Annotation of gforth/gforth.ds, revision 1.26
1.1 anton 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.)
1.4 anton 4: @setfilename gforth.info
1.17 anton 5: @settitle Gforth Manual
1.4 anton 6: @comment @setchapternewpage odd
1.1 anton 7: @comment %**end of header (This is for running Texinfo on a region.)
8:
9: @ifinfo
1.17 anton 10: This file documents Gforth 0.1
1.1 anton 11:
1.21 anton 12: Copyright @copyright{} 1995 Free Software Foundation, Inc.
1.1 anton 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:
1.4 anton 18: @ignore
1.1 anton 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:
1.4 anton 24: @end ignore
1.1 anton 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:
1.24 anton 39: @finalout
1.1 anton 40: @titlepage
41: @sp 10
1.17 anton 42: @center @titlefont{Gforth Manual}
1.1 anton 43: @sp 2
1.17 anton 44: @center for version 0.1
1.1 anton 45: @sp 2
46: @center Anton Ertl
1.25 anton 47: @center Bernd Paysan
1.17 anton 48: @sp 3
49: @center This manual is under construction
1.1 anton 50:
51: @comment The following two commands start the copyright page.
52: @page
53: @vskip 0pt plus 1filll
1.21 anton 54: Copyright @copyright{} 1995 Free Software Foundation, Inc.
1.1 anton 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
1.17 anton 79: Gforth is a free implementation of ANS Forth available on many
1.21 anton 80: personal machines. This manual corresponds to version 0.1.
1.1 anton 81: @end ifinfo
82:
83: @menu
1.4 anton 84: * License::
1.17 anton 85: * Goals:: About the Gforth Project
1.4 anton 86: * Other Books:: Things you might want to read
1.17 anton 87: * Invocation:: Starting Gforth
88: * Words:: Forth words available in Gforth
1.4 anton 89: * ANS conformance:: Implementation-defined options etc.
1.17 anton 90: * Model:: The abstract machine of Gforth
91: * Emacs and Gforth:: The Gforth Mode
1.4 anton 92: * Internals:: Implementation details
93: * Bugs:: How to report them
1.17 anton 94: * Pedigree:: Ancestors of Gforth
1.4 anton 95: * Word Index:: An item for each Forth word
96: * Node Index:: An item for each node
1.1 anton 97: @end menu
98:
99: @node License, Goals, Top, Top
1.20 pazsan 100: @unnumbered GNU GENERAL PUBLIC LICENSE
101: @center Version 2, June 1991
102:
103: @display
104: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
105: 675 Mass Ave, Cambridge, MA 02139, USA
106:
107: Everyone is permitted to copy and distribute verbatim copies
108: of this license document, but changing it is not allowed.
109: @end display
110:
111: @unnumberedsec Preamble
112:
113: The licenses for most software are designed to take away your
114: freedom to share and change it. By contrast, the GNU General Public
115: License is intended to guarantee your freedom to share and change free
116: software---to make sure the software is free for all its users. This
117: General Public License applies to most of the Free Software
118: Foundation's software and to any other program whose authors commit to
119: using it. (Some other Free Software Foundation software is covered by
120: the GNU Library General Public License instead.) You can apply it to
121: your programs, too.
122:
123: When we speak of free software, we are referring to freedom, not
124: price. Our General Public Licenses are designed to make sure that you
125: have the freedom to distribute copies of free software (and charge for
126: this service if you wish), that you receive source code or can get it
127: if you want it, that you can change the software or use pieces of it
128: in new free programs; and that you know you can do these things.
129:
130: To protect your rights, we need to make restrictions that forbid
131: anyone to deny you these rights or to ask you to surrender the rights.
132: These restrictions translate to certain responsibilities for you if you
133: distribute copies of the software, or if you modify it.
134:
135: For example, if you distribute copies of such a program, whether
136: gratis or for a fee, you must give the recipients all the rights that
137: you have. You must make sure that they, too, receive or can get the
138: source code. And you must show them these terms so they know their
139: rights.
140:
141: We protect your rights with two steps: (1) copyright the software, and
142: (2) offer you this license which gives you legal permission to copy,
143: distribute and/or modify the software.
144:
145: Also, for each author's protection and ours, we want to make certain
146: that everyone understands that there is no warranty for this free
147: software. If the software is modified by someone else and passed on, we
148: want its recipients to know that what they have is not the original, so
149: that any problems introduced by others will not reflect on the original
150: authors' reputations.
151:
152: Finally, any free program is threatened constantly by software
153: patents. We wish to avoid the danger that redistributors of a free
154: program will individually obtain patent licenses, in effect making the
155: program proprietary. To prevent this, we have made it clear that any
156: patent must be licensed for everyone's free use or not licensed at all.
157:
158: The precise terms and conditions for copying, distribution and
159: modification follow.
160:
161: @iftex
162: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
163: @end iftex
164: @ifinfo
165: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
166: @end ifinfo
167:
168: @enumerate 0
169: @item
170: This License applies to any program or other work which contains
171: a notice placed by the copyright holder saying it may be distributed
172: under the terms of this General Public License. The ``Program'', below,
173: refers to any such program or work, and a ``work based on the Program''
174: means either the Program or any derivative work under copyright law:
175: that is to say, a work containing the Program or a portion of it,
176: either verbatim or with modifications and/or translated into another
177: language. (Hereinafter, translation is included without limitation in
178: the term ``modification''.) Each licensee is addressed as ``you''.
179:
180: Activities other than copying, distribution and modification are not
181: covered by this License; they are outside its scope. The act of
182: running the Program is not restricted, and the output from the Program
183: is covered only if its contents constitute a work based on the
184: Program (independent of having been made by running the Program).
185: Whether that is true depends on what the Program does.
186:
187: @item
188: You may copy and distribute verbatim copies of the Program's
189: source code as you receive it, in any medium, provided that you
190: conspicuously and appropriately publish on each copy an appropriate
191: copyright notice and disclaimer of warranty; keep intact all the
192: notices that refer to this License and to the absence of any warranty;
193: and give any other recipients of the Program a copy of this License
194: along with the Program.
195:
196: You may charge a fee for the physical act of transferring a copy, and
197: you may at your option offer warranty protection in exchange for a fee.
198:
199: @item
200: You may modify your copy or copies of the Program or any portion
201: of it, thus forming a work based on the Program, and copy and
202: distribute such modifications or work under the terms of Section 1
203: above, provided that you also meet all of these conditions:
204:
205: @enumerate a
206: @item
207: You must cause the modified files to carry prominent notices
208: stating that you changed the files and the date of any change.
209:
210: @item
211: You must cause any work that you distribute or publish, that in
212: whole or in part contains or is derived from the Program or any
213: part thereof, to be licensed as a whole at no charge to all third
214: parties under the terms of this License.
215:
216: @item
217: If the modified program normally reads commands interactively
218: when run, you must cause it, when started running for such
219: interactive use in the most ordinary way, to print or display an
220: announcement including an appropriate copyright notice and a
221: notice that there is no warranty (or else, saying that you provide
222: a warranty) and that users may redistribute the program under
223: these conditions, and telling the user how to view a copy of this
224: License. (Exception: if the Program itself is interactive but
225: does not normally print such an announcement, your work based on
226: the Program is not required to print an announcement.)
227: @end enumerate
228:
229: These requirements apply to the modified work as a whole. If
230: identifiable sections of that work are not derived from the Program,
231: and can be reasonably considered independent and separate works in
232: themselves, then this License, and its terms, do not apply to those
233: sections when you distribute them as separate works. But when you
234: distribute the same sections as part of a whole which is a work based
235: on the Program, the distribution of the whole must be on the terms of
236: this License, whose permissions for other licensees extend to the
237: entire whole, and thus to each and every part regardless of who wrote it.
238:
239: Thus, it is not the intent of this section to claim rights or contest
240: your rights to work written entirely by you; rather, the intent is to
241: exercise the right to control the distribution of derivative or
242: collective works based on the Program.
243:
244: In addition, mere aggregation of another work not based on the Program
245: with the Program (or with a work based on the Program) on a volume of
246: a storage or distribution medium does not bring the other work under
247: the scope of this License.
248:
249: @item
250: You may copy and distribute the Program (or a work based on it,
251: under Section 2) in object code or executable form under the terms of
252: Sections 1 and 2 above provided that you also do one of the following:
253:
254: @enumerate a
255: @item
256: Accompany it with the complete corresponding machine-readable
257: source code, which must be distributed under the terms of Sections
258: 1 and 2 above on a medium customarily used for software interchange; or,
259:
260: @item
261: Accompany it with a written offer, valid for at least three
262: years, to give any third party, for a charge no more than your
263: cost of physically performing source distribution, a complete
264: machine-readable copy of the corresponding source code, to be
265: distributed under the terms of Sections 1 and 2 above on a medium
266: customarily used for software interchange; or,
267:
268: @item
269: Accompany it with the information you received as to the offer
270: to distribute corresponding source code. (This alternative is
271: allowed only for noncommercial distribution and only if you
272: received the program in object code or executable form with such
273: an offer, in accord with Subsection b above.)
274: @end enumerate
275:
276: The source code for a work means the preferred form of the work for
277: making modifications to it. For an executable work, complete source
278: code means all the source code for all modules it contains, plus any
279: associated interface definition files, plus the scripts used to
280: control compilation and installation of the executable. However, as a
281: special exception, the source code distributed need not include
282: anything that is normally distributed (in either source or binary
283: form) with the major components (compiler, kernel, and so on) of the
284: operating system on which the executable runs, unless that component
285: itself accompanies the executable.
286:
287: If distribution of executable or object code is made by offering
288: access to copy from a designated place, then offering equivalent
289: access to copy the source code from the same place counts as
290: distribution of the source code, even though third parties are not
291: compelled to copy the source along with the object code.
292:
293: @item
294: You may not copy, modify, sublicense, or distribute the Program
295: except as expressly provided under this License. Any attempt
296: otherwise to copy, modify, sublicense or distribute the Program is
297: void, and will automatically terminate your rights under this License.
298: However, parties who have received copies, or rights, from you under
299: this License will not have their licenses terminated so long as such
300: parties remain in full compliance.
301:
302: @item
303: You are not required to accept this License, since you have not
304: signed it. However, nothing else grants you permission to modify or
305: distribute the Program or its derivative works. These actions are
306: prohibited by law if you do not accept this License. Therefore, by
307: modifying or distributing the Program (or any work based on the
308: Program), you indicate your acceptance of this License to do so, and
309: all its terms and conditions for copying, distributing or modifying
310: the Program or works based on it.
311:
312: @item
313: Each time you redistribute the Program (or any work based on the
314: Program), the recipient automatically receives a license from the
315: original licensor to copy, distribute or modify the Program subject to
316: these terms and conditions. You may not impose any further
317: restrictions on the recipients' exercise of the rights granted herein.
318: You are not responsible for enforcing compliance by third parties to
319: this License.
320:
321: @item
322: If, as a consequence of a court judgment or allegation of patent
323: infringement or for any other reason (not limited to patent issues),
324: conditions are imposed on you (whether by court order, agreement or
325: otherwise) that contradict the conditions of this License, they do not
326: excuse you from the conditions of this License. If you cannot
327: distribute so as to satisfy simultaneously your obligations under this
328: License and any other pertinent obligations, then as a consequence you
329: may not distribute the Program at all. For example, if a patent
330: license would not permit royalty-free redistribution of the Program by
331: all those who receive copies directly or indirectly through you, then
332: the only way you could satisfy both it and this License would be to
333: refrain entirely from distribution of the Program.
334:
335: If any portion of this section is held invalid or unenforceable under
336: any particular circumstance, the balance of the section is intended to
337: apply and the section as a whole is intended to apply in other
338: circumstances.
339:
340: It is not the purpose of this section to induce you to infringe any
341: patents or other property right claims or to contest validity of any
342: such claims; this section has the sole purpose of protecting the
343: integrity of the free software distribution system, which is
344: implemented by public license practices. Many people have made
345: generous contributions to the wide range of software distributed
346: through that system in reliance on consistent application of that
347: system; it is up to the author/donor to decide if he or she is willing
348: to distribute software through any other system and a licensee cannot
349: impose that choice.
350:
351: This section is intended to make thoroughly clear what is believed to
352: be a consequence of the rest of this License.
353:
354: @item
355: If the distribution and/or use of the Program is restricted in
356: certain countries either by patents or by copyrighted interfaces, the
357: original copyright holder who places the Program under this License
358: may add an explicit geographical distribution limitation excluding
359: those countries, so that distribution is permitted only in or among
360: countries not thus excluded. In such case, this License incorporates
361: the limitation as if written in the body of this License.
362:
363: @item
364: The Free Software Foundation may publish revised and/or new versions
365: of the General Public License from time to time. Such new versions will
366: be similar in spirit to the present version, but may differ in detail to
367: address new problems or concerns.
368:
369: Each version is given a distinguishing version number. If the Program
370: specifies a version number of this License which applies to it and ``any
371: later version'', you have the option of following the terms and conditions
372: either of that version or of any later version published by the Free
373: Software Foundation. If the Program does not specify a version number of
374: this License, you may choose any version ever published by the Free Software
375: Foundation.
376:
377: @item
378: If you wish to incorporate parts of the Program into other free
379: programs whose distribution conditions are different, write to the author
380: to ask for permission. For software which is copyrighted by the Free
381: Software Foundation, write to the Free Software Foundation; we sometimes
382: make exceptions for this. Our decision will be guided by the two goals
383: of preserving the free status of all derivatives of our free software and
384: of promoting the sharing and reuse of software generally.
385:
386: @iftex
387: @heading NO WARRANTY
388: @end iftex
389: @ifinfo
390: @center NO WARRANTY
391: @end ifinfo
392:
393: @item
394: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
395: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
396: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
397: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
398: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
399: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
400: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
401: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
402: REPAIR OR CORRECTION.
403:
404: @item
405: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
406: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
407: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
408: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
409: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
410: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
411: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
412: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
413: POSSIBILITY OF SUCH DAMAGES.
414: @end enumerate
415:
416: @iftex
417: @heading END OF TERMS AND CONDITIONS
418: @end iftex
419: @ifinfo
420: @center END OF TERMS AND CONDITIONS
421: @end ifinfo
422:
423: @page
424: @unnumberedsec How to Apply These Terms to Your New Programs
425:
426: If you develop a new program, and you want it to be of the greatest
427: possible use to the public, the best way to achieve this is to make it
428: free software which everyone can redistribute and change under these terms.
429:
430: To do so, attach the following notices to the program. It is safest
431: to attach them to the start of each source file to most effectively
432: convey the exclusion of warranty; and each file should have at least
433: the ``copyright'' line and a pointer to where the full notice is found.
434:
435: @smallexample
436: @var{one line to give the program's name and a brief idea of what it does.}
437: Copyright (C) 19@var{yy} @var{name of author}
438:
439: This program is free software; you can redistribute it and/or modify
440: it under the terms of the GNU General Public License as published by
441: the Free Software Foundation; either version 2 of the License, or
442: (at your option) any later version.
443:
444: This program is distributed in the hope that it will be useful,
445: but WITHOUT ANY WARRANTY; without even the implied warranty of
446: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
447: GNU General Public License for more details.
448:
449: You should have received a copy of the GNU General Public License
450: along with this program; if not, write to the Free Software
451: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
452: @end smallexample
453:
454: Also add information on how to contact you by electronic and paper mail.
455:
456: If the program is interactive, make it output a short notice like this
457: when it starts in an interactive mode:
458:
459: @smallexample
460: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
461: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
462: type `show w'.
463: This is free software, and you are welcome to redistribute it
464: under certain conditions; type `show c' for details.
465: @end smallexample
466:
467: The hypothetical commands @samp{show w} and @samp{show c} should show
468: the appropriate parts of the General Public License. Of course, the
469: commands you use may be called something other than @samp{show w} and
470: @samp{show c}; they could even be mouse-clicks or menu items---whatever
471: suits your program.
472:
473: You should also get your employer (if you work as a programmer) or your
474: school, if any, to sign a ``copyright disclaimer'' for the program, if
475: necessary. Here is a sample; alter the names:
476:
477: @smallexample
478: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
479: `Gnomovision' (which makes passes at compilers) written by James Hacker.
480:
481: @var{signature of Ty Coon}, 1 April 1989
482: Ty Coon, President of Vice
483: @end smallexample
484:
485: This General Public License does not permit incorporating your program into
486: proprietary programs. If your program is a subroutine library, you may
487: consider it more useful to permit linking proprietary applications with the
488: library. If this is what you want to do, use the GNU Library General
489: Public License instead of this License.
1.1 anton 490:
491: @iftex
1.23 pazsan 492: @node Preface
493: @comment node-name, next, previous, up
1.1 anton 494: @unnumbered Preface
1.23 pazsan 495: @cindex Preface
1.17 anton 496: This manual documents Gforth. The reader is expected to know
1.1 anton 497: Forth. This manual is primarily a reference manual. @xref{Other Books}
498: for introductory material.
499: @end iftex
500:
501: @node Goals, Other Books, License, Top
502: @comment node-name, next, previous, up
1.17 anton 503: @chapter Goals of Gforth
1.1 anton 504: @cindex Goals
1.17 anton 505: The goal of the Gforth Project is to develop a standard model for
1.1 anton 506: ANSI Forth. This can be split into several subgoals:
507:
508: @itemize @bullet
509: @item
1.17 anton 510: Gforth should conform to the ANSI Forth standard.
1.1 anton 511: @item
512: It should be a model, i.e. it should define all the
513: implementation-dependent things.
514: @item
515: It should become standard, i.e. widely accepted and used. This goal
516: is the most difficult one.
517: @end itemize
518:
1.17 anton 519: To achieve these goals Gforth should be
1.1 anton 520: @itemize @bullet
521: @item
522: Similar to previous models (fig-Forth, F83)
523: @item
524: Powerful. It should provide for all the things that are considered
525: necessary today and even some that are not yet considered necessary.
526: @item
527: Efficient. It should not get the reputation of being exceptionally
528: slow.
529: @item
530: Free.
531: @item
532: Available on many machines/easy to port.
533: @end itemize
534:
1.17 anton 535: Have we achieved these goals? Gforth conforms to the ANS Forth
536: standard. It may be considered a model, but we have not yet documented
1.1 anton 537: which parts of the model are stable and which parts we are likely to
1.17 anton 538: change. It certainly has not yet become a de facto standard. It has some
539: similarities and some differences to previous models. It has some
540: powerful features, but not yet everything that we envisioned. We
541: certainly have achieved our execution speed goals (@pxref{Performance}).
542: It is free and available on many machines.
1.1 anton 543:
544: @node Other Books, Invocation, Goals, Top
545: @chapter Other books on ANS Forth
546:
547: As the standard is relatively new, there are not many books out yet. It
1.17 anton 548: is not recommended to learn Forth by using Gforth and a book that is
1.1 anton 549: not written for ANS Forth, as you will not know your mistakes from the
550: deviations of the book.
551:
552: There is, of course, the standard, the definite reference if you want to
1.19 anton 553: write ANS Forth programs. It is available in printed form from the
554: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
555: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
556: can also get it from Global Engineering Documents (Tel.: USA (800)
557: 854-7179; Fax.: (303) 843-9880) for about $300.
558:
559: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
560: for publication is available electronically and for free in some MS Word
561: format, and it has been converted to HTML. Some pointers to these
562: versions can be found through
1.24 anton 563: @*@file{http://www.complang.tuwien.ac.at/projects/forth.html}.
1.1 anton 564:
1.21 anton 565: @cite{Forth: The new model} by Jack Woehr (Prentice-Hall, 1993) is an
1.1 anton 566: introductory book based on a draft version of the standard. It does not
567: cover the whole standard. It also contains interesting background
568: information (Jack Woehr was in the ANS Forth Technical Committe). It is
569: not appropriate for complete newbies, but programmers experienced in
570: other languages should find it ok.
571:
572: @node Invocation, Words, Other Books, Top
573: @chapter Invocation
574:
575: You will usually just say @code{gforth}. In many other cases the default
1.17 anton 576: Gforth image will be invoked like this:
1.1 anton 577:
578: @example
579: gforth [files] [-e forth-code]
580: @end example
581:
582: executing the contents of the files and the Forth code in the order they
583: are given.
584:
585: In general, the command line looks like this:
586:
587: @example
588: gforth [initialization options] [image-specific options]
589: @end example
590:
591: The initialization options must come before the rest of the command
592: line. They are:
593:
594: @table @code
595: @item --image-file @var{file}
1.20 pazsan 596: @item -i @var{file}
1.1 anton 597: Loads the Forth image @var{file} instead of the default
598: @file{gforth.fi}.
599:
600: @item --path @var{path}
1.20 pazsan 601: @item -p @var{path}
1.1 anton 602: Uses @var{path} for searching the image file and Forth source code
603: files instead of the default in the environment variable
604: @code{GFORTHPATH} or the path specified at installation time (typically
605: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
606: list.
607:
608: @item --dictionary-size @var{size}
609: @item -m @var{size}
610: Allocate @var{size} space for the Forth dictionary space instead of
611: using the default specified in the image (typically 256K). The
612: @var{size} specification consists of an integer and a unit (e.g.,
613: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
614: size, in this case Cells), @code{k} (kilobytes), and @code{M}
615: (Megabytes). If no unit is specified, @code{e} is used.
616:
617: @item --data-stack-size @var{size}
618: @item -d @var{size}
619: Allocate @var{size} space for the data stack instead of using the
620: default specified in the image (typically 16K).
621:
622: @item --return-stack-size @var{size}
623: @item -r @var{size}
624: Allocate @var{size} space for the return stack instead of using the
625: default specified in the image (typically 16K).
626:
627: @item --fp-stack-size @var{size}
628: @item -f @var{size}
629: Allocate @var{size} space for the floating point stack instead of
630: using the default specified in the image (typically 16K). In this case
631: the unit specifier @code{e} refers to floating point numbers.
632:
633: @item --locals-stack-size @var{size}
634: @item -l @var{size}
635: Allocate @var{size} space for the locals stack instead of using the
636: default specified in the image (typically 16K).
637:
638: @end table
639:
640: As explained above, the image-specific command-line arguments for the
641: default image @file{gforth.fi} consist of a sequence of filenames and
642: @code{-e @var{forth-code}} options that are interpreted in the seqence
643: in which they are given. The @code{-e @var{forth-code}} or
644: @code{--evaluate @var{forth-code}} option evaluates the forth
645: code. This option takes only one argument; if you want to evaluate more
646: Forth words, you have to quote them or use several @code{-e}s. To exit
647: after processing the command line (instead of entering interactive mode)
648: append @code{-e bye} to the command line.
649:
1.22 anton 650: If you have several versions of Gforth installed, @code{gforth} will
651: invoke the version that was installed last. @code{gforth-@var{version}}
652: invokes a specific version. You may want to use the option
653: @code{--path}, if your environment contains the variable
654: @code{GFORTHPATH}.
655:
1.1 anton 656: Not yet implemented:
657: On startup the system first executes the system initialization file
658: (unless the option @code{--no-init-file} is given; note that the system
659: resulting from using this option may not be ANS Forth conformant). Then
660: the user initialization file @file{.gforth.fs} is executed, unless the
661: option @code{--no-rc} is given; this file is first searched in @file{.},
662: then in @file{~}, then in the normal path (see above).
663:
1.4 anton 664: @node Words, ANS conformance, Invocation, Top
1.1 anton 665: @chapter Forth Words
666:
667: @menu
1.4 anton 668: * Notation::
669: * Arithmetic::
670: * Stack Manipulation::
671: * Memory access::
672: * Control Structures::
673: * Locals::
674: * Defining Words::
675: * Wordlists::
676: * Files::
677: * Blocks::
678: * Other I/O::
679: * Programming Tools::
1.18 anton 680: * Assembler and Code words::
1.4 anton 681: * Threading Words::
1.1 anton 682: @end menu
683:
684: @node Notation, Arithmetic, Words, Words
685: @section Notation
686:
687: The Forth words are described in this section in the glossary notation
688: that has become a de-facto standard for Forth texts, i.e.
689:
1.4 anton 690: @format
1.1 anton 691: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
1.4 anton 692: @end format
1.1 anton 693: @var{Description}
694:
695: @table @var
696: @item word
1.17 anton 697: The name of the word. BTW, Gforth is case insensitive, so you can
1.14 anton 698: type the words in in lower case (However, @pxref{core-idef}).
1.1 anton 699:
700: @item Stack effect
701: The stack effect is written in the notation @code{@var{before} --
702: @var{after}}, where @var{before} and @var{after} describe the top of
703: stack entries before and after the execution of the word. The rest of
704: the stack is not touched by the word. The top of stack is rightmost,
1.17 anton 705: i.e., a stack sequence is written as it is typed in. Note that Gforth
1.1 anton 706: uses a separate floating point stack, but a unified stack
707: notation. Also, return stack effects are not shown in @var{stack
708: effect}, but in @var{Description}. The name of a stack item describes
709: the type and/or the function of the item. See below for a discussion of
710: the types.
711:
1.19 anton 712: All words have two stack effects: A compile-time stack effect and a
713: run-time stack effect. The compile-time stack-effect of most words is
714: @var{ -- }. If the compile-time stack-effect of a word deviates from
715: this standard behaviour, or the word does other unusual things at
716: compile time, both stack effects are shown; otherwise only the run-time
717: stack effect is shown.
718:
1.1 anton 719: @item pronunciation
720: How the word is pronounced
721:
722: @item wordset
723: The ANS Forth standard is divided into several wordsets. A standard
724: system need not support all of them. So, the fewer wordsets your program
725: uses the more portable it will be in theory. However, we suspect that
726: most ANS Forth systems on personal machines will feature all
727: wordsets. Words that are not defined in the ANS standard have
1.19 anton 728: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
729: describes words that will work in future releases of Gforth;
730: @code{gforth-internal} words are more volatile. Environmental query
731: strings are also displayed like words; you can recognize them by the
732: @code{environment} in the wordset field.
1.1 anton 733:
734: @item Description
735: A description of the behaviour of the word.
736: @end table
737:
1.4 anton 738: The type of a stack item is specified by the character(s) the name
739: starts with:
1.1 anton 740:
741: @table @code
742: @item f
743: Bool, i.e. @code{false} or @code{true}.
744: @item c
745: Char
746: @item w
747: Cell, can contain an integer or an address
748: @item n
749: signed integer
750: @item u
751: unsigned integer
752: @item d
753: double sized signed integer
754: @item ud
755: double sized unsigned integer
756: @item r
757: Float
758: @item a_
759: Cell-aligned address
760: @item c_
761: Char-aligned address (note that a Char is two bytes in Windows NT)
762: @item f_
763: Float-aligned address
764: @item df_
765: Address aligned for IEEE double precision float
766: @item sf_
767: Address aligned for IEEE single precision float
768: @item xt
769: Execution token, same size as Cell
770: @item wid
771: Wordlist ID, same size as Cell
772: @item f83name
773: Pointer to a name structure
774: @end table
775:
1.4 anton 776: @node Arithmetic, Stack Manipulation, Notation, Words
1.1 anton 777: @section Arithmetic
778: Forth arithmetic is not checked, i.e., you will not hear about integer
779: overflow on addition or multiplication, you may hear about division by
780: zero if you are lucky. The operator is written after the operands, but
781: the operands are still in the original order. I.e., the infix @code{2-1}
782: corresponds to @code{2 1 -}. Forth offers a variety of division
783: operators. If you perform division with potentially negative operands,
784: you do not want to use @code{/} or @code{/mod} with its undefined
785: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
1.4 anton 786: former, @pxref{Mixed precision}).
787:
788: @menu
789: * Single precision::
790: * Bitwise operations::
791: * Mixed precision:: operations with single and double-cell integers
792: * Double precision:: Double-cell integer arithmetic
793: * Floating Point::
794: @end menu
1.1 anton 795:
1.4 anton 796: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
1.1 anton 797: @subsection Single precision
798: doc-+
799: doc--
800: doc-*
801: doc-/
802: doc-mod
803: doc-/mod
804: doc-negate
805: doc-abs
806: doc-min
807: doc-max
808:
1.4 anton 809: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
1.1 anton 810: @subsection Bitwise operations
811: doc-and
812: doc-or
813: doc-xor
814: doc-invert
815: doc-2*
816: doc-2/
817:
1.4 anton 818: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
1.1 anton 819: @subsection Mixed precision
820: doc-m+
821: doc-*/
822: doc-*/mod
823: doc-m*
824: doc-um*
825: doc-m*/
826: doc-um/mod
827: doc-fm/mod
828: doc-sm/rem
829:
1.4 anton 830: @node Double precision, Floating Point, Mixed precision, Arithmetic
1.1 anton 831: @subsection Double precision
1.16 anton 832:
833: The outer (aka text) interpreter converts numbers containing a dot into
834: a double precision number. Note that only numbers with the dot as last
835: character are standard-conforming.
836:
1.1 anton 837: doc-d+
838: doc-d-
839: doc-dnegate
840: doc-dabs
841: doc-dmin
842: doc-dmax
843:
1.4 anton 844: @node Floating Point, , Double precision, Arithmetic
845: @subsection Floating Point
1.16 anton 846:
847: The format of floating point numbers recognized by the outer (aka text)
848: interpreter is: a signed decimal number, possibly containing a decimal
849: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
850: by a signed integer (the exponent). E.g., @code{1e} ist the same as
851: @code{+1.0e+1}. Note that a number without @code{e}
852: is not interpreted as floating-point number, but as double (if the
853: number contains a @code{.}) or single precision integer. Also,
854: conversions between string and floating point numbers always use base
855: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
856: value greater then 14, the @code{E} may be interpreted as digit and the
857: number will be interpreted as integer, unless it has a signed exponent
858: (both @code{+} and @code{-} are allowed as signs).
1.4 anton 859:
860: Angles in floating point operations are given in radians (a full circle
1.17 anton 861: has 2 pi radians). Note, that Gforth has a separate floating point
1.4 anton 862: stack, but we use the unified notation.
863:
864: Floating point numbers have a number of unpleasant surprises for the
865: unwary (e.g., floating point addition is not associative) and even a few
866: for the wary. You should not use them unless you know what you are doing
867: or you don't care that the results you get are totally bogus. If you
868: want to learn about the problems of floating point numbers (and how to
1.11 anton 869: avoid them), you might start with @cite{David Goldberg, What Every
1.6 anton 870: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
871: Computing Surveys 23(1):5@minus{}48, March 1991}.
1.4 anton 872:
873: doc-f+
874: doc-f-
875: doc-f*
876: doc-f/
877: doc-fnegate
878: doc-fabs
879: doc-fmax
880: doc-fmin
881: doc-floor
882: doc-fround
883: doc-f**
884: doc-fsqrt
885: doc-fexp
886: doc-fexpm1
887: doc-fln
888: doc-flnp1
889: doc-flog
1.6 anton 890: doc-falog
1.4 anton 891: doc-fsin
892: doc-fcos
893: doc-fsincos
894: doc-ftan
895: doc-fasin
896: doc-facos
897: doc-fatan
898: doc-fatan2
899: doc-fsinh
900: doc-fcosh
901: doc-ftanh
902: doc-fasinh
903: doc-facosh
904: doc-fatanh
905:
906: @node Stack Manipulation, Memory access, Arithmetic, Words
1.1 anton 907: @section Stack Manipulation
908:
1.17 anton 909: Gforth has a data stack (aka parameter stack) for characters, cells,
1.1 anton 910: addresses, and double cells, a floating point stack for floating point
911: numbers, a return stack for storing the return addresses of colon
912: definitions and other data, and a locals stack for storing local
913: variables. Note that while every sane Forth has a separate floating
914: point stack, this is not strictly required; an ANS Forth system could
915: theoretically keep floating point numbers on the data stack. As an
916: additional difficulty, you don't know how many cells a floating point
917: number takes. It is reportedly possible to write words in a way that
918: they work also for a unified stack model, but we do not recommend trying
1.4 anton 919: it. Instead, just say that your program has an environmental dependency
920: on a separate FP stack.
921:
922: Also, a Forth system is allowed to keep the local variables on the
1.1 anton 923: return stack. This is reasonable, as local variables usually eliminate
924: the need to use the return stack explicitly. So, if you want to produce
925: a standard complying program and if you are using local variables in a
926: word, forget about return stack manipulations in that word (see the
927: standard document for the exact rules).
928:
1.4 anton 929: @menu
930: * Data stack::
931: * Floating point stack::
932: * Return stack::
933: * Locals stack::
934: * Stack pointer manipulation::
935: @end menu
936:
937: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1.1 anton 938: @subsection Data stack
939: doc-drop
940: doc-nip
941: doc-dup
942: doc-over
943: doc-tuck
944: doc-swap
945: doc-rot
946: doc--rot
947: doc-?dup
948: doc-pick
949: doc-roll
950: doc-2drop
951: doc-2nip
952: doc-2dup
953: doc-2over
954: doc-2tuck
955: doc-2swap
956: doc-2rot
957:
1.4 anton 958: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1.1 anton 959: @subsection Floating point stack
960: doc-fdrop
961: doc-fnip
962: doc-fdup
963: doc-fover
964: doc-ftuck
965: doc-fswap
966: doc-frot
967:
1.4 anton 968: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1.1 anton 969: @subsection Return stack
970: doc->r
971: doc-r>
972: doc-r@
973: doc-rdrop
974: doc-2>r
975: doc-2r>
976: doc-2r@
977: doc-2rdrop
978:
1.4 anton 979: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1.1 anton 980: @subsection Locals stack
981:
1.4 anton 982: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
1.1 anton 983: @subsection Stack pointer manipulation
984: doc-sp@
985: doc-sp!
986: doc-fp@
987: doc-fp!
988: doc-rp@
989: doc-rp!
990: doc-lp@
991: doc-lp!
992:
1.4 anton 993: @node Memory access, Control Structures, Stack Manipulation, Words
1.1 anton 994: @section Memory access
995:
1.4 anton 996: @menu
997: * Stack-Memory transfers::
998: * Address arithmetic::
999: * Memory block access::
1000: @end menu
1001:
1002: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1.1 anton 1003: @subsection Stack-Memory transfers
1004:
1005: doc-@
1006: doc-!
1007: doc-+!
1008: doc-c@
1009: doc-c!
1010: doc-2@
1011: doc-2!
1012: doc-f@
1013: doc-f!
1014: doc-sf@
1015: doc-sf!
1016: doc-df@
1017: doc-df!
1018:
1.4 anton 1019: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1.1 anton 1020: @subsection Address arithmetic
1021:
1022: ANS Forth does not specify the sizes of the data types. Instead, it
1023: offers a number of words for computing sizes and doing address
1024: arithmetic. Basically, address arithmetic is performed in terms of
1025: address units (aus); on most systems the address unit is one byte. Note
1026: that a character may have more than one au, so @code{chars} is no noop
1027: (on systems where it is a noop, it compiles to nothing).
1028:
1029: ANS Forth also defines words for aligning addresses for specific
1030: addresses. Many computers require that accesses to specific data types
1031: must only occur at specific addresses; e.g., that cells may only be
1032: accessed at addresses divisible by 4. Even if a machine allows unaligned
1033: accesses, it can usually perform aligned accesses faster.
1034:
1.17 anton 1035: For the performance-conscious: alignment operations are usually only
1.1 anton 1036: necessary during the definition of a data structure, not during the
1037: (more frequent) accesses to it.
1038:
1039: ANS Forth defines no words for character-aligning addresses. This is not
1040: an oversight, but reflects the fact that addresses that are not
1041: char-aligned have no use in the standard and therefore will not be
1042: created.
1043:
1044: The standard guarantees that addresses returned by @code{CREATE}d words
1.17 anton 1045: are cell-aligned; in addition, Gforth guarantees that these addresses
1.1 anton 1046: are aligned for all purposes.
1047:
1.9 anton 1048: Note that the standard defines a word @code{char}, which has nothing to
1049: do with address arithmetic.
1050:
1.1 anton 1051: doc-chars
1052: doc-char+
1053: doc-cells
1054: doc-cell+
1055: doc-align
1056: doc-aligned
1057: doc-floats
1058: doc-float+
1059: doc-falign
1060: doc-faligned
1061: doc-sfloats
1062: doc-sfloat+
1063: doc-sfalign
1064: doc-sfaligned
1065: doc-dfloats
1066: doc-dfloat+
1067: doc-dfalign
1068: doc-dfaligned
1.10 anton 1069: doc-maxalign
1070: doc-maxaligned
1071: doc-cfalign
1072: doc-cfaligned
1.1 anton 1073: doc-address-unit-bits
1074:
1.4 anton 1075: @node Memory block access, , Address arithmetic, Memory access
1.1 anton 1076: @subsection Memory block access
1077:
1078: doc-move
1079: doc-erase
1080:
1081: While the previous words work on address units, the rest works on
1082: characters.
1083:
1084: doc-cmove
1085: doc-cmove>
1086: doc-fill
1087: doc-blank
1088:
1.4 anton 1089: @node Control Structures, Locals, Memory access, Words
1.1 anton 1090: @section Control Structures
1091:
1092: Control structures in Forth cannot be used in interpret state, only in
1093: compile state, i.e., in a colon definition. We do not like this
1094: limitation, but have not seen a satisfying way around it yet, although
1095: many schemes have been proposed.
1096:
1.4 anton 1097: @menu
1098: * Selection::
1099: * Simple Loops::
1100: * Counted Loops::
1101: * Arbitrary control structures::
1102: * Calls and returns::
1103: * Exception Handling::
1104: @end menu
1105:
1106: @node Selection, Simple Loops, Control Structures, Control Structures
1.1 anton 1107: @subsection Selection
1108:
1109: @example
1110: @var{flag}
1111: IF
1112: @var{code}
1113: ENDIF
1114: @end example
1115: or
1116: @example
1117: @var{flag}
1118: IF
1119: @var{code1}
1120: ELSE
1121: @var{code2}
1122: ENDIF
1123: @end example
1124:
1.4 anton 1125: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1.1 anton 1126: standard, and @code{ENDIF} is not, although it is quite popular. We
1127: recommend using @code{ENDIF}, because it is less confusing for people
1128: who also know other languages (and is not prone to reinforcing negative
1129: prejudices against Forth in these people). Adding @code{ENDIF} to a
1130: system that only supplies @code{THEN} is simple:
1131: @example
1132: : endif POSTPONE then ; immediate
1133: @end example
1134:
1135: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1136: (adv.)} has the following meanings:
1137: @quotation
1138: ... 2b: following next after in order ... 3d: as a necessary consequence
1139: (if you were there, then you saw them).
1140: @end quotation
1141: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1142: and many other programming languages has the meaning 3d.]
1143:
1144: We also provide the words @code{?dup-if} and @code{?dup-0=-if}, so you
1145: can avoid using @code{?dup}.
1146:
1147: @example
1148: @var{n}
1149: CASE
1150: @var{n1} OF @var{code1} ENDOF
1151: @var{n2} OF @var{code2} ENDOF
1.4 anton 1152: @dots{}
1.1 anton 1153: ENDCASE
1154: @end example
1155:
1156: Executes the first @var{codei}, where the @var{ni} is equal to
1157: @var{n}. A default case can be added by simply writing the code after
1158: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1159: but must not consume it.
1160:
1.4 anton 1161: @node Simple Loops, Counted Loops, Selection, Control Structures
1.1 anton 1162: @subsection Simple Loops
1163:
1164: @example
1165: BEGIN
1166: @var{code1}
1167: @var{flag}
1168: WHILE
1169: @var{code2}
1170: REPEAT
1171: @end example
1172:
1173: @var{code1} is executed and @var{flag} is computed. If it is true,
1174: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
1175:
1176: @example
1177: BEGIN
1178: @var{code}
1179: @var{flag}
1180: UNTIL
1181: @end example
1182:
1183: @var{code} is executed. The loop is restarted if @code{flag} is false.
1184:
1185: @example
1186: BEGIN
1187: @var{code}
1188: AGAIN
1189: @end example
1190:
1191: This is an endless loop.
1192:
1.4 anton 1193: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1.1 anton 1194: @subsection Counted Loops
1195:
1196: The basic counted loop is:
1197: @example
1198: @var{limit} @var{start}
1199: ?DO
1200: @var{body}
1201: LOOP
1202: @end example
1203:
1204: This performs one iteration for every integer, starting from @var{start}
1205: and up to, but excluding @var{limit}. The counter, aka index, can be
1206: accessed with @code{i}. E.g., the loop
1207: @example
1208: 10 0 ?DO
1209: i .
1210: LOOP
1211: @end example
1212: prints
1213: @example
1214: 0 1 2 3 4 5 6 7 8 9
1215: @end example
1216: The index of the innermost loop can be accessed with @code{i}, the index
1217: of the next loop with @code{j}, and the index of the third loop with
1218: @code{k}.
1219:
1220: The loop control data are kept on the return stack, so there are some
1221: restrictions on mixing return stack accesses and counted loop
1222: words. E.g., if you put values on the return stack outside the loop, you
1223: cannot read them inside the loop. If you put values on the return stack
1224: within a loop, you have to remove them before the end of the loop and
1225: before accessing the index of the loop.
1226:
1227: There are several variations on the counted loop:
1228:
1229: @code{LEAVE} leaves the innermost counted loop immediately.
1230:
1.18 anton 1231: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1232: (and @code{LOOP} iterates until they become equal by wrap-around
1233: arithmetic). This behaviour is usually not what you want. Therefore,
1234: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1235: @code{?DO}), which do not enter the loop if @var{start} is greater than
1236: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1237: unsigned loop parameters. These words can be implemented easily on
1238: standard systems, so using them does not make your programs hard to
1239: port; e.g.:
1240: @example
1241: : +DO ( compile-time: -- do-sys; run-time: n1 n2 -- )
1242: POSTPONE over POSTPONE min POSTPONE ?DO ; immediate
1243: @end example
1244:
1.1 anton 1245: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1246: index by @var{n} instead of by 1. The loop is terminated when the border
1247: between @var{limit-1} and @var{limit} is crossed. E.g.:
1248:
1.18 anton 1249: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1.1 anton 1250:
1.18 anton 1251: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1.1 anton 1252:
1253: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1254:
1.2 anton 1255: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1.1 anton 1256:
1.2 anton 1257: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1.1 anton 1258:
1.18 anton 1259: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1260: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1261: index by @var{u} each iteration. The loop is terminated when the border
1262: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1263: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1 anton 1264:
1.18 anton 1265: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1.1 anton 1266:
1.18 anton 1267: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1.1 anton 1268:
1.18 anton 1269: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1.1 anton 1270:
1.18 anton 1271: Another alternative is @code{@var{n} S+LOOP}, where the negative
1272: case behaves symmetrical to the positive case:
1.1 anton 1273:
1.18 anton 1274: @code{-2 0 -DO i . -1 S+LOOP} prints @code{0 -1}
1275:
1276: The loop is terminated when the border between @var{limit@minus{}sgn(n)}
1277: and @var{limit} is crossed. Unfortunately, neither @code{-LOOP} nor
1278: @code{S+LOOP} are part of the ANS Forth standard, and they are not easy
1279: to implement using standard words. If you want to write standard
1280: programs, just avoid counting down.
1281:
1282: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1283: the loop, independent of the loop parameters. Do not use @code{DO}, even
1284: if you know that the loop is entered in any case. Such knowledge tends
1285: to become invalid during maintenance of a program, and then the
1286: @code{DO} will make trouble.
1.1 anton 1287:
1288: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1289: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1290: return stack so @code{EXIT} can get to its return address.
1291:
1292: Another counted loop is
1293: @example
1294: @var{n}
1295: FOR
1296: @var{body}
1297: NEXT
1298: @end example
1299: This is the preferred loop of native code compiler writers who are too
1.17 anton 1300: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1 anton 1301: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1302: and ending with 0. Other Forth systems may behave differently, even if
1303: they support @code{FOR} loops.
1304:
1.4 anton 1305: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2 anton 1306: @subsection Arbitrary control structures
1307:
1308: ANS Forth permits and supports using control structures in a non-nested
1309: way. Information about incomplete control structures is stored on the
1310: control-flow stack. This stack may be implemented on the Forth data
1.17 anton 1311: stack, and this is what we have done in Gforth.
1.2 anton 1312:
1313: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1314: entry represents a backward branch target. A few words are the basis for
1315: building any control structure possible (except control structures that
1316: need storage, like calls, coroutines, and backtracking).
1317:
1.3 anton 1318: doc-if
1319: doc-ahead
1320: doc-then
1321: doc-begin
1322: doc-until
1323: doc-again
1324: doc-cs-pick
1325: doc-cs-roll
1.2 anton 1326:
1.17 anton 1327: On many systems control-flow stack items take one word, in Gforth they
1.2 anton 1328: currently take three (this may change in the future). Therefore it is a
1329: really good idea to manipulate the control flow stack with
1330: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1331: words.
1332:
1333: Some standard control structure words are built from these words:
1334:
1.3 anton 1335: doc-else
1336: doc-while
1337: doc-repeat
1.2 anton 1338:
1339: Counted loop words constitute a separate group of words:
1340:
1.3 anton 1341: doc-?do
1.18 anton 1342: doc-+do
1343: doc-u+do
1344: doc--do
1345: doc-u-do
1.3 anton 1346: doc-do
1347: doc-for
1348: doc-loop
1349: doc-s+loop
1350: doc-+loop
1.18 anton 1351: doc--loop
1.3 anton 1352: doc-next
1353: doc-leave
1354: doc-?leave
1355: doc-unloop
1.10 anton 1356: doc-done
1.2 anton 1357:
1358: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1359: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1360: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3 anton 1361: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1362: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7 pazsan 1363: resolved (by using one of the loop-ending words or @code{DONE}).
1.2 anton 1364:
1365: Another group of control structure words are
1366:
1.3 anton 1367: doc-case
1368: doc-endcase
1369: doc-of
1370: doc-endof
1.2 anton 1371:
1372: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1373: @code{cs-roll}.
1374:
1.3 anton 1375: @subsubsection Programming Style
1376:
1377: In order to ensure readability we recommend that you do not create
1378: arbitrary control structures directly, but define new control structure
1379: words for the control structure you want and use these words in your
1380: program.
1381:
1382: E.g., instead of writing
1383:
1384: @example
1385: begin
1386: ...
1387: if [ 1 cs-roll ]
1388: ...
1389: again then
1390: @end example
1391:
1392: we recommend defining control structure words, e.g.,
1393:
1394: @example
1395: : while ( dest -- orig dest )
1396: POSTPONE if
1397: 1 cs-roll ; immediate
1398:
1399: : repeat ( orig dest -- )
1400: POSTPONE again
1401: POSTPONE then ; immediate
1402: @end example
1403:
1404: and then using these to create the control structure:
1405:
1406: @example
1407: begin
1408: ...
1409: while
1410: ...
1411: repeat
1412: @end example
1413:
1414: That's much easier to read, isn't it? Of course, @code{BEGIN} and
1415: @code{WHILE} are predefined, so in this example it would not be
1416: necessary to define them.
1417:
1.4 anton 1418: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3 anton 1419: @subsection Calls and returns
1420:
1421: A definition can be called simply be writing the name of the
1.17 anton 1422: definition. When the end of the definition is reached, it returns. An
1423: earlier return can be forced using
1.3 anton 1424:
1425: doc-exit
1426:
1427: Don't forget to clean up the return stack and @code{UNLOOP} any
1428: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1429: primitive compiled by @code{EXIT} is
1430:
1431: doc-;s
1432:
1.4 anton 1433: @node Exception Handling, , Calls and returns, Control Structures
1.3 anton 1434: @subsection Exception Handling
1435:
1436: doc-catch
1437: doc-throw
1438:
1.4 anton 1439: @node Locals, Defining Words, Control Structures, Words
1.1 anton 1440: @section Locals
1441:
1.2 anton 1442: Local variables can make Forth programming more enjoyable and Forth
1443: programs easier to read. Unfortunately, the locals of ANS Forth are
1444: laden with restrictions. Therefore, we provide not only the ANS Forth
1445: locals wordset, but also our own, more powerful locals wordset (we
1446: implemented the ANS Forth locals wordset through our locals wordset).
1447:
1.24 anton 1448: The ideas in this section have also been published in the paper
1449: @cite{Automatic Scoping of Local Variables} by M. Anton Ertl, presented
1450: at EuroForth '94; it is available at
1451: @*@file{http://www.complang.tuwien.ac.at/papers/ertl94l.ps.gz}.
1452:
1.2 anton 1453: @menu
1.17 anton 1454: * Gforth locals::
1.4 anton 1455: * ANS Forth locals::
1.2 anton 1456: @end menu
1457:
1.17 anton 1458: @node Gforth locals, ANS Forth locals, Locals, Locals
1459: @subsection Gforth locals
1.2 anton 1460:
1461: Locals can be defined with
1462:
1463: @example
1464: @{ local1 local2 ... -- comment @}
1465: @end example
1466: or
1467: @example
1468: @{ local1 local2 ... @}
1469: @end example
1470:
1471: E.g.,
1472: @example
1473: : max @{ n1 n2 -- n3 @}
1474: n1 n2 > if
1475: n1
1476: else
1477: n2
1478: endif ;
1479: @end example
1480:
1481: The similarity of locals definitions with stack comments is intended. A
1482: locals definition often replaces the stack comment of a word. The order
1483: of the locals corresponds to the order in a stack comment and everything
1484: after the @code{--} is really a comment.
1485:
1486: This similarity has one disadvantage: It is too easy to confuse locals
1487: declarations with stack comments, causing bugs and making them hard to
1488: find. However, this problem can be avoided by appropriate coding
1489: conventions: Do not use both notations in the same program. If you do,
1490: they should be distinguished using additional means, e.g. by position.
1491:
1492: The name of the local may be preceded by a type specifier, e.g.,
1493: @code{F:} for a floating point value:
1494:
1495: @example
1496: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1497: \ complex multiplication
1498: Ar Br f* Ai Bi f* f-
1499: Ar Bi f* Ai Br f* f+ ;
1500: @end example
1501:
1.17 anton 1502: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2 anton 1503: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1504: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1505: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1506: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1507: produces its address (which becomes invalid when the variable's scope is
1508: left). E.g., the standard word @code{emit} can be defined in therms of
1509: @code{type} like this:
1510:
1511: @example
1512: : emit @{ C^ char* -- @}
1513: char* 1 type ;
1514: @end example
1515:
1516: A local without type specifier is a @code{W:} local. Both flavours of
1517: locals are initialized with values from the data or FP stack.
1518:
1519: Currently there is no way to define locals with user-defined data
1520: structures, but we are working on it.
1521:
1.17 anton 1522: Gforth allows defining locals everywhere in a colon definition. This
1.7 pazsan 1523: poses the following questions:
1.2 anton 1524:
1.4 anton 1525: @menu
1526: * Where are locals visible by name?::
1.14 anton 1527: * How long do locals live?::
1.4 anton 1528: * Programming Style::
1529: * Implementation::
1530: @end menu
1531:
1.17 anton 1532: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2 anton 1533: @subsubsection Where are locals visible by name?
1534:
1535: Basically, the answer is that locals are visible where you would expect
1536: it in block-structured languages, and sometimes a little longer. If you
1537: want to restrict the scope of a local, enclose its definition in
1538: @code{SCOPE}...@code{ENDSCOPE}.
1539:
1540: doc-scope
1541: doc-endscope
1542:
1543: These words behave like control structure words, so you can use them
1544: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1545: arbitrary ways.
1546:
1547: If you want a more exact answer to the visibility question, here's the
1548: basic principle: A local is visible in all places that can only be
1549: reached through the definition of the local@footnote{In compiler
1550: construction terminology, all places dominated by the definition of the
1551: local.}. In other words, it is not visible in places that can be reached
1552: without going through the definition of the local. E.g., locals defined
1553: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1554: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1555: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1556:
1557: The reasoning behind this solution is: We want to have the locals
1558: visible as long as it is meaningful. The user can always make the
1559: visibility shorter by using explicit scoping. In a place that can
1560: only be reached through the definition of a local, the meaning of a
1561: local name is clear. In other places it is not: How is the local
1562: initialized at the control flow path that does not contain the
1563: definition? Which local is meant, if the same name is defined twice in
1564: two independent control flow paths?
1565:
1566: This should be enough detail for nearly all users, so you can skip the
1567: rest of this section. If you relly must know all the gory details and
1568: options, read on.
1569:
1570: In order to implement this rule, the compiler has to know which places
1571: are unreachable. It knows this automatically after @code{AHEAD},
1572: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1573: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1574: compiler that the control flow never reaches that place. If
1575: @code{UNREACHABLE} is not used where it could, the only consequence is
1576: that the visibility of some locals is more limited than the rule above
1577: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1578: lie to the compiler), buggy code will be produced.
1579:
1580: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3 anton 1581: does not know which locals will be visible on the incoming
1582: back-edge. All problems discussed in the following are due to this
1583: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1584: loops as examples; the discussion also applies to @code{?DO} and other
1.2 anton 1585: loops). Perhaps the most insidious example is:
1586: @example
1587: AHEAD
1588: BEGIN
1589: x
1590: [ 1 CS-ROLL ] THEN
1.4 anton 1591: @{ x @}
1.2 anton 1592: ...
1593: UNTIL
1594: @end example
1595:
1596: This should be legal according to the visibility rule. The use of
1597: @code{x} can only be reached through the definition; but that appears
1598: textually below the use.
1599:
1600: From this example it is clear that the visibility rules cannot be fully
1601: implemented without major headaches. Our implementation treats common
1602: cases as advertised and the exceptions are treated in a safe way: The
1603: compiler makes a reasonable guess about the locals visible after a
1604: @code{BEGIN}; if it is too pessimistic, the
1605: user will get a spurious error about the local not being defined; if the
1606: compiler is too optimistic, it will notice this later and issue a
1607: warning. In the case above the compiler would complain about @code{x}
1608: being undefined at its use. You can see from the obscure examples in
1609: this section that it takes quite unusual control structures to get the
1610: compiler into trouble, and even then it will often do fine.
1611:
1612: If the @code{BEGIN} is reachable from above, the most optimistic guess
1613: is that all locals visible before the @code{BEGIN} will also be
1614: visible after the @code{BEGIN}. This guess is valid for all loops that
1615: are entered only through the @code{BEGIN}, in particular, for normal
1616: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1617: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1618: compiler. When the branch to the @code{BEGIN} is finally generated by
1619: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1620: warns the user if it was too optimisitic:
1621: @example
1622: IF
1.4 anton 1623: @{ x @}
1.2 anton 1624: BEGIN
1625: \ x ?
1626: [ 1 cs-roll ] THEN
1627: ...
1628: UNTIL
1629: @end example
1630:
1631: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1632: optimistically assumes that it lives until the @code{THEN}. It notices
1633: this difference when it compiles the @code{UNTIL} and issues a
1634: warning. The user can avoid the warning, and make sure that @code{x}
1635: is not used in the wrong area by using explicit scoping:
1636: @example
1637: IF
1638: SCOPE
1.4 anton 1639: @{ x @}
1.2 anton 1640: ENDSCOPE
1641: BEGIN
1642: [ 1 cs-roll ] THEN
1643: ...
1644: UNTIL
1645: @end example
1646:
1647: Since the guess is optimistic, there will be no spurious error messages
1648: about undefined locals.
1649:
1650: If the @code{BEGIN} is not reachable from above (e.g., after
1651: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1652: optimistic guess, as the locals visible after the @code{BEGIN} may be
1653: defined later. Therefore, the compiler assumes that no locals are
1.17 anton 1654: visible after the @code{BEGIN}. However, the user can use
1.2 anton 1655: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17 anton 1656: visible at the BEGIN as at the point where the top control-flow stack
1657: item was created.
1.2 anton 1658:
1659: doc-assume-live
1660:
1661: E.g.,
1662: @example
1.4 anton 1663: @{ x @}
1.2 anton 1664: AHEAD
1665: ASSUME-LIVE
1666: BEGIN
1667: x
1668: [ 1 CS-ROLL ] THEN
1669: ...
1670: UNTIL
1671: @end example
1672:
1673: Other cases where the locals are defined before the @code{BEGIN} can be
1674: handled by inserting an appropriate @code{CS-ROLL} before the
1675: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1676: behind the @code{ASSUME-LIVE}).
1677:
1678: Cases where locals are defined after the @code{BEGIN} (but should be
1679: visible immediately after the @code{BEGIN}) can only be handled by
1680: rearranging the loop. E.g., the ``most insidious'' example above can be
1681: arranged into:
1682: @example
1683: BEGIN
1.4 anton 1684: @{ x @}
1.2 anton 1685: ... 0=
1686: WHILE
1687: x
1688: REPEAT
1689: @end example
1690:
1.17 anton 1691: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2 anton 1692: @subsubsection How long do locals live?
1693:
1694: The right answer for the lifetime question would be: A local lives at
1695: least as long as it can be accessed. For a value-flavoured local this
1696: means: until the end of its visibility. However, a variable-flavoured
1697: local could be accessed through its address far beyond its visibility
1698: scope. Ultimately, this would mean that such locals would have to be
1699: garbage collected. Since this entails un-Forth-like implementation
1700: complexities, I adopted the same cowardly solution as some other
1701: languages (e.g., C): The local lives only as long as it is visible;
1702: afterwards its address is invalid (and programs that access it
1703: afterwards are erroneous).
1704:
1.17 anton 1705: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2 anton 1706: @subsubsection Programming Style
1707:
1708: The freedom to define locals anywhere has the potential to change
1709: programming styles dramatically. In particular, the need to use the
1710: return stack for intermediate storage vanishes. Moreover, all stack
1711: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1712: determined arguments) can be eliminated: If the stack items are in the
1713: wrong order, just write a locals definition for all of them; then
1714: write the items in the order you want.
1715:
1716: This seems a little far-fetched and eliminating stack manipulations is
1.4 anton 1717: unlikely to become a conscious programming objective. Still, the number
1718: of stack manipulations will be reduced dramatically if local variables
1.17 anton 1719: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4 anton 1720: a traditional implementation of @code{max}).
1.2 anton 1721:
1722: This shows one potential benefit of locals: making Forth programs more
1723: readable. Of course, this benefit will only be realized if the
1724: programmers continue to honour the principle of factoring instead of
1725: using the added latitude to make the words longer.
1726:
1727: Using @code{TO} can and should be avoided. Without @code{TO},
1728: every value-flavoured local has only a single assignment and many
1729: advantages of functional languages apply to Forth. I.e., programs are
1730: easier to analyse, to optimize and to read: It is clear from the
1731: definition what the local stands for, it does not turn into something
1732: different later.
1733:
1734: E.g., a definition using @code{TO} might look like this:
1735: @example
1736: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1737: u1 u2 min 0
1738: ?do
1739: addr1 c@ addr2 c@ - ?dup
1740: if
1741: unloop exit
1742: then
1743: addr1 char+ TO addr1
1744: addr2 char+ TO addr2
1745: loop
1746: u1 u2 - ;
1747: @end example
1748: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1749: every loop iteration. @code{strcmp} is a typical example of the
1750: readability problems of using @code{TO}. When you start reading
1751: @code{strcmp}, you think that @code{addr1} refers to the start of the
1752: string. Only near the end of the loop you realize that it is something
1753: else.
1754:
1755: This can be avoided by defining two locals at the start of the loop that
1756: are initialized with the right value for the current iteration.
1757: @example
1758: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1759: addr1 addr2
1760: u1 u2 min 0
1761: ?do @{ s1 s2 @}
1762: s1 c@ s2 c@ - ?dup
1763: if
1764: unloop exit
1765: then
1766: s1 char+ s2 char+
1767: loop
1768: 2drop
1769: u1 u2 - ;
1770: @end example
1771: Here it is clear from the start that @code{s1} has a different value
1772: in every loop iteration.
1773:
1.17 anton 1774: @node Implementation, , Programming Style, Gforth locals
1.2 anton 1775: @subsubsection Implementation
1776:
1.17 anton 1777: Gforth uses an extra locals stack. The most compelling reason for
1.2 anton 1778: this is that the return stack is not float-aligned; using an extra stack
1779: also eliminates the problems and restrictions of using the return stack
1780: as locals stack. Like the other stacks, the locals stack grows toward
1781: lower addresses. A few primitives allow an efficient implementation:
1782:
1783: doc-@local#
1784: doc-f@local#
1785: doc-laddr#
1786: doc-lp+!#
1787: doc-lp!
1788: doc->l
1789: doc-f>l
1790:
1791: In addition to these primitives, some specializations of these
1792: primitives for commonly occurring inline arguments are provided for
1793: efficiency reasons, e.g., @code{@@local0} as specialization of
1794: @code{@@local#} for the inline argument 0. The following compiling words
1795: compile the right specialized version, or the general version, as
1796: appropriate:
1797:
1.12 anton 1798: doc-compile-@local
1799: doc-compile-f@local
1.2 anton 1800: doc-compile-lp+!
1801:
1802: Combinations of conditional branches and @code{lp+!#} like
1803: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
1804: is taken) are provided for efficiency and correctness in loops.
1805:
1806: A special area in the dictionary space is reserved for keeping the
1807: local variable names. @code{@{} switches the dictionary pointer to this
1808: area and @code{@}} switches it back and generates the locals
1809: initializing code. @code{W:} etc.@ are normal defining words. This
1810: special area is cleared at the start of every colon definition.
1811:
1.17 anton 1812: A special feature of Gforth's dictionary is used to implement the
1.2 anton 1813: definition of locals without type specifiers: every wordlist (aka
1814: vocabulary) has its own methods for searching
1.4 anton 1815: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2 anton 1816: with a special search method: When it is searched for a word, it
1817: actually creates that word using @code{W:}. @code{@{} changes the search
1818: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
1819: and then the wordlist for defining locals without type specifiers.
1820:
1821: The lifetime rules support a stack discipline within a colon
1822: definition: The lifetime of a local is either nested with other locals
1823: lifetimes or it does not overlap them.
1824:
1825: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
1826: pointer manipulation is generated. Between control structure words
1827: locals definitions can push locals onto the locals stack. @code{AGAIN}
1828: is the simplest of the other three control flow words. It has to
1829: restore the locals stack depth of the corresponding @code{BEGIN}
1830: before branching. The code looks like this:
1831: @format
1832: @code{lp+!#} current-locals-size @minus{} dest-locals-size
1833: @code{branch} <begin>
1834: @end format
1835:
1836: @code{UNTIL} is a little more complicated: If it branches back, it
1837: must adjust the stack just like @code{AGAIN}. But if it falls through,
1838: the locals stack must not be changed. The compiler generates the
1839: following code:
1840: @format
1841: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
1842: @end format
1843: The locals stack pointer is only adjusted if the branch is taken.
1844:
1845: @code{THEN} can produce somewhat inefficient code:
1846: @format
1847: @code{lp+!#} current-locals-size @minus{} orig-locals-size
1848: <orig target>:
1849: @code{lp+!#} orig-locals-size @minus{} new-locals-size
1850: @end format
1851: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4 anton 1852: level at the @var{orig} point to the level after the @code{THEN}. The
1.2 anton 1853: first @code{lp+!#} adjusts the locals stack pointer from the current
1854: level to the level at the orig point, so the complete effect is an
1855: adjustment from the current level to the right level after the
1856: @code{THEN}.
1857:
1858: In a conventional Forth implementation a dest control-flow stack entry
1859: is just the target address and an orig entry is just the address to be
1860: patched. Our locals implementation adds a wordlist to every orig or dest
1861: item. It is the list of locals visible (or assumed visible) at the point
1862: described by the entry. Our implementation also adds a tag to identify
1863: the kind of entry, in particular to differentiate between live and dead
1864: (reachable and unreachable) orig entries.
1865:
1866: A few unusual operations have to be performed on locals wordlists:
1867:
1868: doc-common-list
1869: doc-sub-list?
1870: doc-list-size
1871:
1872: Several features of our locals wordlist implementation make these
1873: operations easy to implement: The locals wordlists are organised as
1874: linked lists; the tails of these lists are shared, if the lists
1875: contain some of the same locals; and the address of a name is greater
1876: than the address of the names behind it in the list.
1877:
1878: Another important implementation detail is the variable
1879: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
1880: determine if they can be reached directly or only through the branch
1881: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
1882: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
1883: definition, by @code{BEGIN} and usually by @code{THEN}.
1884:
1885: Counted loops are similar to other loops in most respects, but
1886: @code{LEAVE} requires special attention: It performs basically the same
1887: service as @code{AHEAD}, but it does not create a control-flow stack
1888: entry. Therefore the information has to be stored elsewhere;
1889: traditionally, the information was stored in the target fields of the
1890: branches created by the @code{LEAVE}s, by organizing these fields into a
1891: linked list. Unfortunately, this clever trick does not provide enough
1892: space for storing our extended control flow information. Therefore, we
1893: introduce another stack, the leave stack. It contains the control-flow
1894: stack entries for all unresolved @code{LEAVE}s.
1895:
1896: Local names are kept until the end of the colon definition, even if
1897: they are no longer visible in any control-flow path. In a few cases
1898: this may lead to increased space needs for the locals name area, but
1899: usually less than reclaiming this space would cost in code size.
1900:
1901:
1.17 anton 1902: @node ANS Forth locals, , Gforth locals, Locals
1.2 anton 1903: @subsection ANS Forth locals
1904:
1905: The ANS Forth locals wordset does not define a syntax for locals, but
1906: words that make it possible to define various syntaxes. One of the
1.17 anton 1907: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2 anton 1908: wordset, i.e.:
1909:
1910: @example
1911: @{ local1 local2 ... -- comment @}
1912: @end example
1913: or
1914: @example
1915: @{ local1 local2 ... @}
1916: @end example
1917:
1918: The order of the locals corresponds to the order in a stack comment. The
1919: restrictions are:
1.1 anton 1920:
1.2 anton 1921: @itemize @bullet
1922: @item
1.17 anton 1923: Locals can only be cell-sized values (no type specifiers are allowed).
1.2 anton 1924: @item
1925: Locals can be defined only outside control structures.
1926: @item
1927: Locals can interfere with explicit usage of the return stack. For the
1928: exact (and long) rules, see the standard. If you don't use return stack
1.17 anton 1929: accessing words in a definition using locals, you will be all right. The
1.2 anton 1930: purpose of this rule is to make locals implementation on the return
1931: stack easier.
1932: @item
1933: The whole definition must be in one line.
1934: @end itemize
1935:
1936: Locals defined in this way behave like @code{VALUE}s
1.4 anton 1937: (@xref{Values}). I.e., they are initialized from the stack. Using their
1.2 anton 1938: name produces their value. Their value can be changed using @code{TO}.
1939:
1.17 anton 1940: Since this syntax is supported by Gforth directly, you need not do
1.2 anton 1941: anything to use it. If you want to port a program using this syntax to
1942: another ANS Forth system, use @file{anslocal.fs} to implement the syntax
1943: on the other system.
1944:
1945: Note that a syntax shown in the standard, section A.13 looks
1946: similar, but is quite different in having the order of locals
1947: reversed. Beware!
1948:
1949: The ANS Forth locals wordset itself consists of the following word
1950:
1951: doc-(local)
1952:
1953: The ANS Forth locals extension wordset defines a syntax, but it is so
1954: awful that we strongly recommend not to use it. We have implemented this
1.17 anton 1955: syntax to make porting to Gforth easy, but do not document it here. The
1.2 anton 1956: problem with this syntax is that the locals are defined in an order
1957: reversed with respect to the standard stack comment notation, making
1958: programs harder to read, and easier to misread and miswrite. The only
1959: merit of this syntax is that it is easy to implement using the ANS Forth
1960: locals wordset.
1.3 anton 1961:
1.4 anton 1962: @node Defining Words, Wordlists, Locals, Words
1963: @section Defining Words
1964:
1.14 anton 1965: @menu
1966: * Values::
1967: @end menu
1968:
1.4 anton 1969: @node Values, , Defining Words, Defining Words
1970: @subsection Values
1971:
1972: @node Wordlists, Files, Defining Words, Words
1973: @section Wordlists
1974:
1975: @node Files, Blocks, Wordlists, Words
1976: @section Files
1977:
1978: @node Blocks, Other I/O, Files, Words
1979: @section Blocks
1980:
1981: @node Other I/O, Programming Tools, Blocks, Words
1982: @section Other I/O
1983:
1.18 anton 1984: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4 anton 1985: @section Programming Tools
1986:
1.5 anton 1987: @menu
1988: * Debugging:: Simple and quick.
1989: * Assertions:: Making your programs self-checking.
1990: @end menu
1991:
1992: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4 anton 1993: @subsection Debugging
1994:
1995: The simple debugging aids provided in @file{debugging.fs}
1996: are meant to support a different style of debugging than the
1997: tracing/stepping debuggers used in languages with long turn-around
1998: times.
1999:
2000: A much better (faster) way in fast-compilig languages is to add
2001: printing code at well-selected places, let the program run, look at
2002: the output, see where things went wrong, add more printing code, etc.,
2003: until the bug is found.
2004:
2005: The word @code{~~} is easy to insert. It just prints debugging
2006: information (by default the source location and the stack contents). It
2007: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
2008: query-replace them with nothing). The deferred words
2009: @code{printdebugdata} and @code{printdebugline} control the output of
2010: @code{~~}. The default source location output format works well with
2011: Emacs' compilation mode, so you can step through the program at the
1.5 anton 2012: source level using @kbd{C-x `} (the advantage over a stepping debugger
2013: is that you can step in any direction and you know where the crash has
2014: happened or where the strange data has occurred).
1.4 anton 2015:
2016: Note that the default actions clobber the contents of the pictured
2017: numeric output string, so you should not use @code{~~}, e.g., between
2018: @code{<#} and @code{#>}.
2019:
2020: doc-~~
2021: doc-printdebugdata
2022: doc-printdebugline
2023:
1.5 anton 2024: @node Assertions, , Debugging, Programming Tools
1.4 anton 2025: @subsection Assertions
2026:
1.5 anton 2027: It is a good idea to make your programs self-checking, in particular, if
2028: you use an assumption (e.g., that a certain field of a data structure is
1.17 anton 2029: never zero) that may become wrong during maintenance. Gforth supports
1.5 anton 2030: assertions for this purpose. They are used like this:
2031:
2032: @example
2033: assert( @var{flag} )
2034: @end example
2035:
2036: The code between @code{assert(} and @code{)} should compute a flag, that
2037: should be true if everything is alright and false otherwise. It should
2038: not change anything else on the stack. The overall stack effect of the
2039: assertion is @code{( -- )}. E.g.
2040:
2041: @example
2042: assert( 1 1 + 2 = ) \ what we learn in school
2043: assert( dup 0<> ) \ assert that the top of stack is not zero
2044: assert( false ) \ this code should not be reached
2045: @end example
2046:
2047: The need for assertions is different at different times. During
2048: debugging, we want more checking, in production we sometimes care more
2049: for speed. Therefore, assertions can be turned off, i.e., the assertion
2050: becomes a comment. Depending on the importance of an assertion and the
2051: time it takes to check it, you may want to turn off some assertions and
1.17 anton 2052: keep others turned on. Gforth provides several levels of assertions for
1.5 anton 2053: this purpose:
2054:
2055: doc-assert0(
2056: doc-assert1(
2057: doc-assert2(
2058: doc-assert3(
2059: doc-assert(
2060: doc-)
2061:
2062: @code{Assert(} is the same as @code{assert1(}. The variable
2063: @code{assert-level} specifies the highest assertions that are turned
2064: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2065: @code{assert1(} assertions perform checking, while @code{assert2(} and
2066: @code{assert3(} assertions are treated as comments.
2067:
2068: Note that the @code{assert-level} is evaluated at compile-time, not at
2069: run-time. I.e., you cannot turn assertions on or off at run-time, you
2070: have to set the @code{assert-level} appropriately before compiling a
2071: piece of code. You can compile several pieces of code at several
2072: @code{assert-level}s (e.g., a trusted library at level 1 and newly
2073: written code at level 3).
2074:
2075: doc-assert-level
2076:
2077: If an assertion fails, a message compatible with Emacs' compilation mode
2078: is produced and the execution is aborted (currently with @code{ABORT"}.
2079: If there is interest, we will introduce a special throw code. But if you
2080: intend to @code{catch} a specific condition, using @code{throw} is
2081: probably more appropriate than an assertion).
2082:
1.18 anton 2083: @node Assembler and Code words, Threading Words, Programming Tools, Words
2084: @section Assembler and Code words
2085:
2086: Gforth provides some words for defining primitives (words written in
2087: machine code), and for defining the the machine-code equivalent of
2088: @code{DOES>}-based defining words. However, the machine-independent
2089: nature of Gforth poses a few problems: First of all. Gforth runs on
2090: several architectures, so it can provide no standard assembler. What's
2091: worse is that the register allocation not only depends on the processor,
1.25 anton 2092: but also on the @code{gcc} version and options used.
1.18 anton 2093:
1.25 anton 2094: The words that Gforth offers encapsulate some system dependences (e.g., the
1.18 anton 2095: header structure), so a system-independent assembler may be used in
2096: Gforth. If you do not have an assembler, you can compile machine code
2097: directly with @code{,} and @code{c,}.
2098:
2099: doc-assembler
2100: doc-code
2101: doc-end-code
2102: doc-;code
2103: doc-flush-icache
2104:
2105: If @code{flush-icache} does not work correctly, @code{code} words
2106: etc. will not work (reliably), either.
2107:
2108: These words are rarely used. Therefore they reside in @code{code.fs},
2109: which is usually not loaded (except @code{flush-icache}, which is always
1.19 anton 2110: present). You can load them with @code{require code.fs}.
1.18 anton 2111:
1.25 anton 2112: In the assembly code you will want to refer to the inner interpreter's
2113: registers (e.g., the data stack pointer) and you may want to use other
2114: registers for temporary storage. Unfortunately, the register allocation
2115: is installation-dependent.
2116:
2117: The easiest solution is to use explicit register declarations
2118: (@pxref{Explicit Reg Vars, , Variables in Specified Registers, gcc.info,
2119: GNU C Manual}) for all of the inner interpreter's registers: You have to
2120: compile Gforth with @code{-DFORCE_REG} (configure option
2121: @code{--enable-force-reg}) and the appropriate declarations must be
2122: present in the @code{machine.h} file (see @code{mips.h} for an example;
2123: you can find a full list of all declarable register symbols with
2124: @code{grep register engine.c}). If you give explicit registers to all
2125: variables that are declared at the beginning of @code{engine()}, you
2126: should be able to use the other caller-saved registers for temporary
2127: storage. Alternatively, you can use the @code{gcc} option
2128: @code{-ffixed-REG} (@pxref{Code Gen Options, , Options for Code
2129: Generation Conventions, gcc.info, GNU C Manual}) to reserve a register
2130: (however, this restriction on register allocation may slow Gforth
2131: significantly).
2132:
2133: If this solution is not viable (e.g., because @code{gcc} does not allow
2134: you to explicitly declare all the registers you need), you have to find
2135: out by looking at the code where the inner interpreter's registers
2136: reside and which registers can be used for temporary storage. You can
2137: get an assembly listing of the engine's code with @code{make engine.s}.
2138:
2139: In any case, it is good practice to abstract your assembly code from the
2140: actual register allocation. E.g., if the data stack pointer resides in
2141: register @code{$17}, create an alias for this register called @code{sp},
2142: and use that in your assembly code.
2143:
1.18 anton 2144: Another option for implementing normal and defining words efficiently
2145: is: adding the wanted functionality to the source of Gforth. For normal
2146: words you just have to edit @file{primitives}, defining words (for fast
1.25 anton 2147: defined words) may require changes in @file{engine.c},
1.18 anton 2148: @file{kernal.fs}, @file{prims2x.fs}, and possibly @file{cross.fs}.
2149:
2150:
2151: @node Threading Words, , Assembler and Code words, Words
1.4 anton 2152: @section Threading Words
2153:
2154: These words provide access to code addresses and other threading stuff
1.17 anton 2155: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4 anton 2156: abstracts away the differences between direct and indirect threading
2157: (and, for direct threading, the machine dependences). However, at
2158: present this wordset is still inclomplete. It is also pretty low-level;
2159: some day it will hopefully be made unnecessary by an internals words set
2160: that abstracts implementation details away completely.
2161:
2162: doc->code-address
2163: doc->does-code
2164: doc-code-address!
2165: doc-does-code!
2166: doc-does-handler!
2167: doc-/does-handler
2168:
1.18 anton 2169: The code addresses produced by various defining words are produced by
2170: the following words:
1.14 anton 2171:
1.18 anton 2172: doc-docol:
2173: doc-docon:
2174: doc-dovar:
2175: doc-douser:
2176: doc-dodefer:
2177: doc-dofield:
2178:
2179: Currently there is no installation-independent way for recogizing words
2180: defined by a @code{CREATE}...@code{DOES>} word; however, once you know
2181: that a word is defined by a @code{CREATE}...@code{DOES>} word, you can
2182: use @code{>DOES-CODE}.
1.14 anton 2183:
1.4 anton 2184: @node ANS conformance, Model, Words, Top
2185: @chapter ANS conformance
2186:
1.17 anton 2187: To the best of our knowledge, Gforth is an
1.14 anton 2188:
1.15 anton 2189: ANS Forth System
2190: @itemize
2191: @item providing the Core Extensions word set
2192: @item providing the Block word set
2193: @item providing the Block Extensions word set
2194: @item providing the Double-Number word set
2195: @item providing the Double-Number Extensions word set
2196: @item providing the Exception word set
2197: @item providing the Exception Extensions word set
2198: @item providing the Facility word set
2199: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2200: @item providing the File Access word set
2201: @item providing the File Access Extensions word set
2202: @item providing the Floating-Point word set
2203: @item providing the Floating-Point Extensions word set
2204: @item providing the Locals word set
2205: @item providing the Locals Extensions word set
2206: @item providing the Memory-Allocation word set
2207: @item providing the Memory-Allocation Extensions word set (that one's easy)
2208: @item providing the Programming-Tools word set
1.18 anton 2209: @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
1.15 anton 2210: @item providing the Search-Order word set
2211: @item providing the Search-Order Extensions word set
2212: @item providing the String word set
2213: @item providing the String Extensions word set (another easy one)
2214: @end itemize
2215:
2216: In addition, ANS Forth systems are required to document certain
2217: implementation choices. This chapter tries to meet these
2218: requirements. In many cases it gives a way to ask the system for the
2219: information instead of providing the information directly, in
2220: particular, if the information depends on the processor, the operating
2221: system or the installation options chosen, or if they are likely to
1.17 anton 2222: change during the maintenance of Gforth.
1.15 anton 2223:
1.14 anton 2224: @comment The framework for the rest has been taken from pfe.
2225:
2226: @menu
2227: * The Core Words::
2228: * The optional Block word set::
2229: * The optional Double Number word set::
2230: * The optional Exception word set::
2231: * The optional Facility word set::
2232: * The optional File-Access word set::
2233: * The optional Floating-Point word set::
2234: * The optional Locals word set::
2235: * The optional Memory-Allocation word set::
2236: * The optional Programming-Tools word set::
2237: * The optional Search-Order word set::
2238: @end menu
2239:
2240:
2241: @c =====================================================================
2242: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2243: @comment node-name, next, previous, up
2244: @section The Core Words
2245: @c =====================================================================
2246:
2247: @menu
1.15 anton 2248: * core-idef:: Implementation Defined Options
2249: * core-ambcond:: Ambiguous Conditions
2250: * core-other:: Other System Documentation
1.14 anton 2251: @end menu
2252:
2253: @c ---------------------------------------------------------------------
2254: @node core-idef, core-ambcond, The Core Words, The Core Words
2255: @subsection Implementation Defined Options
2256: @c ---------------------------------------------------------------------
2257:
2258: @table @i
2259:
2260: @item (Cell) aligned addresses:
1.17 anton 2261: processor-dependent. Gforth's alignment words perform natural alignment
1.14 anton 2262: (e.g., an address aligned for a datum of size 8 is divisible by
2263: 8). Unaligned accesses usually result in a @code{-23 THROW}.
2264:
2265: @item @code{EMIT} and non-graphic characters:
2266: The character is output using the C library function (actually, macro)
2267: @code{putchar}.
2268:
2269: @item character editing of @code{ACCEPT} and @code{EXPECT}:
2270: This is modeled on the GNU readline library (@pxref{Readline
2271: Interaction, , Command Line Editing, readline, The GNU Readline
2272: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
2273: producing a full word completion every time you type it (instead of
2274: producing the common prefix of all completions).
2275:
2276: @item character set:
2277: The character set of your computer and display device. Gforth is
2278: 8-bit-clean (but some other component in your system may make trouble).
2279:
2280: @item Character-aligned address requirements:
2281: installation-dependent. Currently a character is represented by a C
2282: @code{unsigned char}; in the future we might switch to @code{wchar_t}
2283: (Comments on that requested).
2284:
2285: @item character-set extensions and matching of names:
1.17 anton 2286: Any character except the ASCII NUL charcter can be used in a
2287: name. Matching is case-insensitive. The matching is performed using the
2288: C function @code{strncasecmp}, whose function is probably influenced by
2289: the locale. E.g., the @code{C} locale does not know about accents and
1.14 anton 2290: umlauts, so they are matched case-sensitively in that locale. For
2291: portability reasons it is best to write programs such that they work in
2292: the @code{C} locale. Then one can use libraries written by a Polish
2293: programmer (who might use words containing ISO Latin-2 encoded
2294: characters) and by a French programmer (ISO Latin-1) in the same program
2295: (of course, @code{WORDS} will produce funny results for some of the
2296: words (which ones, depends on the font you are using)). Also, the locale
2297: you prefer may not be available in other operating systems. Hopefully,
2298: Unicode will solve these problems one day.
2299:
2300: @item conditions under which control characters match a space delimiter:
2301: If @code{WORD} is called with the space character as a delimiter, all
2302: white-space characters (as identified by the C macro @code{isspace()})
2303: are delimiters. @code{PARSE}, on the other hand, treats space like other
2304: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
2305: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
2306: interpreter (aka text interpreter) by default, treats all white-space
2307: characters as delimiters.
2308:
2309: @item format of the control flow stack:
2310: The data stack is used as control flow stack. The size of a control flow
2311: stack item in cells is given by the constant @code{cs-item-size}. At the
2312: time of this writing, an item consists of a (pointer to a) locals list
2313: (third), an address in the code (second), and a tag for identifying the
2314: item (TOS). The following tags are used: @code{defstart},
2315: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
2316: @code{scopestart}.
2317:
2318: @item conversion of digits > 35
2319: The characters @code{[\]^_'} are the digits with the decimal value
2320: 36@minus{}41. There is no way to input many of the larger digits.
2321:
2322: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
2323: The cursor is moved to the end of the entered string. If the input is
2324: terminated using the @kbd{Return} key, a space is typed.
2325:
2326: @item exception abort sequence of @code{ABORT"}:
2327: The error string is stored into the variable @code{"error} and a
2328: @code{-2 throw} is performed.
2329:
2330: @item input line terminator:
2331: For interactive input, @kbd{C-m} and @kbd{C-j} terminate lines. One of
2332: these characters is typically produced when you type the @kbd{Enter} or
2333: @kbd{Return} key.
2334:
2335: @item maximum size of a counted string:
2336: @code{s" /counted-string" environment? drop .}. Currently 255 characters
2337: on all ports, but this may change.
2338:
2339: @item maximum size of a parsed string:
2340: Given by the constant @code{/line}. Currently 255 characters.
2341:
2342: @item maximum size of a definition name, in characters:
2343: 31
2344:
2345: @item maximum string length for @code{ENVIRONMENT?}, in characters:
2346: 31
2347:
2348: @item method of selecting the user input device:
1.17 anton 2349: The user input device is the standard input. There is currently no way to
2350: change it from within Gforth. However, the input can typically be
2351: redirected in the command line that starts Gforth.
1.14 anton 2352:
2353: @item method of selecting the user output device:
2354: The user output device is the standard output. It cannot be redirected
1.17 anton 2355: from within Gforth, but typically from the command line that starts
2356: Gforth. Gforth uses buffered output, so output on a terminal does not
1.14 anton 2357: become visible before the next newline or buffer overflow. Output on
2358: non-terminals is invisible until the buffer overflows.
2359:
2360: @item methods of dictionary compilation:
1.17 anton 2361: What are we expected to document here?
1.14 anton 2362:
2363: @item number of bits in one address unit:
2364: @code{s" address-units-bits" environment? drop .}. 8 in all current
2365: ports.
2366:
2367: @item number representation and arithmetic:
2368: Processor-dependent. Binary two's complement on all current ports.
2369:
2370: @item ranges for integer types:
2371: Installation-dependent. Make environmental queries for @code{MAX-N},
2372: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
2373: unsigned (and positive) types is 0. The lower bound for signed types on
2374: two's complement and one's complement machines machines can be computed
2375: by adding 1 to the upper bound.
2376:
2377: @item read-only data space regions:
2378: The whole Forth data space is writable.
2379:
2380: @item size of buffer at @code{WORD}:
2381: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2382: shared with the pictured numeric output string. If overwriting
2383: @code{PAD} is acceptable, it is as large as the remaining dictionary
2384: space, although only as much can be sensibly used as fits in a counted
2385: string.
2386:
2387: @item size of one cell in address units:
2388: @code{1 cells .}.
2389:
2390: @item size of one character in address units:
2391: @code{1 chars .}. 1 on all current ports.
2392:
2393: @item size of the keyboard terminal buffer:
2394: Varies. You can determine the size at a specific time using @code{lp@
2395: tib - .}. It is shared with the locals stack and TIBs of files that
2396: include the current file. You can change the amount of space for TIBs
1.17 anton 2397: and locals stack at Gforth startup with the command line option
1.14 anton 2398: @code{-l}.
2399:
2400: @item size of the pictured numeric output buffer:
2401: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2402: shared with @code{WORD}.
2403:
2404: @item size of the scratch area returned by @code{PAD}:
2405: The remainder of dictionary space. You can even use the unused part of
2406: the data stack space. The current size can be computed with @code{sp@
2407: pad - .}.
2408:
2409: @item system case-sensitivity characteristics:
2410: Dictionary searches are case insensitive. However, as explained above
2411: under @i{character-set extensions}, the matching for non-ASCII
2412: characters is determined by the locale you are using. In the default
2413: @code{C} locale all non-ASCII characters are matched case-sensitively.
2414:
2415: @item system prompt:
2416: @code{ ok} in interpret state, @code{ compiled} in compile state.
2417:
2418: @item division rounding:
2419: installation dependent. @code{s" floored" environment? drop .}. We leave
1.25 anton 2420: the choice to @code{gcc} (what to use for @code{/}) and to you (whether to use
1.14 anton 2421: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
2422:
2423: @item values of @code{STATE} when true:
2424: -1.
2425:
2426: @item values returned after arithmetic overflow:
2427: On two's complement machines, arithmetic is performed modulo
2428: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2429: arithmetic (with appropriate mapping for signed types). Division by zero
2430: typically results in a @code{-55 throw} (floatingpoint unidentified
2431: fault), although a @code{-10 throw} (divide by zero) would be more
2432: appropriate.
2433:
2434: @item whether the current definition can be found after @t{DOES>}:
2435: No.
2436:
2437: @end table
2438:
2439: @c ---------------------------------------------------------------------
2440: @node core-ambcond, core-other, core-idef, The Core Words
2441: @subsection Ambiguous conditions
2442: @c ---------------------------------------------------------------------
2443:
2444: @table @i
2445:
2446: @item a name is neither a word nor a number:
2447: @code{-13 throw} (Undefined word)
2448:
2449: @item a definition name exceeds the maximum length allowed:
2450: @code{-19 throw} (Word name too long)
2451:
2452: @item addressing a region not inside the various data spaces of the forth system:
2453: The stacks, code space and name space are accessible. Machine code space is
2454: typically readable. Accessing other addresses gives results dependent on
2455: the operating system. On decent systems: @code{-9 throw} (Invalid memory
2456: address).
2457:
2458: @item argument type incompatible with parameter:
2459: This is usually not caught. Some words perform checks, e.g., the control
2460: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
2461: mismatch).
2462:
2463: @item attempting to obtain the execution token of a word with undefined execution semantics:
2464: You get an execution token representing the compilation semantics
2465: instead.
2466:
2467: @item dividing by zero:
2468: typically results in a @code{-55 throw} (floating point unidentified
2469: fault), although a @code{-10 throw} (divide by zero) would be more
2470: appropriate.
2471:
2472: @item insufficient data stack or return stack space:
2473: Not checked. This typically results in mysterious illegal memory
2474: accesses, producing @code{-9 throw} (Invalid memory address) or
2475: @code{-23 throw} (Address alignment exception).
2476:
2477: @item insufficient space for loop control parameters:
2478: like other return stack overflows.
2479:
2480: @item insufficient space in the dictionary:
2481: Not checked. Similar results as stack overflows. However, typically the
2482: error appears at a different place when one inserts or removes code.
2483:
2484: @item interpreting a word with undefined interpretation semantics:
2485: For some words, we defined interpretation semantics. For the others:
2486: @code{-14 throw} (Interpreting a compile-only word). Note that this is
2487: checked only by the outer (aka text) interpreter; if the word is
2488: @code{execute}d in some other way, it will typically perform it's
2489: compilation semantics even in interpret state. (We could change @code{'}
2490: and relatives not to give the xt of such words, but we think that would
2491: be too restrictive).
2492:
2493: @item modifying the contents of the input buffer or a string literal:
2494: These are located in writable memory and can be modified.
2495:
2496: @item overflow of the pictured numeric output string:
2497: Not checked.
2498:
2499: @item parsed string overflow:
2500: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
2501:
2502: @item producing a result out of range:
2503: On two's complement machines, arithmetic is performed modulo
2504: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2505: arithmetic (with appropriate mapping for signed types). Division by zero
2506: typically results in a @code{-55 throw} (floatingpoint unidentified
2507: fault), although a @code{-10 throw} (divide by zero) would be more
2508: appropriate. @code{convert} and @code{>number} currently overflow
2509: silently.
2510:
2511: @item reading from an empty data or return stack:
2512: The data stack is checked by the outer (aka text) interpreter after
2513: every word executed. If it has underflowed, a @code{-4 throw} (Stack
2514: underflow) is performed. Apart from that, the stacks are not checked and
2515: underflows can result in similar behaviour as overflows (of adjacent
2516: stacks).
2517:
2518: @item unexepected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
2519: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
2520: use zero-length string as a name). Words like @code{'} probably will not
2521: find what they search. Note that it is possible to create zero-length
2522: names with @code{nextname} (should it not?).
2523:
2524: @item @code{>IN} greater than input buffer:
2525: The next invocation of a parsing word returns a string wih length 0.
2526:
2527: @item @code{RECURSE} appears after @code{DOES>}:
2528: Compiles a recursive call to the defining word not to the defined word.
2529:
2530: @item argument input source different than current input source for @code{RESTORE-INPUT}:
1.26 ! anton 2531: If the argument input source is a valid input source then it gets
! 2532: restored. Otherwise the result is undefined.
! 2533: @comment causes @code{-12 THROW}, which, unless caught, issues the message "argument type mismatch" and aborts. !! not all of the state is restored (e.g., sourcefilename).
1.14 anton 2534:
2535: @item data space containing definitions gets de-allocated:
2536: Deallocation with @code{allot} is not checked. This typically resuls in
2537: memory access faults or execution of illegal instructions.
2538:
2539: @item data space read/write with incorrect alignment:
2540: Processor-dependent. Typically results in a @code{-23 throw} (Address
2541: alignment exception). Under Linux on a 486 or later processor with
2542: alignment turned on, incorrect alignment results in a @code{-9 throw}
2543: (Invalid memory address). There are reportedly some processors with
2544: alignment restrictions that do not report them.
2545:
2546: @item data space pointer not properly aligned, @code{,}, @code{C,}:
2547: Like other alignment errors.
2548:
2549: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
2550: Not checked. May cause an illegal memory access.
2551:
2552: @item loop control parameters not available:
2553: Not checked. The counted loop words simply assume that the top of return
2554: stack items are loop control parameters and behave accordingly.
2555:
2556: @item most recent definition does not have a name (@code{IMMEDIATE}):
2557: @code{abort" last word was headerless"}.
2558:
2559: @item name not defined by @code{VALUE} used by @code{TO}:
2560: @code{-32 throw} (Invalid name argument)
2561:
1.15 anton 2562: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.14 anton 2563: @code{-13 throw} (Undefined word)
2564:
2565: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
2566: Gforth behaves as if they were of the same type. I.e., you can predict
2567: the behaviour by interpreting all parameters as, e.g., signed.
2568:
2569: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
2570: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} is equivalent to
2571: @code{TO}.
2572:
2573: @item String longer than a counted string returned by @code{WORD}:
2574: Not checked. The string will be ok, but the count will, of course,
2575: contain only the least significant bits of the length.
2576:
1.15 anton 2577: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.14 anton 2578: Processor-dependent. Typical behaviours are returning 0 and using only
2579: the low bits of the shift count.
2580:
2581: @item word not defined via @code{CREATE}:
2582: @code{>BODY} produces the PFA of the word no matter how it was defined.
2583:
2584: @code{DOES>} changes the execution semantics of the last defined word no
2585: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
2586: @code{CREATE , DOES>}.
2587:
2588: @item words improperly used outside @code{<#} and @code{#>}:
2589: Not checked. As usual, you can expect memory faults.
2590:
2591: @end table
2592:
2593:
2594: @c ---------------------------------------------------------------------
2595: @node core-other, , core-ambcond, The Core Words
2596: @subsection Other system documentation
2597: @c ---------------------------------------------------------------------
2598:
2599: @table @i
2600:
2601: @item nonstandard words using @code{PAD}:
2602: None.
2603:
2604: @item operator's terminal facilities available:
1.26 ! anton 2605: After processing the command line, Gforth goes into interactive mode,
! 2606: and you can give commands to Gforth interactively. The actual facilities
! 2607: available depend on how you invoke Gforth.
1.14 anton 2608:
2609: @item program data space available:
2610: @code{sp@ here - .} gives the space remaining for dictionary and data
2611: stack together.
2612:
2613: @item return stack space available:
1.26 ! anton 2614: By default 16 KBytes. The default can be overridden with the @code{-r}
! 2615: switch (@pxref{Invocation}) when Gforth starts up.
1.14 anton 2616:
2617: @item stack space available:
2618: @code{sp@ here - .} gives the space remaining for dictionary and data
2619: stack together.
2620:
2621: @item system dictionary space required, in address units:
2622: Type @code{here forthstart - .} after startup. At the time of this
2623: writing, this gives 70108 (bytes) on a 32-bit system.
2624: @end table
2625:
2626:
2627: @c =====================================================================
2628: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
2629: @section The optional Block word set
2630: @c =====================================================================
2631:
2632: @menu
1.15 anton 2633: * block-idef:: Implementation Defined Options
2634: * block-ambcond:: Ambiguous Conditions
2635: * block-other:: Other System Documentation
1.14 anton 2636: @end menu
2637:
2638:
2639: @c ---------------------------------------------------------------------
2640: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
2641: @subsection Implementation Defined Options
2642: @c ---------------------------------------------------------------------
2643:
2644: @table @i
2645:
2646: @item the format for display by @code{LIST}:
2647: First the screen number is displayed, then 16 lines of 64 characters,
2648: each line preceded by the line number.
2649:
2650: @item the length of a line affected by @code{\}:
2651: 64 characters.
2652: @end table
2653:
2654:
2655: @c ---------------------------------------------------------------------
2656: @node block-ambcond, block-other, block-idef, The optional Block word set
2657: @subsection Ambiguous conditions
2658: @c ---------------------------------------------------------------------
2659:
2660: @table @i
2661:
2662: @item correct block read was not possible:
2663: Typically results in a @code{throw} of some OS-derived value (between
2664: -512 and -2048). If the blocks file was just not long enough, blanks are
2665: supplied for the missing portion.
2666:
2667: @item I/O exception in block transfer:
2668: Typically results in a @code{throw} of some OS-derived value (between
2669: -512 and -2048).
2670:
2671: @item invalid block number:
2672: @code{-35 throw} (Invalid block number)
2673:
2674: @item a program directly alters the contents of @code{BLK}:
2675: The input stream is switched to that other block, at the same
2676: position. If the storing to @code{BLK} happens when interpreting
2677: non-block input, the system will get quite confused when the block ends.
2678:
2679: @item no current block buffer for @code{UPDATE}:
2680: @code{UPDATE} has no effect.
2681:
2682: @end table
2683:
2684:
2685: @c ---------------------------------------------------------------------
2686: @node block-other, , block-ambcond, The optional Block word set
2687: @subsection Other system documentation
2688: @c ---------------------------------------------------------------------
2689:
2690: @table @i
2691:
2692: @item any restrictions a multiprogramming system places on the use of buffer addresses:
2693: No restrictions (yet).
2694:
2695: @item the number of blocks available for source and data:
2696: depends on your disk space.
2697:
2698: @end table
2699:
2700:
2701: @c =====================================================================
2702: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
2703: @section The optional Double Number word set
2704: @c =====================================================================
2705:
2706: @menu
1.15 anton 2707: * double-ambcond:: Ambiguous Conditions
1.14 anton 2708: @end menu
2709:
2710:
2711: @c ---------------------------------------------------------------------
1.15 anton 2712: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
1.14 anton 2713: @subsection Ambiguous conditions
2714: @c ---------------------------------------------------------------------
2715:
2716: @table @i
2717:
1.15 anton 2718: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.14 anton 2719: The least significant cell of @var{d} is produced.
2720:
2721: @end table
2722:
2723:
2724: @c =====================================================================
2725: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
2726: @section The optional Exception word set
2727: @c =====================================================================
2728:
2729: @menu
1.15 anton 2730: * exception-idef:: Implementation Defined Options
1.14 anton 2731: @end menu
2732:
2733:
2734: @c ---------------------------------------------------------------------
1.15 anton 2735: @node exception-idef, , The optional Exception word set, The optional Exception word set
1.14 anton 2736: @subsection Implementation Defined Options
2737: @c ---------------------------------------------------------------------
2738:
2739: @table @i
2740: @item @code{THROW}-codes used in the system:
2741: The codes -256@minus{}-511 are used for reporting signals (see
2742: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
2743: (for file and memory allocation operations). The mapping from OS error
2744: numbers to throw code is -512@minus{}@var{errno}. One side effect of
2745: this mapping is that undefined OS errors produce a message with a
2746: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
2747: 488} on my system.
2748: @end table
2749:
2750: @c =====================================================================
2751: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
2752: @section The optional Facility word set
2753: @c =====================================================================
2754:
2755: @menu
1.15 anton 2756: * facility-idef:: Implementation Defined Options
2757: * facility-ambcond:: Ambiguous Conditions
1.14 anton 2758: @end menu
2759:
2760:
2761: @c ---------------------------------------------------------------------
2762: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
2763: @subsection Implementation Defined Options
2764: @c ---------------------------------------------------------------------
2765:
2766: @table @i
2767:
2768: @item encoding of keyboard events (@code{EKEY}):
2769: Not yet implemeted.
2770:
2771: @item duration of a system clock tick
2772: System dependent. With respect to @code{MS}, the time is specified in
2773: microseconds. How well the OS and the hardware implement this, is
2774: another question.
2775:
2776: @item repeatability to be expected from the execution of @code{MS}:
2777: System dependent. On Unix, a lot depends on load. If the system is
1.17 anton 2778: lightly loaded, and the delay is short enough that Gforth does not get
1.14 anton 2779: swapped out, the performance should be acceptable. Under MS-DOS and
2780: other single-tasking systems, it should be good.
2781:
2782: @end table
2783:
2784:
2785: @c ---------------------------------------------------------------------
1.15 anton 2786: @node facility-ambcond, , facility-idef, The optional Facility word set
1.14 anton 2787: @subsection Ambiguous conditions
2788: @c ---------------------------------------------------------------------
2789:
2790: @table @i
2791:
2792: @item @code{AT-XY} can't be performed on user output device:
2793: Largely terminal dependant. No range checks are done on the arguments.
2794: No errors are reported. You may see some garbage appearing, you may see
2795: simply nothing happen.
2796:
2797: @end table
2798:
2799:
2800: @c =====================================================================
2801: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
2802: @section The optional File-Access word set
2803: @c =====================================================================
2804:
2805: @menu
1.15 anton 2806: * file-idef:: Implementation Defined Options
2807: * file-ambcond:: Ambiguous Conditions
1.14 anton 2808: @end menu
2809:
2810:
2811: @c ---------------------------------------------------------------------
2812: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
2813: @subsection Implementation Defined Options
2814: @c ---------------------------------------------------------------------
2815:
2816: @table @i
2817:
2818: @item File access methods used:
2819: @code{R/O}, @code{R/W} and @code{BIN} work as you would
2820: expect. @code{W/O} translates into the C file opening mode @code{w} (or
2821: @code{wb}): The file is cleared, if it exists, and created, if it does
1.15 anton 2822: not (both with @code{open-file} and @code{create-file}). Under Unix
1.14 anton 2823: @code{create-file} creates a file with 666 permissions modified by your
2824: umask.
2825:
2826: @item file exceptions:
2827: The file words do not raise exceptions (except, perhaps, memory access
2828: faults when you pass illegal addresses or file-ids).
2829:
2830: @item file line terminator:
2831: System-dependent. Gforth uses C's newline character as line
2832: terminator. What the actual character code(s) of this are is
2833: system-dependent.
2834:
2835: @item file name format
2836: System dependent. Gforth just uses the file name format of your OS.
2837:
2838: @item information returned by @code{FILE-STATUS}:
2839: @code{FILE-STATUS} returns the most powerful file access mode allowed
2840: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
2841: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
2842: along with the retured mode.
2843:
2844: @item input file state after an exception when including source:
2845: All files that are left via the exception are closed.
2846:
2847: @item @var{ior} values and meaning:
1.15 anton 2848: The @var{ior}s returned by the file and memory allocation words are
2849: intended as throw codes. They typically are in the range
2850: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
2851: @var{ior}s is -512@minus{}@var{errno}.
1.14 anton 2852:
2853: @item maximum depth of file input nesting:
2854: limited by the amount of return stack, locals/TIB stack, and the number
2855: of open files available. This should not give you troubles.
2856:
2857: @item maximum size of input line:
2858: @code{/line}. Currently 255.
2859:
2860: @item methods of mapping block ranges to files:
2861: Currently, the block words automatically access the file
2862: @file{blocks.fb} in the currend working directory. More sophisticated
2863: methods could be implemented if there is demand (and a volunteer).
2864:
2865: @item number of string buffers provided by @code{S"}:
2866: 1
2867:
2868: @item size of string buffer used by @code{S"}:
2869: @code{/line}. currently 255.
2870:
2871: @end table
2872:
2873: @c ---------------------------------------------------------------------
1.15 anton 2874: @node file-ambcond, , file-idef, The optional File-Access word set
1.14 anton 2875: @subsection Ambiguous conditions
2876: @c ---------------------------------------------------------------------
2877:
2878: @table @i
2879:
2880: @item attempting to position a file outside it's boundaries:
2881: @code{REPOSITION-FILE} is performed as usual: Afterwards,
2882: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
2883:
2884: @item attempting to read from file positions not yet written:
2885: End-of-file, i.e., zero characters are read and no error is reported.
2886:
2887: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
2888: An appropriate exception may be thrown, but a memory fault or other
2889: problem is more probable.
2890:
2891: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
2892: The @var{ior} produced by the operation, that discovered the problem, is
2893: thrown.
2894:
2895: @item named file cannot be opened (@code{included}):
2896: The @var{ior} produced by @code{open-file} is thrown.
2897:
2898: @item requesting an unmapped block number:
2899: There are no unmapped legal block numbers. On some operating systems,
2900: writing a block with a large number may overflow the file system and
2901: have an error message as consequence.
2902:
2903: @item using @code{source-id} when @code{blk} is non-zero:
2904: @code{source-id} performs its function. Typically it will give the id of
2905: the source which loaded the block. (Better ideas?)
2906:
2907: @end table
2908:
2909:
2910: @c =====================================================================
2911: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15 anton 2912: @section The optional Floating-Point word set
1.14 anton 2913: @c =====================================================================
2914:
2915: @menu
1.15 anton 2916: * floating-idef:: Implementation Defined Options
2917: * floating-ambcond:: Ambiguous Conditions
1.14 anton 2918: @end menu
2919:
2920:
2921: @c ---------------------------------------------------------------------
2922: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
2923: @subsection Implementation Defined Options
2924: @c ---------------------------------------------------------------------
2925:
2926: @table @i
2927:
1.15 anton 2928: @item format and range of floating point numbers:
2929: System-dependent; the @code{double} type of C.
1.14 anton 2930:
1.15 anton 2931: @item results of @code{REPRESENT} when @var{float} is out of range:
2932: System dependent; @code{REPRESENT} is implemented using the C library
2933: function @code{ecvt()} and inherits its behaviour in this respect.
1.14 anton 2934:
1.15 anton 2935: @item rounding or truncation of floating-point numbers:
1.26 ! anton 2936: System dependent; the rounding behaviour is inherited from the hosting C
! 2937: compiler. IEEE-FP-based (i.e., most) systems by default round to
! 2938: nearest, and break ties by rounding to even (i.e., such that the last
! 2939: bit of the mantissa is 0).
1.14 anton 2940:
1.15 anton 2941: @item size of floating-point stack:
2942: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
2943: with the command-line option @code{-f}.
1.14 anton 2944:
1.15 anton 2945: @item width of floating-point stack:
2946: @code{1 floats}.
1.14 anton 2947:
2948: @end table
2949:
2950:
2951: @c ---------------------------------------------------------------------
1.15 anton 2952: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
2953: @subsection Ambiguous conditions
1.14 anton 2954: @c ---------------------------------------------------------------------
2955:
2956: @table @i
2957:
1.15 anton 2958: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
2959: System-dependent. Typically results in an alignment fault like other
2960: alignment violations.
1.14 anton 2961:
1.15 anton 2962: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
2963: System-dependent. Typically results in an alignment fault like other
2964: alignment violations.
1.14 anton 2965:
1.15 anton 2966: @item Floating-point result out of range:
2967: System-dependent. Can result in a @code{-55 THROW} (Floating-point
2968: unidentified fault), or can produce a special value representing, e.g.,
2969: Infinity.
1.14 anton 2970:
1.15 anton 2971: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
2972: System-dependent. Typically results in an alignment fault like other
2973: alignment violations.
1.14 anton 2974:
1.15 anton 2975: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
2976: The floating-point number is converted into decimal nonetheless.
1.14 anton 2977:
1.15 anton 2978: @item Both arguments are equal to zero (@code{FATAN2}):
2979: System-dependent. @code{FATAN2} is implemented using the C library
2980: function @code{atan2()}.
1.14 anton 2981:
1.15 anton 2982: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
2983: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
2984: because of small errors and the tan will be a very large (or very small)
2985: but finite number.
1.14 anton 2986:
1.15 anton 2987: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
2988: The result is rounded to the nearest float.
1.14 anton 2989:
1.15 anton 2990: @item dividing by zero:
2991: @code{-55 throw} (Floating-point unidentified fault)
1.14 anton 2992:
1.15 anton 2993: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
2994: System dependent. On IEEE-FP based systems the number is converted into
2995: an infinity.
1.14 anton 2996:
1.15 anton 2997: @item @var{float}<1 (@code{facosh}):
2998: @code{-55 throw} (Floating-point unidentified fault)
1.14 anton 2999:
1.15 anton 3000: @item @var{float}=<-1 (@code{flnp1}):
3001: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3002: negative infinity is typically produced for @var{float}=-1.
1.14 anton 3003:
1.15 anton 3004: @item @var{float}=<0 (@code{fln}, @code{flog}):
3005: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
3006: negative infinity is typically produced for @var{float}=0.
1.14 anton 3007:
1.15 anton 3008: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
3009: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
3010: produces values for these inputs on my Linux box (Bug in the C library?)
1.14 anton 3011:
1.15 anton 3012: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
3013: @code{-55 throw} (Floating-point unidentified fault).
1.14 anton 3014:
1.15 anton 3015: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
3016: @code{-55 throw} (Floating-point unidentified fault).
1.14 anton 3017:
1.15 anton 3018: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
3019: This does not happen.
3020: @end table
1.14 anton 3021:
3022:
3023:
3024: @c =====================================================================
1.15 anton 3025: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
3026: @section The optional Locals word set
1.14 anton 3027: @c =====================================================================
3028:
3029: @menu
1.15 anton 3030: * locals-idef:: Implementation Defined Options
3031: * locals-ambcond:: Ambiguous Conditions
1.14 anton 3032: @end menu
3033:
3034:
3035: @c ---------------------------------------------------------------------
1.15 anton 3036: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14 anton 3037: @subsection Implementation Defined Options
3038: @c ---------------------------------------------------------------------
3039:
3040: @table @i
3041:
1.15 anton 3042: @item maximum number of locals in a definition:
3043: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
3044: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
3045: characters. The number of locals in a definition is bounded by the size
3046: of locals-buffer, which contains the names of the locals.
1.14 anton 3047:
3048: @end table
3049:
3050:
3051: @c ---------------------------------------------------------------------
1.15 anton 3052: @node locals-ambcond, , locals-idef, The optional Locals word set
1.14 anton 3053: @subsection Ambiguous conditions
3054: @c ---------------------------------------------------------------------
3055:
3056: @table @i
3057:
1.15 anton 3058: @item executing a named local in interpretation state:
3059: @code{-14 throw} (Interpreting a compile-only word).
1.14 anton 3060:
1.15 anton 3061: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
3062: @code{-32 throw} (Invalid name argument)
1.14 anton 3063:
3064: @end table
3065:
3066:
3067: @c =====================================================================
1.15 anton 3068: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
3069: @section The optional Memory-Allocation word set
1.14 anton 3070: @c =====================================================================
3071:
3072: @menu
1.15 anton 3073: * memory-idef:: Implementation Defined Options
1.14 anton 3074: @end menu
3075:
3076:
3077: @c ---------------------------------------------------------------------
1.15 anton 3078: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14 anton 3079: @subsection Implementation Defined Options
3080: @c ---------------------------------------------------------------------
3081:
3082: @table @i
3083:
1.15 anton 3084: @item values and meaning of @var{ior}:
3085: The @var{ior}s returned by the file and memory allocation words are
3086: intended as throw codes. They typically are in the range
3087: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3088: @var{ior}s is -512@minus{}@var{errno}.
1.14 anton 3089:
3090: @end table
3091:
3092: @c =====================================================================
1.15 anton 3093: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
3094: @section The optional Programming-Tools word set
1.14 anton 3095: @c =====================================================================
3096:
3097: @menu
1.15 anton 3098: * programming-idef:: Implementation Defined Options
3099: * programming-ambcond:: Ambiguous Conditions
1.14 anton 3100: @end menu
3101:
3102:
3103: @c ---------------------------------------------------------------------
1.15 anton 3104: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14 anton 3105: @subsection Implementation Defined Options
3106: @c ---------------------------------------------------------------------
3107:
3108: @table @i
3109:
1.15 anton 3110: @item ending sequence for input following @code{;code} and @code{code}:
3111: Not implemented (yet).
1.14 anton 3112:
1.15 anton 3113: @item manner of processing input following @code{;code} and @code{code}:
3114: Not implemented (yet).
3115:
3116: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
3117: Not implemented (yet). If they were implemented, they would use the
3118: search order wordset.
3119:
3120: @item source and format of display by @code{SEE}:
3121: The source for @code{see} is the intermediate code used by the inner
3122: interpreter. The current @code{see} tries to output Forth source code
3123: as well as possible.
3124:
1.14 anton 3125: @end table
3126:
3127: @c ---------------------------------------------------------------------
1.15 anton 3128: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
1.14 anton 3129: @subsection Ambiguous conditions
3130: @c ---------------------------------------------------------------------
3131:
3132: @table @i
3133:
1.15 anton 3134: @item deleting the compilation wordlist (@code{FORGET}):
3135: Not implemented (yet).
1.14 anton 3136:
1.15 anton 3137: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
3138: This typically results in an @code{abort"} with a descriptive error
3139: message (may change into a @code{-22 throw} (Control structure mismatch)
3140: in the future). You may also get a memory access error. If you are
3141: unlucky, this ambiguous condition is not caught.
3142:
3143: @item @var{name} can't be found (@code{forget}):
3144: Not implemented (yet).
1.14 anton 3145:
1.15 anton 3146: @item @var{name} not defined via @code{CREATE}:
3147: @code{;code} is not implemented (yet). If it were, it would behave like
3148: @code{DOES>} in this respect, i.e., change the execution semantics of
3149: the last defined word no matter how it was defined.
1.14 anton 3150:
1.15 anton 3151: @item @code{POSTPONE} applied to @code{[IF]}:
3152: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
3153: equivalent to @code{[IF]}.
1.14 anton 3154:
1.15 anton 3155: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
3156: Continue in the same state of conditional compilation in the next outer
3157: input source. Currently there is no warning to the user about this.
1.14 anton 3158:
1.15 anton 3159: @item removing a needed definition (@code{FORGET}):
3160: Not implemented (yet).
1.14 anton 3161:
3162: @end table
3163:
3164:
3165: @c =====================================================================
1.15 anton 3166: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
3167: @section The optional Search-Order word set
1.14 anton 3168: @c =====================================================================
3169:
3170: @menu
1.15 anton 3171: * search-idef:: Implementation Defined Options
3172: * search-ambcond:: Ambiguous Conditions
1.14 anton 3173: @end menu
3174:
3175:
3176: @c ---------------------------------------------------------------------
1.15 anton 3177: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14 anton 3178: @subsection Implementation Defined Options
3179: @c ---------------------------------------------------------------------
3180:
3181: @table @i
3182:
1.15 anton 3183: @item maximum number of word lists in search order:
3184: @code{s" wordlists" environment? drop .}. Currently 16.
3185:
3186: @item minimum search order:
3187: @code{root root}.
1.14 anton 3188:
3189: @end table
3190:
3191: @c ---------------------------------------------------------------------
1.15 anton 3192: @node search-ambcond, , search-idef, The optional Search-Order word set
1.14 anton 3193: @subsection Ambiguous conditions
3194: @c ---------------------------------------------------------------------
3195:
3196: @table @i
3197:
1.15 anton 3198: @item changing the compilation wordlist (during compilation):
3199: The definition is put into the wordlist that is the compilation wordlist
3200: when @code{REVEAL} is executed (by @code{;}, @code{DOES>},
3201: @code{RECURSIVE}, etc.).
1.14 anton 3202:
1.15 anton 3203: @item search order empty (@code{previous}):
3204: @code{abort" Vocstack empty"}.
1.14 anton 3205:
1.15 anton 3206: @item too many word lists in search order (@code{also}):
3207: @code{abort" Vocstack full"}.
1.14 anton 3208:
3209: @end table
1.13 anton 3210:
3211:
1.17 anton 3212: @node Model, Emacs and Gforth, ANS conformance, Top
1.4 anton 3213: @chapter Model
3214:
1.17 anton 3215: @node Emacs and Gforth, Internals, Model, Top
3216: @chapter Emacs and Gforth
1.4 anton 3217:
1.17 anton 3218: Gforth comes with @file{gforth.el}, an improved version of
1.4 anton 3219: @file{forth.el} by Goran Rydqvist (icluded in the TILE package). The
3220: improvements are a better (but still not perfect) handling of
3221: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8 anton 3222: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
3223: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
3224: stuff I do not use alone, even though some of it only makes sense for
3225: TILE. To get a description of these features, enter Forth mode and type
3226: @kbd{C-h m}.
1.4 anton 3227:
1.17 anton 3228: In addition, Gforth supports Emacs quite well: The source code locations
1.4 anton 3229: given in error messages, debugging output (from @code{~~}) and failed
3230: assertion messages are in the right format for Emacs' compilation mode
3231: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
3232: Manual}) so the source location corresponding to an error or other
3233: message is only a few keystrokes away (@kbd{C-x `} for the next error,
3234: @kbd{C-c C-c} for the error under the cursor).
3235:
3236: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
3237: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
3238: contains the definitions of all words defined afterwards. You can then
3239: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17 anton 3240: several tags files at the same time (e.g., one for the Gforth sources
1.4 anton 3241: and one for your program).
3242:
3243: To get all these benefits, add the following lines to your @file{.emacs}
3244: file:
3245:
3246: @example
3247: (autoload 'forth-mode "gforth.el")
3248: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
3249: @end example
3250:
1.17 anton 3251: @node Internals, Bugs, Emacs and Gforth, Top
1.3 anton 3252: @chapter Internals
3253:
1.17 anton 3254: Reading this section is not necessary for programming with Gforth. It
3255: should be helpful for finding your way in the Gforth sources.
1.3 anton 3256:
1.24 anton 3257: The ideas in this section have also been published in the papers
3258: @cite{ANS fig/GNU/??? Forth} (in German) by Bernd Paysan, presented at
3259: the Forth-Tagung '93 and @cite{A Portable Forth Engine} by M. Anton
3260: Ertl, presented at EuroForth '93; the latter is available at
3261: @*@file{http://www.complang.tuwien.ac.at/papers/ertl93.ps.Z}.
3262:
1.4 anton 3263: @menu
3264: * Portability::
3265: * Threading::
3266: * Primitives::
3267: * System Architecture::
1.17 anton 3268: * Performance::
1.4 anton 3269: @end menu
3270:
3271: @node Portability, Threading, Internals, Internals
1.3 anton 3272: @section Portability
3273:
3274: One of the main goals of the effort is availability across a wide range
3275: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
3276: this goal by manually coding the engine in assembly language for several
3277: then-popular processors. This approach is very labor-intensive and the
3278: results are short-lived due to progress in computer architecture.
3279:
3280: Others have avoided this problem by coding in C, e.g., Mitch Bradley
3281: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
3282: particularly popular for UNIX-based Forths due to the large variety of
3283: architectures of UNIX machines. Unfortunately an implementation in C
3284: does not mix well with the goals of efficiency and with using
3285: traditional techniques: Indirect or direct threading cannot be expressed
3286: in C, and switch threading, the fastest technique available in C, is
3287: significantly slower. Another problem with C is that it's very
3288: cumbersome to express double integer arithmetic.
3289:
3290: Fortunately, there is a portable language that does not have these
3291: limitations: GNU C, the version of C processed by the GNU C compiler
3292: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
3293: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
3294: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
3295: threading possible, its @code{long long} type (@pxref{Long Long, ,
3296: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forths
3297: double numbers. GNU C is available for free on all important (and many
3298: unimportant) UNIX machines, VMS, 80386s running MS-DOS, the Amiga, and
3299: the Atari ST, so a Forth written in GNU C can run on all these
1.17 anton 3300: machines.
1.3 anton 3301:
3302: Writing in a portable language has the reputation of producing code that
3303: is slower than assembly. For our Forth engine we repeatedly looked at
3304: the code produced by the compiler and eliminated most compiler-induced
3305: inefficiencies by appropriate changes in the source-code.
3306:
3307: However, register allocation cannot be portably influenced by the
3308: programmer, leading to some inefficiencies on register-starved
3309: machines. We use explicit register declarations (@pxref{Explicit Reg
3310: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
3311: improve the speed on some machines. They are turned on by using the
3312: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
3313: only depends on the machine, but also on the compiler version: On some
3314: machines some compiler versions produce incorrect code when certain
3315: explicit register declarations are used. So by default
3316: @code{-DFORCE_REG} is not used.
3317:
1.4 anton 3318: @node Threading, Primitives, Portability, Internals
1.3 anton 3319: @section Threading
3320:
3321: GNU C's labels as values extension (available since @code{gcc-2.0},
3322: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
3323: makes it possible to take the address of @var{label} by writing
3324: @code{&&@var{label}}. This address can then be used in a statement like
3325: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
3326: @code{goto x}.
3327:
3328: With this feature an indirect threaded NEXT looks like:
3329: @example
3330: cfa = *ip++;
3331: ca = *cfa;
3332: goto *ca;
3333: @end example
3334: For those unfamiliar with the names: @code{ip} is the Forth instruction
3335: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
3336: execution token and points to the code field of the next word to be
3337: executed; The @code{ca} (code address) fetched from there points to some
3338: executable code, e.g., a primitive or the colon definition handler
3339: @code{docol}.
3340:
3341: Direct threading is even simpler:
3342: @example
3343: ca = *ip++;
3344: goto *ca;
3345: @end example
3346:
3347: Of course we have packaged the whole thing neatly in macros called
3348: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
3349:
1.4 anton 3350: @menu
3351: * Scheduling::
3352: * Direct or Indirect Threaded?::
3353: * DOES>::
3354: @end menu
3355:
3356: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3 anton 3357: @subsection Scheduling
3358:
3359: There is a little complication: Pipelined and superscalar processors,
3360: i.e., RISC and some modern CISC machines can process independent
3361: instructions while waiting for the results of an instruction. The
3362: compiler usually reorders (schedules) the instructions in a way that
3363: achieves good usage of these delay slots. However, on our first tries
3364: the compiler did not do well on scheduling primitives. E.g., for
3365: @code{+} implemented as
3366: @example
3367: n=sp[0]+sp[1];
3368: sp++;
3369: sp[0]=n;
3370: NEXT;
3371: @end example
3372: the NEXT comes strictly after the other code, i.e., there is nearly no
3373: scheduling. After a little thought the problem becomes clear: The
3374: compiler cannot know that sp and ip point to different addresses (and
1.4 anton 3375: the version of @code{gcc} we used would not know it even if it was
3376: possible), so it could not move the load of the cfa above the store to
3377: the TOS. Indeed the pointers could be the same, if code on or very near
3378: the top of stack were executed. In the interest of speed we chose to
3379: forbid this probably unused ``feature'' and helped the compiler in
3380: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
3381: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3 anton 3382: @example
3383: n=sp[0]+sp[1];
3384: sp++;
3385: NEXT_P1;
3386: sp[0]=n;
3387: NEXT_P2;
3388: @end example
1.4 anton 3389: This can be scheduled optimally by the compiler.
1.3 anton 3390:
3391: This division can be turned off with the switch @code{-DCISC_NEXT}. This
3392: switch is on by default on machines that do not profit from scheduling
3393: (e.g., the 80386), in order to preserve registers.
3394:
1.4 anton 3395: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3 anton 3396: @subsection Direct or Indirect Threaded?
3397:
3398: Both! After packaging the nasty details in macro definitions we
3399: realized that we could switch between direct and indirect threading by
3400: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
3401: defining a few machine-specific macros for the direct-threading case.
3402: On the Forth level we also offer access words that hide the
3403: differences between the threading methods (@pxref{Threading Words}).
3404:
3405: Indirect threading is implemented completely
3406: machine-independently. Direct threading needs routines for creating
3407: jumps to the executable code (e.g. to docol or dodoes). These routines
3408: are inherently machine-dependent, but they do not amount to many source
3409: lines. I.e., even porting direct threading to a new machine is a small
3410: effort.
3411:
1.4 anton 3412: @node DOES>, , Direct or Indirect Threaded?, Threading
1.3 anton 3413: @subsection DOES>
3414: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
3415: the chunk of code executed by every word defined by a
3416: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
3417: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
3418: DOES-code)? There are two solutions:
3419:
3420: In fig-Forth the code field points directly to the dodoes and the
3421: DOES-code address is stored in the cell after the code address
3422: (i.e. at cfa cell+). It may seem that this solution is illegal in the
3423: Forth-79 and all later standards, because in fig-Forth this address
3424: lies in the body (which is illegal in these standards). However, by
3425: making the code field larger for all words this solution becomes legal
3426: again. We use this approach for the indirect threaded version. Leaving
3427: a cell unused in most words is a bit wasteful, but on the machines we
3428: are targetting this is hardly a problem. The other reason for having a
3429: code field size of two cells is to avoid having different image files
1.4 anton 3430: for direct and indirect threaded systems (@pxref{System Architecture}).
1.3 anton 3431:
3432: The other approach is that the code field points or jumps to the cell
3433: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
3434: this address. @code{dodoes} can then get the DOES-code address by
3435: computing the code address, i.e., the address of the jump to dodoes,
3436: and add the length of that jump field. A variant of this is to have a
3437: call to @code{dodoes} after the @code{DOES>}; then the return address
3438: (which can be found in the return register on RISCs) is the DOES-code
3439: address. Since the two cells available in the code field are usually
3440: used up by the jump to the code address in direct threading, we use
3441: this approach for direct threading. We did not want to add another
3442: cell to the code field.
3443:
1.4 anton 3444: @node Primitives, System Architecture, Threading, Internals
1.3 anton 3445: @section Primitives
3446:
1.4 anton 3447: @menu
3448: * Automatic Generation::
3449: * TOS Optimization::
3450: * Produced code::
3451: @end menu
3452:
3453: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3 anton 3454: @subsection Automatic Generation
3455:
3456: Since the primitives are implemented in a portable language, there is no
3457: longer any need to minimize the number of primitives. On the contrary,
3458: having many primitives is an advantage: speed. In order to reduce the
3459: number of errors in primitives and to make programming them easier, we
3460: provide a tool, the primitive generator (@file{prims2x.fs}), that
3461: automatically generates most (and sometimes all) of the C code for a
3462: primitive from the stack effect notation. The source for a primitive
3463: has the following form:
3464:
3465: @format
3466: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
3467: [@code{""}@var{glossary entry}@code{""}]
3468: @var{C code}
3469: [@code{:}
3470: @var{Forth code}]
3471: @end format
3472:
3473: The items in brackets are optional. The category and glossary fields
3474: are there for generating the documentation, the Forth code is there
3475: for manual implementations on machines without GNU C. E.g., the source
3476: for the primitive @code{+} is:
3477: @example
3478: + n1 n2 -- n core plus
3479: n = n1+n2;
3480: @end example
3481:
3482: This looks like a specification, but in fact @code{n = n1+n2} is C
3483: code. Our primitive generation tool extracts a lot of information from
3484: the stack effect notations@footnote{We use a one-stack notation, even
3485: though we have separate data and floating-point stacks; The separate
3486: notation can be generated easily from the unified notation.}: The number
3487: of items popped from and pushed on the stack, their type, and by what
3488: name they are referred to in the C code. It then generates a C code
3489: prelude and postlude for each primitive. The final C code for @code{+}
3490: looks like this:
3491:
3492: @example
3493: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
3494: /* */ /* documentation */
1.4 anton 3495: @{
1.3 anton 3496: DEF_CA /* definition of variable ca (indirect threading) */
3497: Cell n1; /* definitions of variables */
3498: Cell n2;
3499: Cell n;
3500: n1 = (Cell) sp[1]; /* input */
3501: n2 = (Cell) TOS;
3502: sp += 1; /* stack adjustment */
3503: NAME("+") /* debugging output (with -DDEBUG) */
1.4 anton 3504: @{
1.3 anton 3505: n = n1+n2; /* C code taken from the source */
1.4 anton 3506: @}
1.3 anton 3507: NEXT_P1; /* NEXT part 1 */
3508: TOS = (Cell)n; /* output */
3509: NEXT_P2; /* NEXT part 2 */
1.4 anton 3510: @}
1.3 anton 3511: @end example
3512:
3513: This looks long and inefficient, but the GNU C compiler optimizes quite
3514: well and produces optimal code for @code{+} on, e.g., the R3000 and the
3515: HP RISC machines: Defining the @code{n}s does not produce any code, and
3516: using them as intermediate storage also adds no cost.
3517:
3518: There are also other optimizations, that are not illustrated by this
3519: example: Assignments between simple variables are usually for free (copy
3520: propagation). If one of the stack items is not used by the primitive
3521: (e.g. in @code{drop}), the compiler eliminates the load from the stack
3522: (dead code elimination). On the other hand, there are some things that
3523: the compiler does not do, therefore they are performed by
3524: @file{prims2x.fs}: The compiler does not optimize code away that stores
3525: a stack item to the place where it just came from (e.g., @code{over}).
3526:
3527: While programming a primitive is usually easy, there are a few cases
3528: where the programmer has to take the actions of the generator into
3529: account, most notably @code{?dup}, but also words that do not (always)
3530: fall through to NEXT.
3531:
1.4 anton 3532: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3 anton 3533: @subsection TOS Optimization
3534:
3535: An important optimization for stack machine emulators, e.g., Forth
3536: engines, is keeping one or more of the top stack items in
1.4 anton 3537: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
3538: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.3 anton 3539: @itemize
3540: @item
3541: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
3542: due to fewer loads from and stores to the stack.
3543: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
3544: @var{y<n}, due to additional moves between registers.
3545: @end itemize
3546:
3547: In particular, keeping one item in a register is never a disadvantage,
3548: if there are enough registers. Keeping two items in registers is a
3549: disadvantage for frequent words like @code{?branch}, constants,
3550: variables, literals and @code{i}. Therefore our generator only produces
3551: code that keeps zero or one items in registers. The generated C code
3552: covers both cases; the selection between these alternatives is made at
3553: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
3554: code for @code{+} is just a simple variable name in the one-item case,
3555: otherwise it is a macro that expands into @code{sp[0]}. Note that the
3556: GNU C compiler tries to keep simple variables like @code{TOS} in
3557: registers, and it usually succeeds, if there are enough registers.
3558:
3559: The primitive generator performs the TOS optimization for the
3560: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
3561: operations the benefit of this optimization is even larger:
3562: floating-point operations take quite long on most processors, but can be
3563: performed in parallel with other operations as long as their results are
3564: not used. If the FP-TOS is kept in a register, this works. If
3565: it is kept on the stack, i.e., in memory, the store into memory has to
3566: wait for the result of the floating-point operation, lengthening the
3567: execution time of the primitive considerably.
3568:
3569: The TOS optimization makes the automatic generation of primitives a
3570: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
3571: @code{TOS} is not sufficient. There are some special cases to
3572: consider:
3573: @itemize
3574: @item In the case of @code{dup ( w -- w w )} the generator must not
3575: eliminate the store to the original location of the item on the stack,
3576: if the TOS optimization is turned on.
1.4 anton 3577: @item Primitives with stack effects of the form @code{--}
3578: @var{out1}...@var{outy} must store the TOS to the stack at the start.
3579: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3 anton 3580: must load the TOS from the stack at the end. But for the null stack
3581: effect @code{--} no stores or loads should be generated.
3582: @end itemize
3583:
1.4 anton 3584: @node Produced code, , TOS Optimization, Primitives
1.3 anton 3585: @subsection Produced code
3586:
3587: To see what assembly code is produced for the primitives on your machine
3588: with your compiler and your flag settings, type @code{make engine.s} and
1.4 anton 3589: look at the resulting file @file{engine.s}.
1.3 anton 3590:
1.17 anton 3591: @node System Architecture, Performance, Primitives, Internals
1.3 anton 3592: @section System Architecture
3593:
3594: Our Forth system consists not only of primitives, but also of
3595: definitions written in Forth. Since the Forth compiler itself belongs
3596: to those definitions, it is not possible to start the system with the
3597: primitives and the Forth source alone. Therefore we provide the Forth
3598: code as an image file in nearly executable form. At the start of the
3599: system a C routine loads the image file into memory, sets up the
3600: memory (stacks etc.) according to information in the image file, and
3601: starts executing Forth code.
3602:
3603: The image file format is a compromise between the goals of making it
3604: easy to generate image files and making them portable. The easiest way
3605: to generate an image file is to just generate a memory dump. However,
3606: this kind of image file cannot be used on a different machine, or on
3607: the next version of the engine on the same machine, it even might not
3608: work with the same engine compiled by a different version of the C
3609: compiler. We would like to have as few versions of the image file as
3610: possible, because we do not want to distribute many versions of the
3611: same image file, and to make it easy for the users to use their image
3612: files on many machines. We currently need to create a different image
3613: file for machines with different cell sizes and different byte order
1.17 anton 3614: (little- or big-endian)@footnote{We are considering adding information to the
1.3 anton 3615: image file that enables the loader to change the byte order.}.
3616:
3617: Forth code that is going to end up in a portable image file has to
1.4 anton 3618: comply to some restrictions: addresses have to be stored in memory with
3619: special words (@code{A!}, @code{A,}, etc.) in order to make the code
3620: relocatable. Cells, floats, etc., have to be stored at the natural
3621: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
3622: dividable by~8. This happens automatically in our system when you use
3623: the ANS Forth alignment words.}, in order to avoid alignment faults on
3624: machines with stricter alignment. The image file is produced by a
3625: metacompiler (@file{cross.fs}).
1.3 anton 3626:
3627: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
3628: file is not directly executable, but has to undergo some manipulations
3629: during loading. Address relocation is performed at image load-time, not
3630: at run-time. The loader also has to replace tokens standing for
3631: primitive calls with the appropriate code-field addresses (or code
3632: addresses in the case of direct threading).
1.4 anton 3633:
1.17 anton 3634: @node Performance, , System Architecture, Internals
3635: @section Performance
3636:
3637: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
3638: impossible to write a significantly faster engine.
3639:
3640: On register-starved machines like the 386 architecture processors
3641: improvements are possible, because @code{gcc} does not utilize the
3642: registers as well as a human, even with explicit register declarations;
3643: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
3644: and hand-tuned it for the 486; this system is 1.19 times faster on the
3645: Sieve benchmark on a 486DX2/66 than Gforth compiled with
3646: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
3647:
3648: However, this potential advantage of assembly language implementations
3649: is not necessarily realized in complete Forth systems: We compared
1.26 ! anton 3650: Gforth (direct threaded, compiled with @code{gcc-2.6.3} and
! 3651: @code{-DFORCE_REG}) with Win32Forth 1.2093, LMI's NT Forth (Beta, May
! 3652: 1994) and Eforth (with and without peephole (aka pinhole) optimization
! 3653: of the threaded code); all these systems were written in assembly
! 3654: language. We also compared Gforth with two systems written in C:
! 3655: PFE-0.9.11 (compiled with @code{gcc-2.6.3} with the default
! 3656: configuration for Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS}) and
! 3657: ThisForth Beta (compiled with gcc-2.6.3 -O3 -fomit-frame-pointer;
! 3658: ThisForth employs peephole optimization of the threaded code). We
! 3659: benchmarked Gforth, PFE and ThisForth on a 486DX2/66 under
! 3660: Linux. Kenneth O'Heskin kindly provided the results for Win32Forth and
! 3661: NT Forth on a 486DX2/66 with similar memory performance under Windows
! 3662: NT. Marcel Hendrix ported Eforth to Linux, then extended it to run the
! 3663: benchmarks, added the peephole optimizer, ran the benchmarks and
! 3664: reported the results.
1.17 anton 3665:
3666: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
3667: matrix multiplication come from the Stanford integer benchmarks and have
3668: been translated into Forth by Martin Fraeman; we used the versions
3669: included in the TILE Forth package; and a recursive Fibonacci number
1.24 anton 3670: computation for benchmarking calling performance. The following table shows
1.17 anton 3671: the time taken for the benchmarks scaled by the time taken by Gforth (in
3672: other words, it shows the speedup factor that Gforth achieved over the
3673: other systems).
3674:
3675: @example
1.26 ! anton 3676: relative Win32- NT eforth This-
! 3677: time Gforth Forth Forth eforth +opt PFE Forth
! 3678: sieve 1.00 1.39 1.14 1.39 0.85 1.78 3.18
! 3679: bubble 1.00 1.33 1.43 1.51 0.89 1.70
! 3680: matmul 1.00 1.43 1.31 1.42 1.12 2.28
! 3681: fib 1.00 1.55 1.36 1.24 1.15 1.97 3.04
1.17 anton 3682: @end example
3683:
3684: You may find the good performance of Gforth compared with the systems
3685: written in assembly language quite surprising. One important reason for
3686: the disappointing performance of these systems is probably that they are
3687: not written optimally for the 486 (e.g., they use the @code{lods}
3688: instruction). In addition, Win32Forth uses a comfortable, but costly
3689: method for relocating the Forth image: like @code{cforth}, it computes
3690: the actual addresses at run time, resulting in two address computations
3691: per NEXT (@pxref{System Architecture}).
3692:
1.26 ! anton 3693: Only Eforth with the peephole optimizer performs comparable to
! 3694: Gforth. The speedups achieved with peephole optimization of threaded
! 3695: code are quite remarkable. Adding a peephole optimizer to Gforth should
! 3696: cause similar speedups.
! 3697:
1.17 anton 3698: The speedup of Gforth over PFE and ThisForth can be easily explained
3699: with the self-imposed restriction to standard C (although the measured
3700: implementation of PFE uses a GNU C extension: global register
3701: variables), which makes efficient threading impossible. Moreover,
3702: current C compilers have a hard time optimizing other aspects of the
3703: ThisForth source.
3704:
3705: Note that the performance of Gforth on 386 architecture processors
3706: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
3707: failed to allocate any of the virtual machine registers into real
3708: machine registers by itself and would not work correctly with explicit
3709: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
3710: running the Sieve) than the one measured above.
3711:
1.26 ! anton 3712: In @cite{Translating Forth to Efficient C} by M. Anton Ertl and Martin
! 3713: Maierhofer (presented at EuroForth '95), an indirect threaded version of
! 3714: Gforth is compared with Win32Forth, NT Forth, PFE, and ThisForth; that
! 3715: version of Gforth is 2\%@minus{}8\% slower on a 486 than the version
! 3716: used here. The paper available at
1.24 anton 3717: @*@file{http://www.complang.tuwien.ac.at/papers/ertl&maierhofer95.ps.gz};
3718: it also contains numbers for some native code systems. You can find
3719: numbers for Gforth on various machines in @file{Benchres}.
3720:
1.4 anton 3721: @node Bugs, Pedigree, Internals, Top
3722: @chapter Bugs
3723:
1.17 anton 3724: Known bugs are described in the file BUGS in the Gforth distribution.
3725:
1.24 anton 3726: If you find a bug, please send a bug report to
3727: @code{gforth-bugs@@mips.complang.tuwien.ac.at}. A bug report should
1.17 anton 3728: describe the Gforth version used (it is announced at the start of an
3729: interactive Gforth session), the machine and operating system (on Unix
3730: systems you can use @code{uname -a} to produce this information), the
1.24 anton 3731: installation options (send the @code{config.status} file), and a
3732: complete list of changes you (or your installer) have made to the Gforth
3733: sources (if any); it should contain a program (or a sequence of keyboard
3734: commands) that reproduces the bug and a description of what you think
3735: constitutes the buggy behaviour.
1.17 anton 3736:
3737: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
3738: to Report Bugs, gcc.info, GNU C Manual}.
3739:
3740:
1.4 anton 3741: @node Pedigree, Word Index, Bugs, Top
3742: @chapter Pedigree
3743:
1.17 anton 3744: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
1.24 anton 3745: Dirk Zoller) will cross-fertilize each other. Of course, a significant
3746: part of the design of Gforth was prescribed by ANS Forth.
1.17 anton 3747:
1.23 pazsan 3748: Bernd Paysan wrote BigForth, a descendent from TurboForth, an unreleased
3749: 32 bit native code version of VolksForth for the Atari ST, written
3750: mostly by Dietrich Weineck.
3751:
3752: VolksForth descends from F83. It was written by Klaus Schleisiek, Bernd
3753: Pennemann, Georg Rehfeld and Dietrich Weineck for the C64 (called
1.24 anton 3754: UltraForth there) in the mid-80s and ported to the Atari ST in 1986.
1.17 anton 3755:
1.25 anton 3756: Hennry Laxen and Mike Perry wrote F83 as a model implementation of the
1.17 anton 3757: Forth-83 standard. !! Pedigree? When?
3758:
3759: A team led by Bill Ragsdale implemented fig-Forth on many processors in
1.24 anton 3760: 1979. Robert Selzer and Bill Ragsdale developed the original
3761: implementation of fig-Forth for the 6502 based on microForth.
3762:
3763: The principal architect of microForth was Dean Sanderson. microForth was
3764: FORTH, Inc.'s first off-the-shelf product. It was developped in 1976 for
3765: the 1802, and subsequently implemented on the 8080, the 6800 and the
3766: Z80.
1.17 anton 3767:
1.24 anton 3768: All earlier Forth systems were custom-made, usually by Charles Moore,
3769: who discovered (as he puts it) Forth in the late 60s.
1.17 anton 3770:
3771: A part of the information in this section comes from @cite{The Evolution
3772: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
3773: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
3774: Notices 28(3), 1993. You can find more historical and genealogical
3775: information about Forth there.
3776:
1.4 anton 3777: @node Word Index, Node Index, Pedigree, Top
3778: @chapter Word Index
3779:
1.18 anton 3780: This index is as incomplete as the manual. Each word is listed with
3781: stack effect and wordset.
1.17 anton 3782:
3783: @printindex fn
3784:
1.4 anton 3785: @node Node Index, , Word Index, Top
3786: @chapter Node Index
1.17 anton 3787:
3788: This index is even less complete than the manual.
1.1 anton 3789:
3790: @contents
3791: @bye
3792:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>