table is to build it in data space,
4925: like this:
4926:
4927: @example
4928: create table ' aa , ' bb , ' cc ,
4929: @end example
4930:
4931: @noindent
4932: A similar technique can be used to build a table of constants:
4933:
4934: @example
4935: create primes 1 , 3 , 5 , 7 , 11 ,
4936: @end example
4937:
4938: doc-state
4939: doc-[
4940: doc-]
4941:
4942: @node Literals, Interpreter Directives, Interpret/Compile states, The Text Interpreter
4943: @subsection Literals
4944: @cindex Literals
4945:
4946: Often, you want to use a number within a colon definition. When you do
4947: this, the text interpreter automatically compiles the number as a
4948: @i{literal}. A literal is a number whose run-time effect is to be pushed
4949: onto the stack. If you had to do some maths to generate the number, you
4950: might write it like this:
4951:
4952: @example
4953: : HOUR-TO-SEC ( n1 -- n2 )
4954: 60 * \ to minutes
4955: 60 * ; \ to seconds
4956: @end example
4957:
4958: It is very clear what this definition is doing, but it's inefficient
4959: since it is performing 2 multiples at run-time. An alternative would be
4960: to write:
4961:
4962: @example
4963: : HOUR-TO-SEC ( n1 -- n2 )
4964: 3600 * ; \ to seconds
4965: @end example
4966:
4967: Which does the same thing, and has the advantage of using a single
4968: multiply. Ideally, we'd like the efficiency of the second with the
4969: readability of the first.
4970:
4971: @code{Literal} allows us to achieve that. It takes a number from the
4972: stack and lays it down in the current definition just as though the
4973: number had been typed directly into the definition. Our first attempt
4974: might look like this:
4975:
4976: @example
4977: 60 \ mins per hour
4978: 60 * \ seconds per minute
4979: : HOUR-TO-SEC ( n1 -- n2 )
4980: Literal * ; \ to seconds
4981: @end example
4982:
4983: But this produces the error message @code{unstructured}. What happened?
4984: The stack notation for @code{:} is (@i{ -- colon-sys}) and the size of
4985: @i{colon-sys} is implementation-defined. In other words, once we start a
4986: colon definition we can't portably access anything that was on the stack
4987: before the definition began@footnote{@cite{Two Problems in ANS Forth},
4988: by Thomas Worthington; Forth Dimensions 20(2) pages 32--34 describes
4989: some situations where you might want to access stack items above
4990: colon-sys, and provides a solution to the problem.}. The correct way of
4991: solving this problem in this instance is to use @code{[ ]} like this:
4992:
4993: @example
4994: : HOUR-TO-SEC ( n1 -- n2 )
4995: [ 60 \ minutes per hour
4996: 60 * ] \ seconds per minute
4997: LITERAL * ; \ to seconds
4998: @end example
4999:
5000: doc-literal
5001: doc-]L
5002: doc-2literal
5003: doc-fliteral
5004:
5005: @node Interpreter Directives, , Literals, The Text Interpreter
5006: @subsection Interpreter Directives
5007: @cindex interpreter directives
5008:
5009: These words are usually used in interpret state; typically to control
5010: which parts of a source file are processed by the text
5011: interpreter. There are only a few ANS Forth Standard words, but Gforth
5012: supplements these with a rich set of immediate control structure words
5013: to compensate for the fact that the non-immediate versions can only be
5014: used in compile state (@pxref{Control Structures}). Typical usages:
5015:
5016: @example
5017: FALSE Constant ASSEMBLER
5018: .
5019: .
5020: ASSEMBLER [IF]
5021: : ASSEMBLER-FEATURE
5022: ...
5023: ;
5024: [ENDIF]
5025: .
5026: .
5027: : SEE
5028: ... \ general-purpose SEE code
5029: [ ASSEMBLER [IF] ]
5030: ... \ assembler-specific SEE code
5031: [ [ENDIF] ]
5032: ;
5033: @end example
5034:
5035: doc-[IF]
5036: doc-[ELSE]
5037: doc-[THEN]
5038: doc-[ENDIF]
5039:
5040: doc-[IFDEF]
5041: doc-[IFUNDEF]
5042:
5043: doc-[?DO]
5044: doc-[DO]
5045: doc-[FOR]
5046: doc-[LOOP]
5047: doc-[+LOOP]
5048: doc-[NEXT]
5049:
5050: doc-[BEGIN]
5051: doc-[UNTIL]
5052: doc-[AGAIN]
5053: doc-[WHILE]
5054: doc-[REPEAT]
5055:
5056:
5057:
5058: @c -------------------------------------------------------------
5059: @node Tokens for Words, Word Lists, The Text Interpreter, Words
5060: @section Tokens for Words
5061: @cindex tokens for words
5062:
5063: This section describes the creation and use of tokens that represent
5064: words.
5065:
5066: Named words have information stored in their name dictionary entries to
5067: indicate any non-default semantics (@pxref{Interpretation and
5068: Compilation Semantics}). The semantics can be modified, using
5069: @code{immediate} and/or @code{compile-only}, at the time that the words
5070: are defined. Unnamed words have (by definition) no name dictionary
5071: entry, and therefore must have default semantics.
5072:
5073: Named words have interpretation and compilation semantics. Unnamed words
5074: just have execution semantics.
5075:
5076: @cindex xt
5077: @cindex execution token
5078: The execution semantics of an unnamed word are represented by an
5079: @dfn{execution token} (@i{xt}). As explained in @ref{Supplying names},
5080: the execution token of the last word defined can be produced with
5081: @code{lastxt}.
5082:
5083: The interpretation semantics of a named word are also represented by an
5084: execution token. You can produce the execution token using @code{'} or
5085: @code{[']}. A simple example shows the difference between the two:
5086:
5087: @example
5088: : greet ( -- ) ." Hello" ;
5089: : foo ( -- xt ) ['] greet ; \ ['] parses greet at compile-time
5090: : bar ( -- ) ' EXECUTE ; \ ' parses at run-time
5091:
5092: \ the next four lines all do the same thing
5093: foo EXECUTE
5094: greet
5095: ' greet EXECUTE
5096: boo greet
5097: @end example
5098:
5099: An execution token occupies one cell.
5100: @cindex code field address
5101: @cindex CFA
5102: In Gforth, the abstract data type @i{execution token} is implemented
5103: as a code field address (CFA).
5104: @comment TODO note that the standard does not say what it represents..
5105: @comment and you cannot necessarily compile it in all Forths (eg native
5106: @comment compilers?).
5107:
5108: For literals, use @code{'} in interpreted code and @code{[']} in
5109: compiled code. Gforth's @code{'} and @code{[']} behave somewhat
5110: unusually by complaining about compile-only words. To get the execution
5111: token for a compile-only word @i{name}, use @code{COMP' @i{name} DROP}
5112: or @code{[COMP'] @i{name} DROP}.
5113:
5114: @cindex compilation token
5115: The compilation semantics of a named word are represented by a
5116: @dfn{compilation token} consisting of two cells: @i{w xt}. The top cell
5117: @i{xt} is an execution token. The compilation semantics represented by
5118: the compilation token can be performed with @code{execute}, which
5119: consumes the whole compilation token, with an additional stack effect
5120: determined by the represented compilation semantics.
5121:
5122: At present, the @i{w} part of a compilation token is an execution token,
5123: and the @i{xt} part represents either @code{execute} or
5124: @code{compile,}@footnote{Depending upon the compilation semantics of the
5125: word. If the word has default compilation semantics, the @i{xt} will
5126: represent @code{compile,}. If the word is @code{immediate}, the @i{xt}
5127: will represent @code{execute}.}. However, don't rely on that knowledge,
5128: unless necessary; future versions of Gforth may introduce unusual
5129: compilation tokens (e.g., a compilation token that represents the
5130: compilation semantics of a literal).
5131:
5132: You can compile the compilation semantics with @code{postpone,}. I.e.,
5133: @code{COMP' @i{word} postpone,} is equivalent to @code{postpone
5134: @i{word}}.
5135:
5136: @cindex name token
5137: @cindex name field address
5138: @cindex NFA
5139: Named words are also represented by the @dfn{name token}, (@i{nt}). In
5140: Gforth, the abstract data type @emph{name token} is implemented as a
5141: name field address (NFA).
5142:
5143: doc-execute
5144: doc-compile,
5145: doc-[']
5146: doc-'
5147: doc-[comp']
5148: doc-comp'
5149: doc-postpone,
5150:
5151: doc-find-name
5152: doc-name>int
5153: doc-name?int
5154: doc-name>comp
5155: doc-name>string
5156:
5157: @c -------------------------------------------------------------
5158: @node Word Lists, Environmental Queries, Tokens for Words, Words
5159: @section Word Lists
5160: @cindex word lists
5161: @cindex name dictionary
5162:
5163: @cindex wid
5164: All definitions other than those created by @code{:noname} have an entry
5165: in the name dictionary. The name dictionary is fragmented into a number
5166: of parts, called @dfn{word lists}. A word list is identified by a
5167: cell-sized word list identifier (@i{wid}) in much the same way as a
5168: file is identified by a file handle. The numerical value of the wid has
5169: no (portable) meaning, and might change from session to session.
5170:
5171: @cindex compilation word list
5172: At any one time, a single word list is defined as the word list to which
5173: all new definitions will be added -- this is called the @dfn{compilation
5174: word list}. When Gforth is started, the compilation word list is the
5175: word list called @code{FORTH-WORDLIST}.
5176:
5177: @cindex search order stack
5178: Forth maintains a stack of word lists, representing the @dfn{search
5179: order}. When the name dictionary is searched (for example, when
5180: attempting to find a word's execution token during compilation), only
5181: those word lists that are currently in the search order are
5182: searched. The most recently-defined word in the word list at the top of
5183: the word list stack is searched first, and the search proceeds until
5184: either the word is located or the oldest definition in the word list at
5185: the bottom of the stack is reached. Definitions of the word may exist in
5186: more than one word lists; the search order determines which version will
5187: be found.
5188:
5189: The ANS Forth ``Search order'' word set is intended to provide a set of
5190: low-level tools that allow various different schemes to be
5191: implemented. Gforth provides @code{vocabulary}, a traditional Forth
5192: word. @file{compat/vocabulary.fs} provides an implementation in ANS
5193: Standard Forth.
5194:
5195: @comment TODO: locals section refers to here, saying that every word list (aka
5196: @comment vocabulary) has its own methods for searching etc. Need to document that.
5197:
5198: @comment the thisone- prefix is used to pick out the true definition of a
5199: @comment word from the source files, rather than some alias.
5200: doc-forth-wordlist
5201: doc-definitions
5202: doc-get-current
5203: doc-set-current
5204: doc-get-order
5205: doc---thisone-set-order
5206: doc-wordlist
5207: doc-table
5208: doc-also
5209: doc---thisone-forth
5210: doc-only
5211: doc---thisone-order
5212: doc-previous
5213:
5214: doc-find
5215: doc-search-wordlist
5216:
5217: doc-words
5218: doc-vlist
5219:
5220: doc-mappedwordlist
5221: doc-root
5222: doc-vocabulary
5223: doc-seal
5224: doc-vocs
5225: doc-current
5226: doc-context
5227:
5228: @menu
5229: * Why use word lists?::
5230: * Word list examples::
5231: @end menu
5232:
5233: @node Why use word lists?, Word list examples, Word Lists, Word Lists
5234: @subsection Why use word lists?
5235: @cindex word lists - why use them?
5236:
5237: Here are some reasons for using multiple word lists:
5238:
5239: @itemize @bullet
5240: @item
5241: To improve compilation speed by reducing the number of name dictionary
5242: entries that must be searched. This is achieved by creating a new
5243: word list that contains all of the definitions that are used in the
5244: definition of a Forth system but which would not usually be used by
5245: programs running on that system. That word list would be on the search
5246: list when the Forth system was compiled but would be removed from the
5247: search list for normal operation. This can be a useful technique for
5248: low-performance systems (for example, 8-bit processors in embedded
5249: systems) but is unlikely to be necessary in high-performance desktop
5250: systems.
5251: @item
5252: To prevent a set of words from being used outside the context in which
5253: they are valid. Two classic examples of this are an integrated editor
5254: (all of the edit commands are defined in a separate word list; the
5255: search order is set to the editor word list when the editor is invoked;
5256: the old search order is restored when the editor is terminated) and an
5257: integrated assembler (the op-codes for the machine are defined in a
5258: separate word list which is used when a @code{CODE} word is defined).
5259: @item
5260: To prevent a name-space clash between multiple definitions with the same
5261: name. For example, when building a cross-compiler you might have a word
5262: @code{IF} that generates conditional code for your target system. By
5263: placing this definition in a different word list you can control whether
5264: the host system's @code{IF} or the target system's @code{IF} get used in
5265: any particular context by controlling the order of the word lists on the
5266: search order stack.
5267: @end itemize
5268:
5269: @node Word list examples, ,Why use word lists?, Word Lists
5270: @subsection Word list examples
5271: @cindex word lists - examples
5272:
5273: Here is an example of creating and using a new wordlist using ANS
5274: Forth Standard words:
5275:
5276: @example
5277: wordlist constant my-new-words-wordlist
5278: : my-new-words get-order nip my-new-words-wordlist swap set-order ;
5279:
5280: \ add it to the search order
5281: also my-new-words
5282:
5283: \ alternatively, add it to the search order and make it
5284: \ the compilation word list
5285: also my-new-words definitions
5286: \ type "order" to see the problem
5287: @end example
5288:
5289: The problem with this example is that @code{order} has no way to
5290: associate the name @code{my-new-words} with the wid of the word list (in
5291: Gforth, @code{order} and @code{vocs} will display @code{???} for a wid
5292: that has no associated name). There is no Standard way of associating a
5293: name with a wid.
5294:
5295: In Gforth, this example can be re-coded using @code{vocabulary}, which
5296: associates a name with a wid:
5297:
5298: @example
5299: vocabulary my-new-words
5300:
5301: \ add it to the search order
5302: my-new-words
5303:
5304: \ alternatively, add it to the search order and make it
5305: \ the compilation word list
5306: my-new-words definitions
5307: \ type "order" to see that the problem is solved
5308: @end example
5309:
5310: @c -------------------------------------------------------------
5311: @node Environmental Queries, Files, Word Lists, Words
5312: @section Environmental Queries
5313: @cindex environmental queries
5314:
5315: ANS Forth introduced the idea of ``environmental queries'' as a way
5316: for a program running on a system to determine certain characteristics of the system.
5317: The Standard specifies a number of strings that might be recognised by a system.
5318:
5319: The Standard requires that the name space used for environmental queries
5320: be distinct from the name space used for definitions.
5321:
5322: Typically, environmental queries are supported by creating a set of
5323: definitions in a word list that is @i{only} used during environmental
5324: queries; that is what Gforth does. There is no Standard way of adding
5325: definitions to the set of recognised environmental queries, but any
5326: implementation that supports the loading of optional word sets must have
5327: some mechanism for doing this (after loading the word set, the
5328: associated environmental query string must return @code{true}). In
5329: Gforth, the word list used to honour environmental queries can be
5330: manipulated just like any other word list.
5331:
5332: doc-environment?
5333: doc-environment-wordlist
5334:
5335: doc-gforth
5336: doc-os-class
5337:
5338: Note that, whilst the documentation for (e.g.) @code{gforth} shows it
5339: returning two items on the stack, querying it using @code{environment?}
5340: will return an additional item; the @code{true} flag that shows that the
5341: string was recognised.
5342:
5343: @comment TODO Document the standard strings or note where they are documented herein
5344:
5345: Here are some examples of using environmental queries:
5346:
5347: @example
5348: s" address-unit-bits" environment? 0=
5349: [IF]
5350: cr .( environmental attribute address-units-bits unknown... ) cr
5351: [THEN]
5352:
5353: s" block" environment? [IF] DROP include block.fs [THEN]
5354:
5355: s" gforth" environment? [IF] 2DROP include compat/vocabulary.fs [THEN]
5356:
5357: s" gforth" environment? [IF] .( Gforth version ) TYPE
5358: [ELSE] .( Not Gforth..) [THEN]
5359: @end example
5360:
5361:
5362: Here is an example of adding a definition to the environment word list:
5363:
5364: @example
5365: get-current environment-wordlist set-current
5366: true constant block
5367: true constant block-ext
5368: set-current
5369: @end example
5370:
5371: You can see what definitions are in the environment word list like this:
5372:
5373: @example
5374: get-order 1+ environment-wordlist swap set-order words previous
5375: @end example
5376:
5377:
5378: @c -------------------------------------------------------------
5379: @node Files, Blocks, Environmental Queries, Words
5380: @section Files
5381: @cindex files
5382: @cindex I/O - file-handling
5383:
5384: Gforth provides facilities for accessing files that are stored in the
5385: host operating system's file-system. Files that are processed by Gforth
5386: can be divided into two categories:
5387:
5388: @itemize @bullet
5389: @item
5390: Files that are processed by the Text Interpreter (@dfn{Forth source files}).
5391: @item
5392: Files that are processed by some other program (@dfn{general files}).
5393: @end itemize
5394:
5395: @menu
5396: * Forth source files::
5397: * General files::
5398: * Search Paths::
5399: * Forth Search Paths::
5400: * General Search Paths::
5401: @end menu
5402:
5403:
5404: @c -------------------------------------------------------------
5405: @node Forth source files, General files, Files, Files
5406: @subsection Forth source files
5407: @cindex including files
5408: @cindex Forth source files
5409:
5410: The simplest way to interpret the contents of a file is to use one of
5411: these two formats:
5412:
5413: @example
5414: include mysource.fs
5415: s" mysource.fs" included
5416: @end example
5417:
5418: Sometimes you want to include a file only if it is not included already
5419: (by, say, another source file). In that case, you can use one of these
5420: fomats:
5421:
5422: @example
5423: require mysource.fs
5424: needs mysource.fs
5425: s" mysource.fs" required
5426: @end example
5427:
5428: @cindex stack effect of included files
5429: @cindex including files, stack effect
5430: I recommend that you write your source files such that interpreting them
5431: does not change the stack. This allows using these files with
5432: @code{required} and friends without complications. For example:
5433:
5434: @example
5435: 1 require foo.fs drop
5436: @end example
5437:
5438: doc-include-file
5439: doc-included
5440: doc-included?
5441: doc-include
5442: doc-required
5443: doc-require
5444: doc-needs
5445: doc-init-included-files
5446:
5447: A definition in ANS Forth for @code{required} is provided in
5448: @file{compat/required.fs}.
5449:
5450: @c -------------------------------------------------------------
5451: @node General files, Search Paths, Forth source files, Files
5452: @subsection General files
5453: @cindex general files
5454: @cindex file-handling
5455:
5456: Files are opened/created by name and type. The following types are
5457: recognised:
5458:
5459: doc-r/o
5460: doc-r/w
5461: doc-w/o
5462: doc-bin
5463:
5464: When a file is opened/created, it returns a file identifier,
5465: @i{wfileid} that is used for all other file commands. All file
5466: commands also return a status value, @i{wior}, that is 0 for a
5467: successful operation and an implementation-defined non-zero value in the
5468: case of an error.
5469:
5470: doc-open-file
5471: doc-create-file
5472:
5473: doc-close-file
5474: doc-delete-file
5475: doc-rename-file
5476: doc-read-file
5477: doc-read-line
5478: doc-write-file
5479: doc-write-line
5480: doc-emit-file
5481: doc-flush-file
5482:
5483: doc-file-status
5484: doc-file-position
5485: doc-reposition-file
5486: doc-file-size
5487: doc-resize-file
5488:
5489: @c ---------------------------------------------------------
5490: @node Search Paths, Forth Search Paths, General files, Files
5491: @subsection Search Paths
5492: @cindex path for @code{included}
5493: @cindex file search path
5494: @cindex @code{include} search path
5495: @cindex search path for files
5496:
5497: If you specify an absolute filename (i.e., a filename starting with
5498: @file{/} or @file{~}, or with @file{:} in the second position (as in
5499: @samp{C:...})) for @code{included} and friends, that file is included
5500: just as you would expect.
5501:
5502: For relative filenames, Gforth uses a search path similar to Forth's
5503: search order (@pxref{Word Lists}). It tries to find the given filename
5504: in the directories present in the path, and includes the first one it
5505: finds. There are separate search paths for Forth source files and
5506: general files.
5507:
5508: If the search path contains the directory @file{.} (as it should), this
5509: refers to the directory that the present file was @code{included}
5510: from. This allows files to include other files relative to their own
5511: position (irrespective of the current working directory or the absolute
5512: position). This feature is essential for libraries consisting of
5513: several files, where a file may include other files from the library.
5514: It corresponds to @code{#include "..."} in C. If the current input
5515: source is not a file, @file{.} refers to the directory of the innermost
5516: file being included, or, if there is no file being included, to the
5517: current working directory.
5518:
5519: Use @file{~+} to refer to the current working directory (as in the
5520: @code{bash}).
5521:
5522: If the filename starts with @file{./}, the search path is not searched
5523: (just as with absolute filenames), and the @file{.} has the same meaning
5524: as described above.
5525:
5526: @c ---------------------------------------------------------
5527: @node Forth Search Paths, General Search Paths, Search Paths, Files
5528: @subsubsection Forth Search Paths
5529: @cindex search path control - Forth
5530:
5531: The search path is initialized when you start Gforth (@pxref{Invoking
5532: Gforth}). You can display it and change it using these words:
5533:
5534: doc-.fpath
5535: doc-fpath+
5536: doc-fpath=
5537: doc-open-fpath-file
5538:
5539: Here is an example of using @code{fpath} and @code{require}:
5540:
5541: @example
5542: fpath= /usr/lib/forth/|./
5543: require timer.fs
5544: @end example
5545:
5546: @c ---------------------------------------------------------
5547: @node General Search Paths, , Forth Search Paths, Files
5548: @subsubsection General Search Paths
5549: @cindex search path control - for user applications
5550:
5551: Your application may need to search files in several directories, like
5552: @code{included} does. To facilitate this, Gforth allows you to define
5553: and use your own search paths, by providing generic equivalents of the
5554: Forth search path words:
5555:
5556: doc-.path
5557: doc-path+
5558: doc-path=
5559: doc-open-path-file
5560:
5561: Here's an example of creating a search path:
5562:
5563: @example
5564: \ Make a buffer for the path:
5565: create mypath 100 chars , \ maximum length (is checked)
5566: 0 , \ real len
5567: 100 chars allot \ space for path
5568: @end example
5569:
5570: @c -------------------------------------------------------------
5571: @node Blocks, Other I/O, Files, Words
5572: @section Blocks
5573: @cindex I/O - blocks
5574: @cindex blocks
5575:
5576: When you run Gforth on a modern desk-top computer, it runs under the
5577: control of an operating system which provides certain services. One of
5578: these services is @var{file services}, which allows Forth source code
5579: and data to be stored in files and read into Gforth (@pxref{Files}).
5580:
5581: Traditionally, Forth has been an important programming language on
5582: systems where it has interfaced directly to the underlying hardware with
5583: no intervening operating system. Forth provides a mechanism, called
5584: @dfn{blocks}, for accessing mass storage on such systems.
5585:
5586: A block is a 1024-byte data area, which can be used to hold data or
5587: Forth source code. No structure is imposed on the contents of the
5588: block. A block is identified by its number; blocks are numbered
5589: contiguously from 1 to an implementation-defined maximum.
5590:
5591: A typical system that used blocks but no operating system might use a
5592: single floppy-disk drive for mass storage, with the disks formatted to
5593: provide 256-byte sectors. Blocks would be implemented by assigning the
5594: first four sectors of the disk to block 1, the second four sectors to
5595: block 2 and so on, up to the limit of the capacity of the disk. The disk
5596: would not contain any file system information, just the set of blocks.
5597:
5598: @cindex blocks file
5599: On systems that do provide file services, blocks are typically
5600: implemented by storing a sequence of blocks within a single @dfn{blocks
5601: file}. The size of the blocks file will be an exact multiple of 1024
5602: bytes, corresponding to the number of blocks it contains. This is the
5603: mechanism that Gforth uses.
5604:
5605: @cindex @file{blocks.fb}
5606: Only 1 blocks file can be open at a time. If you use block words without
5607: having specified a blocks file, Gforth defaults to the blocks file
5608: @file{blocks.fb}. Gforth uses the Forth search path when attempting to
5609: locate a blocks file (@pxref{Forth Search Paths}).
5610:
5611: @cindex block buffers
5612: When you read and write blocks under program control, Gforth uses a
5613: number of @dfn{block buffers} as intermediate storage. These buffers are
5614: not used when you use @code{load} to interpret the contents of a block.
5615:
5616: The behaviour of the block buffers is directly analagous to that of a
5617: cache. Each block buffer has three states:
5618:
5619: @itemize @bullet
5620: @item
5621: Unassigned
5622: @item
5623: Assigned-clean
5624: @item
5625: Assigned-dirty
5626: @end itemize
5627:
5628: Initially, all block buffers are @i{unassigned}. In order to access a
5629: block, the block (specified by its block number) must be assigned to a
5630: block buffer.
5631:
5632: The assignment of a block to a block buffer is performed by @code{block}
5633: or @code{buffer}. Use @code{block} when you wish to modify the existing
5634: contents of a block. Use @code{buffer} when you don't care about the
5635: existing contents of the block@footnote{The ANS Forth definition of
5636: @code{block} is intended not to cause disk I/O; if the data associated
5637: with the particular block is already stored in a block buffer due to an
5638: earlier @code{block} command, @code{buffer} will return that block
5639: buffer and the existing contents of the block will be
5640: available. Otherwise, @code{buffer} will simply assign a new, empty
5641: block buffer for the block.}.
5642:
5643: Once a block has been assigned to a block buffer, the block buffer state
5644: becomes @i{assigned-clean}. Data can now be manipulated within the
5645: block buffer.
5646:
5647: When the contents of a block buffer is changed it is necessary,
5648: @i{before calling} @code{block} @i{or} @code{buffer} @i{again}, to
5649: either abandon the changes (by doing nothing) or commit the changes,
5650: using @code{update}. Using @code{update} does not change the blocks
5651: file; it simply changes a block buffer's state to @i{assigned-dirty}.
5652:
5653: The word @code{flush} causes all @i{assigned-dirty} blocks to be
5654: written back to the blocks file on disk. Leaving Gforth using @code{bye}
5655: also causes a @code{flush} to be performed.
5656:
5657: In Gforth, @code{block} and @code{buffer} use a @i{direct-mapped}
5658: algorithm to assign a block buffer to a block. That means that any
5659: particular block can only be assigned to one specific block buffer,
5660: called (for the particular operation) the @i{victim buffer}. If the
5661: victim buffer is @i{unassigned} or @i{assigned-clean} it can be
5662: allocated to the new block immediately. If it is @i{assigned-dirty}
5663: its current contents must be written out to disk before it can be
5664: allocated to the new block.
5665:
5666: Although no structure is imposed on the contents of a block, it is
5667: traditional to display the contents as 16 lines each of 64 characters. A
5668: block provides a single, continuous stream of input (for example, it
5669: acts as a single parse area) -- there are no end-of-line characters
5670: within a block, and no end-of-file character at the end of a
5671: block. There are two consequences of this:
5672:
5673: @itemize @bullet
5674: @item
5675: The last character of one line wraps straight into the first character
5676: of the following line
5677: @item
5678: The word @code{\} -- comment to end of line -- requires special
5679: treatment; in the context of a block it causes all characters until the
5680: end of the current 64-character ``line'' to be ignored.
5681: @end itemize
5682:
5683: In Gforth, when you use @code{block} with a non-existent block number,
5684: the current block file will be extended to the appropriate size and the
5685: block buffer will be initialised with spaces.
5686:
5687: Gforth doesn't encourage the use of blocks; the mechanism is only
5688: provided for backward compatibility -- ANS Forth requires blocks to be
5689: available when files are.
5690:
5691: Common techniques that are used when working with blocks include:
5692:
5693: @itemize @bullet
5694: @item
5695: A screen editor that allows you to edit blocks without leaving the Forth
5696: environment.
5697: @item
5698: Shadow screens; where every code block has an associated block
5699: containing comments (for example: code in odd block numbers, comments in
5700: even block numbers). Typically, the block editor provides a convenient
5701: mechanism to toggle between code and comments.
5702: @item
5703: Load blocks; a single block (typically block 1) contains a number of
5704: @code{thru} commands which @code{load} the whole of the application.
5705: @item
5706: Chaining blocks; a block terminates with a @code{-->} so that a whole
5707: application can be @code{load}ed by @code{load}ing a single block.
5708: @end itemize
5709:
5710: See Frank Sergeant's Pygmy Forth to see just how well blocks can be
5711: integrated into a Forth programming environment.
5712:
5713: @comment TODO what about errors on open-blocks?
5714: doc-open-blocks
5715: doc-use
5716: doc-get-block-fid
5717: doc-block-position
5718:
5719: doc-scr
5720: doc-list
5721:
5722: doc---block-block
5723: doc-buffer
5724:
5725: doc-update
5726: doc-updated?
5727: doc-save-buffers
5728: doc-empty-buffers
5729: doc-empty-buffer
5730: doc-flush
5731:
5732: doc-load
5733: doc-thru
5734: doc-+load
5735: doc-+thru
5736: doc---block--->
5737: doc-block-included
5738:
5739: @c -------------------------------------------------------------
5740: @node Other I/O, Programming Tools, Blocks, Words
5741: @section Other I/O
5742: @cindex I/O - keyboard and display
5743:
5744: @menu
5745: * Simple numeric output:: Predefined formats
5746: * Formatted numeric output:: Formatted (pictured) output
5747: * String Formats:: How Forth stores strings in memory
5748: * Displaying characters and strings:: Other stuff
5749: * Input:: Input
5750: @end menu
5751:
5752: @node Simple numeric output, Formatted numeric output, Other I/O, Other I/O
5753: @subsection Simple numeric output
5754: @cindex numeric output - simple/free-format
5755:
5756: The simplest output functions are those that display numbers from the
5757: data or floating-point stacks. Floating-point output is always displayed
5758: using base 10. Numbers displayed from the data stack use the value stored
5759: in @code{base}.
5760:
5761: doc-.
5762: doc-dec.
5763: doc-hex.
5764: doc-u.
5765: doc-.r
5766: doc-u.r
5767: doc-d.
5768: doc-ud.
5769: doc-d.r
5770: doc-ud.r
5771: doc-f.
5772: doc-fe.
5773: doc-fs.
5774:
5775: Examples of printing the number 1234.5678E23 in the different floating-point output
5776: formats are shown below:
5777:
5778: @example
5779: f. 123456779999999000000000000.
5780: fe. 123.456779999999E24
5781: fs. 1.23456779999999E26
5782: @end example
5783:
5784:
5785: @node Formatted numeric output, String Formats, Simple numeric output, Other I/O
5786: @subsection Formatted numeric output
5787: @cindex formatted numeric output
5788: @cindex pictured numeric output
5789: @cindex numeric output - formatted
5790:
5791: Forth traditionally uses a technique called @dfn{pictured numeric
5792: output} for formatted printing of integers. In this technique, digits
5793: are extracted from the number (using the current output radix defined by
5794: @code{base}), converted to ASCII codes and appended to a string that is
5795: built in a scratch-pad area of memory (@pxref{core-idef,
5796: Implementation-defined options, Implementation-defined
5797: options}). Arbitrary characters can be appended to the string during the
5798: extraction process. The completed string is specified by an address
5799: and length and can be manipulated (@code{TYPE}ed, copied, modified)
5800: under program control.
5801:
5802: All of the words described in the previous section for simple numeric
5803: output are implemented in Gforth using pictured numeric output.
5804:
5805: Three important things to remember about Pictured Numeric Output:
5806:
5807: @itemize @bullet
5808: @item
5809: It always operates on double-precision numbers; to display a
5810: single-precision number, convert it first (@pxref{Double precision} for
5811: ways of doing this).
5812: @item
5813: It always treats the double-precision number as though it were
5814: unsigned. The examples below show ways of printing signed numbers.
5815: @item
5816: The string is built up from right to left; least significant digit first.
5817: @end itemize
5818:
5819: doc-<#
5820: doc-#
5821: doc-#s
5822: doc-hold
5823: doc-sign
5824: doc-#>
5825:
5826: doc-represent
5827:
5828: Here are some examples of using pictured numeric output:
5829:
5830: @example
5831: : my-u. ( u -- )
5832: \ Simplest use of pns.. behaves like Standard u.
5833: 0 \ convert to unsigned double
5834: <# \ start conversion
5835: #s \ convert all digits
5836: #> \ complete conversion
5837: TYPE SPACE ; \ display, with trailing space
5838:
5839: : cents-only ( u -- )
5840: 0 \ convert to unsigned double
5841: <# \ start conversion
5842: # # \ convert two least-significant digits
5843: #> \ complete conversion, discard other digits
5844: TYPE SPACE ; \ display, with trailing space
5845:
5846: : dollars-and-cents ( u -- )
5847: 0 \ convert to unsigned double
5848: <# \ start conversion
5849: # # \ convert two least-significant digits
5850: [char] . hold \ insert decimal point
5851: #s \ convert remaining digits
5852: [char] $ hold \ append currency symbol
5853: #> \ complete conversion
5854: TYPE SPACE ; \ display, with trailing space
5855:
5856: : my-. ( n -- )
5857: \ handling negatives.. behaves like Standard .
5858: s>d \ convert to signed double
5859: swap over dabs \ leave sign byte followed by unsigned double
5860: <# \ start conversion
5861: #s \ convert all digits
5862: rot sign \ get at sign byte, append "-" if needed
5863: #> \ complete conversion
5864: TYPE SPACE ; \ display, with trailing space
5865:
5866: : account. ( n -- )
5867: \ accountants don't like minus signs, they use braces
5868: \ for negative numbers
5869: s>d \ convert to signed double
5870: swap over dabs \ leave sign byte followed by unsigned double
5871: <# \ start conversion
5872: 2 pick \ get copy of sign byte
5873: 0< IF [char] ) hold THEN \ right-most character of output
5874: #s \ convert all digits
5875: rot \ get at sign byte
5876: 0< IF [char] ( hold THEN
5877: #> \ complete conversion
5878: TYPE SPACE ; \ display, with trailing space
5879: @end example
5880:
5881: Here are some examples of using these words:
5882:
5883: @example
5884: 1 my-u. 1
5885: hex -1 my-u. decimal FFFFFFFF
5886: 1 cents-only 01
5887: 1234 cents-only 34
5888: 2 dollars-and-cents $0.02
5889: 1234 dollars-and-cents $12.34
5890: 123 my-. 123
5891: -123 my. -123
5892: 123 account. 123
5893: -456 account. (456)
5894: @end example
5895:
5896:
5897: @node String Formats, Displaying characters and strings, Formatted numeric output, Other I/O
5898: @subsection String Formats
5899: @cindex strings - see character strings
5900: @cindex character strings - formats
5901: @cindex I/O - see character strings
5902:
5903: Forth commonly uses two different methods for representing character
5904: strings:
5905:
5906: @itemize @bullet
5907: @item
5908: @cindex address of counted string
5909: As a @dfn{counted string}, represented by a @i{c-addr}. The char
5910: addressed by @i{c-addr} contains a character-count, @i{n}, of the
5911: string and the string occupies the subsequent @i{n} char addresses in
5912: memory.
5913: @item
5914: As cell pair on the stack; @i{c-addr u}, where @i{u} is the length
5915: of the string in characters, and @i{c-addr} is the address of the
5916: first byte of the string.
5917: @end itemize
5918:
5919: ANS Forth encourages the use of the second format when representing
5920: strings on the stack, whilst conceeding that the counted string format
5921: remains useful as a way of storing strings in memory.
5922:
5923: doc-count
5924:
5925: @xref{Memory Blocks} for words that move, copy and search
5926: for strings. @xref{Displaying characters and strings,} for words that
5927: display characters and strings.
5928:
5929:
5930: @node Displaying characters and strings, Input, String Formats, Other I/O
5931: @subsection Displaying characters and strings
5932: @cindex characters - compiling and displaying
5933: @cindex character strings - compiling and displaying
5934:
5935: This section starts with a glossary of Forth words and ends with a set
5936: of examples.
5937:
5938: doc-bl
5939: doc-space
5940: doc-spaces
5941: doc-emit
5942: doc-toupper
5943: doc-."
5944: doc-.(
5945: doc-type
5946: doc-cr
5947: @cindex cursor control
5948: doc-at-xy
5949: doc-page
5950: doc-s"
5951: doc-c"
5952: doc-char
5953: doc-[char]
5954: doc-sliteral
5955:
5956: As an example, consider the following text, stored in a file @file{test.fs}:
5957:
5958: @example
5959: .( text-1)
5960: : my-word
5961: ." text-2" cr
5962: .( text-3)
5963: ;
5964:
5965: ." text-4"
5966:
5967: : my-char
5968: [char] ALPHABET emit
5969: char emit
5970: ;
5971: @end example
5972:
5973: When you load this code into Gforth, the following output is generated:
5974:
5975: @example
5976: @kbd{include test.fs @key{RET}} text-1text-3text-4 ok
5977: @end example
5978:
5979: @itemize @bullet
5980: @item
5981: Messages @code{text-1} and @code{text-3} are displayed because @code{.(}
5982: is an immediate word; it behaves in the same way whether it is used inside
5983: or outside a colon definition.
5984: @item
5985: Message @code{text-4} is displayed because of Gforth's added interpretation
5986: semantics for @code{."}.
5987: @item
5988: Message @code{text-2} is @i{not} displayed, because the text interpreter
5989: performs the compilation semantics for @code{."} within the definition of
5990: @code{my-word}.
5991: @end itemize
5992:
5993: Here are some examples of executing @code{my-word} and @code{my-char}:
5994:
5995: @example
5996: @kbd{my-word @key{RET}} text-2
5997: ok
5998: @kbd{my-char fred @key{RET}} Af ok
5999: @kbd{my-char jim @key{RET}} Aj ok
6000: @end example
6001:
6002: @itemize @bullet
6003: @item
6004: Message @code{text-2} is displayed because of the run-time behaviour of
6005: @code{."}.
6006: @item
6007: @code{[char]} compiles the ``A'' from ``ALPHABET'' and puts its display code
6008: on the stack at run-time. @code{emit} always displays the character
6009: when @code{my-char} is executed.
6010: @item
6011: @code{char} parses a string at run-time and the second @code{emit} displays
6012: the first character of the string.
6013: @item
6014: If you type @code{see my-char} you can see that @code{[char]} discarded
6015: the text ``LPHABET'' and only compiled the display code for ``A'' into the
6016: definition of @code{my-char}.
6017: @end itemize
6018:
6019:
6020:
6021: @node Input, , Displaying characters and strings, Other I/O
6022: @subsection Input
6023: @cindex input
6024: @cindex I/O - see input
6025: @cindex parsing a string
6026:
6027: @xref{String Formats} for ways of storing character strings in memory.
6028:
6029: @comment TODO examples for >number >float accept key key? pad parse word refill
6030: @comment then index them
6031:
6032: doc-key
6033: doc-key?
6034: doc->number
6035: doc->float
6036: doc-accept
6037: doc-pad
6038: doc-parse
6039: doc-word
6040: doc-sword
6041: doc-refill
6042: @comment obsolescent words..
6043: doc-convert
6044: doc-query
6045: doc-expect
6046: doc-span
6047:
6048:
6049: @c -------------------------------------------------------------
6050: @node Programming Tools, Assembler and Code Words, Other I/O, Words
6051: @section Programming Tools
6052: @cindex programming tools
6053:
6054: @menu
6055: * Debugging:: Simple and quick.
6056: * Assertions:: Making your programs self-checking.
6057: * Singlestep Debugger:: Executing your program word by word.
6058: @end menu
6059:
6060: @node Debugging, Assertions, Programming Tools, Programming Tools
6061: @subsection Debugging
6062: @cindex debugging
6063:
6064: Languages with a slow edit/compile/link/test development loop tend to
6065: require sophisticated tracing/stepping debuggers to facilate
6066: productive debugging.
6067:
6068: A much better (faster) way in fast-compiling languages is to add
6069: printing code at well-selected places, let the program run, look at
6070: the output, see where things went wrong, add more printing code, etc.,
6071: until the bug is found.
6072:
6073: The simple debugging aids provided in @file{debugs.fs}
6074: are meant to support this style of debugging. In addition, there are
6075: words for non-destructively inspecting the stack and memory:
6076:
6077: doc-.s
6078: doc-f.s
6079:
6080: There is a word @code{.r} but it does @i{not} display the return
6081: stack! It is used for formatted numeric output.
6082:
6083: doc-depth
6084: doc-fdepth
6085: doc-clearstack
6086: doc-?
6087: doc-dump
6088:
6089: The word @code{~~} prints debugging information (by default the source
6090: location and the stack contents). It is easy to insert. If you use Emacs
6091: it is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
6092: query-replace them with nothing). The deferred words
6093: @code{printdebugdata} and @code{printdebugline} control the output of
6094: @code{~~}. The default source location output format works well with
6095: Emacs' compilation mode, so you can step through the program at the
6096: source level using @kbd{C-x `} (the advantage over a stepping debugger
6097: is that you can step in any direction and you know where the crash has
6098: happened or where the strange data has occurred).
6099:
6100: The default actions of @code{~~} clobber the contents of the pictured
6101: numeric output string, so you should not use @code{~~}, e.g., between
6102: @code{<#} and @code{#>}.
6103:
6104: doc-~~
6105: doc-printdebugdata
6106: doc-printdebugline
6107:
6108: doc-see
6109: doc-marker
6110:
6111: Here's an example of using @code{marker} at the start of a source file
6112: that you are debugging; it ensures that you only ever have one copy of
6113: the file's definitions compiled at any time:
6114:
6115: @example
6116: [IFDEF] my-code
6117: my-code
6118: [ENDIF]
6119:
6120: marker my-code
6121: init-included-files
6122:
6123: \ .. definitions start here
6124: \ .
6125: \ .
6126: \ end
6127: @end example
6128:
6129:
6130:
6131: @node Assertions, Singlestep Debugger, Debugging, Programming Tools
6132: @subsection Assertions
6133: @cindex assertions
6134:
6135: It is a good idea to make your programs self-checking, especially if you
6136: make an assumption that may become invalid during maintenance (for
6137: example, that a certain field of a data structure is never zero). Gforth
6138: supports @dfn{assertions} for this purpose. They are used like this:
6139:
6140: @example
6141: assert( @i{flag} )
6142: @end example
6143:
6144: The code between @code{assert(} and @code{)} should compute a flag, that
6145: should be true if everything is alright and false otherwise. It should
6146: not change anything else on the stack. The overall stack effect of the
6147: assertion is @code{( -- )}. E.g.
6148:
6149: @example
6150: assert( 1 1 + 2 = ) \ what we learn in school
6151: assert( dup 0<> ) \ assert that the top of stack is not zero
6152: assert( false ) \ this code should not be reached
6153: @end example
6154:
6155: The need for assertions is different at different times. During
6156: debugging, we want more checking, in production we sometimes care more
6157: for speed. Therefore, assertions can be turned off, i.e., the assertion
6158: becomes a comment. Depending on the importance of an assertion and the
6159: time it takes to check it, you may want to turn off some assertions and
6160: keep others turned on. Gforth provides several levels of assertions for
6161: this purpose:
6162:
6163: doc-assert0(
6164: doc-assert1(
6165: doc-assert2(
6166: doc-assert3(
6167: doc-assert(
6168: doc-)
6169:
6170: The variable @code{assert-level} specifies the highest assertions that
6171: are turned on. I.e., at the default @code{assert-level} of one,
6172: @code{assert0(} and @code{assert1(} assertions perform checking, while
6173: @code{assert2(} and @code{assert3(} assertions are treated as comments.
6174:
6175: The value of @code{assert-level} is evaluated at compile-time, not at
6176: run-time. Therefore you cannot turn assertions on or off at run-time;
6177: you have to set the @code{assert-level} appropriately before compiling a
6178: piece of code. You can compile different pieces of code at different
6179: @code{assert-level}s (e.g., a trusted library at level 1 and
6180: newly-written code at level 3).
6181:
6182: doc-assert-level
6183:
6184: If an assertion fails, a message compatible with Emacs' compilation mode
6185: is produced and the execution is aborted (currently with @code{ABORT"}.
6186: If there is interest, we will introduce a special throw code. But if you
6187: intend to @code{catch} a specific condition, using @code{throw} is
6188: probably more appropriate than an assertion).
6189:
6190: Definitions in ANS Forth for these assertion words are provided
6191: in @file{compat/assert.fs}.
6192:
6193:
6194: @node Singlestep Debugger, , Assertions, Programming Tools
6195: @subsection Singlestep Debugger
6196: @cindex singlestep Debugger
6197: @cindex debugging Singlestep
6198: @cindex @code{dbg}
6199: @cindex @code{BREAK:}
6200: @cindex @code{BREAK"}
6201:
6202: When you create a new word there's often the need to check whether it
6203: behaves correctly or not. You can do this by typing @code{dbg
6204: badword}. A debug session might look like this:
6205:
6206: @example
6207: : badword 0 DO i . LOOP ; ok
6208: 2 dbg badword
6209: : badword
6210: Scanning code...
6211:
6212: Nesting debugger ready!
6213:
6214: 400D4738 8049BC4 0 -> [ 2 ] 00002 00000
6215: 400D4740 8049F68 DO -> [ 0 ]
6216: 400D4744 804A0C8 i -> [ 1 ] 00000
6217: 400D4748 400C5E60 . -> 0 [ 0 ]
6218: 400D474C 8049D0C LOOP -> [ 0 ]
6219: 400D4744 804A0C8 i -> [ 1 ] 00001
6220: 400D4748 400C5E60 . -> 1 [ 0 ]
6221: 400D474C 8049D0C LOOP -> [ 0 ]
6222: 400D4758 804B384 ; -> ok
6223: @end example
6224:
6225: Each line displayed is one step. You always have to hit return to
6226: execute the next word that is displayed. If you don't want to execute
6227: the next word in a whole, you have to type @kbd{n} for @code{nest}. Here is
6228: an overview what keys are available:
6229:
6230: @table @i
6231:
6232: @item @key{RET}
6233: Next; Execute the next word.
6234:
6235: @item n
6236: Nest; Single step through next word.
6237:
6238: @item u
6239: Unnest; Stop debugging and execute rest of word. If we got to this word
6240: with nest, continue debugging with the calling word.
6241:
6242: @item d
6243: Done; Stop debugging and execute rest.
6244:
6245: @item s
6246: Stop; Abort immediately.
6247:
6248: @end table
6249:
6250: Debugging large application with this mechanism is very difficult, because
6251: you have to nest very deeply into the program before the interesting part
6252: begins. This takes a lot of time.
6253:
6254: To do it more directly put a @code{BREAK:} command into your source code.
6255: When program execution reaches @code{BREAK:} the single step debugger is
6256: invoked and you have all the features described above.
6257:
6258: If you have more than one part to debug it is useful to know where the
6259: program has stopped at the moment. You can do this by the
6260: @code{BREAK" string"} command. This behaves like @code{BREAK:} except that
6261: string is typed out when the ``breakpoint'' is reached.
6262:
6263: doc-dbg
6264: doc-BREAK:
6265: doc-BREAK"
6266:
6267:
6268: @c -------------------------------------------------------------
6269: @node Assembler and Code Words, Threading Words, Programming Tools, Words
6270: @section Assembler and Code Words
6271: @cindex assembler
6272: @cindex code words
6273:
6274: Gforth provides some words for defining primitives (words written in
6275: machine code), and for defining the machine-code equivalent of
6276: @code{DOES>}-based defining words. However, the machine-independent
6277: nature of Gforth poses a few problems: First of all, Gforth runs on
6278: several architectures, so it can provide no standard assembler. What's
6279: worse is that the register allocation not only depends on the processor,
6280: but also on the @code{gcc} version and options used.
6281:
6282: The words that Gforth offers encapsulate some system dependences (e.g.,
6283: the header structure), so a system-independent assembler may be used in
6284: Gforth. If you do not have an assembler, you can compile machine code
6285: directly with @code{,} and @code{c,}@footnote{This isn't portable,
6286: because these words emit stuff in @i{data} space; it works because
6287: Gforth has unified code/data spaces. Assembler isn't likely to be
6288: portable anyway.}.
6289:
6290: doc-assembler
6291: doc-code
6292: doc-end-code
6293: doc-;code
6294: doc-flush-icache
6295:
6296: If @code{flush-icache} does not work correctly, @code{code} words
6297: etc. will not work (reliably), either.
6298:
6299: The typical usage of these @code{code} words can be shown most easily by
6300: analogy to the equivalent high-level defining words:
6301:
6302: @example
6303: : foo code foo
6304: <high-level Forth words> <assembler>
6305: ; end-code
6306:
6307: : bar : bar
6308: <high-level Forth words> <high-level Forth words>
6309: CREATE CREATE
6310: <high-level Forth words> <high-level Forth words>
6311: DOES> ;code
6312: <high-level Forth words> <assembler>
6313: ; end-code
6314: @end example
6315:
6316: @code{flush-icache} is always present. The other words are rarely used
6317: and reside in @code{code.fs}, which is usually not loaded. You can load
6318: it with @code{require code.fs}.
6319:
6320: @cindex registers of the inner interpreter
6321: In the assembly code you will want to refer to the inner interpreter's
6322: registers (e.g., the data stack pointer) and you may want to use other
6323: registers for temporary storage. Unfortunately, the register allocation
6324: is installation-dependent.
6325:
6326: The easiest solution is to use explicit register declarations
6327: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
6328: GNU C Manual}) for all of the inner interpreter's registers: You have to
6329: compile Gforth with @code{-DFORCE_REG} (configure option
6330: @code{--enable-force-reg}) and the appropriate declarations must be
6331: present in the @code{machine.h} file (see @code{mips.h} for an example;
6332: you can find a full list of all declarable register symbols with
6333: @code{grep register engine.c}). If you give explicit registers to all
6334: variables that are declared at the beginning of @code{engine()}, you
6335: should be able to use the other caller-saved registers for temporary
6336: storage. Alternatively, you can use the @code{gcc} option
6337: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
6338: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
6339: (however, this restriction on register allocation may slow Gforth
6340: significantly).
6341:
6342: If this solution is not viable (e.g., because @code{gcc} does not allow
6343: you to explicitly declare all the registers you need), you have to find
6344: out by looking at the code where the inner interpreter's registers
6345: reside and which registers can be used for temporary storage. You can
6346: get an assembly listing of the engine's code with @code{make engine.s}.
6347:
6348: In any case, it is good practice to abstract your assembly code from the
6349: actual register allocation. E.g., if the data stack pointer resides in
6350: register @code{$17}, create an alias for this register called @code{sp},
6351: and use that in your assembly code.
6352:
6353: @cindex code words, portable
6354: Another option for implementing normal and defining words efficiently
6355: is to add the desired functionality to the source of Gforth. For normal
6356: words you just have to edit @file{primitives} (@pxref{Automatic
6357: Generation}). Defining words (equivalent to @code{;CODE} words, for fast
6358: defined words) may require changes in @file{engine.c}, @file{kernel.fs},
6359: @file{prims2x.fs}, and possibly @file{cross.fs}.
6360:
6361:
6362: @c -------------------------------------------------------------
6363: @node Threading Words, Locals, Assembler and Code Words, Words
6364: @section Threading Words
6365: @cindex threading words
6366:
6367: @cindex code address
6368: These words provide access to code addresses and other threading stuff
6369: in Gforth (and, possibly, other interpretive Forths). It more or less
6370: abstracts away the differences between direct and indirect threading
6371: (and, for direct threading, the machine dependences). However, at
6372: present this wordset is still incomplete. It is also pretty low-level;
6373: some day it will hopefully be made unnecessary by an internals wordset
6374: that abstracts implementation details away completely.
6375:
6376: doc-threading-method
6377: doc->code-address
6378: doc->does-code
6379: doc-code-address!
6380: doc-does-code!
6381: doc-does-handler!
6382: doc-/does-handler
6383:
6384: The code addresses produced by various defining words are produced by
6385: the following words:
6386:
6387: doc-docol:
6388: doc-docon:
6389: doc-dovar:
6390: doc-douser:
6391: doc-dodefer:
6392: doc-dofield:
6393:
6394: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
6395: with @code{>does-code}. If the word was defined in that way, the value
6396: returned is non-zero and identifies the @code{DOES>} used by the
6397: defining word.
6398: @comment TODO should that be ``identifies the xt of the DOES> ??''
6399:
6400: @c -------------------------------------------------------------
6401: @node Locals, Structures, Threading Words, Words
6402: @section Locals
6403: @cindex locals
6404:
6405: Local variables can make Forth programming more enjoyable and Forth
6406: programs easier to read. Unfortunately, the locals of ANS Forth are
6407: laden with restrictions. Therefore, we provide not only the ANS Forth
6408: locals wordset, but also our own, more powerful locals wordset (we
6409: implemented the ANS Forth locals wordset through our locals wordset).
6410:
6411: The ideas in this section have also been published in the paper
6412: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
6413: at EuroForth '94; it is available at
6414: @*@url{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
6415:
6416: @menu
6417: * Gforth locals::
6418: * ANS Forth locals::
6419: @end menu
6420:
6421: @node Gforth locals, ANS Forth locals, Locals, Locals
6422: @subsection Gforth locals
6423: @cindex Gforth locals
6424: @cindex locals, Gforth style
6425:
6426: Locals can be defined with
6427:
6428: @example
6429: @{ local1 local2 ... -- comment @}
6430: @end example
6431: or
6432: @example
6433: @{ local1 local2 ... @}
6434: @end example
6435:
6436: E.g.,
6437: @example
6438: : max @{ n1 n2 -- n3 @}
6439: n1 n2 > if
6440: n1
6441: else
6442: n2
6443: endif ;
6444: @end example
6445:
6446: The similarity of locals definitions with stack comments is intended. A
6447: locals definition often replaces the stack comment of a word. The order
6448: of the locals corresponds to the order in a stack comment and everything
6449: after the @code{--} is really a comment.
6450:
6451: This similarity has one disadvantage: It is too easy to confuse locals
6452: declarations with stack comments, causing bugs and making them hard to
6453: find. However, this problem can be avoided by appropriate coding
6454: conventions: Do not use both notations in the same program. If you do,
6455: they should be distinguished using additional means, e.g. by position.
6456:
6457: @cindex types of locals
6458: @cindex locals types
6459: The name of the local may be preceded by a type specifier, e.g.,
6460: @code{F:} for a floating point value:
6461:
6462: @example
6463: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
6464: \ complex multiplication
6465: Ar Br f* Ai Bi f* f-
6466: Ar Bi f* Ai Br f* f+ ;
6467: @end example
6468:
6469: @cindex flavours of locals
6470: @cindex locals flavours
6471: @cindex value-flavoured locals
6472: @cindex variable-flavoured locals
6473: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
6474: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
6475: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
6476: with @code{W:}, @code{D:} etc.) produces its value and can be changed
6477: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
6478: produces its address (which becomes invalid when the variable's scope is
6479: left). E.g., the standard word @code{emit} can be defined in terms of
6480: @code{type} like this:
6481:
6482: @example
6483: : emit @{ C^ char* -- @}
6484: char* 1 type ;
6485: @end example
6486:
6487: @cindex default type of locals
6488: @cindex locals, default type
6489: A local without type specifier is a @code{W:} local. Both flavours of
6490: locals are initialized with values from the data or FP stack.
6491:
6492: Currently there is no way to define locals with user-defined data
6493: structures, but we are working on it.
6494:
6495: Gforth allows defining locals everywhere in a colon definition. This
6496: poses the following questions:
6497:
6498: @menu
6499: * Where are locals visible by name?::
6500: * How long do locals live?::
6501: * Programming Style::
6502: * Implementation::
6503: @end menu
6504:
6505: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
6506: @subsubsection Where are locals visible by name?
6507: @cindex locals visibility
6508: @cindex visibility of locals
6509: @cindex scope of locals
6510:
6511: Basically, the answer is that locals are visible where you would expect
6512: it in block-structured languages, and sometimes a little longer. If you
6513: want to restrict the scope of a local, enclose its definition in
6514: @code{SCOPE}...@code{ENDSCOPE}.
6515:
6516: doc-scope
6517: doc-endscope
6518:
6519: These words behave like control structure words, so you can use them
6520: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
6521: arbitrary ways.
6522:
6523: If you want a more exact answer to the visibility question, here's the
6524: basic principle: A local is visible in all places that can only be
6525: reached through the definition of the local@footnote{In compiler
6526: construction terminology, all places dominated by the definition of the
6527: local.}. In other words, it is not visible in places that can be reached
6528: without going through the definition of the local. E.g., locals defined
6529: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
6530: defined in @code{BEGIN}...@code{UNTIL} are visible after the
6531: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
6532:
6533: The reasoning behind this solution is: We want to have the locals
6534: visible as long as it is meaningful. The user can always make the
6535: visibility shorter by using explicit scoping. In a place that can
6536: only be reached through the definition of a local, the meaning of a
6537: local name is clear. In other places it is not: How is the local
6538: initialized at the control flow path that does not contain the
6539: definition? Which local is meant, if the same name is defined twice in
6540: two independent control flow paths?
6541:
6542: This should be enough detail for nearly all users, so you can skip the
6543: rest of this section. If you really must know all the gory details and
6544: options, read on.
6545:
6546: In order to implement this rule, the compiler has to know which places
6547: are unreachable. It knows this automatically after @code{AHEAD},
6548: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
6549: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
6550: compiler that the control flow never reaches that place. If
6551: @code{UNREACHABLE} is not used where it could, the only consequence is
6552: that the visibility of some locals is more limited than the rule above
6553: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
6554: lie to the compiler), buggy code will be produced.
6555:
6556: doc-unreachable
6557:
6558: Another problem with this rule is that at @code{BEGIN}, the compiler
6559: does not know which locals will be visible on the incoming
6560: back-edge. All problems discussed in the following are due to this
6561: ignorance of the compiler (we discuss the problems using @code{BEGIN}
6562: loops as examples; the discussion also applies to @code{?DO} and other
6563: loops). Perhaps the most insidious example is:
6564: @example
6565: AHEAD
6566: BEGIN
6567: x
6568: [ 1 CS-ROLL ] THEN
6569: @{ x @}
6570: ...
6571: UNTIL
6572: @end example
6573:
6574: This should be legal according to the visibility rule. The use of
6575: @code{x} can only be reached through the definition; but that appears
6576: textually below the use.
6577:
6578: From this example it is clear that the visibility rules cannot be fully
6579: implemented without major headaches. Our implementation treats common
6580: cases as advertised and the exceptions are treated in a safe way: The
6581: compiler makes a reasonable guess about the locals visible after a
6582: @code{BEGIN}; if it is too pessimistic, the
6583: user will get a spurious error about the local not being defined; if the
6584: compiler is too optimistic, it will notice this later and issue a
6585: warning. In the case above the compiler would complain about @code{x}
6586: being undefined at its use. You can see from the obscure examples in
6587: this section that it takes quite unusual control structures to get the
6588: compiler into trouble, and even then it will often do fine.
6589:
6590: If the @code{BEGIN} is reachable from above, the most optimistic guess
6591: is that all locals visible before the @code{BEGIN} will also be
6592: visible after the @code{BEGIN}. This guess is valid for all loops that
6593: are entered only through the @code{BEGIN}, in particular, for normal
6594: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
6595: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
6596: compiler. When the branch to the @code{BEGIN} is finally generated by
6597: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
6598: warns the user if it was too optimistic:
6599: @example
6600: IF
6601: @{ x @}
6602: BEGIN
6603: \ x ?
6604: [ 1 cs-roll ] THEN
6605: ...
6606: UNTIL
6607: @end example
6608:
6609: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
6610: optimistically assumes that it lives until the @code{THEN}. It notices
6611: this difference when it compiles the @code{UNTIL} and issues a
6612: warning. The user can avoid the warning, and make sure that @code{x}
6613: is not used in the wrong area by using explicit scoping:
6614: @example
6615: IF
6616: SCOPE
6617: @{ x @}
6618: ENDSCOPE
6619: BEGIN
6620: [ 1 cs-roll ] THEN
6621: ...
6622: UNTIL
6623: @end example
6624:
6625: Since the guess is optimistic, there will be no spurious error messages
6626: about undefined locals.
6627:
6628: If the @code{BEGIN} is not reachable from above (e.g., after
6629: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
6630: optimistic guess, as the locals visible after the @code{BEGIN} may be
6631: defined later. Therefore, the compiler assumes that no locals are
6632: visible after the @code{BEGIN}. However, the user can use
6633: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
6634: visible at the BEGIN as at the point where the top control-flow stack
6635: item was created.
6636:
6637: doc-assume-live
6638:
6639: E.g.,
6640: @example
6641: @{ x @}
6642: AHEAD
6643: ASSUME-LIVE
6644: BEGIN
6645: x
6646: [ 1 CS-ROLL ] THEN
6647: ...
6648: UNTIL
6649: @end example
6650:
6651: Other cases where the locals are defined before the @code{BEGIN} can be
6652: handled by inserting an appropriate @code{CS-ROLL} before the
6653: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
6654: behind the @code{ASSUME-LIVE}).
6655:
6656: Cases where locals are defined after the @code{BEGIN} (but should be
6657: visible immediately after the @code{BEGIN}) can only be handled by
6658: rearranging the loop. E.g., the ``most insidious'' example above can be
6659: arranged into:
6660: @example
6661: BEGIN
6662: @{ x @}
6663: ... 0=
6664: WHILE
6665: x
6666: REPEAT
6667: @end example
6668:
6669: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
6670: @subsubsection How long do locals live?
6671: @cindex locals lifetime
6672: @cindex lifetime of locals
6673:
6674: The right answer for the lifetime question would be: A local lives at
6675: least as long as it can be accessed. For a value-flavoured local this
6676: means: until the end of its visibility. However, a variable-flavoured
6677: local could be accessed through its address far beyond its visibility
6678: scope. Ultimately, this would mean that such locals would have to be
6679: garbage collected. Since this entails un-Forth-like implementation
6680: complexities, I adopted the same cowardly solution as some other
6681: languages (e.g., C): The local lives only as long as it is visible;
6682: afterwards its address is invalid (and programs that access it
6683: afterwards are erroneous).
6684:
6685: @node Programming Style, Implementation, How long do locals live?, Gforth locals
6686: @subsubsection Programming Style
6687: @cindex locals programming style
6688: @cindex programming style, locals
6689:
6690: The freedom to define locals anywhere has the potential to change
6691: programming styles dramatically. In particular, the need to use the
6692: return stack for intermediate storage vanishes. Moreover, all stack
6693: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
6694: determined arguments) can be eliminated: If the stack items are in the
6695: wrong order, just write a locals definition for all of them; then
6696: write the items in the order you want.
6697:
6698: This seems a little far-fetched and eliminating stack manipulations is
6699: unlikely to become a conscious programming objective. Still, the number
6700: of stack manipulations will be reduced dramatically if local variables
6701: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
6702: a traditional implementation of @code{max}).
6703:
6704: This shows one potential benefit of locals: making Forth programs more
6705: readable. Of course, this benefit will only be realized if the
6706: programmers continue to honour the principle of factoring instead of
6707: using the added latitude to make the words longer.
6708:
6709: @cindex single-assignment style for locals
6710: Using @code{TO} can and should be avoided. Without @code{TO},
6711: every value-flavoured local has only a single assignment and many
6712: advantages of functional languages apply to Forth. I.e., programs are
6713: easier to analyse, to optimize and to read: It is clear from the
6714: definition what the local stands for, it does not turn into something
6715: different later.
6716:
6717: E.g., a definition using @code{TO} might look like this:
6718: @example
6719: : strcmp @{ addr1 u1 addr2 u2 -- n @}
6720: u1 u2 min 0
6721: ?do
6722: addr1 c@@ addr2 c@@ -
6723: ?dup-if
6724: unloop exit
6725: then
6726: addr1 char+ TO addr1
6727: addr2 char+ TO addr2
6728: loop
6729: u1 u2 - ;
6730: @end example
6731: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
6732: every loop iteration. @code{strcmp} is a typical example of the
6733: readability problems of using @code{TO}. When you start reading
6734: @code{strcmp}, you think that @code{addr1} refers to the start of the
6735: string. Only near the end of the loop you realize that it is something
6736: else.
6737:
6738: This can be avoided by defining two locals at the start of the loop that
6739: are initialized with the right value for the current iteration.
6740: @example
6741: : strcmp @{ addr1 u1 addr2 u2 -- n @}
6742: addr1 addr2
6743: u1 u2 min 0
6744: ?do @{ s1 s2 @}
6745: s1 c@@ s2 c@@ -
6746: ?dup-if
6747: unloop exit
6748: then
6749: s1 char+ s2 char+
6750: loop
6751: 2drop
6752: u1 u2 - ;
6753: @end example
6754: Here it is clear from the start that @code{s1} has a different value
6755: in every loop iteration.
6756:
6757: @node Implementation, , Programming Style, Gforth locals
6758: @subsubsection Implementation
6759: @cindex locals implementation
6760: @cindex implementation of locals
6761:
6762: @cindex locals stack
6763: Gforth uses an extra locals stack. The most compelling reason for
6764: this is that the return stack is not float-aligned; using an extra stack
6765: also eliminates the problems and restrictions of using the return stack
6766: as locals stack. Like the other stacks, the locals stack grows toward
6767: lower addresses. A few primitives allow an efficient implementation:
6768:
6769: doc-@local#
6770: doc-f@local#
6771: doc-laddr#
6772: doc-lp+!#
6773: doc-lp!
6774: doc->l
6775: doc-f>l
6776:
6777: In addition to these primitives, some specializations of these
6778: primitives for commonly occurring inline arguments are provided for
6779: efficiency reasons, e.g., @code{@@local0} as specialization of
6780: @code{@@local#} for the inline argument 0. The following compiling words
6781: compile the right specialized version, or the general version, as
6782: appropriate:
6783:
6784: doc-compile-@local
6785: doc-compile-f@local
6786: doc-compile-lp+!
6787:
6788: Combinations of conditional branches and @code{lp+!#} like
6789: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
6790: is taken) are provided for efficiency and correctness in loops.
6791:
6792: A special area in the dictionary space is reserved for keeping the
6793: local variable names. @code{@{} switches the dictionary pointer to this
6794: area and @code{@}} switches it back and generates the locals
6795: initializing code. @code{W:} etc.@ are normal defining words. This
6796: special area is cleared at the start of every colon definition.
6797:
6798: @cindex word list for defining locals
6799: A special feature of Gforth's dictionary is used to implement the
6800: definition of locals without type specifiers: every word list (aka
6801: vocabulary) has its own methods for searching
6802: etc. (@pxref{Word Lists}). For the present purpose we defined a word list
6803: with a special search method: When it is searched for a word, it
6804: actually creates that word using @code{W:}. @code{@{} changes the search
6805: order to first search the word list containing @code{@}}, @code{W:} etc.,
6806: and then the word list for defining locals without type specifiers.
6807:
6808: The lifetime rules support a stack discipline within a colon
6809: definition: The lifetime of a local is either nested with other locals
6810: lifetimes or it does not overlap them.
6811:
6812: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
6813: pointer manipulation is generated. Between control structure words
6814: locals definitions can push locals onto the locals stack. @code{AGAIN}
6815: is the simplest of the other three control flow words. It has to
6816: restore the locals stack depth of the corresponding @code{BEGIN}
6817: before branching. The code looks like this:
6818: @format
6819: @code{lp+!#} current-locals-size @minus{} dest-locals-size
6820: @code{branch} <begin>
6821: @end format
6822:
6823: @code{UNTIL} is a little more complicated: If it branches back, it
6824: must adjust the stack just like @code{AGAIN}. But if it falls through,
6825: the locals stack must not be changed. The compiler generates the
6826: following code:
6827: @format
6828: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
6829: @end format
6830: The locals stack pointer is only adjusted if the branch is taken.
6831:
6832: @code{THEN} can produce somewhat inefficient code:
6833: @format
6834: @code{lp+!#} current-locals-size @minus{} orig-locals-size
6835: <orig target>:
6836: @code{lp+!#} orig-locals-size @minus{} new-locals-size
6837: @end format
6838: The second @code{lp+!#} adjusts the locals stack pointer from the
6839: level at the @i{orig} point to the level after the @code{THEN}. The
6840: first @code{lp+!#} adjusts the locals stack pointer from the current
6841: level to the level at the orig point, so the complete effect is an
6842: adjustment from the current level to the right level after the
6843: @code{THEN}.
6844:
6845: @cindex locals information on the control-flow stack
6846: @cindex control-flow stack items, locals information
6847: In a conventional Forth implementation a dest control-flow stack entry
6848: is just the target address and an orig entry is just the address to be
6849: patched. Our locals implementation adds a word list to every orig or dest
6850: item. It is the list of locals visible (or assumed visible) at the point
6851: described by the entry. Our implementation also adds a tag to identify
6852: the kind of entry, in particular to differentiate between live and dead
6853: (reachable and unreachable) orig entries.
6854:
6855: A few unusual operations have to be performed on locals word lists:
6856:
6857: doc-common-list
6858: doc-sub-list?
6859: doc-list-size
6860:
6861: Several features of our locals word list implementation make these
6862: operations easy to implement: The locals word lists are organised as
6863: linked lists; the tails of these lists are shared, if the lists
6864: contain some of the same locals; and the address of a name is greater
6865: than the address of the names behind it in the list.
6866:
6867: Another important implementation detail is the variable
6868: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
6869: determine if they can be reached directly or only through the branch
6870: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
6871: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
6872: definition, by @code{BEGIN} and usually by @code{THEN}.
6873:
6874: Counted loops are similar to other loops in most respects, but
6875: @code{LEAVE} requires special attention: It performs basically the same
6876: service as @code{AHEAD}, but it does not create a control-flow stack
6877: entry. Therefore the information has to be stored elsewhere;
6878: traditionally, the information was stored in the target fields of the
6879: branches created by the @code{LEAVE}s, by organizing these fields into a
6880: linked list. Unfortunately, this clever trick does not provide enough
6881: space for storing our extended control flow information. Therefore, we
6882: introduce another stack, the leave stack. It contains the control-flow
6883: stack entries for all unresolved @code{LEAVE}s.
6884:
6885: Local names are kept until the end of the colon definition, even if
6886: they are no longer visible in any control-flow path. In a few cases
6887: this may lead to increased space needs for the locals name area, but
6888: usually less than reclaiming this space would cost in code size.
6889:
6890:
6891: @node ANS Forth locals, , Gforth locals, Locals
6892: @subsection ANS Forth locals
6893: @cindex locals, ANS Forth style
6894:
6895: The ANS Forth locals wordset does not define a syntax for locals, but
6896: words that make it possible to define various syntaxes. One of the
6897: possible syntaxes is a subset of the syntax we used in the Gforth locals
6898: wordset, i.e.:
6899:
6900: @example
6901: @{ local1 local2 ... -- comment @}
6902: @end example
6903: @noindent
6904: or
6905: @example
6906: @{ local1 local2 ... @}
6907: @end example
6908:
6909: The order of the locals corresponds to the order in a stack comment. The
6910: restrictions are:
6911:
6912: @itemize @bullet
6913: @item
6914: Locals can only be cell-sized values (no type specifiers are allowed).
6915: @item
6916: Locals can be defined only outside control structures.
6917: @item
6918: Locals can interfere with explicit usage of the return stack. For the
6919: exact (and long) rules, see the standard. If you don't use return stack
6920: accessing words in a definition using locals, you will be all right. The
6921: purpose of this rule is to make locals implementation on the return
6922: stack easier.
6923: @item
6924: The whole definition must be in one line.
6925: @end itemize
6926:
6927: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
6928: Defining Words}). I.e., they are initialized from the stack. Using their
6929: name produces their value. Their value can be changed using @code{TO}.
6930:
6931: Since this syntax is supported by Gforth directly, you need not do
6932: anything to use it. If you want to port a program using this syntax to
6933: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
6934: syntax on the other system.
6935:
6936: Note that a syntax shown in the standard, section A.13 looks
6937: similar, but is quite different in having the order of locals
6938: reversed. Beware!
6939:
6940: The ANS Forth locals wordset itself consists of a word:
6941:
6942: doc-(local)
6943:
6944: The ANS Forth locals extension wordset defines a syntax using @code{locals|}, but it is so
6945: awful that we strongly recommend not to use it. We have implemented this
6946: syntax to make porting to Gforth easy, but do not document it here. The
6947: problem with this syntax is that the locals are defined in an order
6948: reversed with respect to the standard stack comment notation, making
6949: programs harder to read, and easier to misread and miswrite. The only
6950: merit of this syntax is that it is easy to implement using the ANS Forth
6951: locals wordset.
6952:
6953:
6954: @c ----------------------------------------------------------
6955: @node Structures, Object-oriented Forth, Locals, Words
6956: @section Structures
6957: @cindex structures
6958: @cindex records
6959:
6960: This section presents the structure package that comes with Gforth. A
6961: version of the package implemented in ANS Forth is available in
6962: @file{compat/struct.fs}. This package was inspired by a posting on
6963: comp.lang.forth in 1989 (unfortunately I don't remember, by whom;
6964: possibly John Hayes). A version of this section has been published in
6965: ???. Marcel Hendrix provided helpful comments.
6966:
6967: @menu
6968: * Why explicit structure support?::
6969: * Structure Usage::
6970: * Structure Naming Convention::
6971: * Structure Implementation::
6972: * Structure Glossary::
6973: @end menu
6974:
6975: @node Why explicit structure support?, Structure Usage, Structures, Structures
6976: @subsection Why explicit structure support?
6977:
6978: @cindex address arithmetic for structures
6979: @cindex structures using address arithmetic
6980: If we want to use a structure containing several fields, we could simply
6981: reserve memory for it, and access the fields using address arithmetic
6982: (@pxref{Address Arithmetic}). As an example, consider a structure with
6983: the following fields
6984:
6985: @table @code
6986: @item a
6987: is a float
6988: @item b
6989: is a cell
6990: @item c
6991: is a float
6992: @end table
6993:
6994: Given the (float-aligned) base address of the structure we get the
6995: address of the field
6996:
6997: @table @code
6998: @item a
6999: without doing anything further.
7000: @item b
7001: with @code{float+}
7002: @item c
7003: with @code{float+ cell+ faligned}
7004: @end table
7005:
7006: It is easy to see that this can become quite tiring.
7007:
7008: Moreover, it is not very readable, because seeing a
7009: @code{cell+} tells us neither which kind of structure is
7010: accessed nor what field is accessed; we have to somehow infer the kind
7011: of structure, and then look up in the documentation, which field of
7012: that structure corresponds to that offset.
7013:
7014: Finally, this kind of address arithmetic also causes maintenance
7015: troubles: If you add or delete a field somewhere in the middle of the
7016: structure, you have to find and change all computations for the fields
7017: afterwards.
7018:
7019: So, instead of using @code{cell+} and friends directly, how
7020: about storing the offsets in constants:
7021:
7022: @example
7023: 0 constant a-offset
7024: 0 float+ constant b-offset
7025: 0 float+ cell+ faligned c-offset
7026: @end example
7027:
7028: Now we can get the address of field @code{x} with @code{x-offset
7029: +}. This is much better in all respects. Of course, you still
7030: have to change all later offset definitions if you add a field. You can
7031: fix this by declaring the offsets in the following way:
7032:
7033: @example
7034: 0 constant a-offset
7035: a-offset float+ constant b-offset
7036: b-offset cell+ faligned constant c-offset
7037: @end example
7038:
7039: Since we always use the offsets with @code{+}, we could use a defining
7040: word @code{cfield} that includes the @code{+} in the action of the
7041: defined word:
7042:
7043: @example
7044: : cfield ( n "name" -- )
7045: create ,
7046: does> ( name execution: addr1 -- addr2 )
7047: @@ + ;
7048:
7049: 0 cfield a
7050: 0 a float+ cfield b
7051: 0 b cell+ faligned cfield c
7052: @end example
7053:
7054: Instead of @code{x-offset +}, we now simply write @code{x}.
7055:
7056: The structure field words now can be used quite nicely. However,
7057: their definition is still a bit cumbersome: We have to repeat the
7058: name, the information about size and alignment is distributed before
7059: and after the field definitions etc. The structure package presented
7060: here addresses these problems.
7061:
7062: @node Structure Usage, Structure Naming Convention, Why explicit structure support?, Structures
7063: @subsection Structure Usage
7064: @cindex structure usage
7065:
7066: @cindex @code{field} usage
7067: @cindex @code{struct} usage
7068: @cindex @code{end-struct} usage
7069: You can define a structure for a (data-less) linked list with:
7070: @example
7071: struct
7072: cell% field list-next
7073: end-struct list%
7074: @end example
7075:
7076: With the address of the list node on the stack, you can compute the
7077: address of the field that contains the address of the next node with
7078: @code{list-next}. E.g., you can determine the length of a list
7079: with:
7080:
7081: @example
7082: : list-length ( list -- n )
7083: \ "list" is a pointer to the first element of a linked list
7084: \ "n" is the length of the list
7085: 0 BEGIN ( list1 n1 )
7086: over
7087: WHILE ( list1 n1 )
7088: 1+ swap list-next @@ swap
7089: REPEAT
7090: nip ;
7091: @end example
7092:
7093: You can reserve memory for a list node in the dictionary with
7094: @code{list% %allot}, which leaves the address of the list node on the
7095: stack. For the equivalent allocation on the heap you can use @code{list%
7096: %alloc} (or, for an @code{allocate}-like stack effect (i.e., with ior),
7097: use @code{list% %allocate}). You can get the the size of a list
7098: node with @code{list% %size} and its alignment with @code{list%
7099: %alignment}.
7100:
7101: Note that in ANS Forth the body of a @code{create}d word is
7102: @code{aligned} but not necessarily @code{faligned};
7103: therefore, if you do a:
7104: @example
7105: create @emph{name} foo% %allot
7106: @end example
7107:
7108: @noindent
7109: then the memory alloted for @code{foo%} is
7110: guaranteed to start at the body of @code{@emph{name}} only if
7111: @code{foo%} contains only character, cell and double fields.
7112:
7113: @cindex strcutures containing structures
7114: You can include a structure @code{foo%} as a field of
7115: another structure, like this:
7116: @example
7117: struct
7118: ...
7119: foo% field ...
7120: ...
7121: end-struct ...
7122: @end example
7123:
7124: @cindex structure extension
7125: @cindex extended records
7126: Instead of starting with an empty structure, you can extend an
7127: existing structure. E.g., a plain linked list without data, as defined
7128: above, is hardly useful; You can extend it to a linked list of integers,
7129: like this:@footnote{This feature is also known as @emph{extended
7130: records}. It is the main innovation in the Oberon language; in other
7131: words, adding this feature to Modula-2 led Wirth to create a new
7132: language, write a new compiler etc. Adding this feature to Forth just
7133: required a few lines of code.}
7134:
7135: @example
7136: list%
7137: cell% field intlist-int
7138: end-struct intlist%
7139: @end example
7140:
7141: @code{intlist%} is a structure with two fields:
7142: @code{list-next} and @code{intlist-int}.
7143:
7144: @cindex structures containing arrays
7145: You can specify an array type containing @emph{n} elements of
7146: type @code{foo%} like this:
7147:
7148: @example
7149: foo% @emph{n} *
7150: @end example
7151:
7152: You can use this array type in any place where you can use a normal
7153: type, e.g., when defining a @code{field}, or with
7154: @code{%allot}.
7155:
7156: @cindex first field optimization
7157: The first field is at the base address of a structure and the word
7158: for this field (e.g., @code{list-next}) actually does not change
7159: the address on the stack. You may be tempted to leave it away in the
7160: interest of run-time and space efficiency. This is not necessary,
7161: because the structure package optimizes this case and compiling such
7162: words does not generate any code. So, in the interest of readability
7163: and maintainability you should include the word for the field when
7164: accessing the field.
7165:
7166: @node Structure Naming Convention, Structure Implementation, Structure Usage, Structures
7167: @subsection Structure Naming Convention
7168: @cindex structure naming convention
7169:
7170: The field names that come to (my) mind are often quite generic, and,
7171: if used, would cause frequent name clashes. E.g., many structures
7172: probably contain a @code{counter} field. The structure names
7173: that come to (my) mind are often also the logical choice for the names
7174: of words that create such a structure.
7175:
7176: Therefore, I have adopted the following naming conventions:
7177:
7178: @itemize @bullet
7179: @cindex field naming convention
7180: @item
7181: The names of fields are of the form
7182: @code{@emph{struct}-@emph{field}}, where
7183: @code{@emph{struct}} is the basic name of the structure, and
7184: @code{@emph{field}} is the basic name of the field. You can
7185: think of field words as converting the (address of the)
7186: structure into the (address of the) field.
7187:
7188: @cindex structure naming convention
7189: @item
7190: The names of structures are of the form
7191: @code{@emph{struct}%}, where
7192: @code{@emph{struct}} is the basic name of the structure.
7193: @end itemize
7194:
7195: This naming convention does not work that well for fields of extended
7196: structures; e.g., the integer list structure has a field
7197: @code{intlist-int}, but has @code{list-next}, not
7198: @code{intlist-next}.
7199:
7200: @node Structure Implementation, Structure Glossary, Structure Naming Convention, Structures
7201: @subsection Structure Implementation
7202: @cindex structure implementation
7203: @cindex implementation of structures
7204:
7205: The central idea in the implementation is to pass the data about the
7206: structure being built on the stack, not in some global
7207: variable. Everything else falls into place naturally once this design
7208: decision is made.
7209:
7210: The type description on the stack is of the form @emph{align
7211: size}. Keeping the size on the top-of-stack makes dealing with arrays
7212: very simple.
7213:
7214: @code{field} is a defining word that uses @code{Create}
7215: and @code{DOES>}. The body of the field contains the offset
7216: of the field, and the normal @code{DOES>} action is simply:
7217:
7218: @example
7219: @ +
7220: @end example
7221:
7222: @noindent
7223: i.e., add the offset to the address, giving the stack effect
7224: @i{addr1 -- addr2} for a field.
7225:
7226: @cindex first field optimization, implementation
7227: This simple structure is slightly complicated by the optimization
7228: for fields with offset 0, which requires a different
7229: @code{DOES>}-part (because we cannot rely on there being
7230: something on the stack if such a field is invoked during
7231: compilation). Therefore, we put the different @code{DOES>}-parts
7232: in separate words, and decide which one to invoke based on the
7233: offset. For a zero offset, the field is basically a noop; it is
7234: immediate, and therefore no code is generated when it is compiled.
7235:
7236: @node Structure Glossary, , Structure Implementation, Structures
7237: @subsection Structure Glossary
7238: @cindex structure glossary
7239:
7240: doc-%align
7241: doc-%alignment
7242: doc-%alloc
7243: doc-%allocate
7244: doc-%allot
7245: doc-cell%
7246: doc-char%
7247: doc-dfloat%
7248: doc-double%
7249: doc-end-struct
7250: doc-field
7251: doc-float%
7252: doc-naligned
7253: doc-sfloat%
7254: doc-%size
7255: doc-struct
7256:
7257: @c -------------------------------------------------------------
7258: @node Object-oriented Forth, Passing Commands to the OS, Structures, Words
7259: @section Object-oriented Forth
7260:
7261: Gforth comes with three packages for object-oriented programming:
7262: @file{objects.fs}, @file{oof.fs}, and @file{mini-oof.fs}; none of them
7263: is preloaded, so you have to @code{include} them before use. The most
7264: important differences between these packages (and others) are discussed
7265: in @ref{Comparison with other object models}. All packages are written
7266: in ANS Forth and can be used with any other ANS Forth.
7267:
7268: @menu
7269: * Why object-oriented programming?::
7270: * Object-Oriented Terminology::
7271: * Objects::
7272: * OOF::
7273: * Mini-OOF::
7274: * Comparison with other object models::
7275: @end menu
7276:
7277:
7278: @node Why object-oriented programming?, Object-Oriented Terminology, , Object-oriented Forth
7279: @subsubsection Why object-oriented programming?
7280: @cindex object-oriented programming motivation
7281: @cindex motivation for object-oriented programming
7282:
7283: Often we have to deal with several data structures (@emph{objects}),
7284: that have to be treated similarly in some respects, but differently in
7285: others. Graphical objects are the textbook example: circles, triangles,
7286: dinosaurs, icons, and others, and we may want to add more during program
7287: development. We want to apply some operations to any graphical object,
7288: e.g., @code{draw} for displaying it on the screen. However, @code{draw}
7289: has to do something different for every kind of object.
7290: @comment TODO add some other operations eg perimeter, area
7291: @comment and tie in to concrete examples later..
7292:
7293: We could implement @code{draw} as a big @code{CASE}
7294: control structure that executes the appropriate code depending on the
7295: kind of object to be drawn. This would be not be very elegant, and,
7296: moreover, we would have to change @code{draw} every time we add
7297: a new kind of graphical object (say, a spaceship).
7298:
7299: What we would rather do is: When defining spaceships, we would tell
7300: the system: ``Here's how you @code{draw} a spaceship; you figure
7301: out the rest''.
7302:
7303: This is the problem that all systems solve that (rightfully) call
7304: themselves object-oriented; the object-oriented packages presented here
7305: solve this problem (and not much else).
7306: @comment TODO ?list properties of oo systems.. oo vs o-based?
7307:
7308: @node Object-Oriented Terminology, Objects, Why object-oriented programming?, Object-oriented Forth
7309: @subsubsection Object-Oriented Terminology
7310: @cindex object-oriented terminology
7311: @cindex terminology for object-oriented programming
7312:
7313: This section is mainly for reference, so you don't have to understand
7314: all of it right away. The terminology is mainly Smalltalk-inspired. In
7315: short:
7316:
7317: @table @emph
7318: @cindex class
7319: @item class
7320: a data structure definition with some extras.
7321:
7322: @cindex object
7323: @item object
7324: an instance of the data structure described by the class definition.
7325:
7326: @cindex instance variables
7327: @item instance variables
7328: fields of the data structure.
7329:
7330: @cindex selector
7331: @cindex method selector
7332: @cindex virtual function
7333: @item selector
7334: (or @emph{method selector}) a word (e.g.,
7335: @code{draw}) that performs an operation on a variety of data
7336: structures (classes). A selector describes @emph{what} operation to
7337: perform. In C++ terminology: a (pure) virtual function.
7338:
7339: @cindex method
7340: @item method
7341: the concrete definition that performs the operation
7342: described by the selector for a specific class. A method specifies
7343: @emph{how} the operation is performed for a specific class.
7344:
7345: @cindex selector invocation
7346: @cindex message send
7347: @cindex invoking a selector
7348: @item selector invocation
7349: a call of a selector. One argument of the call (the TOS (top-of-stack))
7350: is used for determining which method is used. In Smalltalk terminology:
7351: a message (consisting of the selector and the other arguments) is sent
7352: to the object.
7353:
7354: @cindex receiving object
7355: @item receiving object
7356: the object used for determining the method executed by a selector
7357: invocation. In the @file{objects.fs} model, it is the object that is on
7358: the TOS when the selector is invoked. (@emph{Receiving} comes from
7359: the Smalltalk @emph{message} terminology.)
7360:
7361: @cindex child class
7362: @cindex parent class
7363: @cindex inheritance
7364: @item child class
7365: a class that has (@emph{inherits}) all properties (instance variables,
7366: selectors, methods) from a @emph{parent class}. In Smalltalk
7367: terminology: The subclass inherits from the superclass. In C++
7368: terminology: The derived class inherits from the base class.
7369:
7370: @end table
7371:
7372: @c If you wonder about the message sending terminology, it comes from
7373: @c a time when each object had it's own task and objects communicated via
7374: @c message passing; eventually the Smalltalk developers realized that
7375: @c they can do most things through simple (indirect) calls. They kept the
7376: @c terminology.
7377:
7378:
7379: @node Objects, OOF, Object-Oriented Terminology, Object-oriented Forth
7380: @subsection The @file{objects.fs} model
7381: @cindex objects
7382: @cindex object-oriented programming
7383:
7384: @cindex @file{objects.fs}
7385: @cindex @file{oof.fs}
7386:
7387: 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}).
7388: @c McKewan's and Zsoter's packages
7389:
7390: This section assumes that you have read @ref{Structures}.
7391:
7392: The techniques on which this model is based have been used to implement
7393: the parser generator, Gray, and have also been used in Gforth for
7394: implementing the various flavours of word lists (hashed or not,
7395: case-sensitive or not, special-purpose word lists for locals etc.).
7396:
7397:
7398: @menu
7399: * Properties of the Objects model::
7400: * Basic Objects Usage::
7401: * The Objects base class::
7402: * Creating objects::
7403: * Object-Oriented Programming Style::
7404: * Class Binding::
7405: * Method conveniences::
7406: * Classes and Scoping::
7407: * Object Interfaces::
7408: * Objects Implementation::
7409: * Objects Glossary::
7410: @end menu
7411:
7412: Marcel Hendrix provided helpful comments on this section. Andras Zsoter
7413: and Bernd Paysan helped me with the related works section.
7414:
7415: @node Properties of the Objects model, Basic Objects Usage, Objects, Objects
7416: @subsubsection Properties of the @file{objects.fs} model
7417: @cindex @file{objects.fs} properties
7418:
7419: @itemize @bullet
7420: @item
7421: It is straightforward to pass objects on the stack. Passing
7422: selectors on the stack is a little less convenient, but possible.
7423:
7424: @item
7425: Objects are just data structures in memory, and are referenced by their
7426: address. You can create words for objects with normal defining words
7427: like @code{constant}. Likewise, there is no difference between instance
7428: variables that contain objects and those that contain other data.
7429:
7430: @item
7431: Late binding is efficient and easy to use.
7432:
7433: @item
7434: It avoids parsing, and thus avoids problems with state-smartness
7435: and reduced extensibility; for convenience there are a few parsing
7436: words, but they have non-parsing counterparts. There are also a few
7437: defining words that parse. This is hard to avoid, because all standard
7438: defining words parse (except @code{:noname}); however, such
7439: words are not as bad as many other parsing words, because they are not
7440: state-smart.
7441:
7442: @item
7443: It does not try to incorporate everything. It does a few things and does
7444: them well (IMO). In particular, this model was not designed to support
7445: information hiding (although it has features that may help); you can use
7446: a separate package for achieving this.
7447:
7448: @item
7449: It is layered; you don't have to learn and use all features to use this
7450: model. Only a few features are necessary (@xref{Basic Objects Usage},
7451: @xref{The Objects base class}, @xref{Creating objects}.), the others
7452: are optional and independent of each other.
7453:
7454: @item
7455: An implementation in ANS Forth is available.
7456:
7457: @end itemize
7458:
7459:
7460: @node Basic Objects Usage, The Objects base class, Properties of the Objects model, Objects
7461: @subsubsection Basic @file{objects.fs} Usage
7462: @cindex basic objects usage
7463: @cindex objects, basic usage
7464:
7465: You can define a class for graphical objects like this:
7466:
7467: @cindex @code{class} usage
7468: @cindex @code{end-class} usage
7469: @cindex @code{selector} usage
7470: @example
7471: object class \ "object" is the parent class
7472: selector draw ( x y graphical -- )
7473: end-class graphical
7474: @end example
7475:
7476: This code defines a class @code{graphical} with an
7477: operation @code{draw}. We can perform the operation
7478: @code{draw} on any @code{graphical} object, e.g.:
7479:
7480: @example
7481: 100 100 t-rex draw
7482: @end example
7483:
7484: @noindent
7485: where @code{t-rex} is a word (say, a constant) that produces a
7486: graphical object.
7487:
7488: @comment TODO add a 2nd operation eg perimeter.. and use for
7489: @comment a concrete example
7490:
7491: @cindex abstract class
7492: How do we create a graphical object? With the present definitions,
7493: we cannot create a useful graphical object. The class
7494: @code{graphical} describes graphical objects in general, but not
7495: any concrete graphical object type (C++ users would call it an
7496: @emph{abstract class}); e.g., there is no method for the selector
7497: @code{draw} in the class @code{graphical}.
7498:
7499: For concrete graphical objects, we define child classes of the
7500: class @code{graphical}, e.g.:
7501:
7502: @cindex @code{overrides} usage
7503: @cindex @code{field} usage in class definition
7504: @example
7505: graphical class \ "graphical" is the parent class
7506: cell% field circle-radius
7507:
7508: :noname ( x y circle -- )
7509: circle-radius @@ draw-circle ;
7510: overrides draw
7511:
7512: :noname ( n-radius circle -- )
7513: circle-radius ! ;
7514: overrides construct
7515:
7516: end-class circle
7517: @end example
7518:
7519: Here we define a class @code{circle} as a child of @code{graphical},
7520: with field @code{circle-radius} (which behaves just like a field
7521: (@pxref{Structures}); it defines (using @code{overrides}) new methods
7522: for the selectors @code{draw} and @code{construct} (@code{construct} is
7523: defined in @code{object}, the parent class of @code{graphical}).
7524:
7525: Now we can create a circle on the heap (i.e.,
7526: @code{allocate}d memory) with:
7527:
7528: @cindex @code{heap-new} usage
7529: @example
7530: 50 circle heap-new constant my-circle
7531: @end example
7532:
7533: @noindent
7534: @code{heap-new} invokes @code{construct}, thus
7535: initializing the field @code{circle-radius} with 50. We can draw
7536: this new circle at (100,100) with:
7537:
7538: @example
7539: 100 100 my-circle draw
7540: @end example
7541:
7542: @cindex selector invocation, restrictions
7543: @cindex class definition, restrictions
7544: Note: You can only invoke a selector if the object on the TOS
7545: (the receiving object) belongs to the class where the selector was
7546: defined or one of its descendents; e.g., you can invoke
7547: @code{draw} only for objects belonging to @code{graphical}
7548: or its descendents (e.g., @code{circle}). Immediately before
7549: @code{end-class}, the search order has to be the same as
7550: immediately after @code{class}.
7551:
7552: @node The Objects base class, Creating objects, Basic Objects Usage, Objects
7553: @subsubsection The @file{object.fs} base class
7554: @cindex @code{object} class
7555:
7556: When you define a class, you have to specify a parent class. So how do
7557: you start defining classes? There is one class available from the start:
7558: @code{object}. It is ancestor for all classes and so is the
7559: only class that has no parent. It has two selectors: @code{construct}
7560: and @code{print}.
7561:
7562: @node Creating objects, Object-Oriented Programming Style, The Objects base class, Objects
7563: @subsubsection Creating objects
7564: @cindex creating objects
7565: @cindex object creation
7566: @cindex object allocation options
7567:
7568: @cindex @code{heap-new} discussion
7569: @cindex @code{dict-new} discussion
7570: @cindex @code{construct} discussion
7571: You can create and initialize an object of a class on the heap with
7572: @code{heap-new} ( ... class -- object ) and in the dictionary
7573: (allocation with @code{allot}) with @code{dict-new} (
7574: ... class -- object ). Both words invoke @code{construct}, which
7575: consumes the stack items indicated by "..." above.
7576:
7577: @cindex @code{init-object} discussion
7578: @cindex @code{class-inst-size} discussion
7579: If you want to allocate memory for an object yourself, you can get its
7580: alignment and size with @code{class-inst-size 2@@} ( class --
7581: align size ). Once you have memory for an object, you can initialize
7582: it with @code{init-object} ( ... class object -- );
7583: @code{construct} does only a part of the necessary work.
7584:
7585: @node Object-Oriented Programming Style, Class Binding, Creating objects, Objects
7586: @subsubsection Object-Oriented Programming Style
7587: @cindex object-oriented programming style
7588:
7589: This section is not exhaustive.
7590:
7591: @cindex stack effects of selectors
7592: @cindex selectors and stack effects
7593: In general, it is a good idea to ensure that all methods for the
7594: same selector have the same stack effect: when you invoke a selector,
7595: you often have no idea which method will be invoked, so, unless all
7596: methods have the same stack effect, you will not know the stack effect
7597: of the selector invocation.
7598:
7599: One exception to this rule is methods for the selector
7600: @code{construct}. We know which method is invoked, because we
7601: specify the class to be constructed at the same place. Actually, I
7602: defined @code{construct} as a selector only to give the users a
7603: convenient way to specify initialization. The way it is used, a
7604: mechanism different from selector invocation would be more natural
7605: (but probably would take more code and more space to explain).
7606:
7607: @node Class Binding, Method conveniences, Object-Oriented Programming Style, Objects
7608: @subsubsection Class Binding
7609: @cindex class binding
7610: @cindex early binding
7611:
7612: @cindex late binding
7613: Normal selector invocations determine the method at run-time depending
7614: on the class of the receiving object. This run-time selection is called
7615: @i{late binding}.
7616:
7617: Sometimes it's preferable to invoke a different method. For example,
7618: you might want to use the simple method for @code{print}ing
7619: @code{object}s instead of the possibly long-winded @code{print} method
7620: of the receiver class. You can achieve this by replacing the invocation
7621: of @code{print} with:
7622:
7623: @cindex @code{[bind]} usage
7624: @example
7625: [bind] object print
7626: @end example
7627:
7628: @noindent
7629: in compiled code or:
7630:
7631: @cindex @code{bind} usage
7632: @example
7633: bind object print
7634: @end example
7635:
7636: @cindex class binding, alternative to
7637: @noindent
7638: in interpreted code. Alternatively, you can define the method with a
7639: name (e.g., @code{print-object}), and then invoke it through the
7640: name. Class binding is just a (often more convenient) way to achieve
7641: the same effect; it avoids name clutter and allows you to invoke
7642: methods directly without naming them first.
7643:
7644: @cindex superclass binding
7645: @cindex parent class binding
7646: A frequent use of class binding is this: When we define a method
7647: for a selector, we often want the method to do what the selector does
7648: in the parent class, and a little more. There is a special word for
7649: this purpose: @code{[parent]}; @code{[parent]
7650: @emph{selector}} is equivalent to @code{[bind] @emph{parent
7651: selector}}, where @code{@emph{parent}} is the parent
7652: class of the current class. E.g., a method definition might look like:
7653:
7654: @cindex @code{[parent]} usage
7655: @example
7656: :noname
7657: dup [parent] foo \ do parent's foo on the receiving object
7658: ... \ do some more
7659: ; overrides foo
7660: @end example
7661:
7662: @cindex class binding as optimization
7663: In @cite{Object-oriented programming in ANS Forth} (Forth Dimensions,
7664: March 1997), Andrew McKewan presents class binding as an optimization
7665: technique. I recommend not using it for this purpose unless you are in
7666: an emergency. Late binding is pretty fast with this model anyway, so the
7667: benefit of using class binding is small; the cost of using class binding
7668: where it is not appropriate is reduced maintainability.
7669:
7670: While we are at programming style questions: You should bind
7671: selectors only to ancestor classes of the receiving object. E.g., say,
7672: you know that the receiving object is of class @code{foo} or its
7673: descendents; then you should bind only to @code{foo} and its
7674: ancestors.
7675:
7676: @node Method conveniences, Classes and Scoping, Class Binding, Objects
7677: @subsubsection Method conveniences
7678: @cindex method conveniences
7679:
7680: In a method you usually access the receiving object pretty often. If
7681: you define the method as a plain colon definition (e.g., with
7682: @code{:noname}), you may have to do a lot of stack
7683: gymnastics. To avoid this, you can define the method with @code{m:
7684: ... ;m}. E.g., you could define the method for
7685: @code{draw}ing a @code{circle} with
7686:
7687: @cindex @code{this} usage
7688: @cindex @code{m:} usage
7689: @cindex @code{;m} usage
7690: @example
7691: m: ( x y circle -- )
7692: ( x y ) this circle-radius @@ draw-circle ;m
7693: @end example
7694:
7695: @cindex @code{exit} in @code{m: ... ;m}
7696: @cindex @code{exitm} discussion
7697: @cindex @code{catch} in @code{m: ... ;m}
7698: When this method is executed, the receiver object is removed from the
7699: stack; you can access it with @code{this} (admittedly, in this
7700: example the use of @code{m: ... ;m} offers no advantage). Note
7701: that I specify the stack effect for the whole method (i.e. including
7702: the receiver object), not just for the code between @code{m:}
7703: and @code{;m}. You cannot use @code{exit} in
7704: @code{m:...;m}; instead, use
7705: @code{exitm}.@footnote{Moreover, for any word that calls
7706: @code{catch} and was defined before loading
7707: @code{objects.fs}, you have to redefine it like I redefined
7708: @code{catch}: @code{: catch this >r catch r> to-this ;}}
7709:
7710: @cindex @code{inst-var} usage
7711: You will frequently use sequences of the form @code{this
7712: @emph{field}} (in the example above: @code{this
7713: circle-radius}). If you use the field only in this way, you can
7714: define it with @code{inst-var} and eliminate the
7715: @code{this} before the field name. E.g., the @code{circle}
7716: class above could also be defined with:
7717:
7718: @example
7719: graphical class
7720: cell% inst-var radius
7721:
7722: m: ( x y circle -- )
7723: radius @@ draw-circle ;m
7724: overrides draw
7725:
7726: m: ( n-radius circle -- )
7727: radius ! ;m
7728: overrides construct
7729:
7730: end-class circle
7731: @end example
7732:
7733: @code{radius} can only be used in @code{circle} and its
7734: descendent classes and inside @code{m:...;m}.
7735:
7736: @cindex @code{inst-value} usage
7737: You can also define fields with @code{inst-value}, which is
7738: to @code{inst-var} what @code{value} is to
7739: @code{variable}. You can change the value of such a field with
7740: @code{[to-inst]}. E.g., we could also define the class
7741: @code{circle} like this:
7742:
7743: @example
7744: graphical class
7745: inst-value radius
7746:
7747: m: ( x y circle -- )
7748: radius draw-circle ;m
7749: overrides draw
7750:
7751: m: ( n-radius circle -- )
7752: [to-inst] radius ;m
7753: overrides construct
7754:
7755: end-class circle
7756: @end example
7757:
7758:
7759: @node Classes and Scoping, Object Interfaces, Method conveniences, Objects
7760: @subsubsection Classes and Scoping
7761: @cindex classes and scoping
7762: @cindex scoping and classes
7763:
7764: Inheritance is frequent, unlike structure extension. This exacerbates
7765: the problem with the field name convention (@pxref{Structure Naming
7766: Convention}): One always has to remember in which class the field was
7767: originally defined; changing a part of the class structure would require
7768: changes for renaming in otherwise unaffected code.
7769:
7770: @cindex @code{inst-var} visibility
7771: @cindex @code{inst-value} visibility
7772: To solve this problem, I added a scoping mechanism (which was not in my
7773: original charter): A field defined with @code{inst-var} (or
7774: @code{inst-value}) is visible only in the class where it is defined and in
7775: the descendent classes of this class. Using such fields only makes
7776: sense in @code{m:}-defined methods in these classes anyway.
7777:
7778: This scoping mechanism allows us to use the unadorned field name,
7779: because name clashes with unrelated words become much less likely.
7780:
7781: @cindex @code{protected} discussion
7782: @cindex @code{private} discussion
7783: Once we have this mechanism, we can also use it for controlling the
7784: visibility of other words: All words defined after
7785: @code{protected} are visible only in the current class and its
7786: descendents. @code{public} restores the compilation
7787: (i.e. @code{current}) word list that was in effect before. If you
7788: have several @code{protected}s without an intervening
7789: @code{public} or @code{set-current}, @code{public}
7790: will restore the compilation word list in effect before the first of
7791: these @code{protected}s.
7792:
7793: @node Object Interfaces, Objects Implementation, Classes and Scoping, Objects
7794: @subsubsection Object Interfaces
7795: @cindex object interfaces
7796: @cindex interfaces for objects
7797:
7798: In this model you can only call selectors defined in the class of the
7799: receiving objects or in one of its ancestors. If you call a selector
7800: with a receiving object that is not in one of these classes, the
7801: result is undefined; if you are lucky, the program crashes
7802: immediately.
7803:
7804: @cindex selectors common to hardly-related classes
7805: Now consider the case when you want to have a selector (or several)
7806: available in two classes: You would have to add the selector to a
7807: common ancestor class, in the worst case to @code{object}. You
7808: may not want to do this, e.g., because someone else is responsible for
7809: this ancestor class.
7810:
7811: The solution for this problem is interfaces. An interface is a
7812: collection of selectors. If a class implements an interface, the
7813: selectors become available to the class and its descendents. A class
7814: can implement an unlimited number of interfaces. For the problem
7815: discussed above, we would define an interface for the selector(s), and
7816: both classes would implement the interface.
7817:
7818: As an example, consider an interface @code{storage} for
7819: writing objects to disk and getting them back, and a class
7820: @code{foo} that implements it. The code would look like this:
7821:
7822: @cindex @code{interface} usage
7823: @cindex @code{end-interface} usage
7824: @cindex @code{implementation} usage
7825: @example
7826: interface
7827: selector write ( file object -- )
7828: selector read1 ( file object -- )
7829: end-interface storage
7830:
7831: bar class
7832: storage implementation
7833:
7834: ... overrides write
7835: ... overrides read
7836: ...
7837: end-class foo
7838: @end example
7839:
7840: @noindent
7841: (I would add a word @code{read} @i{( file -- object )} that uses
7842: @code{read1} internally, but that's beyond the point illustrated
7843: here.)
7844:
7845: Note that you cannot use @code{protected} in an interface; and
7846: of course you cannot define fields.
7847:
7848: In the Neon model, all selectors are available for all classes;
7849: therefore it does not need interfaces. The price you pay in this model
7850: is slower late binding, and therefore, added complexity to avoid late
7851: binding.
7852:
7853: @node Objects Implementation, Objects Glossary, Object Interfaces, Objects
7854: @subsubsection @file{objects.fs} Implementation
7855: @cindex @file{objects.fs} implementation
7856:
7857: @cindex @code{object-map} discussion
7858: An object is a piece of memory, like one of the data structures
7859: described with @code{struct...end-struct}. It has a field
7860: @code{object-map} that points to the method map for the object's
7861: class.
7862:
7863: @cindex method map
7864: @cindex virtual function table
7865: The @emph{method map}@footnote{This is Self terminology; in C++
7866: terminology: virtual function table.} is an array that contains the
7867: execution tokens (@i{xt}s) of the methods for the object's class. Each
7868: selector contains an offset into a method map.
7869:
7870: @cindex @code{selector} implementation, class
7871: @code{selector} is a defining word that uses
7872: @code{CREATE} and @code{DOES>}. The body of the
7873: selector contains the offset; the @code{does>} action for a
7874: class selector is, basically:
7875:
7876: @example
7877: ( object addr ) @@ over object-map @@ + @@ execute
7878: @end example
7879:
7880: Since @code{object-map} is the first field of the object, it
7881: does not generate any code. As you can see, calling a selector has a
7882: small, constant cost.
7883:
7884: @cindex @code{current-interface} discussion
7885: @cindex class implementation and representation
7886: A class is basically a @code{struct} combined with a method
7887: map. During the class definition the alignment and size of the class
7888: are passed on the stack, just as with @code{struct}s, so
7889: @code{field} can also be used for defining class
7890: fields. However, passing more items on the stack would be
7891: inconvenient, so @code{class} builds a data structure in memory,
7892: which is accessed through the variable
7893: @code{current-interface}. After its definition is complete, the
7894: class is represented on the stack by a pointer (e.g., as parameter for
7895: a child class definition).
7896:
7897: A new class starts off with the alignment and size of its parent,
7898: and a copy of the parent's method map. Defining new fields extends the
7899: size and alignment; likewise, defining new selectors extends the
7900: method map. @code{overrides} just stores a new @i{xt} in the method
7901: map at the offset given by the selector.
7902:
7903: @cindex class binding, implementation
7904: Class binding just gets the @i{xt} at the offset given by the selector
7905: from the class's method map and @code{compile,}s (in the case of
7906: @code{[bind]}) it.
7907:
7908: @cindex @code{this} implementation
7909: @cindex @code{catch} and @code{this}
7910: @cindex @code{this} and @code{catch}
7911: I implemented @code{this} as a @code{value}. At the
7912: start of an @code{m:...;m} method the old @code{this} is
7913: stored to the return stack and restored at the end; and the object on
7914: the TOS is stored @code{TO this}. This technique has one
7915: disadvantage: If the user does not leave the method via
7916: @code{;m}, but via @code{throw} or @code{exit},
7917: @code{this} is not restored (and @code{exit} may
7918: crash). To deal with the @code{throw} problem, I have redefined
7919: @code{catch} to save and restore @code{this}; the same
7920: should be done with any word that can catch an exception. As for
7921: @code{exit}, I simply forbid it (as a replacement, there is
7922: @code{exitm}).
7923:
7924: @cindex @code{inst-var} implementation
7925: @code{inst-var} is just the same as @code{field}, with
7926: a different @code{DOES>} action:
7927: @example
7928: @@ this +
7929: @end example
7930: Similar for @code{inst-value}.
7931:
7932: @cindex class scoping implementation
7933: Each class also has a word list that contains the words defined with
7934: @code{inst-var} and @code{inst-value}, and its protected
7935: words. It also has a pointer to its parent. @code{class} pushes
7936: the word lists of the class and all its ancestors onto the search order stack,
7937: and @code{end-class} drops them.
7938:
7939: @cindex interface implementation
7940: An interface is like a class without fields, parent and protected
7941: words; i.e., it just has a method map. If a class implements an
7942: interface, its method map contains a pointer to the method map of the
7943: interface. The positive offsets in the map are reserved for class
7944: methods, therefore interface map pointers have negative
7945: offsets. Interfaces have offsets that are unique throughout the
7946: system, unlike class selectors, whose offsets are only unique for the
7947: classes where the selector is available (invokable).
7948:
7949: This structure means that interface selectors have to perform one
7950: indirection more than class selectors to find their method. Their body
7951: contains the interface map pointer offset in the class method map, and
7952: the method offset in the interface method map. The
7953: @code{does>} action for an interface selector is, basically:
7954:
7955: @example
7956: ( object selector-body )
7957: 2dup selector-interface @@ ( object selector-body object interface-offset )
7958: swap object-map @@ + @@ ( object selector-body map )
7959: swap selector-offset @@ + @@ execute
7960: @end example
7961:
7962: where @code{object-map} and @code{selector-offset} are
7963: first fields and generate no code.
7964:
7965: As a concrete example, consider the following code:
7966:
7967: @example
7968: interface
7969: selector if1sel1
7970: selector if1sel2
7971: end-interface if1
7972:
7973: object class
7974: if1 implementation
7975: selector cl1sel1
7976: cell% inst-var cl1iv1
7977:
7978: ' m1 overrides construct
7979: ' m2 overrides if1sel1
7980: ' m3 overrides if1sel2
7981: ' m4 overrides cl1sel2
7982: end-class cl1
7983:
7984: create obj1 object dict-new drop
7985: create obj2 cl1 dict-new drop
7986: @end example
7987:
7988: The data structure created by this code (including the data structure
7989: for @code{object}) is shown in the <a
7990: href="objects-implementation.eps">figure</a>, assuming a cell size of 4.
7991: @comment TODO add this diagram..
7992:
7993: @node Objects Glossary, , Objects Implementation, Objects
7994: @subsubsection @file{objects.fs} Glossary
7995: @cindex @file{objects.fs} Glossary
7996:
7997: doc---objects-bind
7998: doc---objects-<bind>
7999: doc---objects-bind'
8000: doc---objects-[bind]
8001: doc---objects-class
8002: doc---objects-class->map
8003: doc---objects-class-inst-size
8004: doc---objects-class-override!
8005: doc---objects-construct
8006: doc---objects-current'
8007: doc---objects-[current]
8008: doc---objects-current-interface
8009: doc---objects-dict-new
8010: doc---objects-drop-order
8011: doc---objects-end-class
8012: doc---objects-end-class-noname
8013: doc---objects-end-interface
8014: doc---objects-end-interface-noname
8015: doc---objects-exitm
8016: doc---objects-heap-new
8017: doc---objects-implementation
8018: doc---objects-init-object
8019: doc---objects-inst-value
8020: doc---objects-inst-var
8021: doc---objects-interface
8022: doc---objects-;m
8023: doc---objects-m:
8024: doc---objects-method
8025: doc---objects-object
8026: doc---objects-overrides
8027: doc---objects-[parent]
8028: doc---objects-print
8029: doc---objects-protected
8030: doc---objects-public
8031: doc---objects-push-order
8032: doc---objects-selector
8033: doc---objects-this
8034: doc---objects-<to-inst>
8035: doc---objects-[to-inst]
8036: doc---objects-to-this
8037: doc---objects-xt-new
8038:
8039: @c -------------------------------------------------------------
8040: @node OOF, Mini-OOF, Objects, Object-oriented Forth
8041: @subsection The @file{oof.fs} model
8042: @cindex oof
8043: @cindex object-oriented programming
8044:
8045: @cindex @file{objects.fs}
8046: @cindex @file{oof.fs}
8047:
8048: This section describes the @file{oof.fs} package.
8049:
8050: The package described in this section has been used in bigFORTH since 1991, and
8051: used for two large applications: a chromatographic system used to
8052: create new medicaments, and a graphic user interface library (MINOS).
8053:
8054: You can find a description (in German) of @file{oof.fs} in @cite{Object
8055: oriented bigFORTH} by Bernd Paysan, published in @cite{Vierte Dimension}
8056: 10(2), 1994.
8057:
8058: @menu
8059: * Properties of the OOF model::
8060: * Basic OOF Usage::
8061: * The OOF base class::
8062: * Class Declaration::
8063: * Class Implementation::
8064: @end menu
8065:
8066: @node Properties of the OOF model, Basic OOF Usage, OOF, OOF
8067: @subsubsection Properties of the @file{oof.fs} model
8068: @cindex @file{oof.fs} properties
8069:
8070: @itemize @bullet
8071: @item
8072: This model combines object oriented programming with information
8073: hiding. It helps you writing large application, where scoping is
8074: necessary, because it provides class-oriented scoping.
8075:
8076: @item
8077: Named objects, object pointers, and object arrays can be created,
8078: selector invocation uses the ``object selector'' syntax. Selector invocation
8079: to objects and/or selectors on the stack is a bit less convenient, but
8080: possible.
8081:
8082: @item
8083: Selector invocation and instance variable usage of the active object is
8084: straightforward, since both make use of the active object.
8085:
8086: @item
8087: Late binding is efficient and easy to use.
8088:
8089: @item
8090: State-smart objects parse selectors. However, extensibility is provided
8091: using a (parsing) selector @code{postpone} and a selector @code{'}.
8092:
8093: @item
8094: An implementation in ANS Forth is available.
8095:
8096: @end itemize
8097:
8098:
8099: @node Basic OOF Usage, The OOF base class, Properties of the OOF model, OOF
8100: @subsubsection Basic @file{oof.fs} Usage
8101: @cindex @file{oof.fs} usage
8102:
8103: This section uses the same example as for @code{objects} (@pxref{Basic Objects Usage}).
8104:
8105: You can define a class for graphical objects like this:
8106:
8107: @cindex @code{class} usage
8108: @cindex @code{class;} usage
8109: @cindex @code{method} usage
8110: @example
8111: object class graphical \ "object" is the parent class
8112: method draw ( x y graphical -- )
8113: class;
8114: @end example
8115:
8116: This code defines a class @code{graphical} with an
8117: operation @code{draw}. We can perform the operation
8118: @code{draw} on any @code{graphical} object, e.g.:
8119:
8120: @example
8121: 100 100 t-rex draw
8122: @end example
8123:
8124: @noindent
8125: where @code{t-rex} is an object or object pointer, created with e.g.
8126: @code{graphical : t-rex}.
8127:
8128: @cindex abstract class
8129: How do we create a graphical object? With the present definitions,
8130: we cannot create a useful graphical object. The class
8131: @code{graphical} describes graphical objects in general, but not
8132: any concrete graphical object type (C++ users would call it an
8133: @emph{abstract class}); e.g., there is no method for the selector
8134: @code{draw} in the class @code{graphical}.
8135:
8136: For concrete graphical objects, we define child classes of the
8137: class @code{graphical}, e.g.:
8138:
8139: @example
8140: graphical class circle \ "graphical" is the parent class
8141: cell var circle-radius
8142: how:
8143: : draw ( x y -- )
8144: circle-radius @@ draw-circle ;
8145:
8146: : init ( n-radius -- (
8147: circle-radius ! ;
8148: class;
8149: @end example
8150:
8151: Here we define a class @code{circle} as a child of @code{graphical},
8152: with a field @code{circle-radius}; it defines new methods for the
8153: selectors @code{draw} and @code{init} (@code{init} is defined in
8154: @code{object}, the parent class of @code{graphical}).
8155:
8156: Now we can create a circle in the dictionary with:
8157:
8158: @example
8159: 50 circle : my-circle
8160: @end example
8161:
8162: @noindent
8163: @code{:} invokes @code{init}, thus initializing the field
8164: @code{circle-radius} with 50. We can draw this new circle at (100,100)
8165: with:
8166:
8167: @example
8168: 100 100 my-circle draw
8169: @end example
8170:
8171: @cindex selector invocation, restrictions
8172: @cindex class definition, restrictions
8173: Note: You can only invoke a selector if the receiving object belongs to
8174: the class where the selector was defined or one of its descendents;
8175: e.g., you can invoke @code{draw} only for objects belonging to
8176: @code{graphical} or its descendents (e.g., @code{circle}). The scoping
8177: mechanism will check if you try to invoke a selector that is not
8178: defined in this class hierarchy, so you'll get an error at compilation
8179: time.
8180:
8181:
8182: @node The OOF base class, Class Declaration, Basic OOF Usage, OOF
8183: @subsubsection The @file{oof.fs} base class
8184: @cindex @file{oof.fs} base class
8185:
8186: When you define a class, you have to specify a parent class. So how do
8187: you start defining classes? There is one class available from the start:
8188: @code{object}. You have to use it as ancestor for all classes. It is the
8189: only class that has no parent. Classes are also objects, except that
8190: they don't have instance variables; class manipulation such as
8191: inheritance or changing definitions of a class is handled through
8192: selectors of the class @code{object}.
8193:
8194: @code{object} provides a number of selectors:
8195:
8196: @itemize @bullet
8197: @item
8198: @code{class} for subclassing, @code{definitions} to add definitions
8199: later on, and @code{class?} to get type informations (is the class a
8200: subclass of the class passed on the stack?).
8201: doc---object-class
8202: doc---object-definitions
8203: doc---object-class?
8204:
8205: @item
8206: @code{init} and @code{dispose} as constructor and destructor of the
8207: object. @code{init} is invocated after the object's memory is allocated,
8208: while @code{dispose} also handles deallocation. Thus if you redefine
8209: @code{dispose}, you have to call the parent's dispose with @code{super
8210: dispose}, too.
8211: doc---object-init
8212: doc---object-dispose
8213:
8214: @item
8215: @code{new}, @code{new[]}, @code{:}, @code{ptr}, @code{asptr}, and
8216: @code{[]} to create named and unnamed objects and object arrays or
8217: object pointers.
8218: doc---object-new
8219: doc---object-new[]
8220: doc---object-:
8221: doc---object-ptr
8222: doc---object-asptr
8223: doc---object-[]
8224:
8225: @item
8226: @code{::} and @code{super} for explicit scoping. You should use explicit
8227: scoping only for super classes or classes with the same set of instance
8228: variables. Explicitly-scoped selectors use early binding.
8229: doc---object-::
8230: doc---object-super
8231:
8232: @item
8233: @code{self} to get the address of the object
8234: doc---object-self
8235:
8236: @item
8237: @code{bind}, @code{bound}, @code{link}, and @code{is} to assign object
8238: pointers and instance defers.
8239: doc---object-bind
8240: doc---object-bound
8241: doc---object-link
8242: doc---object-is
8243:
8244: @item
8245: @code{'} to obtain selector tokens, @code{send} to invocate selectors
8246: form the stack, and @code{postpone} to generate selector invocation code.
8247: doc---object-'
8248: doc---object-postpone
8249:
8250: @item
8251: @code{with} and @code{endwith} to select the active object from the
8252: stack, and enable its scope. Using @code{with} and @code{endwith}
8253: also allows you to create code using selector @code{postpone} without being
8254: trapped by the state-smart objects.
8255: doc---object-with
8256: doc---object-endwith
8257:
8258: @end itemize
8259:
8260: @node Class Declaration, Class Implementation, The OOF base class, OOF
8261: @subsubsection Class Declaration
8262: @cindex class declaration
8263:
8264: @itemize @bullet
8265: @item
8266: Instance variables
8267: doc---oof-var
8268:
8269: @item
8270: Object pointers
8271: doc---oof-ptr
8272: doc---oof-asptr
8273:
8274: @item
8275: Instance defers
8276: doc---oof-defer
8277:
8278: @item
8279: Method selectors
8280: doc---oof-early
8281: doc---oof-method
8282:
8283: @item
8284: Class-wide variables
8285: doc---oof-static
8286:
8287: @item
8288: End declaration
8289: doc---oof-how:
8290: doc---oof-class;
8291:
8292: @end itemize
8293:
8294: @c -------------------------------------------------------------
8295: @node Class Implementation, , Class Declaration, OOF
8296: @subsubsection Class Implementation
8297: @cindex class implementation
8298:
8299: @c -------------------------------------------------------------
8300: @node Mini-OOF, Comparison with other object models, OOF, Object-oriented Forth
8301: @subsection The @file{mini-oof.fs} model
8302: @cindex mini-oof
8303:
8304: Gforth's third object oriented Forth package is a 12-liner. It uses a
8305: mixture of the @file{object.fs} and the @file{oof.fs} syntax,
8306: and reduces to the bare minimum of features. This is based on a posting
8307: of Bernd Paysan in comp.arch.
8308:
8309: @menu
8310: * Basic Mini-OOF Usage::
8311: * Mini-OOF Example::
8312: * Mini-OOF Implementation::
8313: @end menu
8314:
8315: @c -------------------------------------------------------------
8316: @node Basic Mini-OOF Usage, Mini-OOF Example, , Mini-OOF
8317: @subsubsection Basic @file{mini-oof.fs} Usage
8318: @cindex mini-oof usage
8319:
8320: There is a base class (@code{class}, which allocates one cell for the
8321: object pointer) plus seven other words: to define a method, a variable,
8322: a class; to end a class, to resolve binding, to allocate an object and
8323: to compile a class method.
8324: @comment TODO better description of the last one
8325:
8326: doc-object
8327: doc-method
8328: doc-var
8329: doc-class
8330: doc-end-class
8331: doc-defines
8332: doc-new
8333: doc-::
8334:
8335:
8336: @c -------------------------------------------------------------
8337: @node Mini-OOF Example, Mini-OOF Implementation, Basic Mini-OOF Usage, Mini-OOF
8338: @subsubsection Mini-OOF Example
8339: @cindex mini-oof example
8340:
8341: A short example shows how to use this package. This example, in slightly
8342: extended form, is supplied as @file{moof-exm.fs}
8343: @comment TODO could flesh this out with some comments from the Forthwrite article
8344:
8345: @example
8346: object class
8347: method init
8348: method draw
8349: end-class graphical
8350: @end example
8351:
8352: This code defines a class @code{graphical} with an
8353: operation @code{draw}. We can perform the operation
8354: @code{draw} on any @code{graphical} object, e.g.:
8355:
8356: @example
8357: 100 100 t-rex draw
8358: @end example
8359:
8360: where @code{t-rex} is an object or object pointer, created with e.g.
8361: @code{graphical new Constant t-rex}.
8362:
8363: For concrete graphical objects, we define child classes of the
8364: class @code{graphical}, e.g.:
8365:
8366: @example
8367: graphical class
8368: cell var circle-radius
8369: end-class circle \ "graphical" is the parent class
8370:
8371: :noname ( x y -- )
8372: circle-radius @@ draw-circle ; circle defines draw
8373: :noname ( r -- )
8374: circle-radius ! ; circle defines init
8375: @end example
8376:
8377: There is no implicit init method, so we have to define one. The creation
8378: code of the object now has to call init explicitely.
8379:
8380: @example
8381: circle new Constant my-circle
8382: 50 my-circle init
8383: @end example
8384:
8385: It is also possible to add a function to create named objects with
8386: automatic call of @code{init}, given that all objects have @code{init}
8387: on the same place:
8388:
8389: @example
8390: : new: ( .. o "name" -- )
8391: new dup Constant init ;
8392: 80 circle new: large-circle
8393: @end example
8394:
8395: We can draw this new circle at (100,100) with:
8396:
8397: @example
8398: 100 100 my-circle draw
8399: @end example
8400:
8401: @node Mini-OOF Implementation, , Mini-OOF Example, Mini-OOF
8402: @subsubsection @file{mini-oof.fs} Implementation
8403:
8404: Object-oriented systems with late binding typically use a
8405: ``vtable''-approach: the first variable in each object is a pointer to a
8406: table, which contains the methods as function pointers. The vtable
8407: may also contain other information.
8408:
8409: So first, let's declare methods:
8410:
8411: @example
8412: : method ( m v -- m' v ) Create over , swap cell+ swap
8413: DOES> ( ... o -- ... ) @ over @ + @ execute ;
8414: @end example
8415:
8416: During method declaration, the number of methods and instance
8417: variables is on the stack (in address units). @code{method} creates
8418: one method and increments the method number. To execute a method, it
8419: takes the object, fetches the vtable pointer, adds the offset, and
8420: executes the @i{xt} stored there. Each method takes the object it is
8421: invoked from as top of stack parameter. The method itself should
8422: consume that object.
8423:
8424: Now, we also have to declare instance variables
8425:
8426: @example
8427: : var ( m v size -- m v' ) Create over , +
8428: DOES> ( o -- addr ) @ + ;
8429: @end example
8430:
8431: As before, a word is created with the current offset. Instance
8432: variables can have different sizes (cells, floats, doubles, chars), so
8433: all we do is take the size and add it to the offset. If your machine
8434: has alignment restrictions, put the proper @code{aligned} or
8435: @code{faligned} before the variable, to adjust the variable
8436: offset. That's why it is on the top of stack.
8437:
8438: We need a starting point (the base object) and some syntactic sugar:
8439:
8440: @example
8441: Create object 1 cells , 2 cells ,
8442: : class ( class -- class methods vars ) dup 2@ ;
8443: @end example
8444:
8445: For inheritance, the vtable of the parent object has to be
8446: copied when a new, derived class is declared. This gives all the
8447: methods of the parent class, which can be overridden, though.
8448:
8449: @example
8450: : end-class ( class methods vars -- )
8451: Create here >r , dup , 2 cells ?DO ['] noop , 1 cells +LOOP
8452: cell+ dup cell+ r> rot @ 2 cells /string move ;
8453: @end example
8454:
8455: The first line creates the vtable, initialized with
8456: @code{noop}s. The second line is the inheritance mechanism, it
8457: copies the xts from the parent vtable.
8458:
8459: We still have no way to define new methods, let's do that now:
8460:
8461: @example
8462: : defines ( xt class -- ) ' >body @ + ! ;
8463: @end example
8464:
8465: To allocate a new object, we need a word, too:
8466:
8467: @example
8468: : new ( class -- o ) here over @ allot swap over ! ;
8469: @end example
8470:
8471: Sometimes derived classes want to access the method of the
8472: parent object. There are two ways to achieve this with Mini-OOF:
8473: first, you could use named words, and second, you could look up the
8474: vtable of the parent object.
8475:
8476: @example
8477: : :: ( class "name" -- ) ' >body @ + @ compile, ;
8478: @end example
8479:
8480:
8481: Nothing can be more confusing than a good example, so here is
8482: one. First let's declare a text object (called
8483: @code{button}), that stores text and position:
8484:
8485: @example
8486: object class
8487: cell var text
8488: cell var len
8489: cell var x
8490: cell var y
8491: method init
8492: method draw
8493: end-class button
8494: @end example
8495:
8496: @noindent
8497: Now, implement the two methods, @code{draw} and @code{init}:
8498:
8499: @example
8500: :noname ( o -- )
8501: >r r@ x @ r@ y @ at-xy r@ text @ r> len @ type ;
8502: button defines draw
8503: :noname ( addr u o -- )
8504: >r 0 r@ x ! 0 r@ y ! r@ len ! r> text ! ;
8505: button defines init
8506: @end example
8507:
8508: @noindent
8509: To demonstrate inheritance, we define a class @code{bold-button}, with no
8510: new data and no new methods:
8511:
8512: @example
8513: button class
8514: end-class bold-button
8515:
8516: : bold 27 emit ." [1m" ;
8517: : normal 27 emit ." [0m" ;
8518: @end example
8519:
8520: @noindent
8521: The class @code{bold-button} has a different draw method to
8522: @code{button}, but the new method is defined in terms of the draw method
8523: for @code{button}:
8524:
8525: @example
8526: :noname bold [ button :: draw ] normal ; bold-button defines draw
8527: @end example
8528:
8529: @noindent
8530: Finally, create two objects and apply methods:
8531:
8532: @example
8533: button new Constant foo
8534: s" thin foo" foo init
8535: page
8536: foo draw
8537: bold-button new Constant bar
8538: s" fat bar" bar init
8539: 1 bar y !
8540: bar draw
8541: @end example
8542:
8543:
8544: @node Comparison with other object models, , Mini-OOF, Object-oriented Forth
8545: @subsubsection Comparison with other object models
8546: @cindex comparison of object models
8547: @cindex object models, comparison
8548:
8549: Many object-oriented Forth extensions have been proposed (@cite{A survey
8550: of object-oriented Forths} (SIGPLAN Notices, April 1996) by Bradford
8551: J. Rodriguez and W. F. S. Poehlman lists 17). This section discusses the
8552: relation of the object models described here to two well-known and two
8553: closely-related (by the use of method maps) models.
8554:
8555: @cindex Neon model
8556: The most popular model currently seems to be the Neon model (see
8557: @cite{Object-oriented programming in ANS Forth} (Forth Dimensions, March
8558: 1997) by Andrew McKewan) but this model has a number of limitations
8559: @footnote{A longer version of this critique can be
8560: found in @cite{On Standardizing Object-Oriented Forth Extensions} (Forth
8561: Dimensions, May 1997) by Anton Ertl.}:
8562:
8563: @itemize @bullet
8564: @item
8565: It uses a @code{@emph{selector
8566: object}} syntax, which makes it unnatural to pass objects on the
8567: stack.
8568:
8569: @item
8570: It requires that the selector parses the input stream (at
8571: compile time); this leads to reduced extensibility and to bugs that are+
8572: hard to find.
8573:
8574: @item
8575: It allows using every selector to every object;
8576: this eliminates the need for classes, but makes it harder to create
8577: efficient implementations.
8578: @end itemize
8579:
8580: @cindex Pountain's object-oriented model
8581: Another well-known publication is @cite{Object-Oriented Forth} (Academic
8582: Press, London, 1987) by Dick Pountain. However, it is not really about
8583: object-oriented programming, because it hardly deals with late
8584: binding. Instead, it focuses on features like information hiding and
8585: overloading that are characteristic of modular languages like Ada (83).
8586:
8587: @cindex Zsoter's object-oriented model
8588: In @cite{Does late binding have to be slow?} (Forth Dimensions 18(1) 1996, pages 31-35)
8589: Andras Zsoter describes a model that makes heavy use of an active object
8590: (like @code{this} in @file{objects.fs}): The active object is not only
8591: used for accessing all fields, but also specifies the receiving object
8592: of every selector invocation; you have to change the active object
8593: explicitly with @code{@{ ... @}}, whereas in @file{objects.fs} it
8594: changes more or less implicitly at @code{m: ... ;m}. Such a change at
8595: the method entry point is unnecessary with the Zsoter's model, because
8596: the receiving object is the active object already. On the other hand, the explicit
8597: change is absolutely necessary in that model, because otherwise no one
8598: could ever change the active object. An ANS Forth implementation of this
8599: model is available at @url{http://www.forth.org/fig/oopf.html}.
8600:
8601: @cindex @file{oof.fs}, differences to other models
8602: The @file{oof.fs} model combines information hiding and overloading
8603: resolution (by keeping names in various word lists) with object-oriented
8604: programming. It sets the active object implicitly on method entry, but
8605: also allows explicit changing (with @code{>o...o>} or with
8606: @code{with...endwith}). It uses parsing and state-smart objects and
8607: classes for resolving overloading and for early binding: the object or
8608: class parses the selector and determines the method from this. If the
8609: selector is not parsed by an object or class, it performs a call to the
8610: selector for the active object (late binding), like Zsoter's model.
8611: Fields are always accessed through the active object. The big
8612: disadvantage of this model is the parsing and the state-smartness, which
8613: reduces extensibility and increases the opportunities for subtle bugs;
8614: essentially, you are only safe if you never tick or @code{postpone} an
8615: object or class (Bernd disagrees, but I (Anton) am not convinced).
8616:
8617: @cindex @file{mini-oof.fs}, differences to other models
8618: The @file{mini-oof.fs} model is quite similar to a very stripped-down version of
8619: the @file{objects.fs} model, but syntactically it is a mixture of the @file{objects.fs} and
8620: @file{oof.fs} models.
8621:
8622: @c -------------------------------------------------------------
8623: @node Passing Commands to the OS, Miscellaneous Words, Object-oriented Forth, Words
8624: @section Passing Commands to the Operating System
8625: @cindex operating system - passing commands
8626: @cindex shell commands
8627:
8628: Gforth allows you to pass an arbitrary string to the host operating
8629: system shell (if such a thing exists) for execution.
8630:
8631: doc-sh
8632: doc-system
8633: doc-$?
8634: doc-getenv
8635:
8636: @c -------------------------------------------------------------
8637: @node Miscellaneous Words, , Passing Commands to the OS, Words
8638: @section Miscellaneous Words
8639: @cindex miscellaneous words
8640:
8641: @comment TODO find homes for these
8642:
8643: These section lists the ANS Forth words that are not documented
8644: elsewhere in this manual. Ultimately, they all need proper homes.
8645:
8646: doc-ms
8647: doc-time&date
8648:
8649: doc-[compile]
8650:
8651: The following ANS Forth words are not currently supported by Gforth
8652: (@pxref{ANS conformance}):
8653:
8654: @code{EDITOR}
8655: @code{EKEY}
8656: @code{EKEY>CHAR}
8657: @code{EKEY?}
8658: @code{EMIT?}
8659: @code{FORGET}
8660:
8661: @c ******************************************************************
8662: @node Error messages, Tools, Words, Top
8663: @chapter Error messages
8664: @cindex error messages
8665: @cindex backtrace
8666:
8667: A typical Gforth error message looks like this:
8668:
8669: @example
8670: in file included from :-1
8671: in file included from ./yyy.fs:1
8672: ./xxx.fs:4: Invalid memory address
8673: bar
8674: ^^^
8675: $400E664C @@
8676: $400E6664 foo
8677: @end example
8678:
8679: The message identifying the error is @code{Invalid memory address}. The
8680: error happened when text-interpreting line 4 of the file
8681: @file{./xxx.fs}. This line is given (it contains @code{bar}), and the
8682: word on the line where the error happened, is pointed out (with
8683: @code{^^^}).
8684:
8685: The file containing the error was included in line 1 of @file{./yyy.fs},
8686: and @file{yyy.fs} was included from a non-file (in this case, by giving
8687: @file{yyy.fs} as command-line parameter to Gforth).
8688:
8689: At the end of the error message you find a return stack dump that can be
8690: interpreted as a backtrace (possibly empty). On top you find the top of
8691: the return stack when the @code{throw} happened, and at the bottom you
8692: find the return stack entry just above the return stack of the topmost
8693: text interpreter.
8694:
8695: To the right of most return stack entries you see a guess for the word
8696: that pushed that return stack entry as its return address. This gives a
8697: backtrace. In our case we see that @code{bar} called @code{foo}, and
8698: @code{foo} called @code{@@} (and @code{@@} had an @emph{Invalid memory
8699: address} exception).
8700:
8701: Note that the backtrace is not perfect: We don't know which return stack
8702: entries are return addresses (so we may get false positives); and in
8703: some cases (e.g., for @code{abort"}) we cannot determine from the return
8704: address the word that pushed the return address, so for some return
8705: addresses you see no names in the return stack dump.
8706:
8707: @cindex @code{catch} and backtraces
8708: The return stack dump represents the return stack at the time when a
8709: specific @code{throw} was executed. In programs that make use of
8710: @code{catch}, it is not necessarily clear which @code{throw} should be
8711: used for the return stack dump (e.g., consider one @code{throw} that
8712: indicates an error, which is caught, and during recovery another error
8713: happens; which @code{throw} should be used for the stack dump). Gforth
8714: presents the return stack dump for the first @code{throw} after the last
8715: executed (not returned-to) @code{catch}; this works well in the usual
8716: case.
8717:
8718: @cindex @code{gforth-fast} and backtraces
8719: @cindex @code{gforth-fast}, difference from @code{gforth}
8720: @cindex backtraces with @code{gforth-fast}
8721: @cindex return stack dump with @code{gforth-fast}
8722: @code{gforth} is able to do a return stack dump for throws generated
8723: from primitives (e.g., invalid memory address, stack empty etc.);
8724: @code{gforth-fast} is only able to do a return stack dump from a
8725: directly called @code{throw} (including @code{abort} etc.). This is the
8726: only difference (apart from a speed factor of between 1.15 (K6-2) and
8727: 1.6 (21164A)) between @code{gforth} and @code{gforth-fast}. Given an
8728: exception caused by a primitive in @code{gforth-fast}, you will
8729: typically see no return stack dump at all; however, if the exception is
8730: caught by @code{catch} (e.g., for restoring some state), and then
8731: @code{throw}n again, the return stack dump will be for the first such
8732: @code{throw}.
8733:
8734: @c ******************************************************************
8735: @node Tools, ANS conformance, Error messages, Top
8736: @chapter Tools
8737:
8738: @menu
8739: * ANS Report:: Report the words used, sorted by wordset.
8740: @end menu
8741:
8742: See also @ref{Emacs and Gforth}.
8743:
8744: @node ANS Report, , Tools, Tools
8745: @section @file{ans-report.fs}: Report the words used, sorted by wordset
8746: @cindex @file{ans-report.fs}
8747: @cindex report the words used in your program
8748: @cindex words used in your program
8749:
8750: If you want to label a Forth program as ANS Forth Program, you must
8751: document which wordsets the program uses; for extension wordsets, it is
8752: helpful to list the words the program requires from these wordsets
8753: (because Forth systems are allowed to provide only some words of them).
8754:
8755: The @file{ans-report.fs} tool makes it easy for you to determine which
8756: words from which wordset and which non-ANS words your application
8757: uses. You simply have to include @file{ans-report.fs} before loading the
8758: program you want to check. After loading your program, you can get the
8759: report with @code{print-ans-report}. A typical use is to run this as
8760: batch job like this:
8761: @example
8762: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
8763: @end example
8764:
8765: The output looks like this (for @file{compat/control.fs}):
8766: @example
8767: The program uses the following words
8768: from CORE :
8769: : POSTPONE THEN ; immediate ?dup IF 0=
8770: from BLOCK-EXT :
8771: \
8772: from FILE :
8773: (
8774: @end example
8775:
8776: @subsection Caveats
8777:
8778: Note that @file{ans-report.fs} just checks which words are used, not whether
8779: they are used in an ANS Forth conforming way!
8780:
8781: Some words are defined in several wordsets in the
8782: standard. @file{ans-report.fs} reports them for only one of the
8783: wordsets, and not necessarily the one you expect. It depends on usage
8784: which wordset is the right one to specify. E.g., if you only use the
8785: compilation semantics of @code{S"}, it is a Core word; if you also use
8786: its interpretation semantics, it is a File word.
8787:
8788: @c ******************************************************************
8789: @node ANS conformance, Model, Tools, Top
8790: @chapter ANS conformance
8791: @cindex ANS conformance of Gforth
8792:
8793: To the best of our knowledge, Gforth is an
8794:
8795: ANS Forth System
8796: @itemize @bullet
8797: @item providing the Core Extensions word set
8798: @item providing the Block word set
8799: @item providing the Block Extensions word set
8800: @item providing the Double-Number word set
8801: @item providing the Double-Number Extensions word set
8802: @item providing the Exception word set
8803: @item providing the Exception Extensions word set
8804: @item providing the Facility word set
8805: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
8806: @item providing the File Access word set
8807: @item providing the File Access Extensions word set
8808: @item providing the Floating-Point word set
8809: @item providing the Floating-Point Extensions word set
8810: @item providing the Locals word set
8811: @item providing the Locals Extensions word set
8812: @item providing the Memory-Allocation word set
8813: @item providing the Memory-Allocation Extensions word set (that one's easy)
8814: @item providing the Programming-Tools word set
8815: @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
8816: @item providing the Search-Order word set
8817: @item providing the Search-Order Extensions word set
8818: @item providing the String word set
8819: @item providing the String Extensions word set (another easy one)
8820: @end itemize
8821:
8822: @cindex system documentation
8823: In addition, ANS Forth systems are required to document certain
8824: implementation choices. This chapter tries to meet these
8825: requirements. In many cases it gives a way to ask the system for the
8826: information instead of providing the information directly, in
8827: particular, if the information depends on the processor, the operating
8828: system or the installation options chosen, or if they are likely to
8829: change during the maintenance of Gforth.
8830:
8831: @comment The framework for the rest has been taken from pfe.
8832:
8833: @menu
8834: * The Core Words::
8835: * The optional Block word set::
8836: * The optional Double Number word set::
8837: * The optional Exception word set::
8838: * The optional Facility word set::
8839: * The optional File-Access word set::
8840: * The optional Floating-Point word set::
8841: * The optional Locals word set::
8842: * The optional Memory-Allocation word set::
8843: * The optional Programming-Tools word set::
8844: * The optional Search-Order word set::
8845: @end menu
8846:
8847:
8848: @c =====================================================================
8849: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
8850: @comment node-name, next, previous, up
8851: @section The Core Words
8852: @c =====================================================================
8853: @cindex core words, system documentation
8854: @cindex system documentation, core words
8855:
8856: @menu
8857: * core-idef:: Implementation Defined Options
8858: * core-ambcond:: Ambiguous Conditions
8859: * core-other:: Other System Documentation
8860: @end menu
8861:
8862: @c ---------------------------------------------------------------------
8863: @node core-idef, core-ambcond, The Core Words, The Core Words
8864: @subsection Implementation Defined Options
8865: @c ---------------------------------------------------------------------
8866: @cindex core words, implementation-defined options
8867: @cindex implementation-defined options, core words
8868:
8869:
8870: @table @i
8871: @item (Cell) aligned addresses:
8872: @cindex cell-aligned addresses
8873: @cindex aligned addresses
8874: processor-dependent. Gforth's alignment words perform natural alignment
8875: (e.g., an address aligned for a datum of size 8 is divisible by
8876: 8). Unaligned accesses usually result in a @code{-23 THROW}.
8877:
8878: @item @code{EMIT} and non-graphic characters:
8879: @cindex @code{EMIT} and non-graphic characters
8880: @cindex non-graphic characters and @code{EMIT}
8881: The character is output using the C library function (actually, macro)
8882: @code{putc}.
8883:
8884: @item character editing of @code{ACCEPT} and @code{EXPECT}:
8885: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
8886: @cindex editing in @code{ACCEPT} and @code{EXPECT}
8887: @cindex @code{ACCEPT}, editing
8888: @cindex @code{EXPECT}, editing
8889: This is modeled on the GNU readline library (@pxref{Readline
8890: Interaction, , Command Line Editing, readline, The GNU Readline
8891: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
8892: producing a full word completion every time you type it (instead of
8893: producing the common prefix of all completions). @xref{Command-line editing}.
8894:
8895: @item character set:
8896: @cindex character set
8897: The character set of your computer and display device. Gforth is
8898: 8-bit-clean (but some other component in your system may make trouble).
8899:
8900: @item Character-aligned address requirements:
8901: @cindex character-aligned address requirements
8902: installation-dependent. Currently a character is represented by a C
8903: @code{unsigned char}; in the future we might switch to @code{wchar_t}
8904: (Comments on that requested).
8905:
8906: @item character-set extensions and matching of names:
8907: @cindex character-set extensions and matching of names
8908: @cindex case-sensitivity for name lookup
8909: @cindex name lookup, case-sensitivity
8910: @cindex locale and case-sensitivity
8911: Any character except the ASCII NUL character can be used in a
8912: name. Matching is case-insensitive (except in @code{TABLE}s). The
8913: matching is performed using the C function @code{strncasecmp}, whose
8914: function is probably influenced by the locale. E.g., the @code{C} locale
8915: does not know about accents and umlauts, so they are matched
8916: case-sensitively in that locale. For portability reasons it is best to
8917: write programs such that they work in the @code{C} locale. Then one can
8918: use libraries written by a Polish programmer (who might use words
8919: containing ISO Latin-2 encoded characters) and by a French programmer
8920: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
8921: funny results for some of the words (which ones, depends on the font you
8922: are using)). Also, the locale you prefer may not be available in other
8923: operating systems. Hopefully, Unicode will solve these problems one day.
8924:
8925: @item conditions under which control characters match a space delimiter:
8926: @cindex space delimiters
8927: @cindex control characters as delimiters
8928: If @code{WORD} is called with the space character as a delimiter, all
8929: white-space characters (as identified by the C macro @code{isspace()})
8930: are delimiters. @code{PARSE}, on the other hand, treats space like other
8931: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
8932: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
8933: interpreter (aka text interpreter) by default, treats all white-space
8934: characters as delimiters.
8935:
8936: @item format of the control-flow stack:
8937: @cindex control-flow stack, format
8938: The data stack is used as control-flow stack. The size of a control-flow
8939: stack item in cells is given by the constant @code{cs-item-size}. At the
8940: time of this writing, an item consists of a (pointer to a) locals list
8941: (third), an address in the code (second), and a tag for identifying the
8942: item (TOS). The following tags are used: @code{defstart},
8943: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
8944: @code{scopestart}.
8945:
8946: @item conversion of digits > 35
8947: @cindex digits > 35
8948: The characters @code{[\]^_'} are the digits with the decimal value
8949: 36@minus{}41. There is no way to input many of the larger digits.
8950:
8951: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
8952: @cindex @code{EXPECT}, display after end of input
8953: @cindex @code{ACCEPT}, display after end of input
8954: The cursor is moved to the end of the entered string. If the input is
8955: terminated using the @kbd{Return} key, a space is typed.
8956:
8957: @item exception abort sequence of @code{ABORT"}:
8958: @cindex exception abort sequence of @code{ABORT"}
8959: @cindex @code{ABORT"}, exception abort sequence
8960: The error string is stored into the variable @code{"error} and a
8961: @code{-2 throw} is performed.
8962:
8963: @item input line terminator:
8964: @cindex input line terminator
8965: @cindex line terminator on input
8966: @cindex newline character on input
8967: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
8968: lines. One of these characters is typically produced when you type the
8969: @kbd{Enter} or @kbd{Return} key.
8970:
8971: @item maximum size of a counted string:
8972: @cindex maximum size of a counted string
8973: @cindex counted string, maximum size
8974: @code{s" /counted-string" environment? drop .}. Currently 255 characters
8975: on all ports, but this may change.
8976:
8977: @item maximum size of a parsed string:
8978: @cindex maximum size of a parsed string
8979: @cindex parsed string, maximum size
8980: Given by the constant @code{/line}. Currently 255 characters.
8981:
8982: @item maximum size of a definition name, in characters:
8983: @cindex maximum size of a definition name, in characters
8984: @cindex name, maximum length
8985: 31
8986:
8987: @item maximum string length for @code{ENVIRONMENT?}, in characters:
8988: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
8989: @cindex @code{ENVIRONMENT?} string length, maximum
8990: 31
8991:
8992: @item method of selecting the user input device:
8993: @cindex user input device, method of selecting
8994: The user input device is the standard input. There is currently no way to
8995: change it from within Gforth. However, the input can typically be
8996: redirected in the command line that starts Gforth.
8997:
8998: @item method of selecting the user output device:
8999: @cindex user output device, method of selecting
9000: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
9001: @code{outfile-id} (@code{stdout} by default). Gforth uses unbuffered
9002: output when the user output device is a terminal, otherwise the output
9003: is buffered.
9004:
9005: @item methods of dictionary compilation:
9006: What are we expected to document here?
9007:
9008: @item number of bits in one address unit:
9009: @cindex number of bits in one address unit
9010: @cindex address unit, size in bits
9011: @code{s" address-units-bits" environment? drop .}. 8 in all current
9012: ports.
9013:
9014: @item number representation and arithmetic:
9015: @cindex number representation and arithmetic
9016: Processor-dependent. Binary two's complement on all current ports.
9017:
9018: @item ranges for integer types:
9019: @cindex ranges for integer types
9020: @cindex integer types, ranges
9021: Installation-dependent. Make environmental queries for @code{MAX-N},
9022: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
9023: unsigned (and positive) types is 0. The lower bound for signed types on
9024: two's complement and one's complement machines machines can be computed
9025: by adding 1 to the upper bound.
9026:
9027: @item read-only data space regions:
9028: @cindex read-only data space regions
9029: @cindex data-space, read-only regions
9030: The whole Forth data space is writable.
9031:
9032: @item size of buffer at @code{WORD}:
9033: @cindex size of buffer at @code{WORD}
9034: @cindex @code{WORD} buffer size
9035: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9036: shared with the pictured numeric output string. If overwriting
9037: @code{PAD} is acceptable, it is as large as the remaining dictionary
9038: space, although only as much can be sensibly used as fits in a counted
9039: string.
9040:
9041: @item size of one cell in address units:
9042: @cindex cell size
9043: @code{1 cells .}.
9044:
9045: @item size of one character in address units:
9046: @cindex char size
9047: @code{1 chars .}. 1 on all current ports.
9048:
9049: @item size of the keyboard terminal buffer:
9050: @cindex size of the keyboard terminal buffer
9051: @cindex terminal buffer, size
9052: Varies. You can determine the size at a specific time using @code{lp@@
9053: tib - .}. It is shared with the locals stack and TIBs of files that
9054: include the current file. You can change the amount of space for TIBs
9055: and locals stack at Gforth startup with the command line option
9056: @code{-l}.
9057:
9058: @item size of the pictured numeric output buffer:
9059: @cindex size of the pictured numeric output buffer
9060: @cindex pictured numeric output buffer, size
9061: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
9062: shared with @code{WORD}.
9063:
9064: @item size of the scratch area returned by @code{PAD}:
9065: @cindex size of the scratch area returned by @code{PAD}
9066: @cindex @code{PAD} size
9067: The remainder of dictionary space. @code{unused pad here - - .}.
9068:
9069: @item system case-sensitivity characteristics:
9070: @cindex case-sensitivity characteristics
9071: Dictionary searches are case-insensitive (except in
9072: @code{TABLE}s). However, as explained above under @i{character-set
9073: extensions}, the matching for non-ASCII characters is determined by the
9074: locale you are using. In the default @code{C} locale all non-ASCII
9075: characters are matched case-sensitively.
9076:
9077: @item system prompt:
9078: @cindex system prompt
9079: @cindex prompt
9080: @code{ ok} in interpret state, @code{ compiled} in compile state.
9081:
9082: @item division rounding:
9083: @cindex division rounding
9084: installation dependent. @code{s" floored" environment? drop .}. We leave
9085: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
9086: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
9087:
9088: @item values of @code{STATE} when true:
9089: @cindex @code{STATE} values
9090: -1.
9091:
9092: @item values returned after arithmetic overflow:
9093: On two's complement machines, arithmetic is performed modulo
9094: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9095: arithmetic (with appropriate mapping for signed types). Division by zero
9096: typically results in a @code{-55 throw} (Floating-point unidentified
9097: fault), although a @code{-10 throw} (divide by zero) would be more
9098: appropriate.
9099:
9100: @item whether the current definition can be found after @t{DOES>}:
9101: @cindex @t{DOES>}, visibility of current definition
9102: No.
9103:
9104: @end table
9105:
9106: @c ---------------------------------------------------------------------
9107: @node core-ambcond, core-other, core-idef, The Core Words
9108: @subsection Ambiguous conditions
9109: @c ---------------------------------------------------------------------
9110: @cindex core words, ambiguous conditions
9111: @cindex ambiguous conditions, core words
9112:
9113: @table @i
9114:
9115: @item a name is neither a word nor a number:
9116: @cindex name not found
9117: @cindex undefined word
9118: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
9119: preserves the data and FP stack, so you don't lose more work than
9120: necessary.
9121:
9122: @item a definition name exceeds the maximum length allowed:
9123: @cindex word name too long
9124: @code{-19 throw} (Word name too long)
9125:
9126: @item addressing a region not inside the various data spaces of the forth system:
9127: @cindex Invalid memory address
9128: The stacks, code space and name space are accessible. Machine code space is
9129: typically readable. Accessing other addresses gives results dependent on
9130: the operating system. On decent systems: @code{-9 throw} (Invalid memory
9131: address).
9132:
9133: @item argument type incompatible with parameter:
9134: @cindex argument type mismatch
9135: This is usually not caught. Some words perform checks, e.g., the control
9136: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
9137: mismatch).
9138:
9139: @item attempting to obtain the execution token of a word with undefined execution semantics:
9140: @cindex Interpreting a compile-only word, for @code{'} etc.
9141: @cindex execution token of words with undefined execution semantics
9142: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
9143: get an execution token for @code{compile-only-error} (which performs a
9144: @code{-14 throw} when executed).
9145:
9146: @item dividing by zero:
9147: @cindex dividing by zero
9148: @cindex floating point unidentified fault, integer division
9149: On better platforms, this produces a @code{-10 throw} (Division by
9150: zero); on other systems, this typically results in a @code{-55 throw}
9151: (Floating-point unidentified fault).
9152:
9153: @item insufficient data stack or return stack space:
9154: @cindex insufficient data stack or return stack space
9155: @cindex stack overflow
9156: @cindex address alignment exception, stack overflow
9157: @cindex Invalid memory address, stack overflow
9158: Depending on the operating system, the installation, and the invocation
9159: of Gforth, this is either checked by the memory management hardware, or
9160: it is not checked. If it is checked, you typically get a @code{-3 throw}
9161: (Stack overflow), @code{-5 throw} (Return stack overflow), or @code{-9
9162: throw} (Invalid memory address) (depending on the platform and how you
9163: achieved the overflow) as soon as the overflow happens. If it is not
9164: checked, overflows typically result in mysterious illegal memory
9165: accesses, producing @code{-9 throw} (Invalid memory address) or
9166: @code{-23 throw} (Address alignment exception); they might also destroy
9167: the internal data structure of @code{ALLOCATE} and friends, resulting in
9168: various errors in these words.
9169:
9170: @item insufficient space for loop control parameters:
9171: @cindex insufficient space for loop control parameters
9172: like other return stack overflows.
9173:
9174: @item insufficient space in the dictionary:
9175: @cindex insufficient space in the dictionary
9176: @cindex dictionary overflow
9177: If you try to allot (either directly with @code{allot}, or indirectly
9178: with @code{,}, @code{create} etc.) more memory than available in the
9179: dictionary, you get a @code{-8 throw} (Dictionary overflow). If you try
9180: to access memory beyond the end of the dictionary, the results are
9181: similar to stack overflows.
9182:
9183: @item interpreting a word with undefined interpretation semantics:
9184: @cindex interpreting a word with undefined interpretation semantics
9185: @cindex Interpreting a compile-only word
9186: For some words, we have defined interpretation semantics. For the
9187: others: @code{-14 throw} (Interpreting a compile-only word).
9188:
9189: @item modifying the contents of the input buffer or a string literal:
9190: @cindex modifying the contents of the input buffer or a string literal
9191: These are located in writable memory and can be modified.
9192:
9193: @item overflow of the pictured numeric output string:
9194: @cindex overflow of the pictured numeric output string
9195: @cindex pictured numeric output string, overflow
9196: @code{-17 throw} (Pictured numeric ouput string overflow).
9197:
9198: @item parsed string overflow:
9199: @cindex parsed string overflow
9200: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
9201:
9202: @item producing a result out of range:
9203: @cindex result out of range
9204: On two's complement machines, arithmetic is performed modulo
9205: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
9206: arithmetic (with appropriate mapping for signed types). Division by zero
9207: typically results in a @code{-10 throw} (divide by zero) or @code{-55
9208: throw} (floating point unidentified fault). @code{convert} and
9209: @code{>number} currently overflow silently.
9210:
9211: @item reading from an empty data or return stack:
9212: @cindex stack empty
9213: @cindex stack underflow
9214: @cindex return stack underflow
9215: The data stack is checked by the outer (aka text) interpreter after
9216: every word executed. If it has underflowed, a @code{-4 throw} (Stack
9217: underflow) is performed. Apart from that, stacks may be checked or not,
9218: depending on operating system, installation, and invocation. If they are
9219: caught by a check, they typically result in @code{-4 throw} (Stack
9220: underflow), @code{-6 throw} (Return stack underflow) or @code{-9 throw}
9221: (Invalid memory address), depending on the platform and which stack
9222: underflows and by how much. Note that even if the system uses checking
9223: (through the MMU), your program may have to underflow by a significant
9224: number of stack items to trigger the reaction (the reason for this is
9225: that the MMU, and therefore the checking, works with a page-size
9226: granularity). If there is no checking, the symptoms resulting from an
9227: underflow are similar to those from an overflow. Unbalanced return
9228: stack errors result in a variaty of symptoms, including @code{-9 throw}
9229: (Invalid memory address) and Illegal Instruction (typically @code{-260
9230: throw}).
9231:
9232: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
9233: @cindex unexpected end of the input buffer
9234: @cindex zero-length string as a name
9235: @cindex Attempt to use zero-length string as a name
9236: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
9237: use zero-length string as a name). Words like @code{'} probably will not
9238: find what they search. Note that it is possible to create zero-length
9239: names with @code{nextname} (should it not?).
9240:
9241: @item @code{>IN} greater than input buffer:
9242: @cindex @code{>IN} greater than input buffer
9243: The next invocation of a parsing word returns a string with length 0.
9244:
9245: @item @code{RECURSE} appears after @code{DOES>}:
9246: @cindex @code{RECURSE} appears after @code{DOES>}
9247: Compiles a recursive call to the defining word, not to the defined word.
9248:
9249: @item argument input source different than current input source for @code{RESTORE-INPUT}:
9250: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
9251: @cindex argument type mismatch, @code{RESTORE-INPUT}
9252: @cindex @code{RESTORE-INPUT}, Argument type mismatch
9253: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
9254: the end of the file was reached), its source-id may be
9255: reused. Therefore, restoring an input source specification referencing a
9256: closed file may lead to unpredictable results instead of a @code{-12
9257: THROW}.
9258:
9259: In the future, Gforth may be able to restore input source specifications
9260: from other than the current input source.
9261:
9262: @item data space containing definitions gets de-allocated:
9263: @cindex data space containing definitions gets de-allocated
9264: Deallocation with @code{allot} is not checked. This typically results in
9265: memory access faults or execution of illegal instructions.
9266:
9267: @item data space read/write with incorrect alignment:
9268: @cindex data space read/write with incorrect alignment
9269: @cindex alignment faults
9270: @cindex address alignment exception
9271: Processor-dependent. Typically results in a @code{-23 throw} (Address
9272: alignment exception). Under Linux-Intel on a 486 or later processor with
9273: alignment turned on, incorrect alignment results in a @code{-9 throw}
9274: (Invalid memory address). There are reportedly some processors with
9275: alignment restrictions that do not report violations.
9276:
9277: @item data space pointer not properly aligned, @code{,}, @code{C,}:
9278: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
9279: Like other alignment errors.
9280:
9281: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
9282: Like other stack underflows.
9283:
9284: @item loop control parameters not available:
9285: @cindex loop control parameters not available
9286: Not checked. The counted loop words simply assume that the top of return
9287: stack items are loop control parameters and behave accordingly.
9288:
9289: @item most recent definition does not have a name (@code{IMMEDIATE}):
9290: @cindex most recent definition does not have a name (@code{IMMEDIATE})
9291: @cindex last word was headerless
9292: @code{abort" last word was headerless"}.
9293:
9294: @item name not defined by @code{VALUE} used by @code{TO}:
9295: @cindex name not defined by @code{VALUE} used by @code{TO}
9296: @cindex @code{TO} on non-@code{VALUE}s
9297: @cindex Invalid name argument, @code{TO}
9298: @code{-32 throw} (Invalid name argument) (unless name is a local or was
9299: defined by @code{CONSTANT}; in the latter case it just changes the constant).
9300:
9301: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
9302: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
9303: @cindex undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
9304: @code{-13 throw} (Undefined word)
9305:
9306: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
9307: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
9308: Gforth behaves as if they were of the same type. I.e., you can predict
9309: the behaviour by interpreting all parameters as, e.g., signed.
9310:
9311: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
9312: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
9313: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
9314: compilation semantics of @code{TO}.
9315:
9316: @item String longer than a counted string returned by @code{WORD}:
9317: @cindex string longer than a counted string returned by @code{WORD}
9318: @cindex @code{WORD}, string overflow
9319: Not checked. The string will be ok, but the count will, of course,
9320: contain only the least significant bits of the length.
9321:
9322: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
9323: @cindex @code{LSHIFT}, large shift counts
9324: @cindex @code{RSHIFT}, large shift counts
9325: Processor-dependent. Typical behaviours are returning 0 and using only
9326: the low bits of the shift count.
9327:
9328: @item word not defined via @code{CREATE}:
9329: @cindex @code{>BODY} of non-@code{CREATE}d words
9330: @code{>BODY} produces the PFA of the word no matter how it was defined.
9331:
9332: @cindex @code{DOES>} of non-@code{CREATE}d words
9333: @code{DOES>} changes the execution semantics of the last defined word no
9334: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
9335: @code{CREATE , DOES>}.
9336:
9337: @item words improperly used outside @code{<#} and @code{#>}:
9338: Not checked. As usual, you can expect memory faults.
9339:
9340: @end table
9341:
9342:
9343: @c ---------------------------------------------------------------------
9344: @node core-other, , core-ambcond, The Core Words
9345: @subsection Other system documentation
9346: @c ---------------------------------------------------------------------
9347: @cindex other system documentation, core words
9348: @cindex core words, other system documentation
9349:
9350: @table @i
9351: @item nonstandard words using @code{PAD}:
9352: @cindex @code{PAD} use by nonstandard words
9353: None.
9354:
9355: @item operator's terminal facilities available:
9356: @cindex operator's terminal facilities available
9357: After processing the command line, Gforth goes into interactive mode,
9358: and you can give commands to Gforth interactively. The actual facilities
9359: available depend on how you invoke Gforth.
9360:
9361: @item program data space available:
9362: @cindex program data space available
9363: @cindex data space available
9364: @code{UNUSED .} gives the remaining dictionary space. The total
9365: dictionary space can be specified with the @code{-m} switch
9366: (@pxref{Invoking Gforth}) when Gforth starts up.
9367:
9368: @item return stack space available:
9369: @cindex return stack space available
9370: You can compute the total return stack space in cells with
9371: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
9372: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
9373:
9374: @item stack space available:
9375: @cindex stack space available
9376: You can compute the total data stack space in cells with
9377: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
9378: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
9379:
9380: @item system dictionary space required, in address units:
9381: @cindex system dictionary space required, in address units
9382: Type @code{here forthstart - .} after startup. At the time of this
9383: writing, this gives 80080 (bytes) on a 32-bit system.
9384: @end table
9385:
9386:
9387: @c =====================================================================
9388: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
9389: @section The optional Block word set
9390: @c =====================================================================
9391: @cindex system documentation, block words
9392: @cindex block words, system documentation
9393:
9394: @menu
9395: * block-idef:: Implementation Defined Options
9396: * block-ambcond:: Ambiguous Conditions
9397: * block-other:: Other System Documentation
9398: @end menu
9399:
9400:
9401: @c ---------------------------------------------------------------------
9402: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
9403: @subsection Implementation Defined Options
9404: @c ---------------------------------------------------------------------
9405: @cindex implementation-defined options, block words
9406: @cindex block words, implementation-defined options
9407:
9408: @table @i
9409: @item the format for display by @code{LIST}:
9410: @cindex @code{LIST} display format
9411: First the screen number is displayed, then 16 lines of 64 characters,
9412: each line preceded by the line number.
9413:
9414: @item the length of a line affected by @code{\}:
9415: @cindex length of a line affected by @code{\}
9416: @cindex @code{\}, line length in blocks
9417: 64 characters.
9418: @end table
9419:
9420:
9421: @c ---------------------------------------------------------------------
9422: @node block-ambcond, block-other, block-idef, The optional Block word set
9423: @subsection Ambiguous conditions
9424: @c ---------------------------------------------------------------------
9425: @cindex block words, ambiguous conditions
9426: @cindex ambiguous conditions, block words
9427:
9428: @table @i
9429: @item correct block read was not possible:
9430: @cindex block read not possible
9431: Typically results in a @code{throw} of some OS-derived value (between
9432: -512 and -2048). If the blocks file was just not long enough, blanks are
9433: supplied for the missing portion.
9434:
9435: @item I/O exception in block transfer:
9436: @cindex I/O exception in block transfer
9437: @cindex block transfer, I/O exception
9438: Typically results in a @code{throw} of some OS-derived value (between
9439: -512 and -2048).
9440:
9441: @item invalid block number:
9442: @cindex invalid block number
9443: @cindex block number invalid
9444: @code{-35 throw} (Invalid block number)
9445:
9446: @item a program directly alters the contents of @code{BLK}:
9447: @cindex @code{BLK}, altering @code{BLK}
9448: The input stream is switched to that other block, at the same
9449: position. If the storing to @code{BLK} happens when interpreting
9450: non-block input, the system will get quite confused when the block ends.
9451:
9452: @item no current block buffer for @code{UPDATE}:
9453: @cindex @code{UPDATE}, no current block buffer
9454: @code{UPDATE} has no effect.
9455:
9456: @end table
9457:
9458: @c ---------------------------------------------------------------------
9459: @node block-other, , block-ambcond, The optional Block word set
9460: @subsection Other system documentation
9461: @c ---------------------------------------------------------------------
9462: @cindex other system documentation, block words
9463: @cindex block words, other system documentation
9464:
9465: @table @i
9466: @item any restrictions a multiprogramming system places on the use of buffer addresses:
9467: No restrictions (yet).
9468:
9469: @item the number of blocks available for source and data:
9470: depends on your disk space.
9471:
9472: @end table
9473:
9474:
9475: @c =====================================================================
9476: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
9477: @section The optional Double Number word set
9478: @c =====================================================================
9479: @cindex system documentation, double words
9480: @cindex double words, system documentation
9481:
9482: @menu
9483: * double-ambcond:: Ambiguous Conditions
9484: @end menu
9485:
9486:
9487: @c ---------------------------------------------------------------------
9488: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
9489: @subsection Ambiguous conditions
9490: @c ---------------------------------------------------------------------
9491: @cindex double words, ambiguous conditions
9492: @cindex ambiguous conditions, double words
9493:
9494: @table @i
9495: @item @i{d} outside of range of @i{n} in @code{D>S}:
9496: @cindex @code{D>S}, @i{d} out of range of @i{n}
9497: The least significant cell of @i{d} is produced.
9498:
9499: @end table
9500:
9501:
9502: @c =====================================================================
9503: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
9504: @section The optional Exception word set
9505: @c =====================================================================
9506: @cindex system documentation, exception words
9507: @cindex exception words, system documentation
9508:
9509: @menu
9510: * exception-idef:: Implementation Defined Options
9511: @end menu
9512:
9513:
9514: @c ---------------------------------------------------------------------
9515: @node exception-idef, , The optional Exception word set, The optional Exception word set
9516: @subsection Implementation Defined Options
9517: @c ---------------------------------------------------------------------
9518: @cindex implementation-defined options, exception words
9519: @cindex exception words, implementation-defined options
9520:
9521: @table @i
9522: @item @code{THROW}-codes used in the system:
9523: @cindex @code{THROW}-codes used in the system
9524: The codes -256@minus{}-511 are used for reporting signals. The mapping
9525: from OS signal numbers to throw codes is -256@minus{}@i{signal}. The
9526: codes -512@minus{}-2047 are used for OS errors (for file and memory
9527: allocation operations). The mapping from OS error numbers to throw codes
9528: is -512@minus{}@code{errno}. One side effect of this mapping is that
9529: undefined OS errors produce a message with a strange number; e.g.,
9530: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
9531: @end table
9532:
9533: @c =====================================================================
9534: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
9535: @section The optional Facility word set
9536: @c =====================================================================
9537: @cindex system documentation, facility words
9538: @cindex facility words, system documentation
9539:
9540: @menu
9541: * facility-idef:: Implementation Defined Options
9542: * facility-ambcond:: Ambiguous Conditions
9543: @end menu
9544:
9545:
9546: @c ---------------------------------------------------------------------
9547: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
9548: @subsection Implementation Defined Options
9549: @c ---------------------------------------------------------------------
9550: @cindex implementation-defined options, facility words
9551: @cindex facility words, implementation-defined options
9552:
9553: @table @i
9554: @item encoding of keyboard events (@code{EKEY}):
9555: @cindex keyboard events, encoding in @code{EKEY}
9556: @cindex @code{EKEY}, encoding of keyboard events
9557: Not yet implemented.
9558:
9559: @item duration of a system clock tick:
9560: @cindex duration of a system clock tick
9561: @cindex clock tick duration
9562: System dependent. With respect to @code{MS}, the time is specified in
9563: microseconds. How well the OS and the hardware implement this, is
9564: another question.
9565:
9566: @item repeatability to be expected from the execution of @code{MS}:
9567: @cindex repeatability to be expected from the execution of @code{MS}
9568: @cindex @code{MS}, repeatability to be expected
9569: System dependent. On Unix, a lot depends on load. If the system is
9570: lightly loaded, and the delay is short enough that Gforth does not get
9571: swapped out, the performance should be acceptable. Under MS-DOS and
9572: other single-tasking systems, it should be good.
9573:
9574: @end table
9575:
9576:
9577: @c ---------------------------------------------------------------------
9578: @node facility-ambcond, , facility-idef, The optional Facility word set
9579: @subsection Ambiguous conditions
9580: @c ---------------------------------------------------------------------
9581: @cindex facility words, ambiguous conditions
9582: @cindex ambiguous conditions, facility words
9583:
9584: @table @i
9585: @item @code{AT-XY} can't be performed on user output device:
9586: @cindex @code{AT-XY} can't be performed on user output device
9587: Largely terminal dependent. No range checks are done on the arguments.
9588: No errors are reported. You may see some garbage appearing, you may see
9589: simply nothing happen.
9590:
9591: @end table
9592:
9593:
9594: @c =====================================================================
9595: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
9596: @section The optional File-Access word set
9597: @c =====================================================================
9598: @cindex system documentation, file words
9599: @cindex file words, system documentation
9600:
9601: @menu
9602: * file-idef:: Implementation Defined Options
9603: * file-ambcond:: Ambiguous Conditions
9604: @end menu
9605:
9606: @c ---------------------------------------------------------------------
9607: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
9608: @subsection Implementation Defined Options
9609: @c ---------------------------------------------------------------------
9610: @cindex implementation-defined options, file words
9611: @cindex file words, implementation-defined options
9612:
9613: @table @i
9614: @item file access methods used:
9615: @cindex file access methods used
9616: @code{R/O}, @code{R/W} and @code{BIN} work as you would
9617: expect. @code{W/O} translates into the C file opening mode @code{w} (or
9618: @code{wb}): The file is cleared, if it exists, and created, if it does
9619: not (with both @code{open-file} and @code{create-file}). Under Unix
9620: @code{create-file} creates a file with 666 permissions modified by your
9621: umask.
9622:
9623: @item file exceptions:
9624: @cindex file exceptions
9625: The file words do not raise exceptions (except, perhaps, memory access
9626: faults when you pass illegal addresses or file-ids).
9627:
9628: @item file line terminator:
9629: @cindex file line terminator
9630: System-dependent. Gforth uses C's newline character as line
9631: terminator. What the actual character code(s) of this are is
9632: system-dependent.
9633:
9634: @item file name format:
9635: @cindex file name format
9636: System dependent. Gforth just uses the file name format of your OS.
9637:
9638: @item information returned by @code{FILE-STATUS}:
9639: @cindex @code{FILE-STATUS}, returned information
9640: @code{FILE-STATUS} returns the most powerful file access mode allowed
9641: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
9642: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
9643: along with the returned mode.
9644:
9645: @item input file state after an exception when including source:
9646: @cindex exception when including source
9647: All files that are left via the exception are closed.
9648:
9649: @item @i{ior} values and meaning:
9650: @cindex @i{ior} values and meaning
9651: The @i{ior}s returned by the file and memory allocation words are
9652: intended as throw codes. They typically are in the range
9653: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
9654: @i{ior}s is -512@minus{}@i{errno}.
9655:
9656: @item maximum depth of file input nesting:
9657: @cindex maximum depth of file input nesting
9658: @cindex file input nesting, maximum depth
9659: limited by the amount of return stack, locals/TIB stack, and the number
9660: of open files available. This should not give you troubles.
9661:
9662: @item maximum size of input line:
9663: @cindex maximum size of input line
9664: @cindex input line size, maximum
9665: @code{/line}. Currently 255.
9666:
9667: @item methods of mapping block ranges to files:
9668: @cindex mapping block ranges to files
9669: @cindex files containing blocks
9670: @cindex blocks in files
9671: By default, blocks are accessed in the file @file{blocks.fb} in the
9672: current working directory. The file can be switched with @code{USE}.
9673:
9674: @item number of string buffers provided by @code{S"}:
9675: @cindex @code{S"}, number of string buffers
9676: 1
9677:
9678: @item size of string buffer used by @code{S"}:
9679: @cindex @code{S"}, size of string buffer
9680: @code{/line}. currently 255.
9681:
9682: @end table
9683:
9684: @c ---------------------------------------------------------------------
9685: @node file-ambcond, , file-idef, The optional File-Access word set
9686: @subsection Ambiguous conditions
9687: @c ---------------------------------------------------------------------
9688: @cindex file words, ambiguous conditions
9689: @cindex ambiguous conditions, file words
9690:
9691: @table @i
9692: @item attempting to position a file outside its boundaries:
9693: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
9694: @code{REPOSITION-FILE} is performed as usual: Afterwards,
9695: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
9696:
9697: @item attempting to read from file positions not yet written:
9698: @cindex reading from file positions not yet written
9699: End-of-file, i.e., zero characters are read and no error is reported.
9700:
9701: @item @i{file-id} is invalid (@code{INCLUDE-FILE}):
9702: @cindex @code{INCLUDE-FILE}, @i{file-id} is invalid
9703: An appropriate exception may be thrown, but a memory fault or other
9704: problem is more probable.
9705:
9706: @item I/O exception reading or closing @i{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
9707: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @i{file-id}
9708: @cindex @code{INCLUDED}, I/O exception reading or closing @i{file-id}
9709: The @i{ior} produced by the operation, that discovered the problem, is
9710: thrown.
9711:
9712: @item named file cannot be opened (@code{INCLUDED}):
9713: @cindex @code{INCLUDED}, named file cannot be opened
9714: The @i{ior} produced by @code{open-file} is thrown.
9715:
9716: @item requesting an unmapped block number:
9717: @cindex unmapped block numbers
9718: There are no unmapped legal block numbers. On some operating systems,
9719: writing a block with a large number may overflow the file system and
9720: have an error message as consequence.
9721:
9722: @item using @code{source-id} when @code{blk} is non-zero:
9723: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
9724: @code{source-id} performs its function. Typically it will give the id of
9725: the source which loaded the block. (Better ideas?)
9726:
9727: @end table
9728:
9729:
9730: @c =====================================================================
9731: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
9732: @section The optional Floating-Point word set
9733: @c =====================================================================
9734: @cindex system documentation, floating-point words
9735: @cindex floating-point words, system documentation
9736:
9737: @menu
9738: * floating-idef:: Implementation Defined Options
9739: * floating-ambcond:: Ambiguous Conditions
9740: @end menu
9741:
9742:
9743: @c ---------------------------------------------------------------------
9744: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
9745: @subsection Implementation Defined Options
9746: @c ---------------------------------------------------------------------
9747: @cindex implementation-defined options, floating-point words
9748: @cindex floating-point words, implementation-defined options
9749:
9750: @table @i
9751: @item format and range of floating point numbers:
9752: @cindex format and range of floating point numbers
9753: @cindex floating point numbers, format and range
9754: System-dependent; the @code{double} type of C.
9755:
9756: @item results of @code{REPRESENT} when @i{float} is out of range:
9757: @cindex @code{REPRESENT}, results when @i{float} is out of range
9758: System dependent; @code{REPRESENT} is implemented using the C library
9759: function @code{ecvt()} and inherits its behaviour in this respect.
9760:
9761: @item rounding or truncation of floating-point numbers:
9762: @cindex rounding of floating-point numbers
9763: @cindex truncation of floating-point numbers
9764: @cindex floating-point numbers, rounding or truncation
9765: System dependent; the rounding behaviour is inherited from the hosting C
9766: compiler. IEEE-FP-based (i.e., most) systems by default round to
9767: nearest, and break ties by rounding to even (i.e., such that the last
9768: bit of the mantissa is 0).
9769:
9770: @item size of floating-point stack:
9771: @cindex floating-point stack size
9772: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
9773: the floating-point stack (in floats). You can specify this on startup
9774: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
9775:
9776: @item width of floating-point stack:
9777: @cindex floating-point stack width
9778: @code{1 floats}.
9779:
9780: @end table
9781:
9782:
9783: @c ---------------------------------------------------------------------
9784: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
9785: @subsection Ambiguous conditions
9786: @c ---------------------------------------------------------------------
9787: @cindex floating-point words, ambiguous conditions
9788: @cindex ambiguous conditions, floating-point words
9789:
9790: @table @i
9791: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
9792: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
9793: System-dependent. Typically results in a @code{-23 THROW} like other
9794: alignment violations.
9795:
9796: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
9797: @cindex @code{f@@} used with an address that is not float aligned
9798: @cindex @code{f!} used with an address that is not float aligned
9799: System-dependent. Typically results in a @code{-23 THROW} like other
9800: alignment violations.
9801:
9802: @item floating-point result out of range:
9803: @cindex floating-point result out of range
9804: System-dependent. Can result in a @code{-55 THROW} (Floating-point
9805: unidentified fault), or can produce a special value representing, e.g.,
9806: Infinity.
9807:
9808: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
9809: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
9810: System-dependent. Typically results in an alignment fault like other
9811: alignment violations.
9812:
9813: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
9814: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
9815: The floating-point number is converted into decimal nonetheless.
9816:
9817: @item Both arguments are equal to zero (@code{FATAN2}):
9818: @cindex @code{FATAN2}, both arguments are equal to zero
9819: System-dependent. @code{FATAN2} is implemented using the C library
9820: function @code{atan2()}.
9821:
9822: @item Using @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero:
9823: @cindex @code{FTAN} on an argument @i{r1} where cos(@i{r1}) is zero
9824: System-dependent. Anyway, typically the cos of @i{r1} will not be zero
9825: because of small errors and the tan will be a very large (or very small)
9826: but finite number.
9827:
9828: @item @i{d} cannot be presented precisely as a float in @code{D>F}:
9829: @cindex @code{D>F}, @i{d} cannot be presented precisely as a float
9830: The result is rounded to the nearest float.
9831:
9832: @item dividing by zero:
9833: @cindex dividing by zero, floating-point
9834: @cindex floating-point dividing by zero
9835: @cindex floating-point unidentified fault, FP divide-by-zero
9836: @code{-55 throw} (Floating-point unidentified fault)
9837:
9838: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
9839: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
9840: System dependent. On IEEE-FP based systems the number is converted into
9841: an infinity.
9842:
9843: @item @i{float}<1 (@code{FACOSH}):
9844: @cindex @code{FACOSH}, @i{float}<1
9845: @cindex floating-point unidentified fault, @code{FACOSH}
9846: @code{-55 throw} (Floating-point unidentified fault)
9847:
9848: @item @i{float}=<-1 (@code{FLNP1}):
9849: @cindex @code{FLNP1}, @i{float}=<-1
9850: @cindex floating-point unidentified fault, @code{FLNP1}
9851: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
9852: negative infinity is typically produced for @i{float}=-1.
9853:
9854: @item @i{float}=<0 (@code{FLN}, @code{FLOG}):
9855: @cindex @code{FLN}, @i{float}=<0
9856: @cindex @code{FLOG}, @i{float}=<0
9857: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
9858: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
9859: negative infinity is typically produced for @i{float}=0.
9860:
9861: @item @i{float}<0 (@code{FASINH}, @code{FSQRT}):
9862: @cindex @code{FASINH}, @i{float}<0
9863: @cindex @code{FSQRT}, @i{float}<0
9864: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
9865: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
9866: produces values for these inputs on my Linux box (Bug in the C library?)
9867:
9868: @item |@i{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
9869: @cindex @code{FACOS}, |@i{float}|>1
9870: @cindex @code{FASIN}, |@i{float}|>1
9871: @cindex @code{FATANH}, |@i{float}|>1
9872: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
9873: @code{-55 throw} (Floating-point unidentified fault).
9874:
9875: @item integer part of float cannot be represented by @i{d} in @code{F>D}:
9876: @cindex @code{F>D}, integer part of float cannot be represented by @i{d}
9877: @cindex floating-point unidentified fault, @code{F>D}
9878: @code{-55 throw} (Floating-point unidentified fault).
9879:
9880: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
9881: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
9882: This does not happen.
9883: @end table
9884:
9885: @c =====================================================================
9886: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
9887: @section The optional Locals word set
9888: @c =====================================================================
9889: @cindex system documentation, locals words
9890: @cindex locals words, system documentation
9891:
9892: @menu
9893: * locals-idef:: Implementation Defined Options
9894: * locals-ambcond:: Ambiguous Conditions
9895: @end menu
9896:
9897:
9898: @c ---------------------------------------------------------------------
9899: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
9900: @subsection Implementation Defined Options
9901: @c ---------------------------------------------------------------------
9902: @cindex implementation-defined options, locals words
9903: @cindex locals words, implementation-defined options
9904:
9905: @table @i
9906: @item maximum number of locals in a definition:
9907: @cindex maximum number of locals in a definition
9908: @cindex locals, maximum number in a definition
9909: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
9910: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
9911: characters. The number of locals in a definition is bounded by the size
9912: of locals-buffer, which contains the names of the locals.
9913:
9914: @end table
9915:
9916:
9917: @c ---------------------------------------------------------------------
9918: @node locals-ambcond, , locals-idef, The optional Locals word set
9919: @subsection Ambiguous conditions
9920: @c ---------------------------------------------------------------------
9921: @cindex locals words, ambiguous conditions
9922: @cindex ambiguous conditions, locals words
9923:
9924: @table @i
9925: @item executing a named local in interpretation state:
9926: @cindex local in interpretation state
9927: @cindex Interpreting a compile-only word, for a local
9928: Locals have no interpretation semantics. If you try to perform the
9929: interpretation semantics, you will get a @code{-14 throw} somewhere
9930: (Interpreting a compile-only word). If you perform the compilation
9931: semantics, the locals access will be compiled (irrespective of state).
9932:
9933: @item @i{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
9934: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
9935: @cindex @code{TO} on non-@code{VALUE}s and non-locals
9936: @cindex Invalid name argument, @code{TO}
9937: @code{-32 throw} (Invalid name argument)
9938:
9939: @end table
9940:
9941:
9942: @c =====================================================================
9943: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
9944: @section The optional Memory-Allocation word set
9945: @c =====================================================================
9946: @cindex system documentation, memory-allocation words
9947: @cindex memory-allocation words, system documentation
9948:
9949: @menu
9950: * memory-idef:: Implementation Defined Options
9951: @end menu
9952:
9953:
9954: @c ---------------------------------------------------------------------
9955: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
9956: @subsection Implementation Defined Options
9957: @c ---------------------------------------------------------------------
9958: @cindex implementation-defined options, memory-allocation words
9959: @cindex memory-allocation words, implementation-defined options
9960:
9961: @table @i
9962: @item values and meaning of @i{ior}:
9963: @cindex @i{ior} values and meaning
9964: The @i{ior}s returned by the file and memory allocation words are
9965: intended as throw codes. They typically are in the range
9966: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
9967: @i{ior}s is -512@minus{}@i{errno}.
9968:
9969: @end table
9970:
9971: @c =====================================================================
9972: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
9973: @section The optional Programming-Tools word set
9974: @c =====================================================================
9975: @cindex system documentation, programming-tools words
9976: @cindex programming-tools words, system documentation
9977:
9978: @menu
9979: * programming-idef:: Implementation Defined Options
9980: * programming-ambcond:: Ambiguous Conditions
9981: @end menu
9982:
9983:
9984: @c ---------------------------------------------------------------------
9985: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
9986: @subsection Implementation Defined Options
9987: @c ---------------------------------------------------------------------
9988: @cindex implementation-defined options, programming-tools words
9989: @cindex programming-tools words, implementation-defined options
9990:
9991: @table @i
9992: @item ending sequence for input following @code{;CODE} and @code{CODE}:
9993: @cindex @code{;CODE} ending sequence
9994: @cindex @code{CODE} ending sequence
9995: @code{END-CODE}
9996:
9997: @item manner of processing input following @code{;CODE} and @code{CODE}:
9998: @cindex @code{;CODE}, processing input
9999: @cindex @code{CODE}, processing input
10000: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
10001: the input is processed by the text interpreter, (starting) in interpret
10002: state.
10003:
10004: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
10005: @cindex @code{ASSEMBLER}, search order capability
10006: The ANS Forth search order word set.
10007:
10008: @item source and format of display by @code{SEE}:
10009: @cindex @code{SEE}, source and format of output
10010: The source for @code{see} is the intermediate code used by the inner
10011: interpreter. The current @code{see} tries to output Forth source code
10012: as well as possible.
10013:
10014: @end table
10015:
10016: @c ---------------------------------------------------------------------
10017: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
10018: @subsection Ambiguous conditions
10019: @c ---------------------------------------------------------------------
10020: @cindex programming-tools words, ambiguous conditions
10021: @cindex ambiguous conditions, programming-tools words
10022:
10023: @table @i
10024:
10025: @item deleting the compilation word list (@code{FORGET}):
10026: @cindex @code{FORGET}, deleting the compilation word list
10027: Not implemented (yet).
10028:
10029: @item fewer than @i{u}+1 items on the control-flow stack (@code{CS-PICK}, @code{CS-ROLL}):
10030: @cindex @code{CS-PICK}, fewer than @i{u}+1 items on the control flow-stack
10031: @cindex @code{CS-ROLL}, fewer than @i{u}+1 items on the control flow-stack
10032: @cindex control-flow stack underflow
10033: This typically results in an @code{abort"} with a descriptive error
10034: message (may change into a @code{-22 throw} (Control structure mismatch)
10035: in the future). You may also get a memory access error. If you are
10036: unlucky, this ambiguous condition is not caught.
10037:
10038: @item @i{name} can't be found (@code{FORGET}):
10039: @cindex @code{FORGET}, @i{name} can't be found
10040: Not implemented (yet).
10041:
10042: @item @i{name} not defined via @code{CREATE}:
10043: @cindex @code{;CODE}, @i{name} not defined via @code{CREATE}
10044: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
10045: the execution semantics of the last defined word no matter how it was
10046: defined.
10047:
10048: @item @code{POSTPONE} applied to @code{[IF]}:
10049: @cindex @code{POSTPONE} applied to @code{[IF]}
10050: @cindex @code{[IF]} and @code{POSTPONE}
10051: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
10052: equivalent to @code{[IF]}.
10053:
10054: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
10055: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
10056: Continue in the same state of conditional compilation in the next outer
10057: input source. Currently there is no warning to the user about this.
10058:
10059: @item removing a needed definition (@code{FORGET}):
10060: @cindex @code{FORGET}, removing a needed definition
10061: Not implemented (yet).
10062:
10063: @end table
10064:
10065:
10066: @c =====================================================================
10067: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
10068: @section The optional Search-Order word set
10069: @c =====================================================================
10070: @cindex system documentation, search-order words
10071: @cindex search-order words, system documentation
10072:
10073: @menu
10074: * search-idef:: Implementation Defined Options
10075: * search-ambcond:: Ambiguous Conditions
10076: @end menu
10077:
10078:
10079: @c ---------------------------------------------------------------------
10080: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
10081: @subsection Implementation Defined Options
10082: @c ---------------------------------------------------------------------
10083: @cindex implementation-defined options, search-order words
10084: @cindex search-order words, implementation-defined options
10085:
10086: @table @i
10087: @item maximum number of word lists in search order:
10088: @cindex maximum number of word lists in search order
10089: @cindex search order, maximum depth
10090: @code{s" wordlists" environment? drop .}. Currently 16.
10091:
10092: @item minimum search order:
10093: @cindex minimum search order
10094: @cindex search order, minimum
10095: @code{root root}.
10096:
10097: @end table
10098:
10099: @c ---------------------------------------------------------------------
10100: @node search-ambcond, , search-idef, The optional Search-Order word set
10101: @subsection Ambiguous conditions
10102: @c ---------------------------------------------------------------------
10103: @cindex search-order words, ambiguous conditions
10104: @cindex ambiguous conditions, search-order words
10105:
10106: @table @i
10107: @item changing the compilation word list (during compilation):
10108: @cindex changing the compilation word list (during compilation)
10109: @cindex compilation word list, change before definition ends
10110: The word is entered into the word list that was the compilation word list
10111: at the start of the definition. Any changes to the name field (e.g.,
10112: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
10113: are applied to the latest defined word (as reported by @code{last} or
10114: @code{lastxt}), if possible, irrespective of the compilation word list.
10115:
10116: @item search order empty (@code{previous}):
10117: @cindex @code{previous}, search order empty
10118: @cindex vocstack empty, @code{previous}
10119: @code{abort" Vocstack empty"}.
10120:
10121: @item too many word lists in search order (@code{also}):
10122: @cindex @code{also}, too many word lists in search order
10123: @cindex vocstack full, @code{also}
10124: @code{abort" Vocstack full"}.
10125:
10126: @end table
10127:
10128: @c ***************************************************************
10129: @node Model, Integrating Gforth, ANS conformance, Top
10130: @chapter Model
10131:
10132: This chapter has yet to be written. It will contain information, on
10133: which internal structures you can rely.
10134:
10135: @c ***************************************************************
10136: @node Integrating Gforth, Emacs and Gforth, Model, Top
10137: @chapter Integrating Gforth into C programs
10138:
10139: This is not yet implemented.
10140:
10141: Several people like to use Forth as scripting language for applications
10142: that are otherwise written in C, C++, or some other language.
10143:
10144: The Forth system ATLAST provides facilities for embedding it into
10145: applications; unfortunately it has several disadvantages: most
10146: importantly, it is not based on ANS Forth, and it is apparently dead
10147: (i.e., not developed further and not supported). The facilities
10148: provided by Gforth in this area are inspired by ATLAST's facilities, so
10149: making the switch should not be hard.
10150:
10151: We also tried to design the interface such that it can easily be
10152: implemented by other Forth systems, so that we may one day arrive at a
10153: standardized interface. Such a standard interface would allow you to
10154: replace the Forth system without having to rewrite C code.
10155:
10156: You embed the Gforth interpreter by linking with the library
10157: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
10158: global symbols in this library that belong to the interface, have the
10159: prefix @code{forth_}. (Global symbols that are used internally have the
10160: prefix @code{gforth_}).
10161:
10162: You can include the declarations of Forth types and the functions and
10163: variables of the interface with @code{#include <forth.h>}.
10164:
10165: Types.
10166:
10167: Variables.
10168:
10169: Data and FP Stack pointer. Area sizes.
10170:
10171: functions.
10172:
10173: forth_init(imagefile)
10174: forth_evaluate(string) exceptions?
10175: forth_goto(address) (or forth_execute(xt)?)
10176: forth_continue() (a corountining mechanism)
10177:
10178: Adding primitives.
10179:
10180: No checking.
10181:
10182: Signals?
10183:
10184: Accessing the Stacks
10185:
10186: @c ******************************************************************
10187: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
10188: @chapter Emacs and Gforth
10189: @cindex Emacs and Gforth
10190:
10191: @cindex @file{gforth.el}
10192: @cindex @file{forth.el}
10193: @cindex Rydqvist, Goran
10194: @cindex comment editing commands
10195: @cindex @code{\}, editing with Emacs
10196: @cindex debug tracer editing commands
10197: @cindex @code{~~}, removal with Emacs
10198: @cindex Forth mode in Emacs
10199: Gforth comes with @file{gforth.el}, an improved version of
10200: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
10201: improvements are:
10202:
10203: @itemize @bullet
10204: @item
10205: A better (but still not perfect) handling of indentation.
10206: @item
10207: Comment paragraph filling (@kbd{M-q})
10208: @item
10209: Commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) of regions
10210: @item
10211: Removal of debugging tracers (@kbd{C-x ~}, @pxref{Debugging}).
10212: @end itemize
10213:
10214: I left the stuff I do not use alone, even though some of it only makes
10215: sense for TILE. To get a description of these features, enter Forth mode
10216: and type @kbd{C-h m}.
10217:
10218: @cindex source location of error or debugging output in Emacs
10219: @cindex error output, finding the source location in Emacs
10220: @cindex debugging output, finding the source location in Emacs
10221: In addition, Gforth supports Emacs quite well: The source code locations
10222: given in error messages, debugging output (from @code{~~}) and failed
10223: assertion messages are in the right format for Emacs' compilation mode
10224: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
10225: Manual}) so the source location corresponding to an error or other
10226: message is only a few keystrokes away (@kbd{C-x `} for the next error,
10227: @kbd{C-c C-c} for the error under the cursor).
10228:
10229: @cindex @file{TAGS} file
10230: @cindex @file{etags.fs}
10231: @cindex viewing the source of a word in Emacs
10232: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file will
10233: be produced (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) that
10234: contains the definitions of all words defined afterwards. You can then
10235: find the source for a word using @kbd{M-.}. Note that emacs can use
10236: several tags files at the same time (e.g., one for the Gforth sources
10237: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
10238: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
10239: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
10240: @file{/usr/local/share/gforth/0.2.0/TAGS}).
10241:
10242: @cindex @file{.emacs}
10243: To get all these benefits, add the following lines to your @file{.emacs}
10244: file:
10245:
10246: @example
10247: (autoload 'forth-mode "gforth.el")
10248: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
10249: @end example
10250:
10251: @c ******************************************************************
10252: @node Image Files, Engine, Emacs and Gforth, Top
10253: @chapter Image Files
10254: @cindex image file
10255: @cindex @file{.fi} files
10256: @cindex precompiled Forth code
10257: @cindex dictionary in persistent form
10258: @cindex persistent form of dictionary
10259:
10260: An image file is a file containing an image of the Forth dictionary,
10261: i.e., compiled Forth code and data residing in the dictionary. By
10262: convention, we use the extension @code{.fi} for image files.
10263:
10264: @menu
10265: * Image Licensing Issues:: Distribution terms for images.
10266: * Image File Background:: Why have image files?
10267: * Non-Relocatable Image Files:: don't always work.
10268: * Data-Relocatable Image Files:: are better.
10269: * Fully Relocatable Image Files:: better yet.
10270: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
10271: * Running Image Files:: @code{gforth -i @i{file}} or @i{file}.
10272: * Modifying the Startup Sequence:: and turnkey applications.
10273: @end menu
10274:
10275: @node Image Licensing Issues, Image File Background, Image Files, Image Files
10276: @section Image Licensing Issues
10277: @cindex license for images
10278: @cindex image license
10279:
10280: An image created with @code{gforthmi} (@pxref{gforthmi}) or
10281: @code{savesystem} (@pxref{Non-Relocatable Image Files}) includes the
10282: original image; i.e., according to copyright law it is a derived work of
10283: the original image.
10284:
10285: Since Gforth is distributed under the GNU GPL, the newly created image
10286: falls under the GNU GPL, too. In particular, this means that if you
10287: distribute the image, you have to make all of the sources for the image
10288: available, including those you wrote. For details see @ref{License, ,
10289: GNU General Public License (Section 3)}.
10290:
10291: If you create an image with @code{cross} (@pxref{cross.fs}), the image
10292: contains only code compiled from the sources you gave it; if none of
10293: these sources is under the GPL, the terms discussed above do not apply
10294: to the image. However, if your image needs an engine (a gforth binary)
10295: that is under the GPL, you should make sure that you distribute both in
10296: a way that is at most a @emph{mere aggregation}, if you don't want the
10297: terms of the GPL to apply to the image.
10298:
10299: @node Image File Background, Non-Relocatable Image Files, Image Licensing Issues, Image Files
10300: @section Image File Background
10301: @cindex image file background
10302:
10303: Our Forth system consists not only of primitives, but also of
10304: definitions written in Forth. Since the Forth compiler itself belongs to
10305: those definitions, it is not possible to start the system with the
10306: primitives and the Forth source alone. Therefore we provide the Forth
10307: code as an image file in nearly executable form. When Gforth starts up,
10308: a C routine loads the image file into memory, optionally relocates the
10309: addresses, then sets up the memory (stacks etc.) according to
10310: information in the image file, and (finally) starts executing Forth
10311: code.
10312:
10313: The image file variants represent different compromises between the
10314: goals of making it easy to generate image files and making them
10315: portable.
10316:
10317: @cindex relocation at run-time
10318: Win32Forth 3.4 and Mitch Bradley's @code{cforth} use relocation at
10319: run-time. This avoids many of the complications discussed below (image
10320: files are data relocatable without further ado), but costs performance
10321: (one addition per memory access).
10322:
10323: @cindex relocation at load-time
10324: By contrast, the Gforth loader performs relocation at image load time. The
10325: loader also has to replace tokens that represent primitive calls with the
10326: appropriate code-field addresses (or code addresses in the case of
10327: direct threading).
10328:
10329: There are three kinds of image files, with different degrees of
10330: relocatability: non-relocatable, data-relocatable, and fully relocatable
10331: image files.
10332:
10333: @cindex image file loader
10334: @cindex relocating loader
10335: @cindex loader for image files
10336: These image file variants have several restrictions in common; they are
10337: caused by the design of the image file loader:
10338:
10339: @itemize @bullet
10340: @item
10341: There is only one segment; in particular, this means, that an image file
10342: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
10343: them). The contents of the stacks are not represented, either.
10344:
10345: @item
10346: The only kinds of relocation supported are: adding the same offset to
10347: all cells that represent data addresses; and replacing special tokens
10348: with code addresses or with pieces of machine code.
10349:
10350: If any complex computations involving addresses are performed, the
10351: results cannot be represented in the image file. Several applications that
10352: use such computations come to mind:
10353: @itemize @minus
10354: @item
10355: Hashing addresses (or data structures which contain addresses) for table
10356: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
10357: purpose, you will have no problem, because the hash tables are
10358: recomputed automatically when the system is started. If you use your own
10359: hash tables, you will have to do something similar.
10360:
10361: @item
10362: There's a cute implementation of doubly-linked lists that uses
10363: @code{XOR}ed addresses. You could represent such lists as singly-linked
10364: in the image file, and restore the doubly-linked representation on
10365: startup.@footnote{In my opinion, though, you should think thrice before
10366: using a doubly-linked list (whatever implementation).}
10367:
10368: @item
10369: The code addresses of run-time routines like @code{docol:} cannot be
10370: represented in the image file (because their tokens would be replaced by
10371: machine code in direct threaded implementations). As a workaround,
10372: compute these addresses at run-time with @code{>code-address} from the
10373: executions tokens of appropriate words (see the definitions of
10374: @code{docol:} and friends in @file{kernel.fs}).
10375:
10376: @item
10377: On many architectures addresses are represented in machine code in some
10378: shifted or mangled form. You cannot put @code{CODE} words that contain
10379: absolute addresses in this form in a relocatable image file. Workarounds
10380: are representing the address in some relative form (e.g., relative to
10381: the CFA, which is present in some register), or loading the address from
10382: a place where it is stored in a non-mangled form.
10383: @end itemize
10384: @end itemize
10385:
10386: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
10387: @section Non-Relocatable Image Files
10388: @cindex non-relocatable image files
10389: @cindex image file, non-relocatable
10390:
10391: These files are simple memory dumps of the dictionary. They are specific
10392: to the executable (i.e., @file{gforth} file) they were created
10393: with. What's worse, they are specific to the place on which the
10394: dictionary resided when the image was created. Now, there is no
10395: guarantee that the dictionary will reside at the same place the next
10396: time you start Gforth, so there's no guarantee that a non-relocatable
10397: image will work the next time (Gforth will complain instead of crashing,
10398: though).
10399:
10400: You can create a non-relocatable image file with
10401:
10402: doc-savesystem
10403:
10404: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
10405: @section Data-Relocatable Image Files
10406: @cindex data-relocatable image files
10407: @cindex image file, data-relocatable
10408:
10409: These files contain relocatable data addresses, but fixed code addresses
10410: (instead of tokens). They are specific to the executable (i.e.,
10411: @file{gforth} file) they were created with. For direct threading on some
10412: architectures (e.g., the i386), data-relocatable images do not work. You
10413: get a data-relocatable image, if you use @file{gforthmi} with a
10414: Gforth binary that is not doubly indirect threaded (@pxref{Fully
10415: Relocatable Image Files}).
10416:
10417: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
10418: @section Fully Relocatable Image Files
10419: @cindex fully relocatable image files
10420: @cindex image file, fully relocatable
10421:
10422: @cindex @file{kern*.fi}, relocatability
10423: @cindex @file{gforth.fi}, relocatability
10424: These image files have relocatable data addresses, and tokens for code
10425: addresses. They can be used with different binaries (e.g., with and
10426: without debugging) on the same machine, and even across machines with
10427: the same data formats (byte order, cell size, floating point
10428: format). However, they are usually specific to the version of Gforth
10429: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
10430: are fully relocatable.
10431:
10432: There are two ways to create a fully relocatable image file:
10433:
10434: @menu
10435: * gforthmi:: The normal way
10436: * cross.fs:: The hard way
10437: @end menu
10438:
10439: @node gforthmi, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
10440: @subsection @file{gforthmi}
10441: @cindex @file{comp-i.fs}
10442: @cindex @file{gforthmi}
10443:
10444: You will usually use @file{gforthmi}. If you want to create an
10445: image @i{file} that contains everything you would load by invoking
10446: Gforth with @code{gforth @i{options}}, you simply say:
10447: @example
10448: gforthmi @i{file} @i{options}
10449: @end example
10450:
10451: E.g., if you want to create an image @file{asm.fi} that has the file
10452: @file{asm.fs} loaded in addition to the usual stuff, you could do it
10453: like this:
10454:
10455: @example
10456: gforthmi asm.fi asm.fs
10457: @end example
10458:
10459: @file{gforthmi} is implemented as a sh script and works like this: It
10460: produces two non-relocatable images for different addresses and then
10461: compares them. Its output reflects this: first you see the output (if
10462: any) of the two Gforth invocations that produce the nonrelocatable image
10463: files, then you see the output of the comparing program: It displays the
10464: offset used for data addresses and the offset used for code addresses;
10465: moreover, for each cell that cannot be represented correctly in the
10466: image files, it displays a line like the following one:
10467:
10468: @example
10469: 78DC BFFFFA50 BFFFFA40
10470: @end example
10471:
10472: This means that at offset $78dc from @code{forthstart}, one input image
10473: contains $bffffa50, and the other contains $bffffa40. Since these cells
10474: cannot be represented correctly in the output image, you should examine
10475: these places in the dictionary and verify that these cells are dead
10476: (i.e., not read before they are written).
10477:
10478: If you type @file{gforthmi} with no arguments, it prints some usage
10479: instructions.
10480:
10481: @cindex @code{savesystem} during @file{gforthmi}
10482: @cindex @code{bye} during @file{gforthmi}
10483: @cindex doubly indirect threaded code
10484: @cindex environment variable @code{GFORTHD}
10485: @cindex @code{GFORTHD} environment variable
10486: @cindex @code{gforth-ditc}
10487: There are a few wrinkles: After processing the passed @i{options}, the
10488: words @code{savesystem} and @code{bye} must be visible. A special doubly
10489: indirect threaded version of the @file{gforth} executable is used for
10490: creating the nonrelocatable images; you can pass the exact filename of
10491: this executable through the environment variable @code{GFORTHD}
10492: (default: @file{gforth-ditc}); if you pass a version that is not doubly
10493: indirect threaded, you will not get a fully relocatable image, but a
10494: data-relocatable image (because there is no code address offset). The
10495: normal @file{gforth} executable is used for creating the relocatable
10496: image; you can pass the exact filename of this executable through the
10497: environment variable @code{GFORTH}.
10498:
10499: @node cross.fs, , gforthmi, Fully Relocatable Image Files
10500: @subsection @file{cross.fs}
10501: @cindex @file{cross.fs}
10502: @cindex cross-compiler
10503: @cindex metacompiler
10504:
10505: You can also use @code{cross}, a batch compiler that accepts a Forth-like
10506: programming language. This @code{cross} language has to be documented
10507: yet.
10508:
10509: @cindex target compiler
10510: @code{cross} also allows you to create image files for machines with
10511: different data sizes and data formats than the one used for generating
10512: the image file. You can also use it to create an application image that
10513: does not contain a Forth compiler. These features are bought with
10514: restrictions and inconveniences in programming. E.g., addresses have to
10515: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
10516: order to make the code relocatable.
10517:
10518:
10519: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
10520: @section Stack and Dictionary Sizes
10521: @cindex image file, stack and dictionary sizes
10522: @cindex dictionary size default
10523: @cindex stack size default
10524:
10525: If you invoke Gforth with a command line flag for the size
10526: (@pxref{Invoking Gforth}), the size you specify is stored in the
10527: dictionary. If you save the dictionary with @code{savesystem} or create
10528: an image with @file{gforthmi}, this size will become the default
10529: for the resulting image file. E.g., the following will create a
10530: fully relocatable version of @file{gforth.fi} with a 1MB dictionary:
10531:
10532: @example
10533: gforthmi gforth.fi -m 1M
10534: @end example
10535:
10536: In other words, if you want to set the default size for the dictionary
10537: and the stacks of an image, just invoke @file{gforthmi} with the
10538: appropriate options when creating the image.
10539:
10540: @cindex stack size, cache-friendly
10541: Note: For cache-friendly behaviour (i.e., good performance), you should
10542: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
10543: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
10544: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
10545:
10546: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
10547: @section Running Image Files
10548: @cindex running image files
10549: @cindex invoking image files
10550: @cindex image file invocation
10551:
10552: @cindex -i, invoke image file
10553: @cindex --image file, invoke image file
10554: You can invoke Gforth with an image file @i{image} instead of the
10555: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
10556: @example
10557: gforth -i @i{image}
10558: @end example
10559:
10560: @cindex executable image file
10561: @cindex image file, executable
10562: If your operating system supports starting scripts with a line of the
10563: form @code{#! ...}, you just have to type the image file name to start
10564: Gforth with this image file (note that the file extension @code{.fi} is
10565: just a convention). I.e., to run Gforth with the image file @i{image},
10566: you can just type @i{image} instead of @code{gforth -i @i{image}}.
10567: This works because every @code{.fi} file starts with a line of this
10568: format:
10569:
10570: @example
10571: #! /usr/local/bin/gforth-0.4.0 -i
10572: @end example
10573:
10574: The file and pathname for the Gforth engine specified on this line is
10575: the specific Gforth executable that it was built against; i.e. the value
10576: of the environment variable @code{GFORTH} at the time that
10577: @file{gforthmi} was executed.
10578:
10579: You can make use of the same shell capability to make a Forth source
10580: file into an executable. For example, if you place this text in a file:
10581:
10582: @example
10583: #! /usr/local/bin/gforth
10584:
10585: ." Hello, world" CR
10586: bye
10587: @end example
10588:
10589: @noindent
10590: and then make the file executable (chmod +x in Unix), you can run it
10591: directly from the command line. The sequence @code{#!} is used in two
10592: ways; firstly, it is recognised as a ``magic sequence'' by the operating
10593: system@footnote{The Unix kernel actually recognises two types of files:
10594: executable files and files of data, where the data is processed by an
10595: interpreter that is specified on the ``interpreter line'' -- the first
10596: line of the file, starting with the sequence #!. There may be a small
10597: limit (e.g., 32) on the number of characters that may be specified on
10598: the interpreter line.} secondly it is treated as a comment character by
10599: Gforth. Because of the second usage, a space is required between
10600: @code{#!} and the path to the executable.
10601:
10602: The disadvantage of this latter technique, compared with using
10603: @file{gforthmi}, is that it is slower; the Forth source code is compiled
10604: on-the-fly, each time the program is invoked.
10605:
10606: @comment TODO describe the #! magic with reference to the Power Tools book.
10607:
10608: doc-#!
10609:
10610: @node Modifying the Startup Sequence, , Running Image Files, Image Files
10611: @section Modifying the Startup Sequence
10612: @cindex startup sequence for image file
10613: @cindex image file initialization sequence
10614: @cindex initialization sequence of image file
10615:
10616: You can add your own initialization to the startup sequence through the
10617: deferred word @code{'cold}. @code{'cold} is invoked just before the
10618: image-specific command line processing (by default, loading files and
10619: evaluating (@code{-e}) strings) starts.
10620:
10621: A sequence for adding your initialization usually looks like this:
10622:
10623: @example
10624: :noname
10625: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
10626: ... \ your stuff
10627: ; IS 'cold
10628: @end example
10629:
10630: @cindex turnkey image files
10631: @cindex image file, turnkey applications
10632: You can make a turnkey image by letting @code{'cold} execute a word
10633: (your turnkey application) that never returns; instead, it exits Gforth
10634: via @code{bye} or @code{throw}.
10635:
10636: @cindex command-line arguments, access
10637: @cindex arguments on the command line, access
10638: You can access the (image-specific) command-line arguments through the
10639: variables @code{argc} and @code{argv}. @code{arg} provides convenient
10640: access to @code{argv}.
10641:
10642: If @code{'cold} exits normally, Gforth processes the command-line
10643: arguments as files to be loaded and strings to be evaluated. Therefore,
10644: @code{'cold} should remove the arguments it has used in this case.
10645:
10646: doc-'cold
10647: doc-argc
10648: doc-argv
10649: doc-arg
10650:
10651:
10652: @c ******************************************************************
10653: @node Engine, Binding to System Library, Image Files, Top
10654: @chapter Engine
10655: @cindex engine
10656: @cindex virtual machine
10657:
10658: Reading this chapter is not necessary for programming with Gforth. It
10659: may be helpful for finding your way in the Gforth sources.
10660:
10661: The ideas in this section have also been published in the papers
10662: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
10663: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
10664: Ertl, presented at EuroForth '93; the latter is available at
10665: @*@url{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
10666:
10667: @menu
10668: * Portability::
10669: * Threading::
10670: * Primitives::
10671: * Performance::
10672: @end menu
10673:
10674: @node Portability, Threading, Engine, Engine
10675: @section Portability
10676: @cindex engine portability
10677:
10678: An important goal of the Gforth Project is availability across a wide
10679: range of personal machines. fig-Forth, and, to a lesser extent, F83,
10680: achieved this goal by manually coding the engine in assembly language
10681: for several then-popular processors. This approach is very
10682: labor-intensive and the results are short-lived due to progress in
10683: computer architecture.
10684:
10685: @cindex C, using C for the engine
10686: Others have avoided this problem by coding in C, e.g., Mitch Bradley
10687: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
10688: particularly popular for UNIX-based Forths due to the large variety of
10689: architectures of UNIX machines. Unfortunately an implementation in C
10690: does not mix well with the goals of efficiency and with using
10691: traditional techniques: Indirect or direct threading cannot be expressed
10692: in C, and switch threading, the fastest technique available in C, is
10693: significantly slower. Another problem with C is that it is very
10694: cumbersome to express double integer arithmetic.
10695:
10696: @cindex GNU C for the engine
10697: @cindex long long
10698: Fortunately, there is a portable language that does not have these
10699: limitations: GNU C, the version of C processed by the GNU C compiler
10700: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
10701: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
10702: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
10703: threading possible, its @code{long long} type (@pxref{Long Long, ,
10704: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
10705: double numbers@footnote{Unfortunately, long longs are not implemented
10706: properly on all machines (e.g., on alpha-osf1, long longs are only 64
10707: bits, the same size as longs (and pointers), but they should be twice as
10708: long according to @pxref{Long Long, , Double-Word Integers, gcc.info, GNU
10709: C Manual}). So, we had to implement doubles in C after all. Still, on
10710: most machines we can use long longs and achieve better performance than
10711: with the emulation package.}. GNU C is available for free on all
10712: important (and many unimportant) UNIX machines, VMS, 80386s running
10713: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
10714: on all these machines.
10715:
10716: Writing in a portable language has the reputation of producing code that
10717: is slower than assembly. For our Forth engine we repeatedly looked at
10718: the code produced by the compiler and eliminated most compiler-induced
10719: inefficiencies by appropriate changes in the source code.
10720:
10721: @cindex explicit register declarations
10722: @cindex --enable-force-reg, configuration flag
10723: @cindex -DFORCE_REG
10724: However, register allocation cannot be portably influenced by the
10725: programmer, leading to some inefficiencies on register-starved
10726: machines. We use explicit register declarations (@pxref{Explicit Reg
10727: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
10728: improve the speed on some machines. They are turned on by using the
10729: configuration flag @code{--enable-force-reg} (@code{gcc} switch
10730: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
10731: machine, but also on the compiler version: On some machines some
10732: compiler versions produce incorrect code when certain explicit register
10733: declarations are used. So by default @code{-DFORCE_REG} is not used.
10734:
10735: @node Threading, Primitives, Portability, Engine
10736: @section Threading
10737: @cindex inner interpreter implementation
10738: @cindex threaded code implementation
10739:
10740: @cindex labels as values
10741: GNU C's labels as values extension (available since @code{gcc-2.0},
10742: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
10743: makes it possible to take the address of @i{label} by writing
10744: @code{&&@i{label}}. This address can then be used in a statement like
10745: @code{goto *@i{address}}. I.e., @code{goto *&&x} is the same as
10746: @code{goto x}.
10747:
10748: @cindex @code{NEXT}, indirect threaded
10749: @cindex indirect threaded inner interpreter
10750: @cindex inner interpreter, indirect threaded
10751: With this feature an indirect threaded @code{NEXT} looks like:
10752: @example
10753: cfa = *ip++;
10754: ca = *cfa;
10755: goto *ca;
10756: @end example
10757: @cindex instruction pointer
10758: For those unfamiliar with the names: @code{ip} is the Forth instruction
10759: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
10760: execution token and points to the code field of the next word to be
10761: executed; The @code{ca} (code address) fetched from there points to some
10762: executable code, e.g., a primitive or the colon definition handler
10763: @code{docol}.
10764:
10765: @cindex @code{NEXT}, direct threaded
10766: @cindex direct threaded inner interpreter
10767: @cindex inner interpreter, direct threaded
10768: Direct threading is even simpler:
10769: @example
10770: ca = *ip++;
10771: goto *ca;
10772: @end example
10773:
10774: Of course we have packaged the whole thing neatly in macros called
10775: @code{NEXT} and @code{NEXT1} (the part of @code{NEXT} after fetching the cfa).
10776:
10777: @menu
10778: * Scheduling::
10779: * Direct or Indirect Threaded?::
10780: * DOES>::
10781: @end menu
10782:
10783: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
10784: @subsection Scheduling
10785: @cindex inner interpreter optimization
10786:
10787: There is a little complication: Pipelined and superscalar processors,
10788: i.e., RISC and some modern CISC machines can process independent
10789: instructions while waiting for the results of an instruction. The
10790: compiler usually reorders (schedules) the instructions in a way that
10791: achieves good usage of these delay slots. However, on our first tries
10792: the compiler did not do well on scheduling primitives. E.g., for
10793: @code{+} implemented as
10794: @example
10795: n=sp[0]+sp[1];
10796: sp++;
10797: sp[0]=n;
10798: NEXT;
10799: @end example
10800: the @code{NEXT} comes strictly after the other code, i.e., there is nearly no
10801: scheduling. After a little thought the problem becomes clear: The
10802: compiler cannot know that @code{sp} and @code{ip} point to different
10803: addresses (and the version of @code{gcc} we used would not know it even
10804: if it was possible), so it could not move the load of the cfa above the
10805: store to the TOS. Indeed the pointers could be the same, if code on or
10806: very near the top of stack were executed. In the interest of speed we
10807: chose to forbid this probably unused ``feature'' and helped the compiler
10808: in scheduling: @code{NEXT} is divided into the loading part (@code{NEXT_P1})
10809: and the goto part (@code{NEXT_P2}). @code{+} now looks like:
10810: @example
10811: n=sp[0]+sp[1];
10812: sp++;
10813: NEXT_P1;
10814: sp[0]=n;
10815: NEXT_P2;
10816: @end example
10817: This can be scheduled optimally by the compiler.
10818:
10819: This division can be turned off with the switch @code{-DCISC_NEXT}. This
10820: switch is on by default on machines that do not profit from scheduling
10821: (e.g., the 80386), in order to preserve registers.
10822:
10823: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
10824: @subsection Direct or Indirect Threaded?
10825: @cindex threading, direct or indirect?
10826:
10827: @cindex -DDIRECT_THREADED
10828: Both! After packaging the nasty details in macro definitions we
10829: realized that we could switch between direct and indirect threading by
10830: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
10831: defining a few machine-specific macros for the direct-threading case.
10832: On the Forth level we also offer access words that hide the
10833: differences between the threading methods (@pxref{Threading Words}).
10834:
10835: Indirect threading is implemented completely machine-independently.
10836: Direct threading needs routines for creating jumps to the executable
10837: code (e.g. to @code{docol} or @code{dodoes}). These routines are inherently
10838: machine-dependent, but they do not amount to many source lines. Therefore,
10839: even porting direct threading to a new machine requires little effort.
10840:
10841: @cindex --enable-indirect-threaded, configuration flag
10842: @cindex --enable-direct-threaded, configuration flag
10843: The default threading method is machine-dependent. You can enforce a
10844: specific threading method when building Gforth with the configuration
10845: flag @code{--enable-direct-threaded} or
10846: @code{--enable-indirect-threaded}. Note that direct threading is not
10847: supported on all machines.
10848:
10849: @node DOES>, , Direct or Indirect Threaded?, Threading
10850: @subsection DOES>
10851: @cindex @code{DOES>} implementation
10852:
10853: @cindex @code{dodoes} routine
10854: @cindex @code{DOES>}-code
10855: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
10856: the chunk of code executed by every word defined by a
10857: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
10858: the Forth code to be executed, i.e. the code after the
10859: @code{DOES>} (the @code{DOES>}-code)? There are two solutions:
10860:
10861: In fig-Forth the code field points directly to the @code{dodoes} and the
10862: @code{DOES>}code address is stored in the cell after the code address (i.e. at
10863: @code{@i{CFA} cell+}). It may seem that this solution is illegal in
10864: the Forth-79 and all later standards, because in fig-Forth this address
10865: lies in the body (which is illegal in these standards). However, by
10866: making the code field larger for all words this solution becomes legal
10867: again. We use this approach for the indirect threaded version and for
10868: direct threading on some machines. Leaving a cell unused in most words
10869: is a bit wasteful, but on the machines we are targeting this is hardly a
10870: problem. The other reason for having a code field size of two cells is
10871: to avoid having different image files for direct and indirect threaded
10872: systems (direct threaded systems require two-cell code fields on many
10873: machines).
10874:
10875: @cindex @code{DOES>}-handler
10876: The other approach is that the code field points or jumps to the cell
10877: after @code{DOES>}. In this variant there is a jump to @code{dodoes} at
10878: this address (the @code{DOES>}-handler). @code{dodoes} can then get the
10879: @code{DOES>}-code address by computing the code address, i.e., the address of
10880: the jump to dodoes, and add the length of that jump field. A variant of
10881: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
10882: return address (which can be found in the return register on RISCs) is
10883: the @code{DOES>}-code address. Since the two cells available in the code field
10884: are used up by the jump to the code address in direct threading on many
10885: architectures, we use this approach for direct threading on these
10886: architectures. We did not want to add another cell to the code field.
10887:
10888: @node Primitives, Performance, Threading, Engine
10889: @section Primitives
10890: @cindex primitives, implementation
10891: @cindex virtual machine instructions, implementation
10892:
10893: @menu
10894: * Automatic Generation::
10895: * TOS Optimization::
10896: * Produced code::
10897: @end menu
10898:
10899: @node Automatic Generation, TOS Optimization, Primitives, Primitives
10900: @subsection Automatic Generation
10901: @cindex primitives, automatic generation
10902:
10903: @cindex @file{prims2x.fs}
10904: Since the primitives are implemented in a portable language, there is no
10905: longer any need to minimize the number of primitives. On the contrary,
10906: having many primitives has an advantage: speed. In order to reduce the
10907: number of errors in primitives and to make programming them easier, we
10908: provide a tool, the primitive generator (@file{prims2x.fs}), that
10909: automatically generates most (and sometimes all) of the C code for a
10910: primitive from the stack effect notation. The source for a primitive
10911: has the following form:
10912:
10913: @cindex primitive source format
10914: @format
10915: @i{Forth-name} @i{stack-effect} @i{category} [@i{pronounc.}]
10916: [@code{""}@i{glossary entry}@code{""}]
10917: @i{C code}
10918: [@code{:}
10919: @i{Forth code}]
10920: @end format
10921:
10922: The items in brackets are optional. The category and glossary fields
10923: are there for generating the documentation, the Forth code is there
10924: for manual implementations on machines without GNU C. E.g., the source
10925: for the primitive @code{+} is:
10926: @example
10927: + n1 n2 -- n core plus
10928: n = n1+n2;
10929: @end example
10930:
10931: This looks like a specification, but in fact @code{n = n1+n2} is C
10932: code. Our primitive generation tool extracts a lot of information from
10933: the stack effect notations@footnote{We use a one-stack notation, even
10934: though we have separate data and floating-point stacks; The separate
10935: notation can be generated easily from the unified notation.}: The number
10936: of items popped from and pushed on the stack, their type, and by what
10937: name they are referred to in the C code. It then generates a C code
10938: prelude and postlude for each primitive. The final C code for @code{+}
10939: looks like this:
10940:
10941: @example
10942: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
10943: /* */ /* documentation */
10944: @{
10945: DEF_CA /* definition of variable ca (indirect threading) */
10946: Cell n1; /* definitions of variables */
10947: Cell n2;
10948: Cell n;
10949: n1 = (Cell) sp[1]; /* input */
10950: n2 = (Cell) TOS;
10951: sp += 1; /* stack adjustment */
10952: NAME("+") /* debugging output (with -DDEBUG) */
10953: @{
10954: n = n1+n2; /* C code taken from the source */
10955: @}
10956: NEXT_P1; /* NEXT part 1 */
10957: TOS = (Cell)n; /* output */
10958: NEXT_P2; /* NEXT part 2 */
10959: @}
10960: @end example
10961:
10962: This looks long and inefficient, but the GNU C compiler optimizes quite
10963: well and produces optimal code for @code{+} on, e.g., the R3000 and the
10964: HP RISC machines: Defining the @code{n}s does not produce any code, and
10965: using them as intermediate storage also adds no cost.
10966:
10967: There are also other optimizations that are not illustrated by this
10968: example: assignments between simple variables are usually for free (copy
10969: propagation). If one of the stack items is not used by the primitive
10970: (e.g. in @code{drop}), the compiler eliminates the load from the stack
10971: (dead code elimination). On the other hand, there are some things that
10972: the compiler does not do, therefore they are performed by
10973: @file{prims2x.fs}: The compiler does not optimize code away that stores
10974: a stack item to the place where it just came from (e.g., @code{over}).
10975:
10976: While programming a primitive is usually easy, there are a few cases
10977: where the programmer has to take the actions of the generator into
10978: account, most notably @code{?dup}, but also words that do not (always)
10979: fall through to @code{NEXT}.
10980:
10981: @node TOS Optimization, Produced code, Automatic Generation, Primitives
10982: @subsection TOS Optimization
10983: @cindex TOS optimization for primitives
10984: @cindex primitives, keeping the TOS in a register
10985:
10986: An important optimization for stack machine emulators, e.g., Forth
10987: engines, is keeping one or more of the top stack items in
10988: registers. If a word has the stack effect @i{in1}...@i{inx} @code{--}
10989: @i{out1}...@i{outy}, keeping the top @i{n} items in registers
10990: @itemize @bullet
10991: @item
10992: is better than keeping @i{n-1} items, if @i{x>=n} and @i{y>=n},
10993: due to fewer loads from and stores to the stack.
10994: @item is slower than keeping @i{n-1} items, if @i{x<>y} and @i{x<n} and
10995: @i{y<n}, due to additional moves between registers.
10996: @end itemize
10997:
10998: @cindex -DUSE_TOS
10999: @cindex -DUSE_NO_TOS
11000: In particular, keeping one item in a register is never a disadvantage,
11001: if there are enough registers. Keeping two items in registers is a
11002: disadvantage for frequent words like @code{?branch}, constants,
11003: variables, literals and @code{i}. Therefore our generator only produces
11004: code that keeps zero or one items in registers. The generated C code
11005: covers both cases; the selection between these alternatives is made at
11006: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
11007: code for @code{+} is just a simple variable name in the one-item case,
11008: otherwise it is a macro that expands into @code{sp[0]}. Note that the
11009: GNU C compiler tries to keep simple variables like @code{TOS} in
11010: registers, and it usually succeeds, if there are enough registers.
11011:
11012: @cindex -DUSE_FTOS
11013: @cindex -DUSE_NO_FTOS
11014: The primitive generator performs the TOS optimization for the
11015: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
11016: operations the benefit of this optimization is even larger:
11017: floating-point operations take quite long on most processors, but can be
11018: performed in parallel with other operations as long as their results are
11019: not used. If the FP-TOS is kept in a register, this works. If
11020: it is kept on the stack, i.e., in memory, the store into memory has to
11021: wait for the result of the floating-point operation, lengthening the
11022: execution time of the primitive considerably.
11023:
11024: The TOS optimization makes the automatic generation of primitives a
11025: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
11026: @code{TOS} is not sufficient. There are some special cases to
11027: consider:
11028: @itemize @bullet
11029: @item In the case of @code{dup ( w -- w w )} the generator must not
11030: eliminate the store to the original location of the item on the stack,
11031: if the TOS optimization is turned on.
11032: @item Primitives with stack effects of the form @code{--}
11033: @i{out1}...@i{outy} must store the TOS to the stack at the start.
11034: Likewise, primitives with the stack effect @i{in1}...@i{inx} @code{--}
11035: must load the TOS from the stack at the end. But for the null stack
11036: effect @code{--} no stores or loads should be generated.
11037: @end itemize
11038:
11039: @node Produced code, , TOS Optimization, Primitives
11040: @subsection Produced code
11041: @cindex primitives, assembly code listing
11042:
11043: @cindex @file{engine.s}
11044: To see what assembly code is produced for the primitives on your machine
11045: with your compiler and your flag settings, type @code{make engine.s} and
11046: look at the resulting file @file{engine.s}.
11047:
11048: @node Performance, , Primitives, Engine
11049: @section Performance
11050: @cindex performance of some Forth interpreters
11051: @cindex engine performance
11052: @cindex benchmarking Forth systems
11053: @cindex Gforth performance
11054:
11055: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
11056: impossible to write a significantly faster engine.
11057:
11058: On register-starved machines like the 386 architecture processors
11059: improvements are possible, because @code{gcc} does not utilize the
11060: registers as well as a human, even with explicit register declarations;
11061: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
11062: and hand-tuned it for the 486; this system is 1.19 times faster on the
11063: Sieve benchmark on a 486DX2/66 than Gforth compiled with
11064: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
11065:
11066: @cindex Win32Forth performance
11067: @cindex NT Forth performance
11068: @cindex eforth performance
11069: @cindex ThisForth performance
11070: @cindex PFE performance
11071: @cindex TILE performance
11072: However, this potential advantage of assembly language implementations
11073: is not necessarily realized in complete Forth systems: We compared
11074: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
11075: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
11076: 1994) and Eforth (with and without peephole (aka pinhole) optimization
11077: of the threaded code); all these systems were written in assembly
11078: language. We also compared Gforth with three systems written in C:
11079: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
11080: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
11081: -DUNROLL_NEXT}), ThisForth Beta (compiled with @code{gcc-2.6.3 -O3
11082: -fomit-frame-pointer}; ThisForth employs peephole optimization of the
11083: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
11084: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
11085: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
11086: 486DX2/66 with similar memory performance under Windows NT. Marcel
11087: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
11088: added the peephole optimizer, ran the benchmarks and reported the
11089: results.
11090:
11091: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
11092: matrix multiplication come from the Stanford integer benchmarks and have
11093: been translated into Forth by Martin Fraeman; we used the versions
11094: included in the TILE Forth package, but with bigger data set sizes; and
11095: a recursive Fibonacci number computation for benchmarking calling
11096: performance. The following table shows the time taken for the benchmarks
11097: scaled by the time taken by Gforth (in other words, it shows the speedup
11098: factor that Gforth achieved over the other systems).
11099:
11100: @example
11101: relative Win32- NT eforth This-
11102: time Gforth Forth Forth eforth +opt PFE Forth TILE
11103: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
11104: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
11105: matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
11106: fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
11107: @end example
11108:
11109: You may be quite surprised by the good performance of Gforth when
11110: compared with systems written in assembly language. One important reason
11111: for the disappointing performance of these other systems is probably
11112: that they are not written optimally for the 486 (e.g., they use the
11113: @code{lods} instruction). In addition, Win32Forth uses a comfortable,
11114: but costly method for relocating the Forth image: like @code{cforth}, it
11115: computes the actual addresses at run time, resulting in two address
11116: computations per @code{NEXT} (@pxref{Image File Background}).
11117:
11118: Only Eforth with the peephole optimizer has a performance that is
11119: comparable to Gforth. The speedups achieved with peephole optimization
11120: of threaded code are quite remarkable. Adding a peephole optimizer to
11121: Gforth should cause similar speedups.
11122:
11123: The speedup of Gforth over PFE, ThisForth and TILE can be easily
11124: explained with the self-imposed restriction of the latter systems to
11125: standard C, which makes efficient threading impossible (however, the
11126: measured implementation of PFE uses a GNU C extension: @pxref{Global Reg
11127: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
11128: Moreover, current C compilers have a hard time optimizing other aspects
11129: of the ThisForth and the TILE source.
11130:
11131: The performance of Gforth on 386 architecture processors varies widely
11132: with the version of @code{gcc} used. E.g., @code{gcc-2.5.8} failed to
11133: allocate any of the virtual machine registers into real machine
11134: registers by itself and would not work correctly with explicit register
11135: declarations, giving a 1.3 times slower engine (on a 486DX2/66 running
11136: the Sieve) than the one measured above.
11137:
11138: Note that there have been several releases of Win32Forth since the
11139: release presented here, so the results presented above may have little
11140: predictive value for the performance of Win32Forth today.
11141:
11142: @cindex @file{Benchres}
11143: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
11144: Maierhofer (presented at EuroForth '95), an indirect threaded version of
11145: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
11146: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
11147: threaded version used here. The paper available at
11148: @*@url{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
11149: it also contains numbers for some native code systems. You can find a
11150: newer version of these measurements at
11151: @url{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
11152: find numbers for Gforth on various machines in @file{Benchres}.
11153:
11154: @c ******************************************************************
11155: @node Binding to System Library, Cross Compiler, Engine, Top
11156: @chapter Binding to System Library
11157:
11158: @node Cross Compiler, Bugs, Binding to System Library, Top
11159: @chapter Cross Compiler
11160:
11161: Cross Compiler
11162:
11163: @menu
11164: * Using the Cross Compiler::
11165: * How the Cross Compiler Works::
11166: @end menu
11167:
11168: @node Using the Cross Compiler, How the Cross Compiler Works, Cross Compiler, Cross Compiler
11169: @section Using the Cross Compiler
11170:
11171: @node How the Cross Compiler Works, , Using the Cross Compiler, Cross Compiler
11172: @section How the Cross Compiler Works
11173:
11174: @node Bugs, Origin, Cross Compiler, Top
11175: @appendix Bugs
11176: @cindex bug reporting
11177:
11178: Known bugs are described in the file @file{BUGS} in the Gforth distribution.
11179:
11180: If you find a bug, please send a bug report to
11181: @email{bug-gforth@@gnu.ai.mit.edu}. A bug report should include this
11182: information:
11183:
11184: @itemize @bullet
11185: @item
11186: The Gforth version used (it is announced at the start of an
11187: interactive Gforth session).
11188: @item
11189: The machine and operating system (on Unix
11190: systems @code{uname -a} will report this information).
11191: @item
11192: The installation options (send the file @file{config.status}).
11193: @item
11194: A complete list of changes (if any) you (or your installer) have made to the
11195: Gforth sources.
11196: @item
11197: A program (or a sequence of keyboard commands) that reproduces the bug.
11198: @item
11199: A description of what you think constitutes the buggy behaviour.
11200: @end itemize
11201:
11202: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
11203: to Report Bugs, gcc.info, GNU C Manual}.
11204:
11205:
11206: @node Origin, Forth-related information, Bugs, Top
11207: @appendix Authors and Ancestors of Gforth
11208:
11209: @section Authors and Contributors
11210: @cindex authors of Gforth
11211: @cindex contributors to Gforth
11212:
11213: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
11214: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
11215: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
11216: with their continuous feedback. Lennart Benshop contributed
11217: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
11218: support for calling C libraries. Helpful comments also came from Paul
11219: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
11220: Wavrik, Barrie Stott, Marc de Groot, and Jorge Acerada. Since the
11221: release of Gforth-0.2.1 there were also helpful comments from many
11222: others; thank you all, sorry for not listing you here (but digging
11223: through my mailbox to extract your names is on my to-do list). Since the
11224: release of Gforth-0.4.0 Neal Crook worked on the manual.
11225:
11226: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
11227: and autoconf, among others), and to the creators of the Internet: Gforth
11228: was developed across the Internet, and its authors did not meet
11229: physically for the first 4 years of development.
11230:
11231: @section Pedigree
11232: @cindex pedigree of Gforth
11233:
11234: Gforth descends from bigFORTH (1993) and fig-Forth. Gforth and PFE (by
11235: Dirk Zoller) will cross-fertilize each other. Of course, a significant
11236: part of the design of Gforth was prescribed by ANS Forth.
11237:
11238: Bernd Paysan wrote bigFORTH, a descendent from TurboForth, an unreleased
11239: 32 bit native code version of VolksForth for the Atari ST, written
11240: mostly by Dietrich Weineck.
11241:
11242: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
11243: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
11244: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
11245:
11246: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
11247: Forth-83 standard. !! Pedigree? When?
11248:
11249: A team led by Bill Ragsdale implemented fig-Forth on many processors in
11250: 1979. Robert Selzer and Bill Ragsdale developed the original
11251: implementation of fig-Forth for the 6502 based on microForth.
11252:
11253: The principal architect of microForth was Dean Sanderson. microForth was
11254: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
11255: the 1802, and subsequently implemented on the 8080, the 6800 and the
11256: Z80.
11257:
11258: All earlier Forth systems were custom-made, usually by Charles Moore,
11259: who discovered (as he puts it) Forth during the late 60s. The first full
11260: Forth existed in 1971.
11261:
11262: A part of the information in this section comes from @cite{The Evolution
11263: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
11264: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
11265: Notices 28(3), 1993. You can find more historical and genealogical
11266: information about Forth there.
11267:
11268: @node Forth-related information, Word Index, Origin, Top
11269: @appendix Other Forth-related information
11270: @cindex Forth-related information
11271:
11272: @menu
11273: * Internet resources::
11274: * Books::
11275: * The Forth Interest Group::
11276: * Conferences::
11277: @end menu
11278:
11279:
11280: @node Internet resources, Books, Forth-related information, Forth-related information
11281: @section Internet resources
11282: @cindex internet resources
11283:
11284: @cindex comp.lang.forth
11285: @cindex frequently asked questions
11286: There is an active newsgroup (comp.lang.forth) discussing Forth and
11287: Forth-related issues. A frequently-asked-questions (FAQ) list
11288: is posted to the newsgroup regulary, and archived at these sites:
11289:
11290: @itemize @bullet
11291: @item
11292: @url{ftp://rtfm.mit.edu/pub/usenet-by-group/comp.lang.forth/}
11293: @item
11294: @url{ftp://ftp.forth.org/pub/Forth/FAQ/}
11295: @end itemize
11296:
11297: The FAQ list should be considered mandatory reading before posting to
11298: the newsgroup.
11299:
11300: Here are some other web sites holding Forth-related material:
11301:
11302: @itemize @bullet
11303: @item
11304: @url{http://www.taygeta.com/forth.html} -- Skip Carter's Forth pages.
11305: @item
11306: @url{http://www.jwdt.com/~paysan/gforth.html} -- the Gforth home page.
11307: @item
11308: @url{http://www.minerva.com/uathena.htm} -- home of ANS Forth Standard.
11309: @item
11310: @url{http://dec.bournemouth.ac.uk/forth/index.html} -- the Forth
11311: Research page, including links to the Journal of Forth Application and
11312: Research (JFAR) and a searchable Forth bibliography.
11313: @end itemize
11314:
11315:
11316: @node Books, The Forth Interest Group, Internet resources, Forth-related information
11317: @section Books
11318: @cindex books on Forth
11319:
11320: As the Standard is relatively new, there are not many books out yet. It
11321: is not recommended to learn Forth by using Gforth and a book that is not
11322: written for ANS Forth, as you will not know your mistakes from the
11323: deviations of the book. However, books based on the Forth-83 standard
11324: should be ok, because ANS Forth is primarily an extension of Forth-83.
11325:
11326: @cindex standard document for ANS Forth
11327: @cindex ANS Forth document
11328: The definite reference if you want to write ANS Forth programs is, of
11329: course, the ANS Forth document. It is available in printed form from the
11330: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
11331: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about
11332: $200. You can also get it from Global Engineering Documents (Tel.: USA
11333: (800) 854-7179; Fax.: (303) 843-9880) for about $300.
11334:
11335: @cite{dpANS6}, the last draft of the standard, which was then submitted
11336: to ANSI for publication is available electronically and for free in some
11337: MS Word format, and it has been converted to HTML
11338: (@url{http://www.taygeta.com/forth/dpans.html}; this is my favourite
11339: format); this HTML version also includes the answers to Requests for
11340: Interpretation (RFIs). Some pointers to these versions can be found
11341: through @*@url{http://www.complang.tuwien.ac.at/projects/forth.html}.
11342:
11343: @cindex introductory book on Forth
11344: @cindex book on Forth, introductory
11345: @cindex Woehr, Jack: @cite{Forth: The New Model}
11346: @cindex @cite{Forth: The new model} (book)
11347: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
11348: introductory book based on a draft version of the standard. It does not
11349: cover the whole standard. It also contains interesting background
11350: information (Jack Woehr was in the ANS Forth Technical Committee). It is
11351: not appropriate for complete newbies, but programmers experienced in
11352: other languages should find it ok.
11353:
11354: @cindex Conklin, Edward K., and Elizabeth Rather: @cite{Forth Programmer's Handbook}
11355: @cindex Rather, Elizabeth and Edward K. Conklin: @cite{Forth Programmer's Handbook}
11356: @cindex @cite{Forth Programmer's Handbook} (book)
11357: @cite{Forth Programmer's Handbook} by Edward K. Conklin, Elizabeth
11358: D. Rather and the technical staff of Forth, Inc. (Forth, Inc., 1997;
11359: ISBN 0-9662156-0-5) contains little introductory material. The majority
11360: of the book is similar to @ref{Words}, but the book covers most of the
11361: standard words and some non-standard words (whereas this manual is
11362: quite incomplete). In addition, the book contains a chapter on
11363: programming style. The major drawback of this book is that it usually
11364: does not identify what is standard and what is specific to the Forth
11365: system described in the book (probably one of Forth, Inc.'s systems).
11366: Fortunately, many of the non-standard programming practices described in
11367: the book work in Gforth, too. Still, this drawback makes the book
11368: hardly more useful than a pre-ANS book.
11369:
11370: @node The Forth Interest Group, Conferences, Books, Forth-related information
11371: @section The Forth Interest Group
11372: @cindex Forth interest group (FIG)
11373:
11374: The Forth Interest Group (FIG) is a world-wide, non-profit,
11375: member-supported organisation. It publishes a regular magazine,
11376: @var{FORTH Dimensions}, and offers other benefits of membership. You can
11377: contact the FIG through their office email address:
11378: @email{office@@forth.org} or by visiting their web site at
11379: @url{http://www.forth.org/}. This web site also includes links to FIG
11380: chapters in other countries and American cities
11381: (@url{http://www.forth.org/chapters.html}).
11382:
11383: @node Conferences, , The Forth Interest Group, Forth-related information
11384: @section Conferences
11385: @cindex Conferences
11386:
11387: There are several regular conferences related to Forth. They are all
11388: well-publicised in @var{FORTH Dimensions} and on the comp.lang.forth
11389: news group:
11390:
11391: @itemize @bullet
11392: @item
11393: FORML -- the Forth modification laboratory convenes every year near
11394: Monterey, California.
11395: @item
11396: The Rochester Forth Conference -- an annual conference traditionally
11397: held in Rochester, New York.
11398: @item
11399: EuroForth -- this European conference takes place annually.
11400: @end itemize
11401:
11402:
11403: @node Word Index, Concept Index, Forth-related information, Top
11404: @unnumbered Word Index
11405:
11406: This index is a list of Forth words that have ``glossary'' entries
11407: within this manual. Each word is listed with its stack effect and
11408: wordset.
11409:
11410: @printindex fn
11411:
11412: @node Concept Index, , Word Index, Top
11413: @unnumbered Concept and Word Index
11414:
11415: Not all entries listed in this index are present verbatim in the
11416: text. This index also duplicates, in abbreviated form, all of the words
11417: listed in the Word Index (only the names are listed for the words here).
11418:
11419: @printindex cp
11420:
11421: @contents
11422: @bye
11423:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>
![[BACK]](/cvsweb/icons/back.gif){kind=icon}
Return to gforth.ds CVS log ![[TXT]](/cvsweb/icons/text.gif){kind=icon}
![[DIR]](/cvsweb/icons/dir.gif){kind=icon}
Up to [gforth] / gforth / doc