1: \input texinfo @c -*-texinfo-*-
2: @comment The source is gforth.ds, from which gforth.texi is generated
3: @comment %**start of header (This is for running Texinfo on a region.)
4: @setfilename gforth.info
5: @settitle Gforth Manual
6: @comment @setchapternewpage odd
7: @comment %**end of header (This is for running Texinfo on a region.)
8:
9: @ifinfo
10: This file documents Gforth 0.3
11:
12: Copyright @copyright{} 1995-1997 Free Software Foundation, Inc.
13:
14: Permission is granted to make and distribute verbatim copies of
15: this manual provided the copyright notice and this permission notice
16: are preserved on all copies.
17:
18: @ignore
19: Permission is granted to process this file through TeX and print the
20: results, provided the printed document carries a copying permission
21: notice identical to this one except for the removal of this paragraph
22: (this paragraph not being relevant to the printed manual).
23:
24: @end ignore
25: Permission is granted to copy and distribute modified versions of this
26: manual under the conditions for verbatim copying, provided also that the
27: sections entitled "Distribution" and "General Public License" are
28: included exactly as in the original, and provided that the entire
29: resulting derived work is distributed under the terms of a permission
30: notice identical to this one.
31:
32: Permission is granted to copy and distribute translations of this manual
33: into another language, under the above conditions for modified versions,
34: except that the sections entitled "Distribution" and "General Public
35: License" may be included in a translation approved by the author instead
36: of in the original English.
37: @end ifinfo
38:
39: @finalout
40: @titlepage
41: @sp 10
42: @center @titlefont{Gforth Manual}
43: @sp 2
44: @center for version 0.3
45: @sp 2
46: @center Anton Ertl
47: @center Bernd Paysan
48: @sp 3
49: @center This manual is under construction
50:
51: @comment The following two commands start the copyright page.
52: @page
53: @vskip 0pt plus 1filll
54: Copyright @copyright{} 1995--1997 Free Software Foundation, Inc.
55:
56: @comment !! Published by ... or You can get a copy of this manual ...
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: Permission is granted to copy and distribute modified versions of this
63: manual under the conditions for verbatim copying, provided also that the
64: sections entitled "Distribution" and "General Public License" are
65: included exactly as in the original, and provided that the entire
66: resulting derived work is distributed under the terms of a permission
67: notice identical to this one.
68:
69: Permission is granted to copy and distribute translations of this manual
70: into another language, under the above conditions for modified versions,
71: except that the sections entitled "Distribution" and "General Public
72: License" may be included in a translation approved by the author instead
73: of in the original English.
74: @end titlepage
75:
76:
77: @node Top, License, (dir), (dir)
78: @ifinfo
79: Gforth is a free implementation of ANS Forth available on many
80: personal machines. This manual corresponds to version 0.3.
81: @end ifinfo
82:
83: @menu
84: * License::
85: * Goals:: About the Gforth Project
86: * Other Books:: Things you might want to read
87: * Invoking Gforth:: Starting Gforth
88: * Words:: Forth words available in Gforth
89: * Tools:: Programming tools
90: * ANS conformance:: Implementation-defined options etc.
91: * Model:: The abstract machine of Gforth
92: * Integrating Gforth:: Forth as scripting language for applications
93: * Emacs and Gforth:: The Gforth Mode
94: * Image Files:: @code{.fi} files contain compiled code
95: * Engine:: The inner interpreter and the primitives
96: * Bugs:: How to report them
97: * Origin:: Authors and ancestors of Gforth
98: * Word Index:: An item for each Forth word
99: * Concept Index:: A menu covering many topics
100: @end menu
101:
102: @node License, Goals, Top, Top
103: @unnumbered GNU GENERAL PUBLIC LICENSE
104: @center Version 2, June 1991
105:
106: @display
107: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
108: 675 Mass Ave, Cambridge, MA 02139, USA
109:
110: Everyone is permitted to copy and distribute verbatim copies
111: of this license document, but changing it is not allowed.
112: @end display
113:
114: @unnumberedsec Preamble
115:
116: The licenses for most software are designed to take away your
117: freedom to share and change it. By contrast, the GNU General Public
118: License is intended to guarantee your freedom to share and change free
119: software---to make sure the software is free for all its users. This
120: General Public License applies to most of the Free Software
121: Foundation's software and to any other program whose authors commit to
122: using it. (Some other Free Software Foundation software is covered by
123: the GNU Library General Public License instead.) You can apply it to
124: your programs, too.
125:
126: When we speak of free software, we are referring to freedom, not
127: price. Our General Public Licenses are designed to make sure that you
128: have the freedom to distribute copies of free software (and charge for
129: this service if you wish), that you receive source code or can get it
130: if you want it, that you can change the software or use pieces of it
131: in new free programs; and that you know you can do these things.
132:
133: To protect your rights, we need to make restrictions that forbid
134: anyone to deny you these rights or to ask you to surrender the rights.
135: These restrictions translate to certain responsibilities for you if you
136: distribute copies of the software, or if you modify it.
137:
138: For example, if you distribute copies of such a program, whether
139: gratis or for a fee, you must give the recipients all the rights that
140: you have. You must make sure that they, too, receive or can get the
141: source code. And you must show them these terms so they know their
142: rights.
143:
144: We protect your rights with two steps: (1) copyright the software, and
145: (2) offer you this license which gives you legal permission to copy,
146: distribute and/or modify the software.
147:
148: Also, for each author's protection and ours, we want to make certain
149: that everyone understands that there is no warranty for this free
150: software. If the software is modified by someone else and passed on, we
151: want its recipients to know that what they have is not the original, so
152: that any problems introduced by others will not reflect on the original
153: authors' reputations.
154:
155: Finally, any free program is threatened constantly by software
156: patents. We wish to avoid the danger that redistributors of a free
157: program will individually obtain patent licenses, in effect making the
158: program proprietary. To prevent this, we have made it clear that any
159: patent must be licensed for everyone's free use or not licensed at all.
160:
161: The precise terms and conditions for copying, distribution and
162: modification follow.
163:
164: @iftex
165: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
166: @end iftex
167: @ifinfo
168: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
169: @end ifinfo
170:
171: @enumerate 0
172: @item
173: This License applies to any program or other work which contains
174: a notice placed by the copyright holder saying it may be distributed
175: under the terms of this General Public License. The ``Program'', below,
176: refers to any such program or work, and a ``work based on the Program''
177: means either the Program or any derivative work under copyright law:
178: that is to say, a work containing the Program or a portion of it,
179: either verbatim or with modifications and/or translated into another
180: language. (Hereinafter, translation is included without limitation in
181: the term ``modification''.) Each licensee is addressed as ``you''.
182:
183: Activities other than copying, distribution and modification are not
184: covered by this License; they are outside its scope. The act of
185: running the Program is not restricted, and the output from the Program
186: is covered only if its contents constitute a work based on the
187: Program (independent of having been made by running the Program).
188: Whether that is true depends on what the Program does.
189:
190: @item
191: You may copy and distribute verbatim copies of the Program's
192: source code as you receive it, in any medium, provided that you
193: conspicuously and appropriately publish on each copy an appropriate
194: copyright notice and disclaimer of warranty; keep intact all the
195: notices that refer to this License and to the absence of any warranty;
196: and give any other recipients of the Program a copy of this License
197: along with the Program.
198:
199: You may charge a fee for the physical act of transferring a copy, and
200: you may at your option offer warranty protection in exchange for a fee.
201:
202: @item
203: You may modify your copy or copies of the Program or any portion
204: of it, thus forming a work based on the Program, and copy and
205: distribute such modifications or work under the terms of Section 1
206: above, provided that you also meet all of these conditions:
207:
208: @enumerate a
209: @item
210: You must cause the modified files to carry prominent notices
211: stating that you changed the files and the date of any change.
212:
213: @item
214: You must cause any work that you distribute or publish, that in
215: whole or in part contains or is derived from the Program or any
216: part thereof, to be licensed as a whole at no charge to all third
217: parties under the terms of this License.
218:
219: @item
220: If the modified program normally reads commands interactively
221: when run, you must cause it, when started running for such
222: interactive use in the most ordinary way, to print or display an
223: announcement including an appropriate copyright notice and a
224: notice that there is no warranty (or else, saying that you provide
225: a warranty) and that users may redistribute the program under
226: these conditions, and telling the user how to view a copy of this
227: License. (Exception: if the Program itself is interactive but
228: does not normally print such an announcement, your work based on
229: the Program is not required to print an announcement.)
230: @end enumerate
231:
232: These requirements apply to the modified work as a whole. If
233: identifiable sections of that work are not derived from the Program,
234: and can be reasonably considered independent and separate works in
235: themselves, then this License, and its terms, do not apply to those
236: sections when you distribute them as separate works. But when you
237: distribute the same sections as part of a whole which is a work based
238: on the Program, the distribution of the whole must be on the terms of
239: this License, whose permissions for other licensees extend to the
240: entire whole, and thus to each and every part regardless of who wrote it.
241:
242: Thus, it is not the intent of this section to claim rights or contest
243: your rights to work written entirely by you; rather, the intent is to
244: exercise the right to control the distribution of derivative or
245: collective works based on the Program.
246:
247: In addition, mere aggregation of another work not based on the Program
248: with the Program (or with a work based on the Program) on a volume of
249: a storage or distribution medium does not bring the other work under
250: the scope of this License.
251:
252: @item
253: You may copy and distribute the Program (or a work based on it,
254: under Section 2) in object code or executable form under the terms of
255: Sections 1 and 2 above provided that you also do one of the following:
256:
257: @enumerate a
258: @item
259: Accompany it with the complete corresponding machine-readable
260: source code, which must be distributed under the terms of Sections
261: 1 and 2 above on a medium customarily used for software interchange; or,
262:
263: @item
264: Accompany it with a written offer, valid for at least three
265: years, to give any third party, for a charge no more than your
266: cost of physically performing source distribution, a complete
267: machine-readable copy of the corresponding source code, to be
268: distributed under the terms of Sections 1 and 2 above on a medium
269: customarily used for software interchange; or,
270:
271: @item
272: Accompany it with the information you received as to the offer
273: to distribute corresponding source code. (This alternative is
274: allowed only for noncommercial distribution and only if you
275: received the program in object code or executable form with such
276: an offer, in accord with Subsection b above.)
277: @end enumerate
278:
279: The source code for a work means the preferred form of the work for
280: making modifications to it. For an executable work, complete source
281: code means all the source code for all modules it contains, plus any
282: associated interface definition files, plus the scripts used to
283: control compilation and installation of the executable. However, as a
284: special exception, the source code distributed need not include
285: anything that is normally distributed (in either source or binary
286: form) with the major components (compiler, kernel, and so on) of the
287: operating system on which the executable runs, unless that component
288: itself accompanies the executable.
289:
290: If distribution of executable or object code is made by offering
291: access to copy from a designated place, then offering equivalent
292: access to copy the source code from the same place counts as
293: distribution of the source code, even though third parties are not
294: compelled to copy the source along with the object code.
295:
296: @item
297: You may not copy, modify, sublicense, or distribute the Program
298: except as expressly provided under this License. Any attempt
299: otherwise to copy, modify, sublicense or distribute the Program is
300: void, and will automatically terminate your rights under this License.
301: However, parties who have received copies, or rights, from you under
302: this License will not have their licenses terminated so long as such
303: parties remain in full compliance.
304:
305: @item
306: You are not required to accept this License, since you have not
307: signed it. However, nothing else grants you permission to modify or
308: distribute the Program or its derivative works. These actions are
309: prohibited by law if you do not accept this License. Therefore, by
310: modifying or distributing the Program (or any work based on the
311: Program), you indicate your acceptance of this License to do so, and
312: all its terms and conditions for copying, distributing or modifying
313: the Program or works based on it.
314:
315: @item
316: Each time you redistribute the Program (or any work based on the
317: Program), the recipient automatically receives a license from the
318: original licensor to copy, distribute or modify the Program subject to
319: these terms and conditions. You may not impose any further
320: restrictions on the recipients' exercise of the rights granted herein.
321: You are not responsible for enforcing compliance by third parties to
322: this License.
323:
324: @item
325: If, as a consequence of a court judgment or allegation of patent
326: infringement or for any other reason (not limited to patent issues),
327: conditions are imposed on you (whether by court order, agreement or
328: otherwise) that contradict the conditions of this License, they do not
329: excuse you from the conditions of this License. If you cannot
330: distribute so as to satisfy simultaneously your obligations under this
331: License and any other pertinent obligations, then as a consequence you
332: may not distribute the Program at all. For example, if a patent
333: license would not permit royalty-free redistribution of the Program by
334: all those who receive copies directly or indirectly through you, then
335: the only way you could satisfy both it and this License would be to
336: refrain entirely from distribution of the Program.
337:
338: If any portion of this section is held invalid or unenforceable under
339: any particular circumstance, the balance of the section is intended to
340: apply and the section as a whole is intended to apply in other
341: circumstances.
342:
343: It is not the purpose of this section to induce you to infringe any
344: patents or other property right claims or to contest validity of any
345: such claims; this section has the sole purpose of protecting the
346: integrity of the free software distribution system, which is
347: implemented by public license practices. Many people have made
348: generous contributions to the wide range of software distributed
349: through that system in reliance on consistent application of that
350: system; it is up to the author/donor to decide if he or she is willing
351: to distribute software through any other system and a licensee cannot
352: impose that choice.
353:
354: This section is intended to make thoroughly clear what is believed to
355: be a consequence of the rest of this License.
356:
357: @item
358: If the distribution and/or use of the Program is restricted in
359: certain countries either by patents or by copyrighted interfaces, the
360: original copyright holder who places the Program under this License
361: may add an explicit geographical distribution limitation excluding
362: those countries, so that distribution is permitted only in or among
363: countries not thus excluded. In such case, this License incorporates
364: the limitation as if written in the body of this License.
365:
366: @item
367: The Free Software Foundation may publish revised and/or new versions
368: of the General Public License from time to time. Such new versions will
369: be similar in spirit to the present version, but may differ in detail to
370: address new problems or concerns.
371:
372: Each version is given a distinguishing version number. If the Program
373: specifies a version number of this License which applies to it and ``any
374: later version'', you have the option of following the terms and conditions
375: either of that version or of any later version published by the Free
376: Software Foundation. If the Program does not specify a version number of
377: this License, you may choose any version ever published by the Free Software
378: Foundation.
379:
380: @item
381: If you wish to incorporate parts of the Program into other free
382: programs whose distribution conditions are different, write to the author
383: to ask for permission. For software which is copyrighted by the Free
384: Software Foundation, write to the Free Software Foundation; we sometimes
385: make exceptions for this. Our decision will be guided by the two goals
386: of preserving the free status of all derivatives of our free software and
387: of promoting the sharing and reuse of software generally.
388:
389: @iftex
390: @heading NO WARRANTY
391: @end iftex
392: @ifinfo
393: @center NO WARRANTY
394: @end ifinfo
395:
396: @item
397: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
398: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
399: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
400: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
401: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
402: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
403: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
404: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
405: REPAIR OR CORRECTION.
406:
407: @item
408: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
409: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
410: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
411: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
412: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
413: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
414: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
415: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
416: POSSIBILITY OF SUCH DAMAGES.
417: @end enumerate
418:
419: @iftex
420: @heading END OF TERMS AND CONDITIONS
421: @end iftex
422: @ifinfo
423: @center END OF TERMS AND CONDITIONS
424: @end ifinfo
425:
426: @page
427: @unnumberedsec How to Apply These Terms to Your New Programs
428:
429: If you develop a new program, and you want it to be of the greatest
430: possible use to the public, the best way to achieve this is to make it
431: free software which everyone can redistribute and change under these terms.
432:
433: To do so, attach the following notices to the program. It is safest
434: to attach them to the start of each source file to most effectively
435: convey the exclusion of warranty; and each file should have at least
436: the ``copyright'' line and a pointer to where the full notice is found.
437:
438: @smallexample
439: @var{one line to give the program's name and a brief idea of what it does.}
440: Copyright (C) 19@var{yy} @var{name of author}
441:
442: This program is free software; you can redistribute it and/or modify
443: it under the terms of the GNU General Public License as published by
444: the Free Software Foundation; either version 2 of the License, or
445: (at your option) any later version.
446:
447: This program is distributed in the hope that it will be useful,
448: but WITHOUT ANY WARRANTY; without even the implied warranty of
449: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
450: GNU General Public License for more details.
451:
452: You should have received a copy of the GNU General Public License
453: along with this program; if not, write to the Free Software
454: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
455: @end smallexample
456:
457: Also add information on how to contact you by electronic and paper mail.
458:
459: If the program is interactive, make it output a short notice like this
460: when it starts in an interactive mode:
461:
462: @smallexample
463: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
464: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
465: type `show w'.
466: This is free software, and you are welcome to redistribute it
467: under certain conditions; type `show c' for details.
468: @end smallexample
469:
470: The hypothetical commands @samp{show w} and @samp{show c} should show
471: the appropriate parts of the General Public License. Of course, the
472: commands you use may be called something other than @samp{show w} and
473: @samp{show c}; they could even be mouse-clicks or menu items---whatever
474: suits your program.
475:
476: You should also get your employer (if you work as a programmer) or your
477: school, if any, to sign a ``copyright disclaimer'' for the program, if
478: necessary. Here is a sample; alter the names:
479:
480: @smallexample
481: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
482: `Gnomovision' (which makes passes at compilers) written by James Hacker.
483:
484: @var{signature of Ty Coon}, 1 April 1989
485: Ty Coon, President of Vice
486: @end smallexample
487:
488: This General Public License does not permit incorporating your program into
489: proprietary programs. If your program is a subroutine library, you may
490: consider it more useful to permit linking proprietary applications with the
491: library. If this is what you want to do, use the GNU Library General
492: Public License instead of this License.
493:
494: @iftex
495: @unnumbered Preface
496: @cindex Preface
497: This manual documents Gforth. The reader is expected to know
498: Forth. This manual is primarily a reference manual. @xref{Other Books}
499: for introductory material.
500: @end iftex
501:
502: @node Goals, Other Books, License, Top
503: @comment node-name, next, previous, up
504: @chapter Goals of Gforth
505: @cindex Goals
506: The goal of the Gforth Project is to develop a standard model for
507: ANS Forth. This can be split into several subgoals:
508:
509: @itemize @bullet
510: @item
511: Gforth should conform to the Forth standard (ANS Forth).
512: @item
513: It should be a model, i.e. it should define all the
514: implementation-dependent things.
515: @item
516: It should become standard, i.e. widely accepted and used. This goal
517: is the most difficult one.
518: @end itemize
519:
520: To achieve these goals Gforth should be
521: @itemize @bullet
522: @item
523: Similar to previous models (fig-Forth, F83)
524: @item
525: Powerful. It should provide for all the things that are considered
526: necessary today and even some that are not yet considered necessary.
527: @item
528: Efficient. It should not get the reputation of being exceptionally
529: slow.
530: @item
531: Free.
532: @item
533: Available on many machines/easy to port.
534: @end itemize
535:
536: Have we achieved these goals? Gforth conforms to the ANS Forth
537: standard. It may be considered a model, but we have not yet documented
538: which parts of the model are stable and which parts we are likely to
539: change. It certainly has not yet become a de facto standard. It has some
540: similarities and some differences to previous models. It has some
541: powerful features, but not yet everything that we envisioned. We
542: certainly have achieved our execution speed goals (@pxref{Performance}).
543: It is free and available on many machines.
544:
545: @node Other Books, Invoking Gforth, Goals, Top
546: @chapter Other books on ANS Forth
547: @cindex books on Forth
548:
549: As the standard is relatively new, there are not many books out yet. It
550: is not recommended to learn Forth by using Gforth and a book that is
551: not written for ANS Forth, as you will not know your mistakes from the
552: deviations of the book.
553:
554: @cindex standard document for ANS Forth
555: @cindex ANS Forth document
556: There is, of course, the standard, the definite reference if you want to
557: write ANS Forth programs. It is available in printed form from the
558: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
559: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
560: can also get it from Global Engineering Documents (Tel.: USA (800)
561: 854-7179; Fax.: (303) 843-9880) for about $300.
562:
563: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
564: for publication is available electronically and for free in some MS Word
565: format, and it has been converted to HTML. Some pointers to these
566: versions can be found through
567: @*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
568:
569: @cindex introductory book
570: @cindex book, introductory
571: @cindex Woehr, Jack: @cite{Forth: The New Model}
572: @cindex @cite{Forth: The new model} (book)
573: @cite{Forth: The New Model} by Jack Woehr (Prentice-Hall, 1993) is an
574: introductory book based on a draft version of the standard. It does not
575: cover the whole standard. It also contains interesting background
576: information (Jack Woehr was in the ANS Forth Technical Committee). It is
577: not appropriate for complete newbies, but programmers experienced in
578: other languages should find it ok.
579:
580: @node Invoking Gforth, Words, Other Books, Top
581: @chapter Invoking Gforth
582: @cindex invoking Gforth
583: @cindex running Gforth
584: @cindex command-line options
585: @cindex options on the command line
586: @cindex flags on the command line
587:
588: You will usually just say @code{gforth}. In many other cases the default
589: Gforth image will be invoked like this:
590:
591: @example
592: gforth [files] [-e forth-code]
593: @end example
594:
595: executing the contents of the files and the Forth code in the order they
596: are given.
597:
598: In general, the command line looks like this:
599:
600: @example
601: gforth [initialization options] [image-specific options]
602: @end example
603:
604: The initialization options must come before the rest of the command
605: line. They are:
606:
607: @table @code
608: @cindex -i, command-line option
609: @cindex --image-file, command-line option
610: @item --image-file @var{file}
611: @itemx -i @var{file}
612: Loads the Forth image @var{file} instead of the default
613: @file{gforth.fi} (@pxref{Image Files}).
614:
615: @cindex --path, command-line option
616: @cindex -p, command-line option
617: @item --path @var{path}
618: @itemx -p @var{path}
619: Uses @var{path} for searching the image file and Forth source code files
620: instead of the default in the environment variable @code{GFORTHPATH} or
621: the path specified at installation time (e.g.,
622: @file{/usr/local/share/gforth/0.2.0:.}). A path is given as a list of
623: directories, separated by @samp{:} (on Unix) or @samp{;} (on other OSs).
624:
625: @cindex --dictionary-size, command-line option
626: @cindex -m, command-line option
627: @cindex @var{size} parameters for command-line options
628: @cindex size of the dictionary and the stacks
629: @item --dictionary-size @var{size}
630: @itemx -m @var{size}
631: Allocate @var{size} space for the Forth dictionary space instead of
632: using the default specified in the image (typically 256K). The
633: @var{size} specification consists of an integer and a unit (e.g.,
634: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
635: size, in this case Cells), @code{k} (kilobytes), and @code{M}
636: (Megabytes). If no unit is specified, @code{e} is used.
637:
638: @cindex --data-stack-size, command-line option
639: @cindex -d, command-line option
640: @item --data-stack-size @var{size}
641: @itemx -d @var{size}
642: Allocate @var{size} space for the data stack instead of using the
643: default specified in the image (typically 16K).
644:
645: @cindex --return-stack-size, command-line option
646: @cindex -r, command-line option
647: @item --return-stack-size @var{size}
648: @itemx -r @var{size}
649: Allocate @var{size} space for the return stack instead of using the
650: default specified in the image (typically 15K).
651:
652: @cindex --fp-stack-size, command-line option
653: @cindex -f, command-line option
654: @item --fp-stack-size @var{size}
655: @itemx -f @var{size}
656: Allocate @var{size} space for the floating point stack instead of
657: using the default specified in the image (typically 15.5K). In this case
658: the unit specifier @code{e} refers to floating point numbers.
659:
660: @cindex --locals-stack-size, command-line option
661: @cindex -l, command-line option
662: @item --locals-stack-size @var{size}
663: @itemx -l @var{size}
664: Allocate @var{size} space for the locals stack instead of using the
665: default specified in the image (typically 14.5K).
666:
667: @cindex -h, command-line option
668: @cindex --help, command-line option
669: @item --help
670: @itemx -h
671: Print a message about the command-line options
672:
673: @cindex -v, command-line option
674: @cindex --version, command-line option
675: @item --version
676: @itemx -v
677: Print version and exit
678:
679: @cindex --debug, command-line option
680: @item --debug
681: Print some information useful for debugging on startup.
682:
683: @cindex --offset-image, command-line option
684: @item --offset-image
685: Start the dictionary at a slightly different position than would be used
686: otherwise (useful for creating data-relocatable images,
687: @pxref{Data-Relocatable Image Files}).
688:
689: @cindex --clear-dictionary, command-line option
690: @item --clear-dictionary
691: Initialize all bytes in the dictionary to 0 before loading the image
692: (@pxref{Data-Relocatable Image Files}).
693: @end table
694:
695: @cindex loading files at startup
696: @cindex executing code on startup
697: @cindex batch processing with Gforth
698: As explained above, the image-specific command-line arguments for the
699: default image @file{gforth.fi} consist of a sequence of filenames and
700: @code{-e @var{forth-code}} options that are interpreted in the sequence
701: in which they are given. The @code{-e @var{forth-code}} or
702: @code{--evaluate @var{forth-code}} option evaluates the forth
703: code. This option takes only one argument; if you want to evaluate more
704: Forth words, you have to quote them or use several @code{-e}s. To exit
705: after processing the command line (instead of entering interactive mode)
706: append @code{-e bye} to the command line.
707:
708: @cindex versions, invoking other versions of Gforth
709: If you have several versions of Gforth installed, @code{gforth} will
710: invoke the version that was installed last. @code{gforth-@var{version}}
711: invokes a specific version. You may want to use the option
712: @code{--path}, if your environment contains the variable
713: @code{GFORTHPATH}.
714:
715: Not yet implemented:
716: On startup the system first executes the system initialization file
717: (unless the option @code{--no-init-file} is given; note that the system
718: resulting from using this option may not be ANS Forth conformant). Then
719: the user initialization file @file{.gforth.fs} is executed, unless the
720: option @code{--no-rc} is given; this file is first searched in @file{.},
721: then in @file{~}, then in the normal path (see above).
722:
723: @node Words, Tools, Invoking Gforth, Top
724: @chapter Forth Words
725: @cindex Words
726:
727: @menu
728: * Notation::
729: * Arithmetic::
730: * Stack Manipulation::
731: * Memory::
732: * Control Structures::
733: * Locals::
734: * Defining Words::
735: * Tokens for Words::
736: * Wordlists::
737: * Files::
738: * Blocks::
739: * Other I/O::
740: * Programming Tools::
741: * Assembler and Code words::
742: * Threading Words::
743: @end menu
744:
745: @node Notation, Arithmetic, Words, Words
746: @section Notation
747: @cindex notation of glossary entries
748: @cindex format of glossary entries
749: @cindex glossary notation format
750: @cindex word glossary entry format
751:
752: The Forth words are described in this section in the glossary notation
753: that has become a de-facto standard for Forth texts, i.e.,
754:
755: @format
756: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
757: @end format
758: @var{Description}
759:
760: @table @var
761: @item word
762: @cindex case insensitivity
763: The name of the word. BTW, Gforth is case insensitive, so you can
764: type the words in in lower case (However, @pxref{core-idef}).
765:
766: @item Stack effect
767: @cindex stack effect
768: The stack effect is written in the notation @code{@var{before} --
769: @var{after}}, where @var{before} and @var{after} describe the top of
770: stack entries before and after the execution of the word. The rest of
771: the stack is not touched by the word. The top of stack is rightmost,
772: i.e., a stack sequence is written as it is typed in. Note that Gforth
773: uses a separate floating point stack, but a unified stack
774: notation. Also, return stack effects are not shown in @var{stack
775: effect}, but in @var{Description}. The name of a stack item describes
776: the type and/or the function of the item. See below for a discussion of
777: the types.
778:
779: All words have two stack effects: A compile-time stack effect and a
780: run-time stack effect. The compile-time stack-effect of most words is
781: @var{ -- }. If the compile-time stack-effect of a word deviates from
782: this standard behaviour, or the word does other unusual things at
783: compile time, both stack effects are shown; otherwise only the run-time
784: stack effect is shown.
785:
786: @cindex pronounciation of words
787: @item pronunciation
788: How the word is pronounced.
789:
790: @cindex wordset
791: @item wordset
792: The ANS Forth standard is divided into several wordsets. A standard
793: system need not support all of them. So, the fewer wordsets your program
794: uses the more portable it will be in theory. However, we suspect that
795: most ANS Forth systems on personal machines will feature all
796: wordsets. Words that are not defined in the ANS standard have
797: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
798: describes words that will work in future releases of Gforth;
799: @code{gforth-internal} words are more volatile. Environmental query
800: strings are also displayed like words; you can recognize them by the
801: @code{environment} in the wordset field.
802:
803: @item Description
804: A description of the behaviour of the word.
805: @end table
806:
807: @cindex types of stack items
808: @cindex stack item types
809: The type of a stack item is specified by the character(s) the name
810: starts with:
811:
812: @table @code
813: @item f
814: @cindex @code{f}, stack item type
815: Boolean flags, i.e. @code{false} or @code{true}.
816: @item c
817: @cindex @code{c}, stack item type
818: Char
819: @item w
820: @cindex @code{w}, stack item type
821: Cell, can contain an integer or an address
822: @item n
823: @cindex @code{n}, stack item type
824: signed integer
825: @item u
826: @cindex @code{u}, stack item type
827: unsigned integer
828: @item d
829: @cindex @code{d}, stack item type
830: double sized signed integer
831: @item ud
832: @cindex @code{ud}, stack item type
833: double sized unsigned integer
834: @item r
835: @cindex @code{r}, stack item type
836: Float (on the FP stack)
837: @item a_
838: @cindex @code{a_}, stack item type
839: Cell-aligned address
840: @item c_
841: @cindex @code{c_}, stack item type
842: Char-aligned address (note that a Char may have two bytes in Windows NT)
843: @item f_
844: @cindex @code{f_}, stack item type
845: Float-aligned address
846: @item df_
847: @cindex @code{df_}, stack item type
848: Address aligned for IEEE double precision float
849: @item sf_
850: @cindex @code{sf_}, stack item type
851: Address aligned for IEEE single precision float
852: @item xt
853: @cindex @code{xt}, stack item type
854: Execution token, same size as Cell
855: @item wid
856: @cindex @code{wid}, stack item type
857: Wordlist ID, same size as Cell
858: @item f83name
859: @cindex @code{f83name}, stack item type
860: Pointer to a name structure
861: @item "
862: @cindex @code{"}, stack item type
863: string in the input stream (not the stack). The terminating character is
864: a blank by default. If it is not a blank, it is shown in @code{<>}
865: quotes.
866: @end table
867:
868: @node Arithmetic, Stack Manipulation, Notation, Words
869: @section Arithmetic
870: @cindex arithmetic words
871:
872: @cindex division with potentially negative operands
873: Forth arithmetic is not checked, i.e., you will not hear about integer
874: overflow on addition or multiplication, you may hear about division by
875: zero if you are lucky. The operator is written after the operands, but
876: the operands are still in the original order. I.e., the infix @code{2-1}
877: corresponds to @code{2 1 -}. Forth offers a variety of division
878: operators. If you perform division with potentially negative operands,
879: you do not want to use @code{/} or @code{/mod} with its undefined
880: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
881: former, @pxref{Mixed precision}).
882:
883: @menu
884: * Single precision::
885: * Bitwise operations::
886: * Mixed precision:: operations with single and double-cell integers
887: * Double precision:: Double-cell integer arithmetic
888: * Floating Point::
889: @end menu
890:
891: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
892: @subsection Single precision
893: @cindex single precision arithmetic words
894:
895: doc-+
896: doc--
897: doc-*
898: doc-/
899: doc-mod
900: doc-/mod
901: doc-negate
902: doc-abs
903: doc-min
904: doc-max
905:
906: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
907: @subsection Bitwise operations
908: @cindex bitwise operation words
909:
910: doc-and
911: doc-or
912: doc-xor
913: doc-invert
914: doc-2*
915: doc-2/
916:
917: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
918: @subsection Mixed precision
919: @cindex mixed precision arithmetic words
920:
921: doc-m+
922: doc-*/
923: doc-*/mod
924: doc-m*
925: doc-um*
926: doc-m*/
927: doc-um/mod
928: doc-fm/mod
929: doc-sm/rem
930:
931: @node Double precision, Floating Point, Mixed precision, Arithmetic
932: @subsection Double precision
933: @cindex double precision arithmetic words
934:
935: @cindex double-cell numbers, input format
936: @cindex input format for double-cell numbers
937: The outer (aka text) interpreter converts numbers containing a dot into
938: a double precision number. Note that only numbers with the dot as last
939: character are standard-conforming.
940:
941: doc-d+
942: doc-d-
943: doc-dnegate
944: doc-dabs
945: doc-dmin
946: doc-dmax
947:
948: @node Floating Point, , Double precision, Arithmetic
949: @subsection Floating Point
950: @cindex floating point arithmetic words
951:
952: @cindex floating-point numbers, input format
953: @cindex input format for floating-point numbers
954: The format of floating point numbers recognized by the outer (aka text)
955: interpreter is: a signed decimal number, possibly containing a decimal
956: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
957: by a signed integer (the exponent). E.g., @code{1e} is the same as
958: @code{+1.0e+0}. Note that a number without @code{e}
959: is not interpreted as floating-point number, but as double (if the
960: number contains a @code{.}) or single precision integer. Also,
961: conversions between string and floating point numbers always use base
962: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
963: value greater then 14, the @code{E} may be interpreted as digit and the
964: number will be interpreted as integer, unless it has a signed exponent
965: (both @code{+} and @code{-} are allowed as signs).
966:
967: @cindex angles in trigonometric operations
968: @cindex trigonometric operations
969: Angles in floating point operations are given in radians (a full circle
970: has 2 pi radians). Note, that Gforth has a separate floating point
971: stack, but we use the unified notation.
972:
973: @cindex floating-point arithmetic, pitfalls
974: Floating point numbers have a number of unpleasant surprises for the
975: unwary (e.g., floating point addition is not associative) and even a few
976: for the wary. You should not use them unless you know what you are doing
977: or you don't care that the results you get are totally bogus. If you
978: want to learn about the problems of floating point numbers (and how to
979: avoid them), you might start with @cite{David Goldberg, What Every
980: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
981: Computing Surveys 23(1):5@minus{}48, March 1991}.
982:
983: doc-f+
984: doc-f-
985: doc-f*
986: doc-f/
987: doc-fnegate
988: doc-fabs
989: doc-fmax
990: doc-fmin
991: doc-floor
992: doc-fround
993: doc-f**
994: doc-fsqrt
995: doc-fexp
996: doc-fexpm1
997: doc-fln
998: doc-flnp1
999: doc-flog
1000: doc-falog
1001: doc-fsin
1002: doc-fcos
1003: doc-fsincos
1004: doc-ftan
1005: doc-fasin
1006: doc-facos
1007: doc-fatan
1008: doc-fatan2
1009: doc-fsinh
1010: doc-fcosh
1011: doc-ftanh
1012: doc-fasinh
1013: doc-facosh
1014: doc-fatanh
1015:
1016: @node Stack Manipulation, Memory, Arithmetic, Words
1017: @section Stack Manipulation
1018: @cindex stack manipulation words
1019:
1020: @cindex floating-point stack in the standard
1021: Gforth has a data stack (aka parameter stack) for characters, cells,
1022: addresses, and double cells, a floating point stack for floating point
1023: numbers, a return stack for storing the return addresses of colon
1024: definitions and other data, and a locals stack for storing local
1025: variables. Note that while every sane Forth has a separate floating
1026: point stack, this is not strictly required; an ANS Forth system could
1027: theoretically keep floating point numbers on the data stack. As an
1028: additional difficulty, you don't know how many cells a floating point
1029: number takes. It is reportedly possible to write words in a way that
1030: they work also for a unified stack model, but we do not recommend trying
1031: it. Instead, just say that your program has an environmental dependency
1032: on a separate FP stack.
1033:
1034: @cindex return stack and locals
1035: @cindex locals and return stack
1036: Also, a Forth system is allowed to keep the local variables on the
1037: return stack. This is reasonable, as local variables usually eliminate
1038: the need to use the return stack explicitly. So, if you want to produce
1039: a standard complying program and if you are using local variables in a
1040: word, forget about return stack manipulations in that word (see the
1041: standard document for the exact rules).
1042:
1043: @menu
1044: * Data stack::
1045: * Floating point stack::
1046: * Return stack::
1047: * Locals stack::
1048: * Stack pointer manipulation::
1049: @end menu
1050:
1051: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1052: @subsection Data stack
1053: @cindex data stack manipulation words
1054: @cindex stack manipulations words, data stack
1055:
1056: doc-drop
1057: doc-nip
1058: doc-dup
1059: doc-over
1060: doc-tuck
1061: doc-swap
1062: doc-rot
1063: doc--rot
1064: doc-?dup
1065: doc-pick
1066: doc-roll
1067: doc-2drop
1068: doc-2nip
1069: doc-2dup
1070: doc-2over
1071: doc-2tuck
1072: doc-2swap
1073: doc-2rot
1074:
1075: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1076: @subsection Floating point stack
1077: @cindex floating-point stack manipulation words
1078: @cindex stack manipulation words, floating-point stack
1079:
1080: doc-fdrop
1081: doc-fnip
1082: doc-fdup
1083: doc-fover
1084: doc-ftuck
1085: doc-fswap
1086: doc-frot
1087:
1088: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1089: @subsection Return stack
1090: @cindex return stack manipulation words
1091: @cindex stack manipulation words, return stack
1092:
1093: doc->r
1094: doc-r>
1095: doc-r@
1096: doc-rdrop
1097: doc-2>r
1098: doc-2r>
1099: doc-2r@
1100: doc-2rdrop
1101:
1102: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1103: @subsection Locals stack
1104:
1105: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
1106: @subsection Stack pointer manipulation
1107: @cindex stack pointer manipulation words
1108:
1109: doc-sp@
1110: doc-sp!
1111: doc-fp@
1112: doc-fp!
1113: doc-rp@
1114: doc-rp!
1115: doc-lp@
1116: doc-lp!
1117:
1118: @node Memory, Control Structures, Stack Manipulation, Words
1119: @section Memory
1120: @cindex Memory words
1121:
1122: @menu
1123: * Memory Access::
1124: * Address arithmetic::
1125: * Memory Blocks::
1126: @end menu
1127:
1128: @node Memory Access, Address arithmetic, Memory, Memory
1129: @subsection Memory Access
1130: @cindex memory access words
1131:
1132: doc-@
1133: doc-!
1134: doc-+!
1135: doc-c@
1136: doc-c!
1137: doc-2@
1138: doc-2!
1139: doc-f@
1140: doc-f!
1141: doc-sf@
1142: doc-sf!
1143: doc-df@
1144: doc-df!
1145:
1146: @node Address arithmetic, Memory Blocks, Memory Access, Memory
1147: @subsection Address arithmetic
1148: @cindex address arithmetic words
1149:
1150: ANS Forth does not specify the sizes of the data types. Instead, it
1151: offers a number of words for computing sizes and doing address
1152: arithmetic. Basically, address arithmetic is performed in terms of
1153: address units (aus); on most systems the address unit is one byte. Note
1154: that a character may have more than one au, so @code{chars} is no noop
1155: (on systems where it is a noop, it compiles to nothing).
1156:
1157: @cindex alignment of addresses for types
1158: ANS Forth also defines words for aligning addresses for specific
1159: types. Many computers require that accesses to specific data types
1160: must only occur at specific addresses; e.g., that cells may only be
1161: accessed at addresses divisible by 4. Even if a machine allows unaligned
1162: accesses, it can usually perform aligned accesses faster.
1163:
1164: For the performance-conscious: alignment operations are usually only
1165: necessary during the definition of a data structure, not during the
1166: (more frequent) accesses to it.
1167:
1168: ANS Forth defines no words for character-aligning addresses. This is not
1169: an oversight, but reflects the fact that addresses that are not
1170: char-aligned have no use in the standard and therefore will not be
1171: created.
1172:
1173: @cindex @code{CREATE} and alignment
1174: The standard guarantees that addresses returned by @code{CREATE}d words
1175: are cell-aligned; in addition, Gforth guarantees that these addresses
1176: are aligned for all purposes.
1177:
1178: Note that the standard defines a word @code{char}, which has nothing to
1179: do with address arithmetic.
1180:
1181: doc-chars
1182: doc-char+
1183: doc-cells
1184: doc-cell+
1185: doc-cell
1186: doc-align
1187: doc-aligned
1188: doc-floats
1189: doc-float+
1190: doc-float
1191: doc-falign
1192: doc-faligned
1193: doc-sfloats
1194: doc-sfloat+
1195: doc-sfalign
1196: doc-sfaligned
1197: doc-dfloats
1198: doc-dfloat+
1199: doc-dfalign
1200: doc-dfaligned
1201: doc-maxalign
1202: doc-maxaligned
1203: doc-cfalign
1204: doc-cfaligned
1205: doc-address-unit-bits
1206:
1207: @node Memory Blocks, , Address arithmetic, Memory
1208: @subsection Memory Blocks
1209: @cindex memory block words
1210:
1211: doc-move
1212: doc-erase
1213:
1214: While the previous words work on address units, the rest works on
1215: characters.
1216:
1217: doc-cmove
1218: doc-cmove>
1219: doc-fill
1220: doc-blank
1221:
1222: @node Control Structures, Locals, Memory, Words
1223: @section Control Structures
1224: @cindex control structures
1225:
1226: Control structures in Forth cannot be used in interpret state, only in
1227: compile state@footnote{More precisely, they have no interpretation
1228: semantics (@pxref{Interpretation and Compilation Semantics})}, i.e., in
1229: a colon definition. We do not like this limitation, but have not seen a
1230: satisfying way around it yet, although many schemes have been proposed.
1231:
1232: @menu
1233: * Selection::
1234: * Simple Loops::
1235: * Counted Loops::
1236: * Arbitrary control structures::
1237: * Calls and returns::
1238: * Exception Handling::
1239: @end menu
1240:
1241: @node Selection, Simple Loops, Control Structures, Control Structures
1242: @subsection Selection
1243: @cindex selection control structures
1244: @cindex control structures for selection
1245:
1246: @cindex @code{IF} control structure
1247: @example
1248: @var{flag}
1249: IF
1250: @var{code}
1251: ENDIF
1252: @end example
1253: or
1254: @example
1255: @var{flag}
1256: IF
1257: @var{code1}
1258: ELSE
1259: @var{code2}
1260: ENDIF
1261: @end example
1262:
1263: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1264: standard, and @code{ENDIF} is not, although it is quite popular. We
1265: recommend using @code{ENDIF}, because it is less confusing for people
1266: who also know other languages (and is not prone to reinforcing negative
1267: prejudices against Forth in these people). Adding @code{ENDIF} to a
1268: system that only supplies @code{THEN} is simple:
1269: @example
1270: : endif POSTPONE then ; immediate
1271: @end example
1272:
1273: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1274: (adv.)} has the following meanings:
1275: @quotation
1276: ... 2b: following next after in order ... 3d: as a necessary consequence
1277: (if you were there, then you saw them).
1278: @end quotation
1279: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1280: and many other programming languages has the meaning 3d.]
1281:
1282: Gforth also provides the words @code{?dup-if} and @code{?dup-0=-if}, so
1283: you can avoid using @code{?dup}. Using these alternatives is also more
1284: efficient than using @code{?dup}. Definitions in plain standard Forth
1285: for @code{ENDIF}, @code{?DUP-IF} and @code{?DUP-0=-IF} are provided in
1286: @file{compat/control.fs}.
1287:
1288: @cindex @code{CASE} control structure
1289: @example
1290: @var{n}
1291: CASE
1292: @var{n1} OF @var{code1} ENDOF
1293: @var{n2} OF @var{code2} ENDOF
1294: @dots{}
1295: ENDCASE
1296: @end example
1297:
1298: Executes the first @var{codei}, where the @var{ni} is equal to
1299: @var{n}. A default case can be added by simply writing the code after
1300: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1301: but must not consume it.
1302:
1303: @node Simple Loops, Counted Loops, Selection, Control Structures
1304: @subsection Simple Loops
1305: @cindex simple loops
1306: @cindex loops without count
1307:
1308: @cindex @code{WHILE} loop
1309: @example
1310: BEGIN
1311: @var{code1}
1312: @var{flag}
1313: WHILE
1314: @var{code2}
1315: REPEAT
1316: @end example
1317:
1318: @var{code1} is executed and @var{flag} is computed. If it is true,
1319: @var{code2} is executed and the loop is restarted; If @var{flag} is
1320: false, execution continues after the @code{REPEAT}.
1321:
1322: @cindex @code{UNTIL} loop
1323: @example
1324: BEGIN
1325: @var{code}
1326: @var{flag}
1327: UNTIL
1328: @end example
1329:
1330: @var{code} is executed. The loop is restarted if @code{flag} is false.
1331:
1332: @cindex endless loop
1333: @cindex loops, endless
1334: @example
1335: BEGIN
1336: @var{code}
1337: AGAIN
1338: @end example
1339:
1340: This is an endless loop.
1341:
1342: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1343: @subsection Counted Loops
1344: @cindex counted loops
1345: @cindex loops, counted
1346: @cindex @code{DO} loops
1347:
1348: The basic counted loop is:
1349: @example
1350: @var{limit} @var{start}
1351: ?DO
1352: @var{body}
1353: LOOP
1354: @end example
1355:
1356: This performs one iteration for every integer, starting from @var{start}
1357: and up to, but excluding @var{limit}. The counter, aka index, can be
1358: accessed with @code{i}. E.g., the loop
1359: @example
1360: 10 0 ?DO
1361: i .
1362: LOOP
1363: @end example
1364: prints
1365: @example
1366: 0 1 2 3 4 5 6 7 8 9
1367: @end example
1368: The index of the innermost loop can be accessed with @code{i}, the index
1369: of the next loop with @code{j}, and the index of the third loop with
1370: @code{k}.
1371:
1372: doc-i
1373: doc-j
1374: doc-k
1375:
1376: The loop control data are kept on the return stack, so there are some
1377: restrictions on mixing return stack accesses and counted loop
1378: words. E.g., if you put values on the return stack outside the loop, you
1379: cannot read them inside the loop. If you put values on the return stack
1380: within a loop, you have to remove them before the end of the loop and
1381: before accessing the index of the loop.
1382:
1383: There are several variations on the counted loop:
1384:
1385: @code{LEAVE} leaves the innermost counted loop immediately.
1386:
1387: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1388: (and @code{LOOP} iterates until they become equal by wrap-around
1389: arithmetic). This behaviour is usually not what you want. Therefore,
1390: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1391: @code{?DO}), which do not enter the loop if @var{start} is greater than
1392: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1393: unsigned loop parameters.
1394:
1395: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1396: index by @var{n} instead of by 1. The loop is terminated when the border
1397: between @var{limit-1} and @var{limit} is crossed. E.g.:
1398:
1399: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1400:
1401: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1402:
1403: @cindex negative increment for counted loops
1404: @cindex counted loops with negative increment
1405: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1406:
1407: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1408:
1409: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1410:
1411: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1412: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1413: index by @var{u} each iteration. The loop is terminated when the border
1414: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1415: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1416:
1417: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1418:
1419: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1420:
1421: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1422:
1423: Unfortunately, @code{+DO}, @code{U+DO}, @code{-DO}, @code{U-DO} and
1424: @code{-LOOP} are not in the ANS Forth standard. However, an
1425: implementation for these words that uses only standard words is provided
1426: in @file{compat/loops.fs}.
1427:
1428: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1429: the loop, independent of the loop parameters. Do not use @code{DO}, even
1430: if you know that the loop is entered in any case. Such knowledge tends
1431: to become invalid during maintenance of a program, and then the
1432: @code{DO} will make trouble.
1433:
1434: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1435: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1436: return stack so @code{EXIT} can get to its return address.
1437:
1438: @cindex @code{FOR} loops
1439: Another counted loop is
1440: @example
1441: @var{n}
1442: FOR
1443: @var{body}
1444: NEXT
1445: @end example
1446: This is the preferred loop of native code compiler writers who are too
1447: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1448: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1449: and ending with 0. Other Forth systems may behave differently, even if
1450: they support @code{FOR} loops. To avoid problems, don't use @code{FOR}
1451: loops.
1452:
1453: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1454: @subsection Arbitrary control structures
1455: @cindex control structures, user-defined
1456:
1457: @cindex control-flow stack
1458: ANS Forth permits and supports using control structures in a non-nested
1459: way. Information about incomplete control structures is stored on the
1460: control-flow stack. This stack may be implemented on the Forth data
1461: stack, and this is what we have done in Gforth.
1462:
1463: @cindex @code{orig}, control-flow stack item
1464: @cindex @code{dest}, control-flow stack item
1465: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1466: entry represents a backward branch target. A few words are the basis for
1467: building any control structure possible (except control structures that
1468: need storage, like calls, coroutines, and backtracking).
1469:
1470: doc-if
1471: doc-ahead
1472: doc-then
1473: doc-begin
1474: doc-until
1475: doc-again
1476: doc-cs-pick
1477: doc-cs-roll
1478:
1479: On many systems control-flow stack items take one word, in Gforth they
1480: currently take three (this may change in the future). Therefore it is a
1481: really good idea to manipulate the control flow stack with
1482: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1483: words.
1484:
1485: Some standard control structure words are built from these words:
1486:
1487: doc-else
1488: doc-while
1489: doc-repeat
1490:
1491: Gforth adds some more control-structure words:
1492:
1493: doc-endif
1494: doc-?dup-if
1495: doc-?dup-0=-if
1496:
1497: Counted loop words constitute a separate group of words:
1498:
1499: doc-?do
1500: doc-+do
1501: doc-u+do
1502: doc--do
1503: doc-u-do
1504: doc-do
1505: doc-for
1506: doc-loop
1507: doc-+loop
1508: doc--loop
1509: doc-next
1510: doc-leave
1511: doc-?leave
1512: doc-unloop
1513: doc-done
1514:
1515: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1516: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1517: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1518: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1519: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1520: resolved (by using one of the loop-ending words or @code{DONE}).
1521:
1522: Another group of control structure words are
1523:
1524: doc-case
1525: doc-endcase
1526: doc-of
1527: doc-endof
1528:
1529: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1530: @code{cs-roll}.
1531:
1532: @subsubsection Programming Style
1533:
1534: In order to ensure readability we recommend that you do not create
1535: arbitrary control structures directly, but define new control structure
1536: words for the control structure you want and use these words in your
1537: program.
1538:
1539: E.g., instead of writing
1540:
1541: @example
1542: begin
1543: ...
1544: if [ 1 cs-roll ]
1545: ...
1546: again then
1547: @end example
1548:
1549: we recommend defining control structure words, e.g.,
1550:
1551: @example
1552: : while ( dest -- orig dest )
1553: POSTPONE if
1554: 1 cs-roll ; immediate
1555:
1556: : repeat ( orig dest -- )
1557: POSTPONE again
1558: POSTPONE then ; immediate
1559: @end example
1560:
1561: and then using these to create the control structure:
1562:
1563: @example
1564: begin
1565: ...
1566: while
1567: ...
1568: repeat
1569: @end example
1570:
1571: That's much easier to read, isn't it? Of course, @code{REPEAT} and
1572: @code{WHILE} are predefined, so in this example it would not be
1573: necessary to define them.
1574:
1575: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1576: @subsection Calls and returns
1577: @cindex calling a definition
1578: @cindex returning from a definition
1579:
1580: A definition can be called simply be writing the name of the
1581: definition. When the end of the definition is reached, it returns. An
1582: earlier return can be forced using
1583:
1584: doc-exit
1585:
1586: Don't forget to clean up the return stack and @code{UNLOOP} any
1587: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1588: primitive compiled by @code{EXIT} is
1589:
1590: doc-;s
1591:
1592: @node Exception Handling, , Calls and returns, Control Structures
1593: @subsection Exception Handling
1594: @cindex Exceptions
1595:
1596: doc-catch
1597: doc-throw
1598:
1599: @node Locals, Defining Words, Control Structures, Words
1600: @section Locals
1601: @cindex locals
1602:
1603: Local variables can make Forth programming more enjoyable and Forth
1604: programs easier to read. Unfortunately, the locals of ANS Forth are
1605: laden with restrictions. Therefore, we provide not only the ANS Forth
1606: locals wordset, but also our own, more powerful locals wordset (we
1607: implemented the ANS Forth locals wordset through our locals wordset).
1608:
1609: The ideas in this section have also been published in the paper
1610: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1611: at EuroForth '94; it is available at
1612: @*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1613:
1614: @menu
1615: * Gforth locals::
1616: * ANS Forth locals::
1617: @end menu
1618:
1619: @node Gforth locals, ANS Forth locals, Locals, Locals
1620: @subsection Gforth locals
1621: @cindex Gforth locals
1622: @cindex locals, Gforth style
1623:
1624: Locals can be defined with
1625:
1626: @example
1627: @{ local1 local2 ... -- comment @}
1628: @end example
1629: or
1630: @example
1631: @{ local1 local2 ... @}
1632: @end example
1633:
1634: E.g.,
1635: @example
1636: : max @{ n1 n2 -- n3 @}
1637: n1 n2 > if
1638: n1
1639: else
1640: n2
1641: endif ;
1642: @end example
1643:
1644: The similarity of locals definitions with stack comments is intended. A
1645: locals definition often replaces the stack comment of a word. The order
1646: of the locals corresponds to the order in a stack comment and everything
1647: after the @code{--} is really a comment.
1648:
1649: This similarity has one disadvantage: It is too easy to confuse locals
1650: declarations with stack comments, causing bugs and making them hard to
1651: find. However, this problem can be avoided by appropriate coding
1652: conventions: Do not use both notations in the same program. If you do,
1653: they should be distinguished using additional means, e.g. by position.
1654:
1655: @cindex types of locals
1656: @cindex locals types
1657: The name of the local may be preceded by a type specifier, e.g.,
1658: @code{F:} for a floating point value:
1659:
1660: @example
1661: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1662: \ complex multiplication
1663: Ar Br f* Ai Bi f* f-
1664: Ar Bi f* Ai Br f* f+ ;
1665: @end example
1666:
1667: @cindex flavours of locals
1668: @cindex locals flavours
1669: @cindex value-flavoured locals
1670: @cindex variable-flavoured locals
1671: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1672: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1673: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1674: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1675: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1676: produces its address (which becomes invalid when the variable's scope is
1677: left). E.g., the standard word @code{emit} can be defined in terms of
1678: @code{type} like this:
1679:
1680: @example
1681: : emit @{ C^ char* -- @}
1682: char* 1 type ;
1683: @end example
1684:
1685: @cindex default type of locals
1686: @cindex locals, default type
1687: A local without type specifier is a @code{W:} local. Both flavours of
1688: locals are initialized with values from the data or FP stack.
1689:
1690: Currently there is no way to define locals with user-defined data
1691: structures, but we are working on it.
1692:
1693: Gforth allows defining locals everywhere in a colon definition. This
1694: poses the following questions:
1695:
1696: @menu
1697: * Where are locals visible by name?::
1698: * How long do locals live?::
1699: * Programming Style::
1700: * Implementation::
1701: @end menu
1702:
1703: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1704: @subsubsection Where are locals visible by name?
1705: @cindex locals visibility
1706: @cindex visibility of locals
1707: @cindex scope of locals
1708:
1709: Basically, the answer is that locals are visible where you would expect
1710: it in block-structured languages, and sometimes a little longer. If you
1711: want to restrict the scope of a local, enclose its definition in
1712: @code{SCOPE}...@code{ENDSCOPE}.
1713:
1714: doc-scope
1715: doc-endscope
1716:
1717: These words behave like control structure words, so you can use them
1718: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1719: arbitrary ways.
1720:
1721: If you want a more exact answer to the visibility question, here's the
1722: basic principle: A local is visible in all places that can only be
1723: reached through the definition of the local@footnote{In compiler
1724: construction terminology, all places dominated by the definition of the
1725: local.}. In other words, it is not visible in places that can be reached
1726: without going through the definition of the local. E.g., locals defined
1727: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1728: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1729: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1730:
1731: The reasoning behind this solution is: We want to have the locals
1732: visible as long as it is meaningful. The user can always make the
1733: visibility shorter by using explicit scoping. In a place that can
1734: only be reached through the definition of a local, the meaning of a
1735: local name is clear. In other places it is not: How is the local
1736: initialized at the control flow path that does not contain the
1737: definition? Which local is meant, if the same name is defined twice in
1738: two independent control flow paths?
1739:
1740: This should be enough detail for nearly all users, so you can skip the
1741: rest of this section. If you really must know all the gory details and
1742: options, read on.
1743:
1744: In order to implement this rule, the compiler has to know which places
1745: are unreachable. It knows this automatically after @code{AHEAD},
1746: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1747: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1748: compiler that the control flow never reaches that place. If
1749: @code{UNREACHABLE} is not used where it could, the only consequence is
1750: that the visibility of some locals is more limited than the rule above
1751: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1752: lie to the compiler), buggy code will be produced.
1753:
1754: doc-unreachable
1755:
1756: Another problem with this rule is that at @code{BEGIN}, the compiler
1757: does not know which locals will be visible on the incoming
1758: back-edge. All problems discussed in the following are due to this
1759: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1760: loops as examples; the discussion also applies to @code{?DO} and other
1761: loops). Perhaps the most insidious example is:
1762: @example
1763: AHEAD
1764: BEGIN
1765: x
1766: [ 1 CS-ROLL ] THEN
1767: @{ x @}
1768: ...
1769: UNTIL
1770: @end example
1771:
1772: This should be legal according to the visibility rule. The use of
1773: @code{x} can only be reached through the definition; but that appears
1774: textually below the use.
1775:
1776: From this example it is clear that the visibility rules cannot be fully
1777: implemented without major headaches. Our implementation treats common
1778: cases as advertised and the exceptions are treated in a safe way: The
1779: compiler makes a reasonable guess about the locals visible after a
1780: @code{BEGIN}; if it is too pessimistic, the
1781: user will get a spurious error about the local not being defined; if the
1782: compiler is too optimistic, it will notice this later and issue a
1783: warning. In the case above the compiler would complain about @code{x}
1784: being undefined at its use. You can see from the obscure examples in
1785: this section that it takes quite unusual control structures to get the
1786: compiler into trouble, and even then it will often do fine.
1787:
1788: If the @code{BEGIN} is reachable from above, the most optimistic guess
1789: is that all locals visible before the @code{BEGIN} will also be
1790: visible after the @code{BEGIN}. This guess is valid for all loops that
1791: are entered only through the @code{BEGIN}, in particular, for normal
1792: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1793: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1794: compiler. When the branch to the @code{BEGIN} is finally generated by
1795: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1796: warns the user if it was too optimistic:
1797: @example
1798: IF
1799: @{ x @}
1800: BEGIN
1801: \ x ?
1802: [ 1 cs-roll ] THEN
1803: ...
1804: UNTIL
1805: @end example
1806:
1807: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1808: optimistically assumes that it lives until the @code{THEN}. It notices
1809: this difference when it compiles the @code{UNTIL} and issues a
1810: warning. The user can avoid the warning, and make sure that @code{x}
1811: is not used in the wrong area by using explicit scoping:
1812: @example
1813: IF
1814: SCOPE
1815: @{ x @}
1816: ENDSCOPE
1817: BEGIN
1818: [ 1 cs-roll ] THEN
1819: ...
1820: UNTIL
1821: @end example
1822:
1823: Since the guess is optimistic, there will be no spurious error messages
1824: about undefined locals.
1825:
1826: If the @code{BEGIN} is not reachable from above (e.g., after
1827: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1828: optimistic guess, as the locals visible after the @code{BEGIN} may be
1829: defined later. Therefore, the compiler assumes that no locals are
1830: visible after the @code{BEGIN}. However, the user can use
1831: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1832: visible at the BEGIN as at the point where the top control-flow stack
1833: item was created.
1834:
1835: doc-assume-live
1836:
1837: E.g.,
1838: @example
1839: @{ x @}
1840: AHEAD
1841: ASSUME-LIVE
1842: BEGIN
1843: x
1844: [ 1 CS-ROLL ] THEN
1845: ...
1846: UNTIL
1847: @end example
1848:
1849: Other cases where the locals are defined before the @code{BEGIN} can be
1850: handled by inserting an appropriate @code{CS-ROLL} before the
1851: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1852: behind the @code{ASSUME-LIVE}).
1853:
1854: Cases where locals are defined after the @code{BEGIN} (but should be
1855: visible immediately after the @code{BEGIN}) can only be handled by
1856: rearranging the loop. E.g., the ``most insidious'' example above can be
1857: arranged into:
1858: @example
1859: BEGIN
1860: @{ x @}
1861: ... 0=
1862: WHILE
1863: x
1864: REPEAT
1865: @end example
1866:
1867: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1868: @subsubsection How long do locals live?
1869: @cindex locals lifetime
1870: @cindex lifetime of locals
1871:
1872: The right answer for the lifetime question would be: A local lives at
1873: least as long as it can be accessed. For a value-flavoured local this
1874: means: until the end of its visibility. However, a variable-flavoured
1875: local could be accessed through its address far beyond its visibility
1876: scope. Ultimately, this would mean that such locals would have to be
1877: garbage collected. Since this entails un-Forth-like implementation
1878: complexities, I adopted the same cowardly solution as some other
1879: languages (e.g., C): The local lives only as long as it is visible;
1880: afterwards its address is invalid (and programs that access it
1881: afterwards are erroneous).
1882:
1883: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1884: @subsubsection Programming Style
1885: @cindex locals programming style
1886: @cindex programming style, locals
1887:
1888: The freedom to define locals anywhere has the potential to change
1889: programming styles dramatically. In particular, the need to use the
1890: return stack for intermediate storage vanishes. Moreover, all stack
1891: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1892: determined arguments) can be eliminated: If the stack items are in the
1893: wrong order, just write a locals definition for all of them; then
1894: write the items in the order you want.
1895:
1896: This seems a little far-fetched and eliminating stack manipulations is
1897: unlikely to become a conscious programming objective. Still, the number
1898: of stack manipulations will be reduced dramatically if local variables
1899: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1900: a traditional implementation of @code{max}).
1901:
1902: This shows one potential benefit of locals: making Forth programs more
1903: readable. Of course, this benefit will only be realized if the
1904: programmers continue to honour the principle of factoring instead of
1905: using the added latitude to make the words longer.
1906:
1907: @cindex single-assignment style for locals
1908: Using @code{TO} can and should be avoided. Without @code{TO},
1909: every value-flavoured local has only a single assignment and many
1910: advantages of functional languages apply to Forth. I.e., programs are
1911: easier to analyse, to optimize and to read: It is clear from the
1912: definition what the local stands for, it does not turn into something
1913: different later.
1914:
1915: E.g., a definition using @code{TO} might look like this:
1916: @example
1917: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1918: u1 u2 min 0
1919: ?do
1920: addr1 c@@ addr2 c@@ -
1921: ?dup-if
1922: unloop exit
1923: then
1924: addr1 char+ TO addr1
1925: addr2 char+ TO addr2
1926: loop
1927: u1 u2 - ;
1928: @end example
1929: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1930: every loop iteration. @code{strcmp} is a typical example of the
1931: readability problems of using @code{TO}. When you start reading
1932: @code{strcmp}, you think that @code{addr1} refers to the start of the
1933: string. Only near the end of the loop you realize that it is something
1934: else.
1935:
1936: This can be avoided by defining two locals at the start of the loop that
1937: are initialized with the right value for the current iteration.
1938: @example
1939: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1940: addr1 addr2
1941: u1 u2 min 0
1942: ?do @{ s1 s2 @}
1943: s1 c@@ s2 c@@ -
1944: ?dup-if
1945: unloop exit
1946: then
1947: s1 char+ s2 char+
1948: loop
1949: 2drop
1950: u1 u2 - ;
1951: @end example
1952: Here it is clear from the start that @code{s1} has a different value
1953: in every loop iteration.
1954:
1955: @node Implementation, , Programming Style, Gforth locals
1956: @subsubsection Implementation
1957: @cindex locals implementation
1958: @cindex implementation of locals
1959:
1960: @cindex locals stack
1961: Gforth uses an extra locals stack. The most compelling reason for
1962: this is that the return stack is not float-aligned; using an extra stack
1963: also eliminates the problems and restrictions of using the return stack
1964: as locals stack. Like the other stacks, the locals stack grows toward
1965: lower addresses. A few primitives allow an efficient implementation:
1966:
1967: doc-@local#
1968: doc-f@local#
1969: doc-laddr#
1970: doc-lp+!#
1971: doc-lp!
1972: doc->l
1973: doc-f>l
1974:
1975: In addition to these primitives, some specializations of these
1976: primitives for commonly occurring inline arguments are provided for
1977: efficiency reasons, e.g., @code{@@local0} as specialization of
1978: @code{@@local#} for the inline argument 0. The following compiling words
1979: compile the right specialized version, or the general version, as
1980: appropriate:
1981:
1982: doc-compile-@local
1983: doc-compile-f@local
1984: doc-compile-lp+!
1985:
1986: Combinations of conditional branches and @code{lp+!#} like
1987: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
1988: is taken) are provided for efficiency and correctness in loops.
1989:
1990: A special area in the dictionary space is reserved for keeping the
1991: local variable names. @code{@{} switches the dictionary pointer to this
1992: area and @code{@}} switches it back and generates the locals
1993: initializing code. @code{W:} etc.@ are normal defining words. This
1994: special area is cleared at the start of every colon definition.
1995:
1996: @cindex wordlist for defining locals
1997: A special feature of Gforth's dictionary is used to implement the
1998: definition of locals without type specifiers: every wordlist (aka
1999: vocabulary) has its own methods for searching
2000: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
2001: with a special search method: When it is searched for a word, it
2002: actually creates that word using @code{W:}. @code{@{} changes the search
2003: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
2004: and then the wordlist for defining locals without type specifiers.
2005:
2006: The lifetime rules support a stack discipline within a colon
2007: definition: The lifetime of a local is either nested with other locals
2008: lifetimes or it does not overlap them.
2009:
2010: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
2011: pointer manipulation is generated. Between control structure words
2012: locals definitions can push locals onto the locals stack. @code{AGAIN}
2013: is the simplest of the other three control flow words. It has to
2014: restore the locals stack depth of the corresponding @code{BEGIN}
2015: before branching. The code looks like this:
2016: @format
2017: @code{lp+!#} current-locals-size @minus{} dest-locals-size
2018: @code{branch} <begin>
2019: @end format
2020:
2021: @code{UNTIL} is a little more complicated: If it branches back, it
2022: must adjust the stack just like @code{AGAIN}. But if it falls through,
2023: the locals stack must not be changed. The compiler generates the
2024: following code:
2025: @format
2026: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
2027: @end format
2028: The locals stack pointer is only adjusted if the branch is taken.
2029:
2030: @code{THEN} can produce somewhat inefficient code:
2031: @format
2032: @code{lp+!#} current-locals-size @minus{} orig-locals-size
2033: <orig target>:
2034: @code{lp+!#} orig-locals-size @minus{} new-locals-size
2035: @end format
2036: The second @code{lp+!#} adjusts the locals stack pointer from the
2037: level at the @var{orig} point to the level after the @code{THEN}. The
2038: first @code{lp+!#} adjusts the locals stack pointer from the current
2039: level to the level at the orig point, so the complete effect is an
2040: adjustment from the current level to the right level after the
2041: @code{THEN}.
2042:
2043: @cindex locals information on the control-flow stack
2044: @cindex control-flow stack items, locals information
2045: In a conventional Forth implementation a dest control-flow stack entry
2046: is just the target address and an orig entry is just the address to be
2047: patched. Our locals implementation adds a wordlist to every orig or dest
2048: item. It is the list of locals visible (or assumed visible) at the point
2049: described by the entry. Our implementation also adds a tag to identify
2050: the kind of entry, in particular to differentiate between live and dead
2051: (reachable and unreachable) orig entries.
2052:
2053: A few unusual operations have to be performed on locals wordlists:
2054:
2055: doc-common-list
2056: doc-sub-list?
2057: doc-list-size
2058:
2059: Several features of our locals wordlist implementation make these
2060: operations easy to implement: The locals wordlists are organised as
2061: linked lists; the tails of these lists are shared, if the lists
2062: contain some of the same locals; and the address of a name is greater
2063: than the address of the names behind it in the list.
2064:
2065: Another important implementation detail is the variable
2066: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
2067: determine if they can be reached directly or only through the branch
2068: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
2069: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
2070: definition, by @code{BEGIN} and usually by @code{THEN}.
2071:
2072: Counted loops are similar to other loops in most respects, but
2073: @code{LEAVE} requires special attention: It performs basically the same
2074: service as @code{AHEAD}, but it does not create a control-flow stack
2075: entry. Therefore the information has to be stored elsewhere;
2076: traditionally, the information was stored in the target fields of the
2077: branches created by the @code{LEAVE}s, by organizing these fields into a
2078: linked list. Unfortunately, this clever trick does not provide enough
2079: space for storing our extended control flow information. Therefore, we
2080: introduce another stack, the leave stack. It contains the control-flow
2081: stack entries for all unresolved @code{LEAVE}s.
2082:
2083: Local names are kept until the end of the colon definition, even if
2084: they are no longer visible in any control-flow path. In a few cases
2085: this may lead to increased space needs for the locals name area, but
2086: usually less than reclaiming this space would cost in code size.
2087:
2088:
2089: @node ANS Forth locals, , Gforth locals, Locals
2090: @subsection ANS Forth locals
2091: @cindex locals, ANS Forth style
2092:
2093: The ANS Forth locals wordset does not define a syntax for locals, but
2094: words that make it possible to define various syntaxes. One of the
2095: possible syntaxes is a subset of the syntax we used in the Gforth locals
2096: wordset, i.e.:
2097:
2098: @example
2099: @{ local1 local2 ... -- comment @}
2100: @end example
2101: or
2102: @example
2103: @{ local1 local2 ... @}
2104: @end example
2105:
2106: The order of the locals corresponds to the order in a stack comment. The
2107: restrictions are:
2108:
2109: @itemize @bullet
2110: @item
2111: Locals can only be cell-sized values (no type specifiers are allowed).
2112: @item
2113: Locals can be defined only outside control structures.
2114: @item
2115: Locals can interfere with explicit usage of the return stack. For the
2116: exact (and long) rules, see the standard. If you don't use return stack
2117: accessing words in a definition using locals, you will be all right. The
2118: purpose of this rule is to make locals implementation on the return
2119: stack easier.
2120: @item
2121: The whole definition must be in one line.
2122: @end itemize
2123:
2124: Locals defined in this way behave like @code{VALUE}s (@xref{Simple
2125: Defining Words}). I.e., they are initialized from the stack. Using their
2126: name produces their value. Their value can be changed using @code{TO}.
2127:
2128: Since this syntax is supported by Gforth directly, you need not do
2129: anything to use it. If you want to port a program using this syntax to
2130: another ANS Forth system, use @file{compat/anslocal.fs} to implement the
2131: syntax on the other system.
2132:
2133: Note that a syntax shown in the standard, section A.13 looks
2134: similar, but is quite different in having the order of locals
2135: reversed. Beware!
2136:
2137: The ANS Forth locals wordset itself consists of the following word
2138:
2139: doc-(local)
2140:
2141: The ANS Forth locals extension wordset defines a syntax, but it is so
2142: awful that we strongly recommend not to use it. We have implemented this
2143: syntax to make porting to Gforth easy, but do not document it here. The
2144: problem with this syntax is that the locals are defined in an order
2145: reversed with respect to the standard stack comment notation, making
2146: programs harder to read, and easier to misread and miswrite. The only
2147: merit of this syntax is that it is easy to implement using the ANS Forth
2148: locals wordset.
2149:
2150: @node Defining Words, Tokens for Words, Locals, Words
2151: @section Defining Words
2152: @cindex defining words
2153:
2154: @menu
2155: * Simple Defining Words::
2156: * Colon Definitions::
2157: * User-defined Defining Words::
2158: * Supplying names::
2159: * Interpretation and Compilation Semantics::
2160: @end menu
2161:
2162: @node Simple Defining Words, Colon Definitions, Defining Words, Defining Words
2163: @subsection Simple Defining Words
2164: @cindex simple defining words
2165: @cindex defining words, simple
2166:
2167: doc-constant
2168: doc-2constant
2169: doc-fconstant
2170: doc-variable
2171: doc-2variable
2172: doc-fvariable
2173: doc-create
2174: doc-user
2175: doc-value
2176: doc-to
2177: doc-defer
2178: doc-is
2179:
2180: @node Colon Definitions, User-defined Defining Words, Simple Defining Words, Defining Words
2181: @subsection Colon Definitions
2182: @cindex colon definitions
2183:
2184: @example
2185: : name ( ... -- ... )
2186: word1 word2 word3 ;
2187: @end example
2188:
2189: creates a word called @code{name}, that, upon execution, executes
2190: @code{word1 word2 word3}. @code{name} is a @dfn{(colon) definition}.
2191:
2192: The explanation above is somewhat superficial. @xref{Interpretation and
2193: Compilation Semantics} for an in-depth discussion of some of the issues
2194: involved.
2195:
2196: doc-:
2197: doc-;
2198:
2199: @node User-defined Defining Words, Supplying names, Colon Definitions, Defining Words
2200: @subsection User-defined Defining Words
2201: @cindex user-defined defining words
2202: @cindex defining words, user-defined
2203:
2204: You can create new defining words simply by wrapping defining-time code
2205: around existing defining words and putting the sequence in a colon
2206: definition.
2207:
2208: @cindex @code{CREATE} ... @code{DOES>}
2209: If you want the words defined with your defining words to behave
2210: differently from words defined with standard defining words, you can
2211: write your defining word like this:
2212:
2213: @example
2214: : def-word ( "name" -- )
2215: Create @var{code1}
2216: DOES> ( ... -- ... )
2217: @var{code2} ;
2218:
2219: def-word name
2220: @end example
2221:
2222: Technically, this fragment defines a defining word @code{def-word}, and
2223: a word @code{name}; when you execute @code{name}, the address of the
2224: body of @code{name} is put on the data stack and @var{code2} is executed
2225: (the address of the body of @code{name} is the address @code{HERE}
2226: returns immediately after the @code{CREATE}).
2227:
2228: In other words, if you make the following definitions:
2229:
2230: @example
2231: : def-word1 ( "name" -- )
2232: Create @var{code1} ;
2233:
2234: : action1 ( ... -- ... )
2235: @var{code2} ;
2236:
2237: def-word name1
2238: @end example
2239:
2240: Using @code{name1 action1} is equivalent to using @code{name}.
2241:
2242: E.g., you can implement @code{Constant} in this way:
2243:
2244: @example
2245: : constant ( w "name" -- )
2246: create ,
2247: DOES> ( -- w )
2248: @@ ;
2249: @end example
2250:
2251: When you create a constant with @code{5 constant five}, first a new word
2252: @code{five} is created, then the value 5 is laid down in the body of
2253: @code{five} with @code{,}. When @code{five} is invoked, the address of
2254: the body is put on the stack, and @code{@@} retrieves the value 5.
2255:
2256: @cindex stack effect of @code{DOES>}-parts
2257: @cindex @code{DOES>}-parts, stack effect
2258: In the example above the stack comment after the @code{DOES>} specifies
2259: the stack effect of the defined words, not the stack effect of the
2260: following code (the following code expects the address of the body on
2261: the top of stack, which is not reflected in the stack comment). This is
2262: the convention that I use and recommend (it clashes a bit with using
2263: locals declarations for stack effect specification, though).
2264:
2265: @subsubsection Applications of @code{CREATE..DOES>}
2266: @cindex @code{CREATE} ... @code{DOES>}, applications
2267:
2268: You may wonder how to use this feature. Here are some usage patterns:
2269:
2270: @cindex factoring similar colon definitions
2271: When you see a sequence of code occurring several times, and you can
2272: identify a meaning, you will factor it out as a colon definition. When
2273: you see similar colon definitions, you can factor them using
2274: @code{CREATE..DOES>}. E.g., an assembler usually defines several words
2275: that look very similar:
2276: @example
2277: : ori, ( reg-target reg-source n -- )
2278: 0 asm-reg-reg-imm ;
2279: : andi, ( reg-target reg-source n -- )
2280: 1 asm-reg-reg-imm ;
2281: @end example
2282:
2283: This could be factored with:
2284: @example
2285: : reg-reg-imm ( op-code -- )
2286: create ,
2287: DOES> ( reg-target reg-source n -- )
2288: @@ asm-reg-reg-imm ;
2289:
2290: 0 reg-reg-imm ori,
2291: 1 reg-reg-imm andi,
2292: @end example
2293:
2294: @cindex currying
2295: Another view of @code{CREATE..DOES>} is to consider it as a crude way to
2296: supply a part of the parameters for a word (known as @dfn{currying} in
2297: the functional language community). E.g., @code{+} needs two
2298: parameters. Creating versions of @code{+} with one parameter fixed can
2299: be done like this:
2300: @example
2301: : curry+ ( n1 -- )
2302: create ,
2303: DOES> ( n2 -- n1+n2 )
2304: @@ + ;
2305:
2306: 3 curry+ 3+
2307: -2 curry+ 2-
2308: @end example
2309:
2310: @subsubsection The gory details of @code{CREATE..DOES>}
2311: @cindex @code{CREATE} ... @code{DOES>}, details
2312:
2313: doc-does>
2314:
2315: @cindex @code{DOES>} in a separate definition
2316: This means that you need not use @code{CREATE} and @code{DOES>} in the
2317: same definition; E.g., you can put the @code{DOES>}-part in a separate
2318: definition. This allows us to, e.g., select among different DOES>-parts:
2319: @example
2320: : does1
2321: DOES> ( ... -- ... )
2322: ... ;
2323:
2324: : does2
2325: DOES> ( ... -- ... )
2326: ... ;
2327:
2328: : def-word ( ... -- ... )
2329: create ...
2330: IF
2331: does1
2332: ELSE
2333: does2
2334: ENDIF ;
2335: @end example
2336:
2337: @cindex @code{DOES>} in interpretation state
2338: In a standard program you can apply a @code{DOES>}-part only if the last
2339: word was defined with @code{CREATE}. In Gforth, the @code{DOES>}-part
2340: will override the behaviour of the last word defined in any case. In a
2341: standard program, you can use @code{DOES>} only in a colon
2342: definition. In Gforth, you can also use it in interpretation state, in a
2343: kind of one-shot mode:
2344: @example
2345: CREATE name ( ... -- ... )
2346: @var{initialization}
2347: DOES>
2348: @var{code} ;
2349: @end example
2350: This is equivalent to the standard
2351: @example
2352: :noname
2353: DOES>
2354: @var{code} ;
2355: CREATE name EXECUTE ( ... -- ... )
2356: @var{initialization}
2357: @end example
2358:
2359: You can get the address of the body of a word with
2360:
2361: doc->body
2362:
2363: @node Supplying names, Interpretation and Compilation Semantics, User-defined Defining Words, Defining Words
2364: @subsection Supplying names for the defined words
2365: @cindex names for defined words
2366: @cindex defining words, name parameter
2367:
2368: @cindex defining words, name given in a string
2369: By default, defining words take the names for the defined words from the
2370: input stream. Sometimes you want to supply the name from a string. You
2371: can do this with
2372:
2373: doc-nextname
2374:
2375: E.g.,
2376:
2377: @example
2378: s" foo" nextname create
2379: @end example
2380: is equivalent to
2381: @example
2382: create foo
2383: @end example
2384:
2385: @cindex defining words without name
2386: Sometimes you want to define a word without a name. You can do this with
2387:
2388: doc-noname
2389:
2390: @cindex execution token of last defined word
2391: To make any use of the newly defined word, you need its execution
2392: token. You can get it with
2393:
2394: doc-lastxt
2395:
2396: E.g., you can initialize a deferred word with an anonymous colon
2397: definition:
2398: @example
2399: Defer deferred
2400: noname : ( ... -- ... )
2401: ... ;
2402: lastxt IS deferred
2403: @end example
2404:
2405: @code{lastxt} also works when the last word was not defined as
2406: @code{noname}.
2407:
2408: The standard has also recognized the need for anonymous words and
2409: provides
2410:
2411: doc-:noname
2412:
2413: This leaves the execution token for the word on the stack after the
2414: closing @code{;}. You can rewrite the last example with @code{:noname}:
2415: @example
2416: Defer deferred
2417: :noname ( ... -- ... )
2418: ... ;
2419: IS deferred
2420: @end example
2421:
2422: @node Interpretation and Compilation Semantics, , Supplying names, Defining Words
2423: @subsection Interpretation and Compilation Semantics
2424: @cindex semantics, interpretation and compilation
2425:
2426: @cindex interpretation semantics
2427: The @dfn{interpretation semantics} of a word are what the text
2428: interpreter does when it encounters the word in interpret state. It also
2429: appears in some other contexts, e.g., the execution token returned by
2430: @code{' @var{word}} identifies the interpretation semantics of
2431: @var{word} (in other words, @code{' @var{word} execute} is equivalent to
2432: interpret-state text interpretation of @code{@var{word}}).
2433:
2434: @cindex compilation semantics
2435: The @dfn{compilation semantics} of a word are what the text interpreter
2436: does when it encounters the word in compile state. It also appears in
2437: other contexts, e.g, @code{POSTPONE @var{word}} compiles@footnote{In
2438: standard terminology, ``appends to the current definition''.} the
2439: compilation semantics of @var{word}.
2440:
2441: @cindex execution semantics
2442: The standard also talks about @dfn{execution semantics}. They are used
2443: only for defining the interpretation and compilation semantics of many
2444: words. By default, the interpretation semantics of a word are to
2445: @code{execute} its execution semantics, and the compilation semantics of
2446: a word are to @code{compile,} its execution semantics.@footnote{In
2447: standard terminology: The default interpretation semantics are its
2448: execution semantics; the default compilation semantics are to append its
2449: execution semantics to the execution semantics of the current
2450: definition.}
2451:
2452: @cindex immediate words
2453: You can change the compilation semantics into @code{execute}ing the
2454: execution semantics with
2455:
2456: doc-immediate
2457:
2458: @cindex compile-only words
2459: You can remove the interpretation semantics of a word with
2460:
2461: doc-compile-only
2462: doc-restrict
2463:
2464: Note that ticking (@code{'}) compile-only words gives an error
2465: (``Interpreting a compile-only word'').
2466:
2467: Gforth also allows you to define words with arbitrary combinations of
2468: interpretation and compilation semantics.
2469:
2470: doc-interpret/compile:
2471:
2472: This feature was introduced for implementing @code{TO} and @code{S"}. I
2473: recommend that you do not define such words, as cute as they may be:
2474: they make it hard to get at both parts of the word in some contexts.
2475: E.g., assume you want to get an execution token for the compilation
2476: part. Instead, define two words, one that embodies the interpretation
2477: part, and one that embodies the compilation part.
2478:
2479: There is, however, a potentially useful application of this feature:
2480: Providing differing implementations for the default semantics. While
2481: this introduces redundancy and is therefore usually a bad idea, a
2482: performance improvement may be worth the trouble. E.g., consider the
2483: word @code{foobar}:
2484:
2485: @example
2486: : foobar
2487: foo bar ;
2488: @end example
2489:
2490: Let us assume that @code{foobar} is called so frequently that the
2491: calling overhead would take a significant amount of the run-time. We can
2492: optimize it with @code{interpret/compile:}:
2493:
2494: @example
2495: :noname
2496: foo bar ;
2497: :noname
2498: POSTPONE foo POSTPONE bar ;
2499: interpret/compile: foobar
2500: @end example
2501:
2502: This definition has the same interpretation semantics and essentially
2503: the same compilation semantics as the simple definition of
2504: @code{foobar}, but the implementation of the compilation semantics is
2505: more efficient with respect to run-time.
2506:
2507: @cindex state-smart words are a bad idea
2508: Some people try to use state-smart words to emulate the feature provided
2509: by @code{interpret/compile:} (words are state-smart if they check
2510: @code{STATE} during execution). E.g., they would try to code
2511: @code{foobar} like this:
2512:
2513: @example
2514: : foobar
2515: STATE @@
2516: IF ( compilation state )
2517: POSTPONE foo POSTPONE bar
2518: ELSE
2519: foo bar
2520: ENDIF ; immediate
2521: @end example
2522:
2523: While this works if @code{foobar} is processed only by the text
2524: interpreter, it does not work in other contexts (like @code{'} or
2525: @code{POSTPONE}). E.g., @code{' foobar} will produce an execution token
2526: for a state-smart word, not for the interpretation semantics of the
2527: original @code{foobar}; when you execute this execution token (directly
2528: with @code{EXECUTE} or indirectly through @code{COMPILE,}) in compile
2529: state, the result will not be what you expected (i.e., it will not
2530: perform @code{foo bar}). State-smart words are a bad idea. Simply don't
2531: write them!
2532:
2533: @cindex defining words with arbitrary semantics combinations
2534: It is also possible to write defining words that define words with
2535: arbitrary combinations of interpretation and compilation semantics (or,
2536: preferably, arbitrary combinations of implementations of the default
2537: semantics). In general, this looks like:
2538:
2539: @example
2540: : def-word
2541: create-interpret/compile
2542: @var{code1}
2543: interpretation>
2544: @var{code2}
2545: <interpretation
2546: compilation>
2547: @var{code3}
2548: <compilation ;
2549: @end example
2550:
2551: For a @var{word} defined with @code{def-word}, the interpretation
2552: semantics are to push the address of the body of @var{word} and perform
2553: @var{code2}, and the compilation semantics are to push the address of
2554: the body of @var{word} and perform @var{code3}. E.g., @code{constant}
2555: can also be defined like this:
2556:
2557: @example
2558: : constant ( n "name" -- )
2559: create-interpret/compile
2560: ,
2561: interpretation> ( -- n )
2562: @@
2563: <interpretation
2564: compilation> ( compilation. -- ; run-time. -- n )
2565: @@ postpone literal
2566: <compilation ;
2567: @end example
2568:
2569: doc-create-interpret/compile
2570: doc-interpretation>
2571: doc-<interpretation
2572: doc-compilation>
2573: doc-<compilation
2574:
2575: Note that words defined with @code{interpret/compile:} and
2576: @code{create-interpret/compile} have an extended header structure that
2577: differs from other words; however, unless you try to access them with
2578: plain address arithmetic, you should not notice this. Words for
2579: accessing the header structure usually know how to deal with this; e.g.,
2580: @code{' word >body} also gives you the body of a word created with
2581: @code{create-interpret/compile}.
2582:
2583: @node Tokens for Words, Wordlists, Defining Words, Words
2584: @section Tokens for Words
2585: @cindex tokens for words
2586:
2587: This chapter describes the creation and use of tokens that represent
2588: words on the stack (and in data space).
2589:
2590: Named words have interpretation and compilation semantics. Unnamed words
2591: just have execution semantics.
2592:
2593: @cindex execution token
2594: An @dfn{execution token} represents the execution semantics of an
2595: unnamed word. An execution token occupies one cell. As explained in
2596: section @ref{Supplying names}, the execution token of the last words
2597: defined can be produced with
2598:
2599: short-lastxt
2600:
2601: You can perform the semantics represented by an execution token with
2602: doc-execute
2603: You can compile the word with
2604: doc-compile,
2605:
2606: @cindex code field address
2607: @cindex CFA
2608: In Gforth, the abstract data type @emph{execution token} is implemented
2609: as CFA (code field address).
2610:
2611: The interpretation semantics of a named word are also represented by an
2612: execution token. You can get it with
2613:
2614: doc-[']
2615: doc-'
2616:
2617: For literals, you use @code{'} in interpreted code and @code{[']} in
2618: compiled code. Gforth's @code{'} and @code{[']} behave somewhat unusual
2619: by complaining about compile-only words. To get an execution token for a
2620: compiling word @var{X}, use @code{COMP' @var{X} drop} or @code{[COMP']
2621: @var{X} drop}.
2622:
2623: @cindex compilation token
2624: The compilation semantics are represented by a @dfn{compilation token}
2625: consisting of two cells: @var{w xt}. The top cell @var{xt} is an
2626: execution token. The compilation semantics represented by the
2627: compilation token can be performed with @code{execute}, which consumes
2628: the whole compilation token, with an additional stack effect determined
2629: by the represented compilation semantics.
2630:
2631: doc-[comp']
2632: doc-comp'
2633:
2634: You can compile the compilation semantics with @code{postpone,}. I.e.,
2635: @code{COMP' @var{word} POSTPONE,} is equivalent to @code{POSTPONE
2636: @var{word}}.
2637:
2638: doc-postpone,
2639:
2640: At present, the @var{w} part of a compilation token is an execution
2641: token, and the @var{xt} part represents either @code{execute} or
2642: @code{compile,}. However, don't rely on that knowledge, unless necessary;
2643: we may introduce unusual compilation tokens in the future (e.g.,
2644: compilation tokens representing the compilation semantics of literals).
2645:
2646: @cindex name token
2647: @cindex name field address
2648: @cindex NFA
2649: Named words are also represented by the @dfn{name token}. The abstract
2650: data type @emph{name token} is implemented as NFA (name field address).
2651:
2652: doc-find-name
2653: doc-name>int
2654: doc-name?int
2655: doc-name>comp
2656: doc-name>string
2657:
2658: @node Wordlists, Files, Tokens for Words, Words
2659: @section Wordlists
2660:
2661: @node Files, Blocks, Wordlists, Words
2662: @section Files
2663:
2664: @node Blocks, Other I/O, Files, Words
2665: @section Blocks
2666:
2667: @node Other I/O, Programming Tools, Blocks, Words
2668: @section Other I/O
2669:
2670: @node Programming Tools, Assembler and Code words, Other I/O, Words
2671: @section Programming Tools
2672: @cindex programming tools
2673:
2674: @menu
2675: * Debugging:: Simple and quick.
2676: * Assertions:: Making your programs self-checking.
2677: @end menu
2678:
2679: @node Debugging, Assertions, Programming Tools, Programming Tools
2680: @subsection Debugging
2681: @cindex debugging
2682:
2683: The simple debugging aids provided in @file{debugging.fs}
2684: are meant to support a different style of debugging than the
2685: tracing/stepping debuggers used in languages with long turn-around
2686: times.
2687:
2688: A much better (faster) way in fast-compiling languages is to add
2689: printing code at well-selected places, let the program run, look at
2690: the output, see where things went wrong, add more printing code, etc.,
2691: until the bug is found.
2692:
2693: The word @code{~~} is easy to insert. It just prints debugging
2694: information (by default the source location and the stack contents). It
2695: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
2696: query-replace them with nothing). The deferred words
2697: @code{printdebugdata} and @code{printdebugline} control the output of
2698: @code{~~}. The default source location output format works well with
2699: Emacs' compilation mode, so you can step through the program at the
2700: source level using @kbd{C-x `} (the advantage over a stepping debugger
2701: is that you can step in any direction and you know where the crash has
2702: happened or where the strange data has occurred).
2703:
2704: Note that the default actions clobber the contents of the pictured
2705: numeric output string, so you should not use @code{~~}, e.g., between
2706: @code{<#} and @code{#>}.
2707:
2708: doc-~~
2709: doc-printdebugdata
2710: doc-printdebugline
2711:
2712: @node Assertions, , Debugging, Programming Tools
2713: @subsection Assertions
2714: @cindex assertions
2715:
2716: It is a good idea to make your programs self-checking, in particular, if
2717: you use an assumption (e.g., that a certain field of a data structure is
2718: never zero) that may become wrong during maintenance. Gforth supports
2719: assertions for this purpose. They are used like this:
2720:
2721: @example
2722: assert( @var{flag} )
2723: @end example
2724:
2725: The code between @code{assert(} and @code{)} should compute a flag, that
2726: should be true if everything is alright and false otherwise. It should
2727: not change anything else on the stack. The overall stack effect of the
2728: assertion is @code{( -- )}. E.g.
2729:
2730: @example
2731: assert( 1 1 + 2 = ) \ what we learn in school
2732: assert( dup 0<> ) \ assert that the top of stack is not zero
2733: assert( false ) \ this code should not be reached
2734: @end example
2735:
2736: The need for assertions is different at different times. During
2737: debugging, we want more checking, in production we sometimes care more
2738: for speed. Therefore, assertions can be turned off, i.e., the assertion
2739: becomes a comment. Depending on the importance of an assertion and the
2740: time it takes to check it, you may want to turn off some assertions and
2741: keep others turned on. Gforth provides several levels of assertions for
2742: this purpose:
2743:
2744: doc-assert0(
2745: doc-assert1(
2746: doc-assert2(
2747: doc-assert3(
2748: doc-assert(
2749: doc-)
2750:
2751: @code{Assert(} is the same as @code{assert1(}. The variable
2752: @code{assert-level} specifies the highest assertions that are turned
2753: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2754: @code{assert1(} assertions perform checking, while @code{assert2(} and
2755: @code{assert3(} assertions are treated as comments.
2756:
2757: Note that the @code{assert-level} is evaluated at compile-time, not at
2758: run-time. I.e., you cannot turn assertions on or off at run-time, you
2759: have to set the @code{assert-level} appropriately before compiling a
2760: piece of code. You can compile several pieces of code at several
2761: @code{assert-level}s (e.g., a trusted library at level 1 and newly
2762: written code at level 3).
2763:
2764: doc-assert-level
2765:
2766: If an assertion fails, a message compatible with Emacs' compilation mode
2767: is produced and the execution is aborted (currently with @code{ABORT"}.
2768: If there is interest, we will introduce a special throw code. But if you
2769: intend to @code{catch} a specific condition, using @code{throw} is
2770: probably more appropriate than an assertion).
2771:
2772: @node Assembler and Code words, Threading Words, Programming Tools, Words
2773: @section Assembler and Code words
2774: @cindex assembler
2775: @cindex code words
2776:
2777: Gforth provides some words for defining primitives (words written in
2778: machine code), and for defining the the machine-code equivalent of
2779: @code{DOES>}-based defining words. However, the machine-independent
2780: nature of Gforth poses a few problems: First of all, Gforth runs on
2781: several architectures, so it can provide no standard assembler. What's
2782: worse is that the register allocation not only depends on the processor,
2783: but also on the @code{gcc} version and options used.
2784:
2785: The words that Gforth offers encapsulate some system dependences (e.g., the
2786: header structure), so a system-independent assembler may be used in
2787: Gforth. If you do not have an assembler, you can compile machine code
2788: directly with @code{,} and @code{c,}.
2789:
2790: doc-assembler
2791: doc-code
2792: doc-end-code
2793: doc-;code
2794: doc-flush-icache
2795:
2796: If @code{flush-icache} does not work correctly, @code{code} words
2797: etc. will not work (reliably), either.
2798:
2799: These words are rarely used. Therefore they reside in @code{code.fs},
2800: which is usually not loaded (except @code{flush-icache}, which is always
2801: present). You can load them with @code{require code.fs}.
2802:
2803: @cindex registers of the inner interpreter
2804: In the assembly code you will want to refer to the inner interpreter's
2805: registers (e.g., the data stack pointer) and you may want to use other
2806: registers for temporary storage. Unfortunately, the register allocation
2807: is installation-dependent.
2808:
2809: The easiest solution is to use explicit register declarations
2810: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
2811: GNU C Manual}) for all of the inner interpreter's registers: You have to
2812: compile Gforth with @code{-DFORCE_REG} (configure option
2813: @code{--enable-force-reg}) and the appropriate declarations must be
2814: present in the @code{machine.h} file (see @code{mips.h} for an example;
2815: you can find a full list of all declarable register symbols with
2816: @code{grep register engine.c}). If you give explicit registers to all
2817: variables that are declared at the beginning of @code{engine()}, you
2818: should be able to use the other caller-saved registers for temporary
2819: storage. Alternatively, you can use the @code{gcc} option
2820: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
2821: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
2822: (however, this restriction on register allocation may slow Gforth
2823: significantly).
2824:
2825: If this solution is not viable (e.g., because @code{gcc} does not allow
2826: you to explicitly declare all the registers you need), you have to find
2827: out by looking at the code where the inner interpreter's registers
2828: reside and which registers can be used for temporary storage. You can
2829: get an assembly listing of the engine's code with @code{make engine.s}.
2830:
2831: In any case, it is good practice to abstract your assembly code from the
2832: actual register allocation. E.g., if the data stack pointer resides in
2833: register @code{$17}, create an alias for this register called @code{sp},
2834: and use that in your assembly code.
2835:
2836: @cindex code words, portable
2837: Another option for implementing normal and defining words efficiently
2838: is: adding the wanted functionality to the source of Gforth. For normal
2839: words you just have to edit @file{primitives} (@pxref{Automatic
2840: Generation}), defining words (equivalent to @code{;CODE} words, for fast
2841: defined words) may require changes in @file{engine.c}, @file{kernal.fs},
2842: @file{prims2x.fs}, and possibly @file{cross.fs}.
2843:
2844:
2845: @node Threading Words, , Assembler and Code words, Words
2846: @section Threading Words
2847: @cindex threading words
2848:
2849: @cindex code address
2850: These words provide access to code addresses and other threading stuff
2851: in Gforth (and, possibly, other interpretive Forths). It more or less
2852: abstracts away the differences between direct and indirect threading
2853: (and, for direct threading, the machine dependences). However, at
2854: present this wordset is still incomplete. It is also pretty low-level;
2855: some day it will hopefully be made unnecessary by an internals wordset
2856: that abstracts implementation details away completely.
2857:
2858: doc->code-address
2859: doc->does-code
2860: doc-code-address!
2861: doc-does-code!
2862: doc-does-handler!
2863: doc-/does-handler
2864:
2865: The code addresses produced by various defining words are produced by
2866: the following words:
2867:
2868: doc-docol:
2869: doc-docon:
2870: doc-dovar:
2871: doc-douser:
2872: doc-dodefer:
2873: doc-dofield:
2874:
2875: You can recognize words defined by a @code{CREATE}...@code{DOES>} word
2876: with @code{>DOES-CODE}. If the word was defined in that way, the value
2877: returned is different from 0 and identifies the @code{DOES>} used by the
2878: defining word.
2879:
2880: @node Tools, ANS conformance, Words, Top
2881: @chapter Tools
2882:
2883: @menu
2884: * ANS Report:: Report the words used, sorted by wordset.
2885: @end menu
2886:
2887: See also @ref{Emacs and Gforth}.
2888:
2889: @node ANS Report, , Tools, Tools
2890: @section @file{ans-report.fs}: Report the words used, sorted by wordset
2891: @cindex @file{ans-report.fs}
2892: @cindex report the words used in your program
2893: @cindex words used in your program
2894:
2895: If you want to label a Forth program as ANS Forth Program, you must
2896: document which wordsets the program uses; for extension wordsets, it is
2897: helpful to list the words the program requires from these wordsets
2898: (because Forth systems are allowed to provide only some words of them).
2899:
2900: The @file{ans-report.fs} tool makes it easy for you to determine which
2901: words from which wordset and which non-ANS words your application
2902: uses. You simply have to include @file{ans-report.fs} before loading the
2903: program you want to check. After loading your program, you can get the
2904: report with @code{print-ans-report}. A typical use is to run this as
2905: batch job like this:
2906: @example
2907: gforth ans-report.fs myprog.fs -e "print-ans-report bye"
2908: @end example
2909:
2910: The output looks like this (for @file{compat/control.fs}):
2911: @example
2912: The program uses the following words
2913: from CORE :
2914: : POSTPONE THEN ; immediate ?dup IF 0=
2915: from BLOCK-EXT :
2916: \
2917: from FILE :
2918: (
2919: @end example
2920:
2921: @subsection Caveats
2922:
2923: Note that @file{ans-report.fs} just checks which words are used, not whether
2924: they are used in an ANS Forth conforming way!
2925:
2926: Some words are defined in several wordsets in the
2927: standard. @file{ans-report.fs} reports them for only one of the
2928: wordsets, and not necessarily the one you expect. It depends on usage
2929: which wordset is the right one to specify. E.g., if you only use the
2930: compilation semantics of @code{S"}, it is a Core word; if you also use
2931: its interpretation semantics, it is a File word.
2932:
2933: @c ******************************************************************
2934: @node ANS conformance, Model, Tools, Top
2935: @chapter ANS conformance
2936: @cindex ANS conformance of Gforth
2937:
2938: To the best of our knowledge, Gforth is an
2939:
2940: ANS Forth System
2941: @itemize @bullet
2942: @item providing the Core Extensions word set
2943: @item providing the Block word set
2944: @item providing the Block Extensions word set
2945: @item providing the Double-Number word set
2946: @item providing the Double-Number Extensions word set
2947: @item providing the Exception word set
2948: @item providing the Exception Extensions word set
2949: @item providing the Facility word set
2950: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2951: @item providing the File Access word set
2952: @item providing the File Access Extensions word set
2953: @item providing the Floating-Point word set
2954: @item providing the Floating-Point Extensions word set
2955: @item providing the Locals word set
2956: @item providing the Locals Extensions word set
2957: @item providing the Memory-Allocation word set
2958: @item providing the Memory-Allocation Extensions word set (that one's easy)
2959: @item providing the Programming-Tools word set
2960: @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
2961: @item providing the Search-Order word set
2962: @item providing the Search-Order Extensions word set
2963: @item providing the String word set
2964: @item providing the String Extensions word set (another easy one)
2965: @end itemize
2966:
2967: @cindex system documentation
2968: In addition, ANS Forth systems are required to document certain
2969: implementation choices. This chapter tries to meet these
2970: requirements. In many cases it gives a way to ask the system for the
2971: information instead of providing the information directly, in
2972: particular, if the information depends on the processor, the operating
2973: system or the installation options chosen, or if they are likely to
2974: change during the maintenance of Gforth.
2975:
2976: @comment The framework for the rest has been taken from pfe.
2977:
2978: @menu
2979: * The Core Words::
2980: * The optional Block word set::
2981: * The optional Double Number word set::
2982: * The optional Exception word set::
2983: * The optional Facility word set::
2984: * The optional File-Access word set::
2985: * The optional Floating-Point word set::
2986: * The optional Locals word set::
2987: * The optional Memory-Allocation word set::
2988: * The optional Programming-Tools word set::
2989: * The optional Search-Order word set::
2990: @end menu
2991:
2992:
2993: @c =====================================================================
2994: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2995: @comment node-name, next, previous, up
2996: @section The Core Words
2997: @c =====================================================================
2998: @cindex core words, system documentation
2999: @cindex system documentation, core words
3000:
3001: @menu
3002: * core-idef:: Implementation Defined Options
3003: * core-ambcond:: Ambiguous Conditions
3004: * core-other:: Other System Documentation
3005: @end menu
3006:
3007: @c ---------------------------------------------------------------------
3008: @node core-idef, core-ambcond, The Core Words, The Core Words
3009: @subsection Implementation Defined Options
3010: @c ---------------------------------------------------------------------
3011: @cindex core words, implementation-defined options
3012: @cindex implementation-defined options, core words
3013:
3014:
3015: @table @i
3016: @item (Cell) aligned addresses:
3017: @cindex cell-aligned addresses
3018: @cindex aligned addresses
3019: processor-dependent. Gforth's alignment words perform natural alignment
3020: (e.g., an address aligned for a datum of size 8 is divisible by
3021: 8). Unaligned accesses usually result in a @code{-23 THROW}.
3022:
3023: @item @code{EMIT} and non-graphic characters:
3024: @cindex @code{EMIT} and non-graphic characters
3025: @cindex non-graphic characters and @code{EMIT}
3026: The character is output using the C library function (actually, macro)
3027: @code{putc}.
3028:
3029: @item character editing of @code{ACCEPT} and @code{EXPECT}:
3030: @cindex character editing of @code{ACCEPT} and @code{EXPECT}
3031: @cindex editing in @code{ACCEPT} and @code{EXPECT}
3032: @cindex @code{ACCEPT}, editing
3033: @cindex @code{EXPECT}, editing
3034: This is modeled on the GNU readline library (@pxref{Readline
3035: Interaction, , Command Line Editing, readline, The GNU Readline
3036: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
3037: producing a full word completion every time you type it (instead of
3038: producing the common prefix of all completions).
3039:
3040: @item character set:
3041: @cindex character set
3042: The character set of your computer and display device. Gforth is
3043: 8-bit-clean (but some other component in your system may make trouble).
3044:
3045: @item Character-aligned address requirements:
3046: @cindex character-aligned address requirements
3047: installation-dependent. Currently a character is represented by a C
3048: @code{unsigned char}; in the future we might switch to @code{wchar_t}
3049: (Comments on that requested).
3050:
3051: @item character-set extensions and matching of names:
3052: @cindex character-set extensions and matching of names
3053: @cindex case sensitivity for name lookup
3054: @cindex name lookup, case sensitivity
3055: @cindex locale and case sensitivity
3056: Any character except the ASCII NUL charcter can be used in a
3057: name. Matching is case-insensitive (except in @code{TABLE}s). The
3058: matching is performed using the C function @code{strncasecmp}, whose
3059: function is probably influenced by the locale. E.g., the @code{C} locale
3060: does not know about accents and umlauts, so they are matched
3061: case-sensitively in that locale. For portability reasons it is best to
3062: write programs such that they work in the @code{C} locale. Then one can
3063: use libraries written by a Polish programmer (who might use words
3064: containing ISO Latin-2 encoded characters) and by a French programmer
3065: (ISO Latin-1) in the same program (of course, @code{WORDS} will produce
3066: funny results for some of the words (which ones, depends on the font you
3067: are using)). Also, the locale you prefer may not be available in other
3068: operating systems. Hopefully, Unicode will solve these problems one day.
3069:
3070: @item conditions under which control characters match a space delimiter:
3071: @cindex space delimiters
3072: @cindex control characters as delimiters
3073: If @code{WORD} is called with the space character as a delimiter, all
3074: white-space characters (as identified by the C macro @code{isspace()})
3075: are delimiters. @code{PARSE}, on the other hand, treats space like other
3076: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
3077: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
3078: interpreter (aka text interpreter) by default, treats all white-space
3079: characters as delimiters.
3080:
3081: @item format of the control flow stack:
3082: @cindex control flow stack, format
3083: The data stack is used as control flow stack. The size of a control flow
3084: stack item in cells is given by the constant @code{cs-item-size}. At the
3085: time of this writing, an item consists of a (pointer to a) locals list
3086: (third), an address in the code (second), and a tag for identifying the
3087: item (TOS). The following tags are used: @code{defstart},
3088: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
3089: @code{scopestart}.
3090:
3091: @item conversion of digits > 35
3092: @cindex digits > 35
3093: The characters @code{[\]^_'} are the digits with the decimal value
3094: 36@minus{}41. There is no way to input many of the larger digits.
3095:
3096: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
3097: @cindex @code{EXPECT}, display after end of input
3098: @cindex @code{ACCEPT}, display after end of input
3099: The cursor is moved to the end of the entered string. If the input is
3100: terminated using the @kbd{Return} key, a space is typed.
3101:
3102: @item exception abort sequence of @code{ABORT"}:
3103: @cindex exception abort sequence of @code{ABORT"}
3104: @cindex @code{ABORT"}, exception abort sequence
3105: The error string is stored into the variable @code{"error} and a
3106: @code{-2 throw} is performed.
3107:
3108: @item input line terminator:
3109: @cindex input line terminator
3110: @cindex line terminator on input
3111: @cindex newline charcter on input
3112: For interactive input, @kbd{C-m} (CR) and @kbd{C-j} (LF) terminate
3113: lines. One of these characters is typically produced when you type the
3114: @kbd{Enter} or @kbd{Return} key.
3115:
3116: @item maximum size of a counted string:
3117: @cindex maximum size of a counted string
3118: @cindex counted string, maximum size
3119: @code{s" /counted-string" environment? drop .}. Currently 255 characters
3120: on all ports, but this may change.
3121:
3122: @item maximum size of a parsed string:
3123: @cindex maximum size of a parsed string
3124: @cindex parsed string, maximum size
3125: Given by the constant @code{/line}. Currently 255 characters.
3126:
3127: @item maximum size of a definition name, in characters:
3128: @cindex maximum size of a definition name, in characters
3129: @cindex name, maximum length
3130: 31
3131:
3132: @item maximum string length for @code{ENVIRONMENT?}, in characters:
3133: @cindex maximum string length for @code{ENVIRONMENT?}, in characters
3134: @cindex @code{ENVIRONMENT?} string length, maximum
3135: 31
3136:
3137: @item method of selecting the user input device:
3138: @cindex user input device, method of selecting
3139: The user input device is the standard input. There is currently no way to
3140: change it from within Gforth. However, the input can typically be
3141: redirected in the command line that starts Gforth.
3142:
3143: @item method of selecting the user output device:
3144: @cindex user output device, method of selecting
3145: @code{EMIT} and @code{TYPE} output to the file-id stored in the value
3146: @code{outfile-id} (@code{stdout} by default). Gforth uses buffered
3147: output, so output on a terminal does not become visible before the next
3148: newline or buffer overflow. Output on non-terminals is invisible until
3149: the buffer overflows.
3150:
3151: @item methods of dictionary compilation:
3152: What are we expected to document here?
3153:
3154: @item number of bits in one address unit:
3155: @cindex number of bits in one address unit
3156: @cindex address unit, size in bits
3157: @code{s" address-units-bits" environment? drop .}. 8 in all current
3158: ports.
3159:
3160: @item number representation and arithmetic:
3161: @cindex number representation and arithmetic
3162: Processor-dependent. Binary two's complement on all current ports.
3163:
3164: @item ranges for integer types:
3165: @cindex ranges for integer types
3166: @cindex integer types, ranges
3167: Installation-dependent. Make environmental queries for @code{MAX-N},
3168: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
3169: unsigned (and positive) types is 0. The lower bound for signed types on
3170: two's complement and one's complement machines machines can be computed
3171: by adding 1 to the upper bound.
3172:
3173: @item read-only data space regions:
3174: @cindex read-only data space regions
3175: @cindex data-space, read-only regions
3176: The whole Forth data space is writable.
3177:
3178: @item size of buffer at @code{WORD}:
3179: @cindex size of buffer at @code{WORD}
3180: @cindex @code{WORD} buffer size
3181: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
3182: shared with the pictured numeric output string. If overwriting
3183: @code{PAD} is acceptable, it is as large as the remaining dictionary
3184: space, although only as much can be sensibly used as fits in a counted
3185: string.
3186:
3187: @item size of one cell in address units:
3188: @cindex cell size
3189: @code{1 cells .}.
3190:
3191: @item size of one character in address units:
3192: @cindex char size
3193: @code{1 chars .}. 1 on all current ports.
3194:
3195: @item size of the keyboard terminal buffer:
3196: @cindex size of the keyboard terminal buffer
3197: @cindex terminal buffer, size
3198: Varies. You can determine the size at a specific time using @code{lp@@
3199: tib - .}. It is shared with the locals stack and TIBs of files that
3200: include the current file. You can change the amount of space for TIBs
3201: and locals stack at Gforth startup with the command line option
3202: @code{-l}.
3203:
3204: @item size of the pictured numeric output buffer:
3205: @cindex size of the pictured numeric output buffer
3206: @cindex pictured numeric output buffer, size
3207: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
3208: shared with @code{WORD}.
3209:
3210: @item size of the scratch area returned by @code{PAD}:
3211: @cindex size of the scratch area returned by @code{PAD}
3212: @cindex @code{PAD} size
3213: The remainder of dictionary space. @code{unused pad here - - .}.
3214:
3215: @item system case-sensitivity characteristics:
3216: @cindex case-sensitivity characteristics
3217: Dictionary searches are case insensitive (except in
3218: @code{TABLE}s). However, as explained above under @i{character-set
3219: extensions}, the matching for non-ASCII characters is determined by the
3220: locale you are using. In the default @code{C} locale all non-ASCII
3221: characters are matched case-sensitively.
3222:
3223: @item system prompt:
3224: @cindex system prompt
3225: @cindex prompt
3226: @code{ ok} in interpret state, @code{ compiled} in compile state.
3227:
3228: @item division rounding:
3229: @cindex division rounding
3230: installation dependent. @code{s" floored" environment? drop .}. We leave
3231: the choice to @code{gcc} (what to use for @code{/}) and to you (whether
3232: to use @code{fm/mod}, @code{sm/rem} or simply @code{/}).
3233:
3234: @item values of @code{STATE} when true:
3235: @cindex @code{STATE} values
3236: -1.
3237:
3238: @item values returned after arithmetic overflow:
3239: On two's complement machines, arithmetic is performed modulo
3240: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
3241: arithmetic (with appropriate mapping for signed types). Division by zero
3242: typically results in a @code{-55 throw} (Floating-point unidentified
3243: fault), although a @code{-10 throw} (divide by zero) would be more
3244: appropriate.
3245:
3246: @item whether the current definition can be found after @t{DOES>}:
3247: @cindex @t{DOES>}, visibility of current definition
3248: No.
3249:
3250: @end table
3251:
3252: @c ---------------------------------------------------------------------
3253: @node core-ambcond, core-other, core-idef, The Core Words
3254: @subsection Ambiguous conditions
3255: @c ---------------------------------------------------------------------
3256: @cindex core words, ambiguous conditions
3257: @cindex ambiguous conditions, core words
3258:
3259: @table @i
3260:
3261: @item a name is neither a word nor a number:
3262: @cindex name not found
3263: @cindex Undefined word
3264: @code{-13 throw} (Undefined word). Actually, @code{-13 bounce}, which
3265: preserves the data and FP stack, so you don't lose more work than
3266: necessary.
3267:
3268: @item a definition name exceeds the maximum length allowed:
3269: @cindex Word name too long
3270: @code{-19 throw} (Word name too long)
3271:
3272: @item addressing a region not inside the various data spaces of the forth system:
3273: @cindex Invalid memory address
3274: The stacks, code space and name space are accessible. Machine code space is
3275: typically readable. Accessing other addresses gives results dependent on
3276: the operating system. On decent systems: @code{-9 throw} (Invalid memory
3277: address).
3278:
3279: @item argument type incompatible with parameter:
3280: @cindex Argument type mismatch
3281: This is usually not caught. Some words perform checks, e.g., the control
3282: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
3283: mismatch).
3284:
3285: @item attempting to obtain the execution token of a word with undefined execution semantics:
3286: @cindex Interpreting a compile-only word, for @code{'} etc.
3287: @cindex execution token of words with undefined execution semantics
3288: @code{-14 throw} (Interpreting a compile-only word). In some cases, you
3289: get an execution token for @code{compile-only-error} (which performs a
3290: @code{-14 throw} when executed).
3291:
3292: @item dividing by zero:
3293: @cindex dividing by zero
3294: @cindex floating point unidentified fault, integer division
3295: @cindex divide by zero
3296: typically results in a @code{-55 throw} (floating point unidentified
3297: fault), although a @code{-10 throw} (divide by zero) would be more
3298: appropriate.
3299:
3300: @item insufficient data stack or return stack space:
3301: @cindex insufficient data stack or return stack space
3302: @cindex stack overflow
3303: @cindex Address alignment exception, stack overflow
3304: @cindex Invalid memory address, stack overflow
3305: Depending on the operating system, the installation, and the invocation
3306: of Gforth, this is either checked by the memory management hardware, or
3307: it is not checked. If it is checked, you typically get a @code{-9 throw}
3308: (Invalid memory address) as soon as the overflow happens. If it is not
3309: check, overflows typically result in mysterious illegal memory accesses,
3310: producing @code{-9 throw} (Invalid memory address) or @code{-23 throw}
3311: (Address alignment exception); they might also destroy the internal data
3312: structure of @code{ALLOCATE} and friends, resulting in various errors in
3313: these words.
3314:
3315: @item insufficient space for loop control parameters:
3316: @cindex insufficient space for loop control parameters
3317: like other return stack overflows.
3318:
3319: @item insufficient space in the dictionary:
3320: @cindex insufficient space in the dictionary
3321: @cindex dictionary overflow
3322: Depending on the operating system, the installation, and the invocation
3323: of Gforth, this is either checked by the memory management hardware, or
3324: it is not checked. Similar results as stack overflows. However,
3325: typically the error appears at a different place when one inserts or
3326: removes code. Also, the @code{THROW} does not relieve the situation (it
3327: does for stack overflows).
3328:
3329: @item interpreting a word with undefined interpretation semantics:
3330: @cindex interpreting a word with undefined interpretation semantics
3331: @cindex Interpreting a compile-only word
3332: For some words, we have defined interpretation semantics. For the
3333: others: @code{-14 throw} (Interpreting a compile-only word).
3334:
3335: @item modifying the contents of the input buffer or a string literal:
3336: @cindex modifying the contents of the input buffer or a string literal
3337: These are located in writable memory and can be modified.
3338:
3339: @item overflow of the pictured numeric output string:
3340: @cindex overflow of the pictured numeric output string
3341: @cindex pictured numeric output string, overflow
3342: Not checked. Runs into the dictionary and destroys it (at least,
3343: partially).
3344:
3345: @item parsed string overflow:
3346: @cindex parsed string overflow
3347: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
3348:
3349: @item producing a result out of range:
3350: @cindex result out of range
3351: On two's complement machines, arithmetic is performed modulo
3352: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
3353: arithmetic (with appropriate mapping for signed types). Division by zero
3354: typically results in a @code{-55 throw} (floatingpoint unidentified
3355: fault), although a @code{-10 throw} (divide by zero) would be more
3356: appropriate. @code{convert} and @code{>number} currently overflow
3357: silently.
3358:
3359: @item reading from an empty data or return stack:
3360: @cindex stack empty
3361: @cindex stack underflow
3362: The data stack is checked by the outer (aka text) interpreter after
3363: every word executed. If it has underflowed, a @code{-4 throw} (Stack
3364: underflow) is performed. Apart from that, stacks may be checked or not,
3365: depending on operating system, installation, and invocation. The
3366: consequences of stack underflows are similar to the consequences of
3367: stack overflows. Note that even if the system uses checking (through the
3368: MMU), your program may have to underflow by a significant number of
3369: stack items to trigger the reaction (the reason for this is that the
3370: MMU, and therefore the checking, works with a page-size granularity).
3371:
3372: @item unexpected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
3373: @cindex unexpected end of the input buffer
3374: @cindex zero-length string as a name
3375: @cindex Attempt to use zero-length string as a name
3376: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
3377: use zero-length string as a name). Words like @code{'} probably will not
3378: find what they search. Note that it is possible to create zero-length
3379: names with @code{nextname} (should it not?).
3380:
3381: @item @code{>IN} greater than input buffer:
3382: @cindex @code{>IN} greater than input buffer
3383: The next invocation of a parsing word returns a string with length 0.
3384:
3385: @item @code{RECURSE} appears after @code{DOES>}:
3386: @cindex @code{RECURSE} appears after @code{DOES>}
3387: Compiles a recursive call to the defining word, not to the defined word.
3388:
3389: @item argument input source different than current input source for @code{RESTORE-INPUT}:
3390: @cindex argument input source different than current input source for @code{RESTORE-INPUT}
3391: @cindex Argument type mismatch, @code{RESTORE-INPUT}
3392: @cindex @code{RESTORE-INPUT}, Argument type mismatch
3393: @code{-12 THROW}. Note that, once an input file is closed (e.g., because
3394: the end of the file was reached), its source-id may be
3395: reused. Therefore, restoring an input source specification referencing a
3396: closed file may lead to unpredictable results instead of a @code{-12
3397: THROW}.
3398:
3399: In the future, Gforth may be able to restore input source specifications
3400: from other than the current input source.
3401:
3402: @item data space containing definitions gets de-allocated:
3403: @cindex data space containing definitions gets de-allocated
3404: Deallocation with @code{allot} is not checked. This typically results in
3405: memory access faults or execution of illegal instructions.
3406:
3407: @item data space read/write with incorrect alignment:
3408: @cindex data space read/write with incorrect alignment
3409: @cindex alignment faults
3410: @cindex Address alignment exception
3411: Processor-dependent. Typically results in a @code{-23 throw} (Address
3412: alignment exception). Under Linux on a 486 or later processor with
3413: alignment turned on, incorrect alignment results in a @code{-9 throw}
3414: (Invalid memory address). There are reportedly some processors with
3415: alignment restrictions that do not report them.
3416:
3417: @item data space pointer not properly aligned, @code{,}, @code{C,}:
3418: @cindex data space pointer not properly aligned, @code{,}, @code{C,}
3419: Like other alignment errors.
3420:
3421: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
3422: Like other stack underflows.
3423:
3424: @item loop control parameters not available:
3425: @cindex loop control parameters not available
3426: Not checked. The counted loop words simply assume that the top of return
3427: stack items are loop control parameters and behave accordingly.
3428:
3429: @item most recent definition does not have a name (@code{IMMEDIATE}):
3430: @cindex most recent definition does not have a name (@code{IMMEDIATE})
3431: @cindex last word was headerless
3432: @code{abort" last word was headerless"}.
3433:
3434: @item name not defined by @code{VALUE} used by @code{TO}:
3435: @cindex name not defined by @code{VALUE} used by @code{TO}
3436: @cindex @code{TO} on non-@code{VALUE}s
3437: @cindex Invalid name argument, @code{TO}
3438: @code{-32 throw} (Invalid name argument) (unless name is a local or was
3439: defined by @code{CONSTANT}; in the latter case it just changes the constant).
3440:
3441: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
3442: @cindex name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]})
3443: @cindex Undefined word, @code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}
3444: @code{-13 throw} (Undefined word)
3445:
3446: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
3447: @cindex parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN})
3448: Gforth behaves as if they were of the same type. I.e., you can predict
3449: the behaviour by interpreting all parameters as, e.g., signed.
3450:
3451: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
3452: @cindex @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}
3453: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} performs the
3454: compilation semantics of @code{TO}.
3455:
3456: @item String longer than a counted string returned by @code{WORD}:
3457: @cindex String longer than a counted string returned by @code{WORD}
3458: @cindex @code{WORD}, string overflow
3459: Not checked. The string will be ok, but the count will, of course,
3460: contain only the least significant bits of the length.
3461:
3462: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
3463: @cindex @code{LSHIFT}, large shift counts
3464: @cindex @code{RSHIFT}, large shift counts
3465: Processor-dependent. Typical behaviours are returning 0 and using only
3466: the low bits of the shift count.
3467:
3468: @item word not defined via @code{CREATE}:
3469: @cindex @code{>BODY} of non-@code{CREATE}d words
3470: @code{>BODY} produces the PFA of the word no matter how it was defined.
3471:
3472: @cindex @code{DOES>} of non-@code{CREATE}d words
3473: @code{DOES>} changes the execution semantics of the last defined word no
3474: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
3475: @code{CREATE , DOES>}.
3476:
3477: @item words improperly used outside @code{<#} and @code{#>}:
3478: Not checked. As usual, you can expect memory faults.
3479:
3480: @end table
3481:
3482:
3483: @c ---------------------------------------------------------------------
3484: @node core-other, , core-ambcond, The Core Words
3485: @subsection Other system documentation
3486: @c ---------------------------------------------------------------------
3487: @cindex other system documentation, core words
3488: @cindex core words, other system documentation
3489:
3490: @table @i
3491: @item nonstandard words using @code{PAD}:
3492: @cindex @code{PAD} use by nonstandard words
3493: None.
3494:
3495: @item operator's terminal facilities available:
3496: @cindex operator's terminal facilities available
3497: After processing the command line, Gforth goes into interactive mode,
3498: and you can give commands to Gforth interactively. The actual facilities
3499: available depend on how you invoke Gforth.
3500:
3501: @item program data space available:
3502: @cindex program data space available
3503: @cindex data space available
3504: @code{UNUSED .} gives the remaining dictionary space. The total
3505: dictionary space can be specified with the @code{-m} switch
3506: (@pxref{Invoking Gforth}) when Gforth starts up.
3507:
3508: @item return stack space available:
3509: @cindex return stack space available
3510: You can compute the total return stack space in cells with
3511: @code{s" RETURN-STACK-CELLS" environment? drop .}. You can specify it at
3512: startup time with the @code{-r} switch (@pxref{Invoking Gforth}).
3513:
3514: @item stack space available:
3515: @cindex stack space available
3516: You can compute the total data stack space in cells with
3517: @code{s" STACK-CELLS" environment? drop .}. You can specify it at
3518: startup time with the @code{-d} switch (@pxref{Invoking Gforth}).
3519:
3520: @item system dictionary space required, in address units:
3521: @cindex system dictionary space required, in address units
3522: Type @code{here forthstart - .} after startup. At the time of this
3523: writing, this gives 80080 (bytes) on a 32-bit system.
3524: @end table
3525:
3526:
3527: @c =====================================================================
3528: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
3529: @section The optional Block word set
3530: @c =====================================================================
3531: @cindex system documentation, block words
3532: @cindex block words, system documentation
3533:
3534: @menu
3535: * block-idef:: Implementation Defined Options
3536: * block-ambcond:: Ambiguous Conditions
3537: * block-other:: Other System Documentation
3538: @end menu
3539:
3540:
3541: @c ---------------------------------------------------------------------
3542: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
3543: @subsection Implementation Defined Options
3544: @c ---------------------------------------------------------------------
3545: @cindex implementation-defined options, block words
3546: @cindex block words, implementation-defined options
3547:
3548: @table @i
3549: @item the format for display by @code{LIST}:
3550: @cindex @code{LIST} display format
3551: First the screen number is displayed, then 16 lines of 64 characters,
3552: each line preceded by the line number.
3553:
3554: @item the length of a line affected by @code{\}:
3555: @cindex length of a line affected by @code{\}
3556: @cindex @code{\}, line length in blocks
3557: 64 characters.
3558: @end table
3559:
3560:
3561: @c ---------------------------------------------------------------------
3562: @node block-ambcond, block-other, block-idef, The optional Block word set
3563: @subsection Ambiguous conditions
3564: @c ---------------------------------------------------------------------
3565: @cindex block words, ambiguous conditions
3566: @cindex ambiguous conditions, block words
3567:
3568: @table @i
3569: @item correct block read was not possible:
3570: @cindex block read not possible
3571: Typically results in a @code{throw} of some OS-derived value (between
3572: -512 and -2048). If the blocks file was just not long enough, blanks are
3573: supplied for the missing portion.
3574:
3575: @item I/O exception in block transfer:
3576: @cindex I/O exception in block transfer
3577: @cindex block transfer, I/O exception
3578: Typically results in a @code{throw} of some OS-derived value (between
3579: -512 and -2048).
3580:
3581: @item invalid block number:
3582: @cindex invalid block number
3583: @cindex block number invalid
3584: @code{-35 throw} (Invalid block number)
3585:
3586: @item a program directly alters the contents of @code{BLK}:
3587: @cindex @code{BLK}, altering @code{BLK}
3588: The input stream is switched to that other block, at the same
3589: position. If the storing to @code{BLK} happens when interpreting
3590: non-block input, the system will get quite confused when the block ends.
3591:
3592: @item no current block buffer for @code{UPDATE}:
3593: @cindex @code{UPDATE}, no current block buffer
3594: @code{UPDATE} has no effect.
3595:
3596: @end table
3597:
3598: @c ---------------------------------------------------------------------
3599: @node block-other, , block-ambcond, The optional Block word set
3600: @subsection Other system documentation
3601: @c ---------------------------------------------------------------------
3602: @cindex other system documentation, block words
3603: @cindex block words, other system documentation
3604:
3605: @table @i
3606: @item any restrictions a multiprogramming system places on the use of buffer addresses:
3607: No restrictions (yet).
3608:
3609: @item the number of blocks available for source and data:
3610: depends on your disk space.
3611:
3612: @end table
3613:
3614:
3615: @c =====================================================================
3616: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
3617: @section The optional Double Number word set
3618: @c =====================================================================
3619: @cindex system documentation, double words
3620: @cindex double words, system documentation
3621:
3622: @menu
3623: * double-ambcond:: Ambiguous Conditions
3624: @end menu
3625:
3626:
3627: @c ---------------------------------------------------------------------
3628: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
3629: @subsection Ambiguous conditions
3630: @c ---------------------------------------------------------------------
3631: @cindex double words, ambiguous conditions
3632: @cindex ambiguous conditions, double words
3633:
3634: @table @i
3635: @item @var{d} outside of range of @var{n} in @code{D>S}:
3636: @cindex @code{D>S}, @var{d} out of range of @var{n}
3637: The least significant cell of @var{d} is produced.
3638:
3639: @end table
3640:
3641:
3642: @c =====================================================================
3643: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
3644: @section The optional Exception word set
3645: @c =====================================================================
3646: @cindex system documentation, exception words
3647: @cindex exception words, system documentation
3648:
3649: @menu
3650: * exception-idef:: Implementation Defined Options
3651: @end menu
3652:
3653:
3654: @c ---------------------------------------------------------------------
3655: @node exception-idef, , The optional Exception word set, The optional Exception word set
3656: @subsection Implementation Defined Options
3657: @c ---------------------------------------------------------------------
3658: @cindex implementation-defined options, exception words
3659: @cindex exception words, implementation-defined options
3660:
3661: @table @i
3662: @item @code{THROW}-codes used in the system:
3663: @cindex @code{THROW}-codes used in the system
3664: The codes -256@minus{}-511 are used for reporting signals. The mapping
3665: from OS signal numbers to throw codes is -256@minus{}@var{signal}. The
3666: codes -512@minus{}-2047 are used for OS errors (for file and memory
3667: allocation operations). The mapping from OS error numbers to throw codes
3668: is -512@minus{}@code{errno}. One side effect of this mapping is that
3669: undefined OS errors produce a message with a strange number; e.g.,
3670: @code{-1000 THROW} results in @code{Unknown error 488} on my system.
3671: @end table
3672:
3673: @c =====================================================================
3674: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
3675: @section The optional Facility word set
3676: @c =====================================================================
3677: @cindex system documentation, facility words
3678: @cindex facility words, system documentation
3679:
3680: @menu
3681: * facility-idef:: Implementation Defined Options
3682: * facility-ambcond:: Ambiguous Conditions
3683: @end menu
3684:
3685:
3686: @c ---------------------------------------------------------------------
3687: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
3688: @subsection Implementation Defined Options
3689: @c ---------------------------------------------------------------------
3690: @cindex implementation-defined options, facility words
3691: @cindex facility words, implementation-defined options
3692:
3693: @table @i
3694: @item encoding of keyboard events (@code{EKEY}):
3695: @cindex keyboard events, encoding in @code{EKEY}
3696: @cindex @code{EKEY}, encoding of keyboard events
3697: Not yet implemented.
3698:
3699: @item duration of a system clock tick:
3700: @cindex duration of a system clock tick
3701: @cindex clock tick duration
3702: System dependent. With respect to @code{MS}, the time is specified in
3703: microseconds. How well the OS and the hardware implement this, is
3704: another question.
3705:
3706: @item repeatability to be expected from the execution of @code{MS}:
3707: @cindex repeatability to be expected from the execution of @code{MS}
3708: @cindex @code{MS}, repeatability to be expected
3709: System dependent. On Unix, a lot depends on load. If the system is
3710: lightly loaded, and the delay is short enough that Gforth does not get
3711: swapped out, the performance should be acceptable. Under MS-DOS and
3712: other single-tasking systems, it should be good.
3713:
3714: @end table
3715:
3716:
3717: @c ---------------------------------------------------------------------
3718: @node facility-ambcond, , facility-idef, The optional Facility word set
3719: @subsection Ambiguous conditions
3720: @c ---------------------------------------------------------------------
3721: @cindex facility words, ambiguous conditions
3722: @cindex ambiguous conditions, facility words
3723:
3724: @table @i
3725: @item @code{AT-XY} can't be performed on user output device:
3726: @cindex @code{AT-XY} can't be performed on user output device
3727: Largely terminal dependent. No range checks are done on the arguments.
3728: No errors are reported. You may see some garbage appearing, you may see
3729: simply nothing happen.
3730:
3731: @end table
3732:
3733:
3734: @c =====================================================================
3735: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
3736: @section The optional File-Access word set
3737: @c =====================================================================
3738: @cindex system documentation, file words
3739: @cindex file words, system documentation
3740:
3741: @menu
3742: * file-idef:: Implementation Defined Options
3743: * file-ambcond:: Ambiguous Conditions
3744: @end menu
3745:
3746: @c ---------------------------------------------------------------------
3747: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
3748: @subsection Implementation Defined Options
3749: @c ---------------------------------------------------------------------
3750: @cindex implementation-defined options, file words
3751: @cindex file words, implementation-defined options
3752:
3753: @table @i
3754: @item file access methods used:
3755: @cindex file access methods used
3756: @code{R/O}, @code{R/W} and @code{BIN} work as you would
3757: expect. @code{W/O} translates into the C file opening mode @code{w} (or
3758: @code{wb}): The file is cleared, if it exists, and created, if it does
3759: not (with both @code{open-file} and @code{create-file}). Under Unix
3760: @code{create-file} creates a file with 666 permissions modified by your
3761: umask.
3762:
3763: @item file exceptions:
3764: @cindex file exceptions
3765: The file words do not raise exceptions (except, perhaps, memory access
3766: faults when you pass illegal addresses or file-ids).
3767:
3768: @item file line terminator:
3769: @cindex file line terminator
3770: System-dependent. Gforth uses C's newline character as line
3771: terminator. What the actual character code(s) of this are is
3772: system-dependent.
3773:
3774: @item file name format:
3775: @cindex file name format
3776: System dependent. Gforth just uses the file name format of your OS.
3777:
3778: @item information returned by @code{FILE-STATUS}:
3779: @cindex @code{FILE-STATUS}, returned information
3780: @code{FILE-STATUS} returns the most powerful file access mode allowed
3781: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
3782: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
3783: along with the returned mode.
3784:
3785: @item input file state after an exception when including source:
3786: @cindex exception when including source
3787: All files that are left via the exception are closed.
3788:
3789: @item @var{ior} values and meaning:
3790: @cindex @var{ior} values and meaning
3791: The @var{ior}s returned by the file and memory allocation words are
3792: intended as throw codes. They typically are in the range
3793: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3794: @var{ior}s is -512@minus{}@var{errno}.
3795:
3796: @item maximum depth of file input nesting:
3797: @cindex maximum depth of file input nesting
3798: @cindex file input nesting, maximum depth
3799: limited by the amount of return stack, locals/TIB stack, and the number
3800: of open files available. This should not give you troubles.
3801:
3802: @item maximum size of input line:
3803: @cindex maximum size of input line
3804: @cindex input line size, maximum
3805: @code{/line}. Currently 255.
3806:
3807: @item methods of mapping block ranges to files:
3808: @cindex mapping block ranges to files
3809: @cindex files containing blocks
3810: @cindex blocks in files
3811: By default, blocks are accessed in the file @file{blocks.fb} in the
3812: current working directory. The file can be switched with @code{USE}.
3813:
3814: @item number of string buffers provided by @code{S"}:
3815: @cindex @code{S"}, number of string buffers
3816: 1
3817:
3818: @item size of string buffer used by @code{S"}:
3819: @cindex @code{S"}, size of string buffer
3820: @code{/line}. currently 255.
3821:
3822: @end table
3823:
3824: @c ---------------------------------------------------------------------
3825: @node file-ambcond, , file-idef, The optional File-Access word set
3826: @subsection Ambiguous conditions
3827: @c ---------------------------------------------------------------------
3828: @cindex file words, ambiguous conditions
3829: @cindex ambiguous conditions, file words
3830:
3831: @table @i
3832: @item attempting to position a file outside its boundaries:
3833: @cindex @code{REPOSITION-FILE}, outside the file's boundaries
3834: @code{REPOSITION-FILE} is performed as usual: Afterwards,
3835: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
3836:
3837: @item attempting to read from file positions not yet written:
3838: @cindex reading from file positions not yet written
3839: End-of-file, i.e., zero characters are read and no error is reported.
3840:
3841: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
3842: @cindex @code{INCLUDE-FILE}, @var{file-id} is invalid
3843: An appropriate exception may be thrown, but a memory fault or other
3844: problem is more probable.
3845:
3846: @item I/O exception reading or closing @var{file-id} (@code{INCLUDE-FILE}, @code{INCLUDED}):
3847: @cindex @code{INCLUDE-FILE}, I/O exception reading or closing @var{file-id}
3848: @cindex @code{INCLUDED}, I/O exception reading or closing @var{file-id}
3849: The @var{ior} produced by the operation, that discovered the problem, is
3850: thrown.
3851:
3852: @item named file cannot be opened (@code{INCLUDED}):
3853: @cindex @code{INCLUDED}, named file cannot be opened
3854: The @var{ior} produced by @code{open-file} is thrown.
3855:
3856: @item requesting an unmapped block number:
3857: @cindex unmapped block numbers
3858: There are no unmapped legal block numbers. On some operating systems,
3859: writing a block with a large number may overflow the file system and
3860: have an error message as consequence.
3861:
3862: @item using @code{source-id} when @code{blk} is non-zero:
3863: @cindex @code{SOURCE-ID}, behaviour when @code{BLK} is non-zero
3864: @code{source-id} performs its function. Typically it will give the id of
3865: the source which loaded the block. (Better ideas?)
3866:
3867: @end table
3868:
3869:
3870: @c =====================================================================
3871: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
3872: @section The optional Floating-Point word set
3873: @c =====================================================================
3874: @cindex system documentation, floating-point words
3875: @cindex floating-point words, system documentation
3876:
3877: @menu
3878: * floating-idef:: Implementation Defined Options
3879: * floating-ambcond:: Ambiguous Conditions
3880: @end menu
3881:
3882:
3883: @c ---------------------------------------------------------------------
3884: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
3885: @subsection Implementation Defined Options
3886: @c ---------------------------------------------------------------------
3887: @cindex implementation-defined options, floating-point words
3888: @cindex floating-point words, implementation-defined options
3889:
3890: @table @i
3891: @item format and range of floating point numbers:
3892: @cindex format and range of floating point numbers
3893: @cindex floating point numbers, format and range
3894: System-dependent; the @code{double} type of C.
3895:
3896: @item results of @code{REPRESENT} when @var{float} is out of range:
3897: @cindex @code{REPRESENT}, results when @var{float} is out of range
3898: System dependent; @code{REPRESENT} is implemented using the C library
3899: function @code{ecvt()} and inherits its behaviour in this respect.
3900:
3901: @item rounding or truncation of floating-point numbers:
3902: @cindex rounding of floating-point numbers
3903: @cindex truncation of floating-point numbers
3904: @cindex floating-point numbers, rounding or truncation
3905: System dependent; the rounding behaviour is inherited from the hosting C
3906: compiler. IEEE-FP-based (i.e., most) systems by default round to
3907: nearest, and break ties by rounding to even (i.e., such that the last
3908: bit of the mantissa is 0).
3909:
3910: @item size of floating-point stack:
3911: @cindex floating-point stack size
3912: @code{s" FLOATING-STACK" environment? drop .} gives the total size of
3913: the floating-point stack (in floats). You can specify this on startup
3914: with the command-line option @code{-f} (@pxref{Invoking Gforth}).
3915:
3916: @item width of floating-point stack:
3917: @cindex floating-point stack width
3918: @code{1 floats}.
3919:
3920: @end table
3921:
3922:
3923: @c ---------------------------------------------------------------------
3924: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
3925: @subsection Ambiguous conditions
3926: @c ---------------------------------------------------------------------
3927: @cindex floating-point words, ambiguous conditions
3928: @cindex ambiguous conditions, floating-point words
3929:
3930: @table @i
3931: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
3932: @cindex @code{df@@} or @code{df!} used with an address that is not double-float aligned
3933: System-dependent. Typically results in a @code{-23 THROW} like other
3934: alignment violations.
3935:
3936: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
3937: @cindex @code{f@@} used with an address that is not float aligned
3938: @cindex @code{f!} used with an address that is not float aligned
3939: System-dependent. Typically results in a @code{-23 THROW} like other
3940: alignment violations.
3941:
3942: @item floating-point result out of range:
3943: @cindex floating-point result out of range
3944: System-dependent. Can result in a @code{-55 THROW} (Floating-point
3945: unidentified fault), or can produce a special value representing, e.g.,
3946: Infinity.
3947:
3948: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
3949: @cindex @code{sf@@} or @code{sf!} used with an address that is not single-float aligned
3950: System-dependent. Typically results in an alignment fault like other
3951: alignment violations.
3952:
3953: @item @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
3954: @cindex @code{BASE} is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.})
3955: The floating-point number is converted into decimal nonetheless.
3956:
3957: @item Both arguments are equal to zero (@code{FATAN2}):
3958: @cindex @code{FATAN2}, both arguments are equal to zero
3959: System-dependent. @code{FATAN2} is implemented using the C library
3960: function @code{atan2()}.
3961:
3962: @item Using @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero:
3963: @cindex @code{FTAN} on an argument @var{r1} where cos(@var{r1}) is zero
3964: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
3965: because of small errors and the tan will be a very large (or very small)
3966: but finite number.
3967:
3968: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
3969: @cindex @code{D>F}, @var{d} cannot be presented precisely as a float
3970: The result is rounded to the nearest float.
3971:
3972: @item dividing by zero:
3973: @cindex dividing by zero, floating-point
3974: @cindex floating-point dividing by zero
3975: @cindex floating-point unidentified fault, FP divide-by-zero
3976: @code{-55 throw} (Floating-point unidentified fault)
3977:
3978: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
3979: @cindex exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@})
3980: System dependent. On IEEE-FP based systems the number is converted into
3981: an infinity.
3982:
3983: @item @var{float}<1 (@code{FACOSH}):
3984: @cindex @code{FACOSH}, @var{float}<1
3985: @cindex floating-point unidentified fault, @code{FACOSH}
3986: @code{-55 throw} (Floating-point unidentified fault)
3987:
3988: @item @var{float}=<-1 (@code{FLNP1}):
3989: @cindex @code{FLNP1}, @var{float}=<-1
3990: @cindex floating-point unidentified fault, @code{FLNP1}
3991: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3992: negative infinity is typically produced for @var{float}=-1.
3993:
3994: @item @var{float}=<0 (@code{FLN}, @code{FLOG}):
3995: @cindex @code{FLN}, @var{float}=<0
3996: @cindex @code{FLOG}, @var{float}=<0
3997: @cindex floating-point unidentified fault, @code{FLN} or @code{FLOG}
3998: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3999: negative infinity is typically produced for @var{float}=0.
4000:
4001: @item @var{float}<0 (@code{FASINH}, @code{FSQRT}):
4002: @cindex @code{FASINH}, @var{float}<0
4003: @cindex @code{FSQRT}, @var{float}<0
4004: @cindex floating-point unidentified fault, @code{FASINH} or @code{FSQRT}
4005: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
4006: produces values for these inputs on my Linux box (Bug in the C library?)
4007:
4008: @item |@var{float}|>1 (@code{FACOS}, @code{FASIN}, @code{FATANH}):
4009: @cindex @code{FACOS}, |@var{float}|>1
4010: @cindex @code{FASIN}, |@var{float}|>1
4011: @cindex @code{FATANH}, |@var{float}|>1
4012: @cindex floating-point unidentified fault, @code{FACOS}, @code{FASIN} or @code{FATANH}
4013: @code{-55 throw} (Floating-point unidentified fault).
4014:
4015: @item integer part of float cannot be represented by @var{d} in @code{F>D}:
4016: @cindex @code{F>D}, integer part of float cannot be represented by @var{d}
4017: @cindex floating-point unidentified fault, @code{F>D}
4018: @code{-55 throw} (Floating-point unidentified fault).
4019:
4020: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
4021: @cindex string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.})
4022: This does not happen.
4023: @end table
4024:
4025: @c =====================================================================
4026: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
4027: @section The optional Locals word set
4028: @c =====================================================================
4029: @cindex system documentation, locals words
4030: @cindex locals words, system documentation
4031:
4032: @menu
4033: * locals-idef:: Implementation Defined Options
4034: * locals-ambcond:: Ambiguous Conditions
4035: @end menu
4036:
4037:
4038: @c ---------------------------------------------------------------------
4039: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
4040: @subsection Implementation Defined Options
4041: @c ---------------------------------------------------------------------
4042: @cindex implementation-defined options, locals words
4043: @cindex locals words, implementation-defined options
4044:
4045: @table @i
4046: @item maximum number of locals in a definition:
4047: @cindex maximum number of locals in a definition
4048: @cindex locals, maximum number in a definition
4049: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
4050: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
4051: characters. The number of locals in a definition is bounded by the size
4052: of locals-buffer, which contains the names of the locals.
4053:
4054: @end table
4055:
4056:
4057: @c ---------------------------------------------------------------------
4058: @node locals-ambcond, , locals-idef, The optional Locals word set
4059: @subsection Ambiguous conditions
4060: @c ---------------------------------------------------------------------
4061: @cindex locals words, ambiguous conditions
4062: @cindex ambiguous conditions, locals words
4063:
4064: @table @i
4065: @item executing a named local in interpretation state:
4066: @cindex local in interpretation state
4067: @cindex Interpreting a compile-only word, for a local
4068: Locals have no interpretation semantics. If you try to perform the
4069: interpretation semantics, you will get a @code{-14 throw} somewhere
4070: (Interpreting a compile-only word). If you perform the compilation
4071: semantics, the locals access will be compiled (irrespective of state).
4072:
4073: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
4074: @cindex name not defined by @code{VALUE} or @code{(LOCAL)} used by @code{TO}
4075: @cindex @code{TO} on non-@code{VALUE}s and non-locals
4076: @cindex Invalid name argument, @code{TO}
4077: @code{-32 throw} (Invalid name argument)
4078:
4079: @end table
4080:
4081:
4082: @c =====================================================================
4083: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
4084: @section The optional Memory-Allocation word set
4085: @c =====================================================================
4086: @cindex system documentation, memory-allocation words
4087: @cindex memory-allocation words, system documentation
4088:
4089: @menu
4090: * memory-idef:: Implementation Defined Options
4091: @end menu
4092:
4093:
4094: @c ---------------------------------------------------------------------
4095: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
4096: @subsection Implementation Defined Options
4097: @c ---------------------------------------------------------------------
4098: @cindex implementation-defined options, memory-allocation words
4099: @cindex memory-allocation words, implementation-defined options
4100:
4101: @table @i
4102: @item values and meaning of @var{ior}:
4103: @cindex @var{ior} values and meaning
4104: The @var{ior}s returned by the file and memory allocation words are
4105: intended as throw codes. They typically are in the range
4106: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
4107: @var{ior}s is -512@minus{}@var{errno}.
4108:
4109: @end table
4110:
4111: @c =====================================================================
4112: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
4113: @section The optional Programming-Tools word set
4114: @c =====================================================================
4115: @cindex system documentation, programming-tools words
4116: @cindex programming-tools words, system documentation
4117:
4118: @menu
4119: * programming-idef:: Implementation Defined Options
4120: * programming-ambcond:: Ambiguous Conditions
4121: @end menu
4122:
4123:
4124: @c ---------------------------------------------------------------------
4125: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
4126: @subsection Implementation Defined Options
4127: @c ---------------------------------------------------------------------
4128: @cindex implementation-defined options, programming-tools words
4129: @cindex programming-tools words, implementation-defined options
4130:
4131: @table @i
4132: @item ending sequence for input following @code{;CODE} and @code{CODE}:
4133: @cindex @code{;CODE} ending sequence
4134: @cindex @code{CODE} ending sequence
4135: @code{END-CODE}
4136:
4137: @item manner of processing input following @code{;CODE} and @code{CODE}:
4138: @cindex @code{;CODE}, processing input
4139: @cindex @code{CODE}, processing input
4140: The @code{ASSEMBLER} vocabulary is pushed on the search order stack, and
4141: the input is processed by the text interpreter, (starting) in interpret
4142: state.
4143:
4144: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
4145: @cindex @code{ASSEMBLER}, search order capability
4146: The ANS Forth search order word set.
4147:
4148: @item source and format of display by @code{SEE}:
4149: @cindex @code{SEE}, source and format of output
4150: The source for @code{see} is the intermediate code used by the inner
4151: interpreter. The current @code{see} tries to output Forth source code
4152: as well as possible.
4153:
4154: @end table
4155:
4156: @c ---------------------------------------------------------------------
4157: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
4158: @subsection Ambiguous conditions
4159: @c ---------------------------------------------------------------------
4160: @cindex programming-tools words, ambiguous conditions
4161: @cindex ambiguous conditions, programming-tools words
4162:
4163: @table @i
4164:
4165: @item deleting the compilation wordlist (@code{FORGET}):
4166: @cindex @code{FORGET}, deleting the compilation wordlist
4167: Not implemented (yet).
4168:
4169: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
4170: @cindex @code{CS-PICK}, fewer than @var{u}+1 items on the control flow stack
4171: @cindex @code{CS-ROLL}, fewer than @var{u}+1 items on the control flow stack
4172: @cindex control-flow stack underflow
4173: This typically results in an @code{abort"} with a descriptive error
4174: message (may change into a @code{-22 throw} (Control structure mismatch)
4175: in the future). You may also get a memory access error. If you are
4176: unlucky, this ambiguous condition is not caught.
4177:
4178: @item @var{name} can't be found (@code{FORGET}):
4179: @cindex @code{FORGET}, @var{name} can't be found
4180: Not implemented (yet).
4181:
4182: @item @var{name} not defined via @code{CREATE}:
4183: @cindex @code{;CODE}, @var{name} not defined via @code{CREATE}
4184: @code{;CODE} behaves like @code{DOES>} in this respect, i.e., it changes
4185: the execution semantics of the last defined word no matter how it was
4186: defined.
4187:
4188: @item @code{POSTPONE} applied to @code{[IF]}:
4189: @cindex @code{POSTPONE} applied to @code{[IF]}
4190: @cindex @code{[IF]} and @code{POSTPONE}
4191: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
4192: equivalent to @code{[IF]}.
4193:
4194: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
4195: @cindex @code{[IF]}, end of the input source before matching @code{[ELSE]} or @code{[THEN]}
4196: Continue in the same state of conditional compilation in the next outer
4197: input source. Currently there is no warning to the user about this.
4198:
4199: @item removing a needed definition (@code{FORGET}):
4200: @cindex @code{FORGET}, removing a needed definition
4201: Not implemented (yet).
4202:
4203: @end table
4204:
4205:
4206: @c =====================================================================
4207: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
4208: @section The optional Search-Order word set
4209: @c =====================================================================
4210: @cindex system documentation, search-order words
4211: @cindex search-order words, system documentation
4212:
4213: @menu
4214: * search-idef:: Implementation Defined Options
4215: * search-ambcond:: Ambiguous Conditions
4216: @end menu
4217:
4218:
4219: @c ---------------------------------------------------------------------
4220: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
4221: @subsection Implementation Defined Options
4222: @c ---------------------------------------------------------------------
4223: @cindex implementation-defined options, search-order words
4224: @cindex search-order words, implementation-defined options
4225:
4226: @table @i
4227: @item maximum number of word lists in search order:
4228: @cindex maximum number of word lists in search order
4229: @cindex search order, maximum depth
4230: @code{s" wordlists" environment? drop .}. Currently 16.
4231:
4232: @item minimum search order:
4233: @cindex minimum search order
4234: @cindex search order, minimum
4235: @code{root root}.
4236:
4237: @end table
4238:
4239: @c ---------------------------------------------------------------------
4240: @node search-ambcond, , search-idef, The optional Search-Order word set
4241: @subsection Ambiguous conditions
4242: @c ---------------------------------------------------------------------
4243: @cindex search-order words, ambiguous conditions
4244: @cindex ambiguous conditions, search-order words
4245:
4246: @table @i
4247: @item changing the compilation wordlist (during compilation):
4248: @cindex changing the compilation wordlist (during compilation)
4249: @cindex compilation wordlist, change before definition ends
4250: The word is entered into the wordlist that was the compilation wordlist
4251: at the start of the definition. Any changes to the name field (e.g.,
4252: @code{immediate}) or the code field (e.g., when executing @code{DOES>})
4253: are applied to the latest defined word (as reported by @code{last} or
4254: @code{lastxt}), if possible, irrespective of the compilation wordlist.
4255:
4256: @item search order empty (@code{previous}):
4257: @cindex @code{previous}, search order empty
4258: @cindex Vocstack empty, @code{previous}
4259: @code{abort" Vocstack empty"}.
4260:
4261: @item too many word lists in search order (@code{also}):
4262: @cindex @code{also}, too many word lists in search order
4263: @cindex Vocstack full, @code{also}
4264: @code{abort" Vocstack full"}.
4265:
4266: @end table
4267:
4268: @c ***************************************************************
4269: @node Model, Integrating Gforth, ANS conformance, Top
4270: @chapter Model
4271:
4272: This chapter has yet to be written. It will contain information, on
4273: which internal structures you can rely.
4274:
4275: @c ***************************************************************
4276: @node Integrating Gforth, Emacs and Gforth, Model, Top
4277: @chapter Integrating Gforth into C programs
4278:
4279: This is not yet implemented.
4280:
4281: Several people like to use Forth as scripting language for applications
4282: that are otherwise written in C, C++, or some other language.
4283:
4284: The Forth system ATLAST provides facilities for embedding it into
4285: applications; unfortunately it has several disadvantages: most
4286: importantly, it is not based on ANS Forth, and it is apparently dead
4287: (i.e., not developed further and not supported). The facilities
4288: provided by Gforth in this area are inspired by ATLASTs facilities, so
4289: making the switch should not be hard.
4290:
4291: We also tried to design the interface such that it can easily be
4292: implemented by other Forth systems, so that we may one day arrive at a
4293: standardized interface. Such a standard interface would allow you to
4294: replace the Forth system without having to rewrite C code.
4295:
4296: You embed the Gforth interpreter by linking with the library
4297: @code{libgforth.a} (give the compiler the option @code{-lgforth}). All
4298: global symbols in this library that belong to the interface, have the
4299: prefix @code{forth_}. (Global symbols that are used internally have the
4300: prefix @code{gforth_}).
4301:
4302: You can include the declarations of Forth types and the functions and
4303: variables of the interface with @code{#include <forth.h>}.
4304:
4305: Types.
4306:
4307: Variables.
4308:
4309: Data and FP Stack pointer. Area sizes.
4310:
4311: functions.
4312:
4313: forth_init(imagefile)
4314: forth_evaluate(string) exceptions?
4315: forth_goto(address) (or forth_execute(xt)?)
4316: forth_continue() (a corountining mechanism)
4317:
4318: Adding primitives.
4319:
4320: No checking.
4321:
4322: Signals?
4323:
4324: Accessing the Stacks
4325:
4326: @node Emacs and Gforth, Image Files, Integrating Gforth, Top
4327: @chapter Emacs and Gforth
4328: @cindex Emacs and Gforth
4329:
4330: @cindex @file{gforth.el}
4331: @cindex @file{forth.el}
4332: @cindex Rydqvist, Goran
4333: @cindex comment editing commands
4334: @cindex @code{\}, editing with Emacs
4335: @cindex debug tracer editing commands
4336: @cindex @code{~~}, removal with Emacs
4337: @cindex Forth mode in Emacs
4338: Gforth comes with @file{gforth.el}, an improved version of
4339: @file{forth.el} by Goran Rydqvist (included in the TILE package). The
4340: improvements are a better (but still not perfect) handling of
4341: indentation. I have also added comment paragraph filling (@kbd{M-q}),
4342: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
4343: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
4344: stuff I do not use alone, even though some of it only makes sense for
4345: TILE. To get a description of these features, enter Forth mode and type
4346: @kbd{C-h m}.
4347:
4348: @cindex source location of error or debugging output in Emacs
4349: @cindex error output, finding the source location in Emacs
4350: @cindex debugging output, finding the source location in Emacs
4351: In addition, Gforth supports Emacs quite well: The source code locations
4352: given in error messages, debugging output (from @code{~~}) and failed
4353: assertion messages are in the right format for Emacs' compilation mode
4354: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
4355: Manual}) so the source location corresponding to an error or other
4356: message is only a few keystrokes away (@kbd{C-x `} for the next error,
4357: @kbd{C-c C-c} for the error under the cursor).
4358:
4359: @cindex @file{TAGS} file
4360: @cindex @file{etags.fs}
4361: @cindex viewing the source of a word in Emacs
4362: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
4363: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
4364: contains the definitions of all words defined afterwards. You can then
4365: find the source for a word using @kbd{M-.}. Note that emacs can use
4366: several tags files at the same time (e.g., one for the Gforth sources
4367: and one for your program, @pxref{Select Tags Table,,Selecting a Tags
4368: Table,emacs, Emacs Manual}). The TAGS file for the preloaded words is
4369: @file{$(datadir)/gforth/$(VERSION)/TAGS} (e.g.,
4370: @file{/usr/local/share/gforth/0.2.0/TAGS}).
4371:
4372: @cindex @file{.emacs}
4373: To get all these benefits, add the following lines to your @file{.emacs}
4374: file:
4375:
4376: @example
4377: (autoload 'forth-mode "gforth.el")
4378: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
4379: @end example
4380:
4381: @node Image Files, Engine, Emacs and Gforth, Top
4382: @chapter Image Files
4383: @cindex image files
4384: @cindex @code{.fi} files
4385: @cindex precompiled Forth code
4386: @cindex dictionary in persistent form
4387: @cindex persistent form of dictionary
4388:
4389: An image file is a file containing an image of the Forth dictionary,
4390: i.e., compiled Forth code and data residing in the dictionary. By
4391: convention, we use the extension @code{.fi} for image files.
4392:
4393: @menu
4394: * Image File Background:: Why have image files?
4395: * Non-Relocatable Image Files:: don't always work.
4396: * Data-Relocatable Image Files:: are better.
4397: * Fully Relocatable Image Files:: better yet.
4398: * Stack and Dictionary Sizes:: Setting the default sizes for an image.
4399: * Running Image Files:: @code{gforth -i @var{file}} or @var{file}.
4400: * Modifying the Startup Sequence:: and turnkey applications.
4401: @end menu
4402:
4403: @node Image File Background, Non-Relocatable Image Files, Image Files, Image Files
4404: @section Image File Background
4405: @cindex image file background
4406:
4407: Our Forth system consists not only of primitives, but also of
4408: definitions written in Forth. Since the Forth compiler itself belongs to
4409: those definitions, it is not possible to start the system with the
4410: primitives and the Forth source alone. Therefore we provide the Forth
4411: code as an image file in nearly executable form. At the start of the
4412: system a C routine loads the image file into memory, optionally
4413: relocates the addresses, then sets up the memory (stacks etc.) according
4414: to information in the image file, and starts executing Forth code.
4415:
4416: The image file variants represent different compromises between the
4417: goals of making it easy to generate image files and making them
4418: portable.
4419:
4420: @cindex relocation at run-time
4421: Win32Forth 3.4 and Mitch Bradleys @code{cforth} use relocation at
4422: run-time. This avoids many of the complications discussed below (image
4423: files are data relocatable without further ado), but costs performance
4424: (one addition per memory access).
4425:
4426: @cindex relocation at load-time
4427: By contrast, our loader performs relocation at image load time. The
4428: loader also has to replace tokens standing for primitive calls with the
4429: appropriate code-field addresses (or code addresses in the case of
4430: direct threading).
4431:
4432: There are three kinds of image files, with different degrees of
4433: relocatability: non-relocatable, data-relocatable, and fully relocatable
4434: image files.
4435:
4436: @cindex image file loader
4437: @cindex relocating loader
4438: @cindex loader for image files
4439: These image file variants have several restrictions in common; they are
4440: caused by the design of the image file loader:
4441:
4442: @itemize @bullet
4443: @item
4444: There is only one segment; in particular, this means, that an image file
4445: cannot represent @code{ALLOCATE}d memory chunks (and pointers to
4446: them). And the contents of the stacks are not represented, either.
4447:
4448: @item
4449: The only kinds of relocation supported are: adding the same offset to
4450: all cells that represent data addresses; and replacing special tokens
4451: with code addresses or with pieces of machine code.
4452:
4453: If any complex computations involving addresses are performed, the
4454: results cannot be represented in the image file. Several applications that
4455: use such computations come to mind:
4456: @itemize @minus
4457: @item
4458: Hashing addresses (or data structures which contain addresses) for table
4459: lookup. If you use Gforth's @code{table}s or @code{wordlist}s for this
4460: purpose, you will have no problem, because the hash tables are
4461: recomputed automatically when the system is started. If you use your own
4462: hash tables, you will have to do something similar.
4463:
4464: @item
4465: There's a cute implementation of doubly-linked lists that uses
4466: @code{XOR}ed addresses. You could represent such lists as singly-linked
4467: in the image file, and restore the doubly-linked representation on
4468: startup.@footnote{In my opinion, though, you should think thrice before
4469: using a doubly-linked list (whatever implementation).}
4470:
4471: @item
4472: The code addresses of run-time routines like @code{docol:} cannot be
4473: represented in the image file (because their tokens would be replaced by
4474: machine code in direct threaded implementations). As a workaround,
4475: compute these addresses at run-time with @code{>code-address} from the
4476: executions tokens of appropriate words (see the definitions of
4477: @code{docol:} and friends in @file{kernel.fs}).
4478:
4479: @item
4480: On many architectures addresses are represented in machine code in some
4481: shifted or mangled form. You cannot put @code{CODE} words that contain
4482: absolute addresses in this form in a relocatable image file. Workarounds
4483: are representing the address in some relative form (e.g., relative to
4484: the CFA, which is present in some register), or loading the address from
4485: a place where it is stored in a non-mangled form.
4486: @end itemize
4487: @end itemize
4488:
4489: @node Non-Relocatable Image Files, Data-Relocatable Image Files, Image File Background, Image Files
4490: @section Non-Relocatable Image Files
4491: @cindex non-relocatable image files
4492: @cindex image files, non-relocatable
4493:
4494: These files are simple memory dumps of the dictionary. They are specific
4495: to the executable (i.e., @file{gforth} file) they were created
4496: with. What's worse, they are specific to the place on which the
4497: dictionary resided when the image was created. Now, there is no
4498: guarantee that the dictionary will reside at the same place the next
4499: time you start Gforth, so there's no guarantee that a non-relocatable
4500: image will work the next time (Gforth will complain instead of crashing,
4501: though).
4502:
4503: You can create a non-relocatable image file with
4504:
4505: doc-savesystem
4506:
4507: @node Data-Relocatable Image Files, Fully Relocatable Image Files, Non-Relocatable Image Files, Image Files
4508: @section Data-Relocatable Image Files
4509: @cindex data-relocatable image files
4510: @cindex image files, data-relocatable
4511:
4512: These files contain relocatable data addresses, but fixed code addresses
4513: (instead of tokens). They are specific to the executable (i.e.,
4514: @file{gforth} file) they were created with. For direct threading on some
4515: architectures (e.g., the i386), data-relocatable images do not work. You
4516: get a data-relocatable image, if you use @file{gforth-makeimage} with a
4517: Gforth binary that is not doubly indirect threaded (@pxref{Fully
4518: Relocatable Image Files}).
4519:
4520: @node Fully Relocatable Image Files, Stack and Dictionary Sizes, Data-Relocatable Image Files, Image Files
4521: @section Fully Relocatable Image Files
4522: @cindex fully relocatable image files
4523: @cindex image files, fully relocatable
4524:
4525: @cindex @file{kern*.fi}, relocatability
4526: @cindex @file{gforth.fi}, relocatability
4527: These image files have relocatable data addresses, and tokens for code
4528: addresses. They can be used with different binaries (e.g., with and
4529: without debugging) on the same machine, and even across machines with
4530: the same data formats (byte order, cell size, floating point
4531: format). However, they are usually specific to the version of Gforth
4532: they were created with. The files @file{gforth.fi} and @file{kernl*.fi}
4533: are fully relocatable.
4534:
4535: There are two ways to create a fully relocatable image file:
4536:
4537: @menu
4538: * gforth-makeimage:: The normal way
4539: * cross.fs:: The hard way
4540: @end menu
4541:
4542: @node gforth-makeimage, cross.fs, Fully Relocatable Image Files, Fully Relocatable Image Files
4543: @subsection @file{gforth-makeimage}
4544: @cindex @file{comp-image.fs}
4545: @cindex @file{gforth-makeimage}
4546:
4547: You will usually use @file{gforth-makeimage}. If you want to create an
4548: image @var{file} that contains everything you would load by invoking
4549: Gforth with @code{gforth @var{options}}, you simply say
4550: @example
4551: gforth-makeimage @var{file} @var{options}
4552: @end example
4553:
4554: E.g., if you want to create an image @file{asm.fi} that has the file
4555: @file{asm.fs} loaded in addition to the usual stuff, you could do it
4556: like this:
4557:
4558: @example
4559: gforth-makeimage asm.fi asm.fs
4560: @end example
4561:
4562: @file{gforth-makeimage} works like this: It produces two non-relocatable
4563: images for different addresses and then compares them. Its output
4564: reflects this: first you see the output (if any) of the two Gforth
4565: invocations that produce the nonrelocatable image files, then you see
4566: the output of the comparing program: It displays the offset used for
4567: data addresses and the offset used for code addresses;
4568: moreover, for each cell that cannot be represented correctly in the
4569: image files, it displays a line like the following one:
4570:
4571: @example
4572: 78DC BFFFFA50 BFFFFA40
4573: @end example
4574:
4575: This means that at offset $78dc from @code{forthstart}, one input image
4576: contains $bffffa50, and the other contains $bffffa40. Since these cells
4577: cannot be represented correctly in the output image, you should examine
4578: these places in the dictionary and verify that these cells are dead
4579: (i.e., not read before they are written).
4580:
4581: \cindex @code{savesystem} during @file{gforth-makeimage}
4582: \cindex @code{bye} during @file{gforth-makeimage}
4583: \cindex doubly indirect threaded code
4584: \cindex environment variable @code{GFORTHD}
4585: \cindex @code{GFORTHD} environment variable
4586: \cindex @code{gforth-ditc}
4587: There are a few wrinkles: After processing the passed @var{options}, the
4588: words @code{savesystem} and @code{bye} must be visible. A special doubly
4589: indirect threaded version of the @file{gforth} executable is used for
4590: creating the nonrelocatable images; you can pass the exact filename of
4591: this executable through the environment variable @code{GFORTHD}
4592: (default: @file{gforth-ditc}); if you pass a version that is not doubly
4593: indirect threaded, you will not get a fully relocatable image, but a
4594: data-relocatable image (because there is no code address offset).
4595:
4596: @node cross.fs, , gforth-makeimage, Fully Relocatable Image Files
4597: @subsection @file{cross.fs}
4598: @cindex @file{cross.fs}
4599: @cindex cross-compiler
4600: @cindex metacompiler
4601:
4602: You can also use @code{cross}, a batch compiler that accepts a Forth-like
4603: programming language. This @code{cross} language has to be documented
4604: yet.
4605:
4606: @cindex target compiler
4607: @code{cross} also allows you to create image files for machines with
4608: different data sizes and data formats than the one used for generating
4609: the image file. You can also use it to create an application image that
4610: does not contain a Forth compiler. These features are bought with
4611: restrictions and inconveniences in programming. E.g., addresses have to
4612: be stored in memory with special words (@code{A!}, @code{A,}, etc.) in
4613: order to make the code relocatable.
4614:
4615:
4616: @node Stack and Dictionary Sizes, Running Image Files, Fully Relocatable Image Files, Image Files
4617: @section Stack and Dictionary Sizes
4618: @cindex image file, stack and dictionary sizes
4619: @cindex dictionary size default
4620: @cindex stack size default
4621:
4622: If you invoke Gforth with a command line flag for the size
4623: (@pxref{Invoking Gforth}), the size you specify is stored in the
4624: dictionary. If you save the dictionary with @code{savesystem} or create
4625: an image with @file{gforth-makeimage}, this size will become the default
4626: for the resulting image file. E.g., the following will create a
4627: fully relocatable version of gforth.fi with a 1MB dictionary:
4628:
4629: @example
4630: gforth-makeimage gforth.fi -m 1M
4631: @end example
4632:
4633: In other words, if you want to set the default size for the dictionary
4634: and the stacks of an image, just invoke @file{gforth-makeimage} with the
4635: appropriate options when creating the image.
4636:
4637: @cindex stack size, cache-friendly
4638: Note: For cache-friendly behaviour (i.e., good performance), you should
4639: make the sizes of the stacks modulo, say, 2K, somewhat different. E.g.,
4640: the default stack sizes are: data: 16k (mod 2k=0); fp: 15.5k (mod
4641: 2k=1.5k); return: 15k(mod 2k=1k); locals: 14.5k (mod 2k=0.5k).
4642:
4643: @node Running Image Files, Modifying the Startup Sequence, Stack and Dictionary Sizes, Image Files
4644: @section Running Image Files
4645: @cindex running image files
4646: @cindex invoking image files
4647: @cindex image file invocation
4648:
4649: @cindex -i, invoke image file
4650: @cindex --image file, invoke image file
4651: You can invoke Gforth with an image file @var{image} instead of the
4652: default @file{gforth.fi} with the @code{-i} flag (@pxref{Invoking Gforth}):
4653: @example
4654: gforth -i @var{image}
4655: @end example
4656:
4657: @cindex executable image file
4658: @cindex image files, executable
4659: If your operating system supports starting scripts with a line of the
4660: form @code{#! ...}, you can make your image file executable, and you'll
4661: just have to type the image file name to start Gforth with this image
4662: file (note that the file extension @code{.fi} is just a convention).
4663:
4664: I.e., in most Unix systems, you just have to make the image file
4665: @var{image} executable with
4666:
4667: @example
4668: chmod +x @var{image}
4669: @end example
4670:
4671: and then you can invoke it by simply typing @var{image} instead of
4672: @code{gforth -i @var{image}}.
4673:
4674: @node Modifying the Startup Sequence, , Running Image Files, Image Files
4675: @section Modifying the Startup Sequence
4676: @cindex startup sequence for image file
4677: @cindex image file initialization sequence
4678: @cindex initialization sequence of image file
4679:
4680: You can add your own initialization to the startup sequence through the
4681: deferred word
4682:
4683: doc-'cold
4684:
4685: @code{'cold} is invoked just before the image-specific command line
4686: processing (by default, loading files and evaluating (@code{-e}) strings)
4687: starts.
4688:
4689: A sequence for adding your initialization usually looks like this:
4690:
4691: @example
4692: :noname
4693: Defers 'cold \ do other initialization stuff (e.g., rehashing wordlists)
4694: ... \ your stuff
4695: ; IS 'cold
4696: @end example
4697:
4698: @cindex turnkey image files
4699: @cindex image files, turnkey applications
4700: You can make a turnkey image by letting @code{'cold} execute a word
4701: (your turnkey application) that never returns; instead, it exits Gforth
4702: via @code{bye} or @code{throw}.
4703:
4704: @c ******************************************************************
4705: @node Engine, Bugs, Image Files, Top
4706: @chapter Engine
4707: @cindex engine
4708: @cindex virtual machine
4709:
4710: Reading this section is not necessary for programming with Gforth. It
4711: may be helpful for finding your way in the Gforth sources.
4712:
4713: The ideas in this section have also been published in the papers
4714: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
4715: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
4716: Ertl, presented at EuroForth '93; the latter is available at
4717: @*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
4718:
4719: @menu
4720: * Portability::
4721: * Threading::
4722: * Primitives::
4723: * Performance::
4724: @end menu
4725:
4726: @node Portability, Threading, Engine, Engine
4727: @section Portability
4728: @cindex engine portability
4729:
4730: One of the main goals of the effort is availability across a wide range
4731: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
4732: this goal by manually coding the engine in assembly language for several
4733: then-popular processors. This approach is very labor-intensive and the
4734: results are short-lived due to progress in computer architecture.
4735:
4736: @cindex C, using C for the engine
4737: Others have avoided this problem by coding in C, e.g., Mitch Bradley
4738: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
4739: particularly popular for UNIX-based Forths due to the large variety of
4740: architectures of UNIX machines. Unfortunately an implementation in C
4741: does not mix well with the goals of efficiency and with using
4742: traditional techniques: Indirect or direct threading cannot be expressed
4743: in C, and switch threading, the fastest technique available in C, is
4744: significantly slower. Another problem with C is that it is very
4745: cumbersome to express double integer arithmetic.
4746:
4747: @cindex GNU C for the engine
4748: @cindex long long
4749: Fortunately, there is a portable language that does not have these
4750: limitations: GNU C, the version of C processed by the GNU C compiler
4751: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
4752: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
4753: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
4754: threading possible, its @code{long long} type (@pxref{Long Long, ,
4755: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forth's
4756: double numbers@footnote{Unfortunately, long longs are not implemented
4757: properly on all machines (e.g., on alpha-osf1, long longs are only 64
4758: bits, the same size as longs (and pointers), but they should be twice as
4759: long according to @ref{Long Long, , Double-Word Integers, gcc.info, GNU
4760: C Manual}). So, we had to implement doubles in C after all. Still, on
4761: most machines we can use long longs and achieve better performance than
4762: with the emulation package.}. GNU C is available for free on all
4763: important (and many unimportant) UNIX machines, VMS, 80386s running
4764: MS-DOS, the Amiga, and the Atari ST, so a Forth written in GNU C can run
4765: on all these machines.
4766:
4767: Writing in a portable language has the reputation of producing code that
4768: is slower than assembly. For our Forth engine we repeatedly looked at
4769: the code produced by the compiler and eliminated most compiler-induced
4770: inefficiencies by appropriate changes in the source code.
4771:
4772: @cindex explicit register declarations
4773: @cindex --enable-force-reg, configuration flag
4774: @cindex -DFORCE_REG
4775: However, register allocation cannot be portably influenced by the
4776: programmer, leading to some inefficiencies on register-starved
4777: machines. We use explicit register declarations (@pxref{Explicit Reg
4778: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
4779: improve the speed on some machines. They are turned on by using the
4780: configuration flag @code{--enable-force-reg} (@code{gcc} switch
4781: @code{-DFORCE_REG}). Unfortunately, this feature not only depends on the
4782: machine, but also on the compiler version: On some machines some
4783: compiler versions produce incorrect code when certain explicit register
4784: declarations are used. So by default @code{-DFORCE_REG} is not used.
4785:
4786: @node Threading, Primitives, Portability, Engine
4787: @section Threading
4788: @cindex inner interpreter implementation
4789: @cindex threaded code implementation
4790:
4791: @cindex labels as values
4792: GNU C's labels as values extension (available since @code{gcc-2.0},
4793: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
4794: makes it possible to take the address of @var{label} by writing
4795: @code{&&@var{label}}. This address can then be used in a statement like
4796: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
4797: @code{goto x}.
4798:
4799: @cindex NEXT, indirect threaded
4800: @cindex indirect threaded inner interpreter
4801: @cindex inner interpreter, indirect threaded
4802: With this feature an indirect threaded NEXT looks like:
4803: @example
4804: cfa = *ip++;
4805: ca = *cfa;
4806: goto *ca;
4807: @end example
4808: @cindex instruction pointer
4809: For those unfamiliar with the names: @code{ip} is the Forth instruction
4810: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
4811: execution token and points to the code field of the next word to be
4812: executed; The @code{ca} (code address) fetched from there points to some
4813: executable code, e.g., a primitive or the colon definition handler
4814: @code{docol}.
4815:
4816: @cindex NEXT, direct threaded
4817: @cindex direct threaded inner interpreter
4818: @cindex inner interpreter, direct threaded
4819: Direct threading is even simpler:
4820: @example
4821: ca = *ip++;
4822: goto *ca;
4823: @end example
4824:
4825: Of course we have packaged the whole thing neatly in macros called
4826: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
4827:
4828: @menu
4829: * Scheduling::
4830: * Direct or Indirect Threaded?::
4831: * DOES>::
4832: @end menu
4833:
4834: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
4835: @subsection Scheduling
4836: @cindex inner interpreter optimization
4837:
4838: There is a little complication: Pipelined and superscalar processors,
4839: i.e., RISC and some modern CISC machines can process independent
4840: instructions while waiting for the results of an instruction. The
4841: compiler usually reorders (schedules) the instructions in a way that
4842: achieves good usage of these delay slots. However, on our first tries
4843: the compiler did not do well on scheduling primitives. E.g., for
4844: @code{+} implemented as
4845: @example
4846: n=sp[0]+sp[1];
4847: sp++;
4848: sp[0]=n;
4849: NEXT;
4850: @end example
4851: the NEXT comes strictly after the other code, i.e., there is nearly no
4852: scheduling. After a little thought the problem becomes clear: The
4853: compiler cannot know that sp and ip point to different addresses (and
4854: the version of @code{gcc} we used would not know it even if it was
4855: possible), so it could not move the load of the cfa above the store to
4856: the TOS. Indeed the pointers could be the same, if code on or very near
4857: the top of stack were executed. In the interest of speed we chose to
4858: forbid this probably unused ``feature'' and helped the compiler in
4859: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
4860: the goto part (@code{NEXT_P2}). @code{+} now looks like:
4861: @example
4862: n=sp[0]+sp[1];
4863: sp++;
4864: NEXT_P1;
4865: sp[0]=n;
4866: NEXT_P2;
4867: @end example
4868: This can be scheduled optimally by the compiler.
4869:
4870: This division can be turned off with the switch @code{-DCISC_NEXT}. This
4871: switch is on by default on machines that do not profit from scheduling
4872: (e.g., the 80386), in order to preserve registers.
4873:
4874: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
4875: @subsection Direct or Indirect Threaded?
4876: @cindex threading, direct or indirect?
4877:
4878: @cindex -DDIRECT_THREADED
4879: Both! After packaging the nasty details in macro definitions we
4880: realized that we could switch between direct and indirect threading by
4881: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
4882: defining a few machine-specific macros for the direct-threading case.
4883: On the Forth level we also offer access words that hide the
4884: differences between the threading methods (@pxref{Threading Words}).
4885:
4886: Indirect threading is implemented completely machine-independently.
4887: Direct threading needs routines for creating jumps to the executable
4888: code (e.g. to docol or dodoes). These routines are inherently
4889: machine-dependent, but they do not amount to many source lines. I.e.,
4890: even porting direct threading to a new machine is a small effort.
4891:
4892: @cindex --enable-indirect-threaded, configuration flag
4893: @cindex --enable-direct-threaded, configuration flag
4894: The default threading method is machine-dependent. You can enforce a
4895: specific threading method when building Gforth with the configuration
4896: flag @code{--enable-direct-threaded} or
4897: @code{--enable-indirect-threaded}. Note that direct threading is not
4898: supported on all machines.
4899:
4900: @node DOES>, , Direct or Indirect Threaded?, Threading
4901: @subsection DOES>
4902: @cindex @code{DOES>} implementation
4903:
4904: @cindex dodoes routine
4905: @cindex DOES-code
4906: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
4907: the chunk of code executed by every word defined by a
4908: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
4909: the Forth code to be executed, i.e. the code after the
4910: @code{DOES>} (the DOES-code)? There are two solutions:
4911:
4912: In fig-Forth the code field points directly to the dodoes and the
4913: DOES-code address is stored in the cell after the code address (i.e. at
4914: @code{@var{cfa} cell+}). It may seem that this solution is illegal in
4915: the Forth-79 and all later standards, because in fig-Forth this address
4916: lies in the body (which is illegal in these standards). However, by
4917: making the code field larger for all words this solution becomes legal
4918: again. We use this approach for the indirect threaded version and for
4919: direct threading on some machines. Leaving a cell unused in most words
4920: is a bit wasteful, but on the machines we are targeting this is hardly a
4921: problem. The other reason for having a code field size of two cells is
4922: to avoid having different image files for direct and indirect threaded
4923: systems (direct threaded systems require two-cell code fields on many
4924: machines).
4925:
4926: @cindex DOES-handler
4927: The other approach is that the code field points or jumps to the cell
4928: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
4929: this address (the DOES-handler). @code{dodoes} can then get the
4930: DOES-code address by computing the code address, i.e., the address of
4931: the jump to dodoes, and add the length of that jump field. A variant of
4932: this is to have a call to @code{dodoes} after the @code{DOES>}; then the
4933: return address (which can be found in the return register on RISCs) is
4934: the DOES-code address. Since the two cells available in the code field
4935: are used up by the jump to the code address in direct threading on many
4936: architectures, we use this approach for direct threading on these
4937: architectures. We did not want to add another cell to the code field.
4938:
4939: @node Primitives, Performance, Threading, Engine
4940: @section Primitives
4941: @cindex primitives, implementation
4942: @cindex virtual machine instructions, implementation
4943:
4944: @menu
4945: * Automatic Generation::
4946: * TOS Optimization::
4947: * Produced code::
4948: @end menu
4949:
4950: @node Automatic Generation, TOS Optimization, Primitives, Primitives
4951: @subsection Automatic Generation
4952: @cindex primitives, automatic generation
4953:
4954: @cindex @file{prims2x.fs}
4955: Since the primitives are implemented in a portable language, there is no
4956: longer any need to minimize the number of primitives. On the contrary,
4957: having many primitives has an advantage: speed. In order to reduce the
4958: number of errors in primitives and to make programming them easier, we
4959: provide a tool, the primitive generator (@file{prims2x.fs}), that
4960: automatically generates most (and sometimes all) of the C code for a
4961: primitive from the stack effect notation. The source for a primitive
4962: has the following form:
4963:
4964: @cindex primitive source format
4965: @format
4966: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
4967: [@code{""}@var{glossary entry}@code{""}]
4968: @var{C code}
4969: [@code{:}
4970: @var{Forth code}]
4971: @end format
4972:
4973: The items in brackets are optional. The category and glossary fields
4974: are there for generating the documentation, the Forth code is there
4975: for manual implementations on machines without GNU C. E.g., the source
4976: for the primitive @code{+} is:
4977: @example
4978: + n1 n2 -- n core plus
4979: n = n1+n2;
4980: @end example
4981:
4982: This looks like a specification, but in fact @code{n = n1+n2} is C
4983: code. Our primitive generation tool extracts a lot of information from
4984: the stack effect notations@footnote{We use a one-stack notation, even
4985: though we have separate data and floating-point stacks; The separate
4986: notation can be generated easily from the unified notation.}: The number
4987: of items popped from and pushed on the stack, their type, and by what
4988: name they are referred to in the C code. It then generates a C code
4989: prelude and postlude for each primitive. The final C code for @code{+}
4990: looks like this:
4991:
4992: @example
4993: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
4994: /* */ /* documentation */
4995: @{
4996: DEF_CA /* definition of variable ca (indirect threading) */
4997: Cell n1; /* definitions of variables */
4998: Cell n2;
4999: Cell n;
5000: n1 = (Cell) sp[1]; /* input */
5001: n2 = (Cell) TOS;
5002: sp += 1; /* stack adjustment */
5003: NAME("+") /* debugging output (with -DDEBUG) */
5004: @{
5005: n = n1+n2; /* C code taken from the source */
5006: @}
5007: NEXT_P1; /* NEXT part 1 */
5008: TOS = (Cell)n; /* output */
5009: NEXT_P2; /* NEXT part 2 */
5010: @}
5011: @end example
5012:
5013: This looks long and inefficient, but the GNU C compiler optimizes quite
5014: well and produces optimal code for @code{+} on, e.g., the R3000 and the
5015: HP RISC machines: Defining the @code{n}s does not produce any code, and
5016: using them as intermediate storage also adds no cost.
5017:
5018: There are also other optimizations, that are not illustrated by this
5019: example: Assignments between simple variables are usually for free (copy
5020: propagation). If one of the stack items is not used by the primitive
5021: (e.g. in @code{drop}), the compiler eliminates the load from the stack
5022: (dead code elimination). On the other hand, there are some things that
5023: the compiler does not do, therefore they are performed by
5024: @file{prims2x.fs}: The compiler does not optimize code away that stores
5025: a stack item to the place where it just came from (e.g., @code{over}).
5026:
5027: While programming a primitive is usually easy, there are a few cases
5028: where the programmer has to take the actions of the generator into
5029: account, most notably @code{?dup}, but also words that do not (always)
5030: fall through to NEXT.
5031:
5032: @node TOS Optimization, Produced code, Automatic Generation, Primitives
5033: @subsection TOS Optimization
5034: @cindex TOS optimization for primitives
5035: @cindex primitives, keeping the TOS in a register
5036:
5037: An important optimization for stack machine emulators, e.g., Forth
5038: engines, is keeping one or more of the top stack items in
5039: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
5040: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
5041: @itemize @bullet
5042: @item
5043: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
5044: due to fewer loads from and stores to the stack.
5045: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
5046: @var{y<n}, due to additional moves between registers.
5047: @end itemize
5048:
5049: @cindex -DUSE_TOS
5050: @cindex -DUSE_NO_TOS
5051: In particular, keeping one item in a register is never a disadvantage,
5052: if there are enough registers. Keeping two items in registers is a
5053: disadvantage for frequent words like @code{?branch}, constants,
5054: variables, literals and @code{i}. Therefore our generator only produces
5055: code that keeps zero or one items in registers. The generated C code
5056: covers both cases; the selection between these alternatives is made at
5057: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
5058: code for @code{+} is just a simple variable name in the one-item case,
5059: otherwise it is a macro that expands into @code{sp[0]}. Note that the
5060: GNU C compiler tries to keep simple variables like @code{TOS} in
5061: registers, and it usually succeeds, if there are enough registers.
5062:
5063: @cindex -DUSE_FTOS
5064: @cindex -DUSE_NO_FTOS
5065: The primitive generator performs the TOS optimization for the
5066: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
5067: operations the benefit of this optimization is even larger:
5068: floating-point operations take quite long on most processors, but can be
5069: performed in parallel with other operations as long as their results are
5070: not used. If the FP-TOS is kept in a register, this works. If
5071: it is kept on the stack, i.e., in memory, the store into memory has to
5072: wait for the result of the floating-point operation, lengthening the
5073: execution time of the primitive considerably.
5074:
5075: The TOS optimization makes the automatic generation of primitives a
5076: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
5077: @code{TOS} is not sufficient. There are some special cases to
5078: consider:
5079: @itemize @bullet
5080: @item In the case of @code{dup ( w -- w w )} the generator must not
5081: eliminate the store to the original location of the item on the stack,
5082: if the TOS optimization is turned on.
5083: @item Primitives with stack effects of the form @code{--}
5084: @var{out1}...@var{outy} must store the TOS to the stack at the start.
5085: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
5086: must load the TOS from the stack at the end. But for the null stack
5087: effect @code{--} no stores or loads should be generated.
5088: @end itemize
5089:
5090: @node Produced code, , TOS Optimization, Primitives
5091: @subsection Produced code
5092: @cindex primitives, assembly code listing
5093:
5094: @cindex @file{engine.s}
5095: To see what assembly code is produced for the primitives on your machine
5096: with your compiler and your flag settings, type @code{make engine.s} and
5097: look at the resulting file @file{engine.s}.
5098:
5099: @node Performance, , Primitives, Engine
5100: @section Performance
5101: @cindex performance of some Forth interpreters
5102: @cindex engine performance
5103: @cindex benchmarking Forth systems
5104: @cindex Gforth performance
5105:
5106: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
5107: impossible to write a significantly faster engine.
5108:
5109: On register-starved machines like the 386 architecture processors
5110: improvements are possible, because @code{gcc} does not utilize the
5111: registers as well as a human, even with explicit register declarations;
5112: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
5113: and hand-tuned it for the 486; this system is 1.19 times faster on the
5114: Sieve benchmark on a 486DX2/66 than Gforth compiled with
5115: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
5116:
5117: @cindex Win32Forth performance
5118: @cindex NT Forth performance
5119: @cindex eforth performance
5120: @cindex ThisForth performance
5121: @cindex PFE performance
5122: @cindex TILE performance
5123: However, this potential advantage of assembly language implementations
5124: is not necessarily realized in complete Forth systems: We compared
5125: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
5126: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
5127: 1994) and Eforth (with and without peephole (aka pinhole) optimization
5128: of the threaded code); all these systems were written in assembly
5129: language. We also compared Gforth with three systems written in C:
5130: PFE-0.9.14 (compiled with @code{gcc-2.6.3} with the default
5131: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS
5132: -DUNROLL_NEXT}), ThisForth Beta (compiled with gcc-2.6.3 -O3
5133: -fomit-frame-pointer; ThisForth employs peephole optimization of the
5134: threaded code) and TILE (compiled with @code{make opt}). We benchmarked
5135: Gforth, PFE, ThisForth and TILE on a 486DX2/66 under Linux. Kenneth
5136: O'Heskin kindly provided the results for Win32Forth and NT Forth on a
5137: 486DX2/66 with similar memory performance under Windows NT. Marcel
5138: Hendrix ported Eforth to Linux, then extended it to run the benchmarks,
5139: added the peephole optimizer, ran the benchmarks and reported the
5140: results.
5141:
5142: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
5143: matrix multiplication come from the Stanford integer benchmarks and have
5144: been translated into Forth by Martin Fraeman; we used the versions
5145: included in the TILE Forth package, but with bigger data set sizes; and
5146: a recursive Fibonacci number computation for benchmarking calling
5147: performance. The following table shows the time taken for the benchmarks
5148: scaled by the time taken by Gforth (in other words, it shows the speedup
5149: factor that Gforth achieved over the other systems).
5150:
5151: @example
5152: relative Win32- NT eforth This-
5153: time Gforth Forth Forth eforth +opt PFE Forth TILE
5154: sieve 1.00 1.39 1.14 1.39 0.85 1.58 3.18 8.58
5155: bubble 1.00 1.31 1.41 1.48 0.88 1.50 3.88
5156: matmul 1.00 1.47 1.35 1.46 0.74 1.58 4.09
5157: fib 1.00 1.52 1.34 1.22 0.86 1.74 2.99 4.30
5158: @end example
5159:
5160: You may find the good performance of Gforth compared with the systems
5161: written in assembly language quite surprising. One important reason for
5162: the disappointing performance of these systems is probably that they are
5163: not written optimally for the 486 (e.g., they use the @code{lods}
5164: instruction). In addition, Win32Forth uses a comfortable, but costly
5165: method for relocating the Forth image: like @code{cforth}, it computes
5166: the actual addresses at run time, resulting in two address computations
5167: per NEXT (@pxref{Image File Background}).
5168:
5169: Only Eforth with the peephole optimizer performs comparable to
5170: Gforth. The speedups achieved with peephole optimization of threaded
5171: code are quite remarkable. Adding a peephole optimizer to Gforth should
5172: cause similar speedups.
5173:
5174: The speedup of Gforth over PFE, ThisForth and TILE can be easily
5175: explained with the self-imposed restriction of the latter systems to
5176: standard C, which makes efficient threading impossible (however, the
5177: measured implementation of PFE uses a GNU C extension: @ref{Global Reg
5178: Vars, , Defining Global Register Variables, gcc.info, GNU C Manual}).
5179: Moreover, current C compilers have a hard time optimizing other aspects
5180: of the ThisForth and the TILE source.
5181:
5182: Note that the performance of Gforth on 386 architecture processors
5183: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
5184: failed to allocate any of the virtual machine registers into real
5185: machine registers by itself and would not work correctly with explicit
5186: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
5187: running the Sieve) than the one measured above.
5188:
5189: Note also that there have been several releases of Win32Forth since the
5190: release presented here, so the results presented here may have little
5191: predictive value for the performance of Win32Forth today.
5192:
5193: @cindex @file{Benchres}
5194: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
5195: Maierhofer (presented at EuroForth '95), an indirect threaded version of
5196: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
5197: version of Gforth is 2%@minus{}8% slower on a 486 than the direct
5198: threaded version used here. The paper available at
5199: @*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
5200: it also contains numbers for some native code systems. You can find a
5201: newer version of these measurements at
5202: @file{http://www.complang.tuwien.ac.at/forth/performance.html}. You can
5203: find numbers for Gforth on various machines in @file{Benchres}.
5204:
5205: @node Bugs, Origin, Engine, Top
5206: @chapter Bugs
5207: @cindex bug reporting
5208:
5209: Known bugs are described in the file BUGS in the Gforth distribution.
5210:
5211: If you find a bug, please send a bug report to
5212: @code{bug-gforth@@gnu.ai.mit.edu}. A bug report should
5213: describe the Gforth version used (it is announced at the start of an
5214: interactive Gforth session), the machine and operating system (on Unix
5215: systems you can use @code{uname -a} to produce this information), the
5216: installation options (send the @file{config.status} file), and a
5217: complete list of changes you (or your installer) have made to the Gforth
5218: sources (if any); it should contain a program (or a sequence of keyboard
5219: commands) that reproduces the bug and a description of what you think
5220: constitutes the buggy behaviour.
5221:
5222: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
5223: to Report Bugs, gcc.info, GNU C Manual}.
5224:
5225:
5226: @node Origin, Word Index, Bugs, Top
5227: @chapter Authors and Ancestors of Gforth
5228:
5229: @section Authors and Contributors
5230: @cindex authors of Gforth
5231: @cindex contributors to Gforth
5232:
5233: The Gforth project was started in mid-1992 by Bernd Paysan and Anton
5234: Ertl. The third major author was Jens Wilke. Lennart Benschop (who was
5235: one of Gforth's first users, in mid-1993) and Stuart Ramsden inspired us
5236: with their continuous feedback. Lennart Benshop contributed
5237: @file{glosgen.fs}, while Stuart Ramsden has been working on automatic
5238: support for calling C libraries. Helpful comments also came from Paul
5239: Kleinrubatscher, Christian Pirker, Dirk Zoller, Marcel Hendrix, John
5240: Wavrik, Barrie Stott and Marc de Groot.
5241:
5242: Gforth also owes a lot to the authors of the tools we used (GCC, CVS,
5243: and autoconf, among others), and to the creators of the Internet: Gforth
5244: was developed across the Internet, and its authors have not met
5245: physically yet.
5246:
5247: @section Pedigree
5248: @cindex Pedigree of Gforth
5249:
5250: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
5251: Dirk Zoller) will cross-fertilize each other. Of course, a significant
5252: part of the design of Gforth was prescribed by ANS Forth.
5253:
5254: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
5255: 32 bit native code version of VolksForth for the Atari ST, written
5256: mostly by Dietrich Weineck.
5257:
5258: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
5259: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
5260: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
5261:
5262: Henry Laxen and Mike Perry wrote F83 as a model implementation of the
5263: Forth-83 standard. !! Pedigree? When?
5264:
5265: A team led by Bill Ragsdale implemented fig-Forth on many processors in
5266: 1979. Robert Selzer and Bill Ragsdale developed the original
5267: implementation of fig-Forth for the 6502 based on microForth.
5268:
5269: The principal architect of microForth was Dean Sanderson. microForth was
5270: FORTH, Inc.'s first off-the-shelf product. It was developed in 1976 for
5271: the 1802, and subsequently implemented on the 8080, the 6800 and the
5272: Z80.
5273:
5274: All earlier Forth systems were custom-made, usually by Charles Moore,
5275: who discovered (as he puts it) Forth during the late 60s. The first full
5276: Forth existed in 1971.
5277:
5278: A part of the information in this section comes from @cite{The Evolution
5279: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
5280: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
5281: Notices 28(3), 1993. You can find more historical and genealogical
5282: information about Forth there.
5283:
5284: @node Word Index, Concept Index, Origin, Top
5285: @unnumbered Word Index
5286:
5287: This index is as incomplete as the manual. Each word is listed with
5288: stack effect and wordset.
5289:
5290: @printindex fn
5291:
5292: @node Concept Index, , Word Index, Top
5293: @unnumbered Concept and Word Index
5294:
5295: This index is as incomplete as the manual. Not all entries listed are
5296: present verbatim in the text. Only the names are listed for the words
5297: here.
5298:
5299: @printindex cp
5300:
5301: @contents
5302: @bye
5303:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>