Annotation of gforth/gforth.ds, revision 1.20
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:
1.17 anton 12: Copyright @copyright{} 1994 Gforth Development Group
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:
39: @titlepage
40: @sp 10
1.17 anton 41: @center @titlefont{Gforth Manual}
1.1 anton 42: @sp 2
1.17 anton 43: @center for version 0.1
1.1 anton 44: @sp 2
45: @center Anton Ertl
1.17 anton 46: @sp 3
47: @center This manual is under construction
1.1 anton 48:
49: @comment The following two commands start the copyright page.
50: @page
51: @vskip 0pt plus 1filll
1.17 anton 52: Copyright @copyright{} 1994 Gforth Development Group
1.1 anton 53:
54: @comment !! Published by ... or You can get a copy of this manual ...
55:
56: Permission is granted to make and distribute verbatim copies of
57: this manual provided the copyright notice and this permission notice
58: are preserved on all copies.
59:
60: Permission is granted to copy and distribute modified versions of this
61: manual under the conditions for verbatim copying, provided also that the
62: sections entitled "Distribution" and "General Public License" are
63: included exactly as in the original, and provided that the entire
64: resulting derived work is distributed under the terms of a permission
65: notice identical to this one.
66:
67: Permission is granted to copy and distribute translations of this manual
68: into another language, under the above conditions for modified versions,
69: except that the sections entitled "Distribution" and "General Public
70: License" may be included in a translation approved by the author instead
71: of in the original English.
72: @end titlepage
73:
74:
75: @node Top, License, (dir), (dir)
76: @ifinfo
1.17 anton 77: Gforth is a free implementation of ANS Forth available on many
1.1 anton 78: personal machines. This manual corresponds to version 0.0.
79: @end ifinfo
80:
81: @menu
1.4 anton 82: * License::
1.17 anton 83: * Goals:: About the Gforth Project
1.4 anton 84: * Other Books:: Things you might want to read
1.17 anton 85: * Invocation:: Starting Gforth
86: * Words:: Forth words available in Gforth
1.4 anton 87: * ANS conformance:: Implementation-defined options etc.
1.17 anton 88: * Model:: The abstract machine of Gforth
89: * Emacs and Gforth:: The Gforth Mode
1.4 anton 90: * Internals:: Implementation details
91: * Bugs:: How to report them
1.17 anton 92: * Pedigree:: Ancestors of Gforth
1.4 anton 93: * Word Index:: An item for each Forth word
94: * Node Index:: An item for each node
1.1 anton 95: @end menu
96:
97: @node License, Goals, Top, Top
1.20 ! pazsan 98: @unnumbered GNU GENERAL PUBLIC LICENSE
! 99: @center Version 2, June 1991
! 100:
! 101: @display
! 102: Copyright @copyright{} 1989, 1991 Free Software Foundation, Inc.
! 103: 675 Mass Ave, Cambridge, MA 02139, USA
! 104:
! 105: Everyone is permitted to copy and distribute verbatim copies
! 106: of this license document, but changing it is not allowed.
! 107: @end display
! 108:
! 109: @unnumberedsec Preamble
! 110:
! 111: The licenses for most software are designed to take away your
! 112: freedom to share and change it. By contrast, the GNU General Public
! 113: License is intended to guarantee your freedom to share and change free
! 114: software---to make sure the software is free for all its users. This
! 115: General Public License applies to most of the Free Software
! 116: Foundation's software and to any other program whose authors commit to
! 117: using it. (Some other Free Software Foundation software is covered by
! 118: the GNU Library General Public License instead.) You can apply it to
! 119: your programs, too.
! 120:
! 121: When we speak of free software, we are referring to freedom, not
! 122: price. Our General Public Licenses are designed to make sure that you
! 123: have the freedom to distribute copies of free software (and charge for
! 124: this service if you wish), that you receive source code or can get it
! 125: if you want it, that you can change the software or use pieces of it
! 126: in new free programs; and that you know you can do these things.
! 127:
! 128: To protect your rights, we need to make restrictions that forbid
! 129: anyone to deny you these rights or to ask you to surrender the rights.
! 130: These restrictions translate to certain responsibilities for you if you
! 131: distribute copies of the software, or if you modify it.
! 132:
! 133: For example, if you distribute copies of such a program, whether
! 134: gratis or for a fee, you must give the recipients all the rights that
! 135: you have. You must make sure that they, too, receive or can get the
! 136: source code. And you must show them these terms so they know their
! 137: rights.
! 138:
! 139: We protect your rights with two steps: (1) copyright the software, and
! 140: (2) offer you this license which gives you legal permission to copy,
! 141: distribute and/or modify the software.
! 142:
! 143: Also, for each author's protection and ours, we want to make certain
! 144: that everyone understands that there is no warranty for this free
! 145: software. If the software is modified by someone else and passed on, we
! 146: want its recipients to know that what they have is not the original, so
! 147: that any problems introduced by others will not reflect on the original
! 148: authors' reputations.
! 149:
! 150: Finally, any free program is threatened constantly by software
! 151: patents. We wish to avoid the danger that redistributors of a free
! 152: program will individually obtain patent licenses, in effect making the
! 153: program proprietary. To prevent this, we have made it clear that any
! 154: patent must be licensed for everyone's free use or not licensed at all.
! 155:
! 156: The precise terms and conditions for copying, distribution and
! 157: modification follow.
! 158:
! 159: @iftex
! 160: @unnumberedsec TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
! 161: @end iftex
! 162: @ifinfo
! 163: @center TERMS AND CONDITIONS FOR COPYING, DISTRIBUTION AND MODIFICATION
! 164: @end ifinfo
! 165:
! 166: @enumerate 0
! 167: @item
! 168: This License applies to any program or other work which contains
! 169: a notice placed by the copyright holder saying it may be distributed
! 170: under the terms of this General Public License. The ``Program'', below,
! 171: refers to any such program or work, and a ``work based on the Program''
! 172: means either the Program or any derivative work under copyright law:
! 173: that is to say, a work containing the Program or a portion of it,
! 174: either verbatim or with modifications and/or translated into another
! 175: language. (Hereinafter, translation is included without limitation in
! 176: the term ``modification''.) Each licensee is addressed as ``you''.
! 177:
! 178: Activities other than copying, distribution and modification are not
! 179: covered by this License; they are outside its scope. The act of
! 180: running the Program is not restricted, and the output from the Program
! 181: is covered only if its contents constitute a work based on the
! 182: Program (independent of having been made by running the Program).
! 183: Whether that is true depends on what the Program does.
! 184:
! 185: @item
! 186: You may copy and distribute verbatim copies of the Program's
! 187: source code as you receive it, in any medium, provided that you
! 188: conspicuously and appropriately publish on each copy an appropriate
! 189: copyright notice and disclaimer of warranty; keep intact all the
! 190: notices that refer to this License and to the absence of any warranty;
! 191: and give any other recipients of the Program a copy of this License
! 192: along with the Program.
! 193:
! 194: You may charge a fee for the physical act of transferring a copy, and
! 195: you may at your option offer warranty protection in exchange for a fee.
! 196:
! 197: @item
! 198: You may modify your copy or copies of the Program or any portion
! 199: of it, thus forming a work based on the Program, and copy and
! 200: distribute such modifications or work under the terms of Section 1
! 201: above, provided that you also meet all of these conditions:
! 202:
! 203: @enumerate a
! 204: @item
! 205: You must cause the modified files to carry prominent notices
! 206: stating that you changed the files and the date of any change.
! 207:
! 208: @item
! 209: You must cause any work that you distribute or publish, that in
! 210: whole or in part contains or is derived from the Program or any
! 211: part thereof, to be licensed as a whole at no charge to all third
! 212: parties under the terms of this License.
! 213:
! 214: @item
! 215: If the modified program normally reads commands interactively
! 216: when run, you must cause it, when started running for such
! 217: interactive use in the most ordinary way, to print or display an
! 218: announcement including an appropriate copyright notice and a
! 219: notice that there is no warranty (or else, saying that you provide
! 220: a warranty) and that users may redistribute the program under
! 221: these conditions, and telling the user how to view a copy of this
! 222: License. (Exception: if the Program itself is interactive but
! 223: does not normally print such an announcement, your work based on
! 224: the Program is not required to print an announcement.)
! 225: @end enumerate
! 226:
! 227: These requirements apply to the modified work as a whole. If
! 228: identifiable sections of that work are not derived from the Program,
! 229: and can be reasonably considered independent and separate works in
! 230: themselves, then this License, and its terms, do not apply to those
! 231: sections when you distribute them as separate works. But when you
! 232: distribute the same sections as part of a whole which is a work based
! 233: on the Program, the distribution of the whole must be on the terms of
! 234: this License, whose permissions for other licensees extend to the
! 235: entire whole, and thus to each and every part regardless of who wrote it.
! 236:
! 237: Thus, it is not the intent of this section to claim rights or contest
! 238: your rights to work written entirely by you; rather, the intent is to
! 239: exercise the right to control the distribution of derivative or
! 240: collective works based on the Program.
! 241:
! 242: In addition, mere aggregation of another work not based on the Program
! 243: with the Program (or with a work based on the Program) on a volume of
! 244: a storage or distribution medium does not bring the other work under
! 245: the scope of this License.
! 246:
! 247: @item
! 248: You may copy and distribute the Program (or a work based on it,
! 249: under Section 2) in object code or executable form under the terms of
! 250: Sections 1 and 2 above provided that you also do one of the following:
! 251:
! 252: @enumerate a
! 253: @item
! 254: Accompany it with the complete corresponding machine-readable
! 255: source code, which must be distributed under the terms of Sections
! 256: 1 and 2 above on a medium customarily used for software interchange; or,
! 257:
! 258: @item
! 259: Accompany it with a written offer, valid for at least three
! 260: years, to give any third party, for a charge no more than your
! 261: cost of physically performing source distribution, a complete
! 262: machine-readable copy of the corresponding source code, to be
! 263: distributed under the terms of Sections 1 and 2 above on a medium
! 264: customarily used for software interchange; or,
! 265:
! 266: @item
! 267: Accompany it with the information you received as to the offer
! 268: to distribute corresponding source code. (This alternative is
! 269: allowed only for noncommercial distribution and only if you
! 270: received the program in object code or executable form with such
! 271: an offer, in accord with Subsection b above.)
! 272: @end enumerate
! 273:
! 274: The source code for a work means the preferred form of the work for
! 275: making modifications to it. For an executable work, complete source
! 276: code means all the source code for all modules it contains, plus any
! 277: associated interface definition files, plus the scripts used to
! 278: control compilation and installation of the executable. However, as a
! 279: special exception, the source code distributed need not include
! 280: anything that is normally distributed (in either source or binary
! 281: form) with the major components (compiler, kernel, and so on) of the
! 282: operating system on which the executable runs, unless that component
! 283: itself accompanies the executable.
! 284:
! 285: If distribution of executable or object code is made by offering
! 286: access to copy from a designated place, then offering equivalent
! 287: access to copy the source code from the same place counts as
! 288: distribution of the source code, even though third parties are not
! 289: compelled to copy the source along with the object code.
! 290:
! 291: @item
! 292: You may not copy, modify, sublicense, or distribute the Program
! 293: except as expressly provided under this License. Any attempt
! 294: otherwise to copy, modify, sublicense or distribute the Program is
! 295: void, and will automatically terminate your rights under this License.
! 296: However, parties who have received copies, or rights, from you under
! 297: this License will not have their licenses terminated so long as such
! 298: parties remain in full compliance.
! 299:
! 300: @item
! 301: You are not required to accept this License, since you have not
! 302: signed it. However, nothing else grants you permission to modify or
! 303: distribute the Program or its derivative works. These actions are
! 304: prohibited by law if you do not accept this License. Therefore, by
! 305: modifying or distributing the Program (or any work based on the
! 306: Program), you indicate your acceptance of this License to do so, and
! 307: all its terms and conditions for copying, distributing or modifying
! 308: the Program or works based on it.
! 309:
! 310: @item
! 311: Each time you redistribute the Program (or any work based on the
! 312: Program), the recipient automatically receives a license from the
! 313: original licensor to copy, distribute or modify the Program subject to
! 314: these terms and conditions. You may not impose any further
! 315: restrictions on the recipients' exercise of the rights granted herein.
! 316: You are not responsible for enforcing compliance by third parties to
! 317: this License.
! 318:
! 319: @item
! 320: If, as a consequence of a court judgment or allegation of patent
! 321: infringement or for any other reason (not limited to patent issues),
! 322: conditions are imposed on you (whether by court order, agreement or
! 323: otherwise) that contradict the conditions of this License, they do not
! 324: excuse you from the conditions of this License. If you cannot
! 325: distribute so as to satisfy simultaneously your obligations under this
! 326: License and any other pertinent obligations, then as a consequence you
! 327: may not distribute the Program at all. For example, if a patent
! 328: license would not permit royalty-free redistribution of the Program by
! 329: all those who receive copies directly or indirectly through you, then
! 330: the only way you could satisfy both it and this License would be to
! 331: refrain entirely from distribution of the Program.
! 332:
! 333: If any portion of this section is held invalid or unenforceable under
! 334: any particular circumstance, the balance of the section is intended to
! 335: apply and the section as a whole is intended to apply in other
! 336: circumstances.
! 337:
! 338: It is not the purpose of this section to induce you to infringe any
! 339: patents or other property right claims or to contest validity of any
! 340: such claims; this section has the sole purpose of protecting the
! 341: integrity of the free software distribution system, which is
! 342: implemented by public license practices. Many people have made
! 343: generous contributions to the wide range of software distributed
! 344: through that system in reliance on consistent application of that
! 345: system; it is up to the author/donor to decide if he or she is willing
! 346: to distribute software through any other system and a licensee cannot
! 347: impose that choice.
! 348:
! 349: This section is intended to make thoroughly clear what is believed to
! 350: be a consequence of the rest of this License.
! 351:
! 352: @item
! 353: If the distribution and/or use of the Program is restricted in
! 354: certain countries either by patents or by copyrighted interfaces, the
! 355: original copyright holder who places the Program under this License
! 356: may add an explicit geographical distribution limitation excluding
! 357: those countries, so that distribution is permitted only in or among
! 358: countries not thus excluded. In such case, this License incorporates
! 359: the limitation as if written in the body of this License.
! 360:
! 361: @item
! 362: The Free Software Foundation may publish revised and/or new versions
! 363: of the General Public License from time to time. Such new versions will
! 364: be similar in spirit to the present version, but may differ in detail to
! 365: address new problems or concerns.
! 366:
! 367: Each version is given a distinguishing version number. If the Program
! 368: specifies a version number of this License which applies to it and ``any
! 369: later version'', you have the option of following the terms and conditions
! 370: either of that version or of any later version published by the Free
! 371: Software Foundation. If the Program does not specify a version number of
! 372: this License, you may choose any version ever published by the Free Software
! 373: Foundation.
! 374:
! 375: @item
! 376: If you wish to incorporate parts of the Program into other free
! 377: programs whose distribution conditions are different, write to the author
! 378: to ask for permission. For software which is copyrighted by the Free
! 379: Software Foundation, write to the Free Software Foundation; we sometimes
! 380: make exceptions for this. Our decision will be guided by the two goals
! 381: of preserving the free status of all derivatives of our free software and
! 382: of promoting the sharing and reuse of software generally.
! 383:
! 384: @iftex
! 385: @heading NO WARRANTY
! 386: @end iftex
! 387: @ifinfo
! 388: @center NO WARRANTY
! 389: @end ifinfo
! 390:
! 391: @item
! 392: BECAUSE THE PROGRAM IS LICENSED FREE OF CHARGE, THERE IS NO WARRANTY
! 393: FOR THE PROGRAM, TO THE EXTENT PERMITTED BY APPLICABLE LAW. EXCEPT WHEN
! 394: OTHERWISE STATED IN WRITING THE COPYRIGHT HOLDERS AND/OR OTHER PARTIES
! 395: PROVIDE THE PROGRAM ``AS IS'' WITHOUT WARRANTY OF ANY KIND, EITHER EXPRESSED
! 396: OR IMPLIED, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF
! 397: MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE. THE ENTIRE RISK AS
! 398: TO THE QUALITY AND PERFORMANCE OF THE PROGRAM IS WITH YOU. SHOULD THE
! 399: PROGRAM PROVE DEFECTIVE, YOU ASSUME THE COST OF ALL NECESSARY SERVICING,
! 400: REPAIR OR CORRECTION.
! 401:
! 402: @item
! 403: IN NO EVENT UNLESS REQUIRED BY APPLICABLE LAW OR AGREED TO IN WRITING
! 404: WILL ANY COPYRIGHT HOLDER, OR ANY OTHER PARTY WHO MAY MODIFY AND/OR
! 405: REDISTRIBUTE THE PROGRAM AS PERMITTED ABOVE, BE LIABLE TO YOU FOR DAMAGES,
! 406: INCLUDING ANY GENERAL, SPECIAL, INCIDENTAL OR CONSEQUENTIAL DAMAGES ARISING
! 407: OUT OF THE USE OR INABILITY TO USE THE PROGRAM (INCLUDING BUT NOT LIMITED
! 408: TO LOSS OF DATA OR DATA BEING RENDERED INACCURATE OR LOSSES SUSTAINED BY
! 409: YOU OR THIRD PARTIES OR A FAILURE OF THE PROGRAM TO OPERATE WITH ANY OTHER
! 410: PROGRAMS), EVEN IF SUCH HOLDER OR OTHER PARTY HAS BEEN ADVISED OF THE
! 411: POSSIBILITY OF SUCH DAMAGES.
! 412: @end enumerate
! 413:
! 414: @iftex
! 415: @heading END OF TERMS AND CONDITIONS
! 416: @end iftex
! 417: @ifinfo
! 418: @center END OF TERMS AND CONDITIONS
! 419: @end ifinfo
! 420:
! 421: @page
! 422: @unnumberedsec How to Apply These Terms to Your New Programs
! 423:
! 424: If you develop a new program, and you want it to be of the greatest
! 425: possible use to the public, the best way to achieve this is to make it
! 426: free software which everyone can redistribute and change under these terms.
! 427:
! 428: To do so, attach the following notices to the program. It is safest
! 429: to attach them to the start of each source file to most effectively
! 430: convey the exclusion of warranty; and each file should have at least
! 431: the ``copyright'' line and a pointer to where the full notice is found.
! 432:
! 433: @smallexample
! 434: @var{one line to give the program's name and a brief idea of what it does.}
! 435: Copyright (C) 19@var{yy} @var{name of author}
! 436:
! 437: This program is free software; you can redistribute it and/or modify
! 438: it under the terms of the GNU General Public License as published by
! 439: the Free Software Foundation; either version 2 of the License, or
! 440: (at your option) any later version.
! 441:
! 442: This program is distributed in the hope that it will be useful,
! 443: but WITHOUT ANY WARRANTY; without even the implied warranty of
! 444: MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
! 445: GNU General Public License for more details.
! 446:
! 447: You should have received a copy of the GNU General Public License
! 448: along with this program; if not, write to the Free Software
! 449: Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA.
! 450: @end smallexample
! 451:
! 452: Also add information on how to contact you by electronic and paper mail.
! 453:
! 454: If the program is interactive, make it output a short notice like this
! 455: when it starts in an interactive mode:
! 456:
! 457: @smallexample
! 458: Gnomovision version 69, Copyright (C) 19@var{yy} @var{name of author}
! 459: Gnomovision comes with ABSOLUTELY NO WARRANTY; for details
! 460: type `show w'.
! 461: This is free software, and you are welcome to redistribute it
! 462: under certain conditions; type `show c' for details.
! 463: @end smallexample
! 464:
! 465: The hypothetical commands @samp{show w} and @samp{show c} should show
! 466: the appropriate parts of the General Public License. Of course, the
! 467: commands you use may be called something other than @samp{show w} and
! 468: @samp{show c}; they could even be mouse-clicks or menu items---whatever
! 469: suits your program.
! 470:
! 471: You should also get your employer (if you work as a programmer) or your
! 472: school, if any, to sign a ``copyright disclaimer'' for the program, if
! 473: necessary. Here is a sample; alter the names:
! 474:
! 475: @smallexample
! 476: Yoyodyne, Inc., hereby disclaims all copyright interest in the program
! 477: `Gnomovision' (which makes passes at compilers) written by James Hacker.
! 478:
! 479: @var{signature of Ty Coon}, 1 April 1989
! 480: Ty Coon, President of Vice
! 481: @end smallexample
! 482:
! 483: This General Public License does not permit incorporating your program into
! 484: proprietary programs. If your program is a subroutine library, you may
! 485: consider it more useful to permit linking proprietary applications with the
! 486: library. If this is what you want to do, use the GNU Library General
! 487: Public License instead of this License.
1.1 anton 488:
489: @iftex
490: @unnumbered Preface
1.17 anton 491: This manual documents Gforth. The reader is expected to know
1.1 anton 492: Forth. This manual is primarily a reference manual. @xref{Other Books}
493: for introductory material.
494: @end iftex
495:
496: @node Goals, Other Books, License, Top
497: @comment node-name, next, previous, up
1.17 anton 498: @chapter Goals of Gforth
1.1 anton 499: @cindex Goals
1.17 anton 500: The goal of the Gforth Project is to develop a standard model for
1.1 anton 501: ANSI Forth. This can be split into several subgoals:
502:
503: @itemize @bullet
504: @item
1.17 anton 505: Gforth should conform to the ANSI Forth standard.
1.1 anton 506: @item
507: It should be a model, i.e. it should define all the
508: implementation-dependent things.
509: @item
510: It should become standard, i.e. widely accepted and used. This goal
511: is the most difficult one.
512: @end itemize
513:
1.17 anton 514: To achieve these goals Gforth should be
1.1 anton 515: @itemize @bullet
516: @item
517: Similar to previous models (fig-Forth, F83)
518: @item
519: Powerful. It should provide for all the things that are considered
520: necessary today and even some that are not yet considered necessary.
521: @item
522: Efficient. It should not get the reputation of being exceptionally
523: slow.
524: @item
525: Free.
526: @item
527: Available on many machines/easy to port.
528: @end itemize
529:
1.17 anton 530: Have we achieved these goals? Gforth conforms to the ANS Forth
531: standard. It may be considered a model, but we have not yet documented
1.1 anton 532: which parts of the model are stable and which parts we are likely to
1.17 anton 533: change. It certainly has not yet become a de facto standard. It has some
534: similarities and some differences to previous models. It has some
535: powerful features, but not yet everything that we envisioned. We
536: certainly have achieved our execution speed goals (@pxref{Performance}).
537: It is free and available on many machines.
1.1 anton 538:
539: @node Other Books, Invocation, Goals, Top
540: @chapter Other books on ANS Forth
541:
542: As the standard is relatively new, there are not many books out yet. It
1.17 anton 543: is not recommended to learn Forth by using Gforth and a book that is
1.1 anton 544: not written for ANS Forth, as you will not know your mistakes from the
545: deviations of the book.
546:
547: There is, of course, the standard, the definite reference if you want to
1.19 anton 548: write ANS Forth programs. It is available in printed form from the
549: National Standards Institute Sales Department (Tel.: USA (212) 642-4900;
550: Fax.: USA (212) 302-1286) as document @cite{X3.215-1994} for about $200. You
551: can also get it from Global Engineering Documents (Tel.: USA (800)
552: 854-7179; Fax.: (303) 843-9880) for about $300.
553:
554: @cite{dpANS6}, the last draft of the standard, which was then submitted to ANSI
555: for publication is available electronically and for free in some MS Word
556: format, and it has been converted to HTML. Some pointers to these
557: versions can be found through
558: http://www.complang.tuwien.ac.at/projects/forth.html.
1.1 anton 559:
560: @cite{Forth: The new model} by Jack Woehr (!! Publisher) is an
561: introductory book based on a draft version of the standard. It does not
562: cover the whole standard. It also contains interesting background
563: information (Jack Woehr was in the ANS Forth Technical Committe). It is
564: not appropriate for complete newbies, but programmers experienced in
565: other languages should find it ok.
566:
567: @node Invocation, Words, Other Books, Top
568: @chapter Invocation
569:
570: You will usually just say @code{gforth}. In many other cases the default
1.17 anton 571: Gforth image will be invoked like this:
1.1 anton 572:
573: @example
574: gforth [files] [-e forth-code]
575: @end example
576:
577: executing the contents of the files and the Forth code in the order they
578: are given.
579:
580: In general, the command line looks like this:
581:
582: @example
583: gforth [initialization options] [image-specific options]
584: @end example
585:
586: The initialization options must come before the rest of the command
587: line. They are:
588:
589: @table @code
590: @item --image-file @var{file}
1.20 ! pazsan 591: @item -i @var{file}
1.1 anton 592: Loads the Forth image @var{file} instead of the default
593: @file{gforth.fi}.
594:
595: @item --path @var{path}
1.20 ! pazsan 596: @item -p @var{path}
1.1 anton 597: Uses @var{path} for searching the image file and Forth source code
598: files instead of the default in the environment variable
599: @code{GFORTHPATH} or the path specified at installation time (typically
600: @file{/usr/local/lib/gforth:.}). A path is given as a @code{:}-separated
601: list.
602:
603: @item --dictionary-size @var{size}
604: @item -m @var{size}
605: Allocate @var{size} space for the Forth dictionary space instead of
606: using the default specified in the image (typically 256K). The
607: @var{size} specification consists of an integer and a unit (e.g.,
608: @code{4M}). The unit can be one of @code{b} (bytes), @code{e} (element
609: size, in this case Cells), @code{k} (kilobytes), and @code{M}
610: (Megabytes). If no unit is specified, @code{e} is used.
611:
612: @item --data-stack-size @var{size}
613: @item -d @var{size}
614: Allocate @var{size} space for the data stack instead of using the
615: default specified in the image (typically 16K).
616:
617: @item --return-stack-size @var{size}
618: @item -r @var{size}
619: Allocate @var{size} space for the return stack instead of using the
620: default specified in the image (typically 16K).
621:
622: @item --fp-stack-size @var{size}
623: @item -f @var{size}
624: Allocate @var{size} space for the floating point stack instead of
625: using the default specified in the image (typically 16K). In this case
626: the unit specifier @code{e} refers to floating point numbers.
627:
628: @item --locals-stack-size @var{size}
629: @item -l @var{size}
630: Allocate @var{size} space for the locals stack instead of using the
631: default specified in the image (typically 16K).
632:
633: @end table
634:
635: As explained above, the image-specific command-line arguments for the
636: default image @file{gforth.fi} consist of a sequence of filenames and
637: @code{-e @var{forth-code}} options that are interpreted in the seqence
638: in which they are given. The @code{-e @var{forth-code}} or
639: @code{--evaluate @var{forth-code}} option evaluates the forth
640: code. This option takes only one argument; if you want to evaluate more
641: Forth words, you have to quote them or use several @code{-e}s. To exit
642: after processing the command line (instead of entering interactive mode)
643: append @code{-e bye} to the command line.
644:
645: Not yet implemented:
646: On startup the system first executes the system initialization file
647: (unless the option @code{--no-init-file} is given; note that the system
648: resulting from using this option may not be ANS Forth conformant). Then
649: the user initialization file @file{.gforth.fs} is executed, unless the
650: option @code{--no-rc} is given; this file is first searched in @file{.},
651: then in @file{~}, then in the normal path (see above).
652:
1.4 anton 653: @node Words, ANS conformance, Invocation, Top
1.1 anton 654: @chapter Forth Words
655:
656: @menu
1.4 anton 657: * Notation::
658: * Arithmetic::
659: * Stack Manipulation::
660: * Memory access::
661: * Control Structures::
662: * Locals::
663: * Defining Words::
664: * Wordlists::
665: * Files::
666: * Blocks::
667: * Other I/O::
668: * Programming Tools::
1.18 anton 669: * Assembler and Code words::
1.4 anton 670: * Threading Words::
1.1 anton 671: @end menu
672:
673: @node Notation, Arithmetic, Words, Words
674: @section Notation
675:
676: The Forth words are described in this section in the glossary notation
677: that has become a de-facto standard for Forth texts, i.e.
678:
1.4 anton 679: @format
1.1 anton 680: @var{word} @var{Stack effect} @var{wordset} @var{pronunciation}
1.4 anton 681: @end format
1.1 anton 682: @var{Description}
683:
684: @table @var
685: @item word
1.17 anton 686: The name of the word. BTW, Gforth is case insensitive, so you can
1.14 anton 687: type the words in in lower case (However, @pxref{core-idef}).
1.1 anton 688:
689: @item Stack effect
690: The stack effect is written in the notation @code{@var{before} --
691: @var{after}}, where @var{before} and @var{after} describe the top of
692: stack entries before and after the execution of the word. The rest of
693: the stack is not touched by the word. The top of stack is rightmost,
1.17 anton 694: i.e., a stack sequence is written as it is typed in. Note that Gforth
1.1 anton 695: uses a separate floating point stack, but a unified stack
696: notation. Also, return stack effects are not shown in @var{stack
697: effect}, but in @var{Description}. The name of a stack item describes
698: the type and/or the function of the item. See below for a discussion of
699: the types.
700:
1.19 anton 701: All words have two stack effects: A compile-time stack effect and a
702: run-time stack effect. The compile-time stack-effect of most words is
703: @var{ -- }. If the compile-time stack-effect of a word deviates from
704: this standard behaviour, or the word does other unusual things at
705: compile time, both stack effects are shown; otherwise only the run-time
706: stack effect is shown.
707:
1.1 anton 708: @item pronunciation
709: How the word is pronounced
710:
711: @item wordset
712: The ANS Forth standard is divided into several wordsets. A standard
713: system need not support all of them. So, the fewer wordsets your program
714: uses the more portable it will be in theory. However, we suspect that
715: most ANS Forth systems on personal machines will feature all
716: wordsets. Words that are not defined in the ANS standard have
1.19 anton 717: @code{gforth} or @code{gforth-internal} as wordset. @code{gforth}
718: describes words that will work in future releases of Gforth;
719: @code{gforth-internal} words are more volatile. Environmental query
720: strings are also displayed like words; you can recognize them by the
721: @code{environment} in the wordset field.
1.1 anton 722:
723: @item Description
724: A description of the behaviour of the word.
725: @end table
726:
1.4 anton 727: The type of a stack item is specified by the character(s) the name
728: starts with:
1.1 anton 729:
730: @table @code
731: @item f
732: Bool, i.e. @code{false} or @code{true}.
733: @item c
734: Char
735: @item w
736: Cell, can contain an integer or an address
737: @item n
738: signed integer
739: @item u
740: unsigned integer
741: @item d
742: double sized signed integer
743: @item ud
744: double sized unsigned integer
745: @item r
746: Float
747: @item a_
748: Cell-aligned address
749: @item c_
750: Char-aligned address (note that a Char is two bytes in Windows NT)
751: @item f_
752: Float-aligned address
753: @item df_
754: Address aligned for IEEE double precision float
755: @item sf_
756: Address aligned for IEEE single precision float
757: @item xt
758: Execution token, same size as Cell
759: @item wid
760: Wordlist ID, same size as Cell
761: @item f83name
762: Pointer to a name structure
763: @end table
764:
1.4 anton 765: @node Arithmetic, Stack Manipulation, Notation, Words
1.1 anton 766: @section Arithmetic
767: Forth arithmetic is not checked, i.e., you will not hear about integer
768: overflow on addition or multiplication, you may hear about division by
769: zero if you are lucky. The operator is written after the operands, but
770: the operands are still in the original order. I.e., the infix @code{2-1}
771: corresponds to @code{2 1 -}. Forth offers a variety of division
772: operators. If you perform division with potentially negative operands,
773: you do not want to use @code{/} or @code{/mod} with its undefined
774: behaviour, but rather @code{fm/mod} or @code{sm/mod} (probably the
1.4 anton 775: former, @pxref{Mixed precision}).
776:
777: @menu
778: * Single precision::
779: * Bitwise operations::
780: * Mixed precision:: operations with single and double-cell integers
781: * Double precision:: Double-cell integer arithmetic
782: * Floating Point::
783: @end menu
1.1 anton 784:
1.4 anton 785: @node Single precision, Bitwise operations, Arithmetic, Arithmetic
1.1 anton 786: @subsection Single precision
787: doc-+
788: doc--
789: doc-*
790: doc-/
791: doc-mod
792: doc-/mod
793: doc-negate
794: doc-abs
795: doc-min
796: doc-max
797:
1.4 anton 798: @node Bitwise operations, Mixed precision, Single precision, Arithmetic
1.1 anton 799: @subsection Bitwise operations
800: doc-and
801: doc-or
802: doc-xor
803: doc-invert
804: doc-2*
805: doc-2/
806:
1.4 anton 807: @node Mixed precision, Double precision, Bitwise operations, Arithmetic
1.1 anton 808: @subsection Mixed precision
809: doc-m+
810: doc-*/
811: doc-*/mod
812: doc-m*
813: doc-um*
814: doc-m*/
815: doc-um/mod
816: doc-fm/mod
817: doc-sm/rem
818:
1.4 anton 819: @node Double precision, Floating Point, Mixed precision, Arithmetic
1.1 anton 820: @subsection Double precision
1.16 anton 821:
822: The outer (aka text) interpreter converts numbers containing a dot into
823: a double precision number. Note that only numbers with the dot as last
824: character are standard-conforming.
825:
1.1 anton 826: doc-d+
827: doc-d-
828: doc-dnegate
829: doc-dabs
830: doc-dmin
831: doc-dmax
832:
1.4 anton 833: @node Floating Point, , Double precision, Arithmetic
834: @subsection Floating Point
1.16 anton 835:
836: The format of floating point numbers recognized by the outer (aka text)
837: interpreter is: a signed decimal number, possibly containing a decimal
838: point (@code{.}), followed by @code{E} or @code{e}, optionally followed
839: by a signed integer (the exponent). E.g., @code{1e} ist the same as
840: @code{+1.0e+1}. Note that a number without @code{e}
841: is not interpreted as floating-point number, but as double (if the
842: number contains a @code{.}) or single precision integer. Also,
843: conversions between string and floating point numbers always use base
844: 10, irrespective of the value of @code{BASE}. If @code{BASE} contains a
845: value greater then 14, the @code{E} may be interpreted as digit and the
846: number will be interpreted as integer, unless it has a signed exponent
847: (both @code{+} and @code{-} are allowed as signs).
1.4 anton 848:
849: Angles in floating point operations are given in radians (a full circle
1.17 anton 850: has 2 pi radians). Note, that Gforth has a separate floating point
1.4 anton 851: stack, but we use the unified notation.
852:
853: Floating point numbers have a number of unpleasant surprises for the
854: unwary (e.g., floating point addition is not associative) and even a few
855: for the wary. You should not use them unless you know what you are doing
856: or you don't care that the results you get are totally bogus. If you
857: want to learn about the problems of floating point numbers (and how to
1.11 anton 858: avoid them), you might start with @cite{David Goldberg, What Every
1.6 anton 859: Computer Scientist Should Know About Floating-Point Arithmetic, ACM
860: Computing Surveys 23(1):5@minus{}48, March 1991}.
1.4 anton 861:
862: doc-f+
863: doc-f-
864: doc-f*
865: doc-f/
866: doc-fnegate
867: doc-fabs
868: doc-fmax
869: doc-fmin
870: doc-floor
871: doc-fround
872: doc-f**
873: doc-fsqrt
874: doc-fexp
875: doc-fexpm1
876: doc-fln
877: doc-flnp1
878: doc-flog
1.6 anton 879: doc-falog
1.4 anton 880: doc-fsin
881: doc-fcos
882: doc-fsincos
883: doc-ftan
884: doc-fasin
885: doc-facos
886: doc-fatan
887: doc-fatan2
888: doc-fsinh
889: doc-fcosh
890: doc-ftanh
891: doc-fasinh
892: doc-facosh
893: doc-fatanh
894:
895: @node Stack Manipulation, Memory access, Arithmetic, Words
1.1 anton 896: @section Stack Manipulation
897:
1.17 anton 898: Gforth has a data stack (aka parameter stack) for characters, cells,
1.1 anton 899: addresses, and double cells, a floating point stack for floating point
900: numbers, a return stack for storing the return addresses of colon
901: definitions and other data, and a locals stack for storing local
902: variables. Note that while every sane Forth has a separate floating
903: point stack, this is not strictly required; an ANS Forth system could
904: theoretically keep floating point numbers on the data stack. As an
905: additional difficulty, you don't know how many cells a floating point
906: number takes. It is reportedly possible to write words in a way that
907: they work also for a unified stack model, but we do not recommend trying
1.4 anton 908: it. Instead, just say that your program has an environmental dependency
909: on a separate FP stack.
910:
911: Also, a Forth system is allowed to keep the local variables on the
1.1 anton 912: return stack. This is reasonable, as local variables usually eliminate
913: the need to use the return stack explicitly. So, if you want to produce
914: a standard complying program and if you are using local variables in a
915: word, forget about return stack manipulations in that word (see the
916: standard document for the exact rules).
917:
1.4 anton 918: @menu
919: * Data stack::
920: * Floating point stack::
921: * Return stack::
922: * Locals stack::
923: * Stack pointer manipulation::
924: @end menu
925:
926: @node Data stack, Floating point stack, Stack Manipulation, Stack Manipulation
1.1 anton 927: @subsection Data stack
928: doc-drop
929: doc-nip
930: doc-dup
931: doc-over
932: doc-tuck
933: doc-swap
934: doc-rot
935: doc--rot
936: doc-?dup
937: doc-pick
938: doc-roll
939: doc-2drop
940: doc-2nip
941: doc-2dup
942: doc-2over
943: doc-2tuck
944: doc-2swap
945: doc-2rot
946:
1.4 anton 947: @node Floating point stack, Return stack, Data stack, Stack Manipulation
1.1 anton 948: @subsection Floating point stack
949: doc-fdrop
950: doc-fnip
951: doc-fdup
952: doc-fover
953: doc-ftuck
954: doc-fswap
955: doc-frot
956:
1.4 anton 957: @node Return stack, Locals stack, Floating point stack, Stack Manipulation
1.1 anton 958: @subsection Return stack
959: doc->r
960: doc-r>
961: doc-r@
962: doc-rdrop
963: doc-2>r
964: doc-2r>
965: doc-2r@
966: doc-2rdrop
967:
1.4 anton 968: @node Locals stack, Stack pointer manipulation, Return stack, Stack Manipulation
1.1 anton 969: @subsection Locals stack
970:
1.4 anton 971: @node Stack pointer manipulation, , Locals stack, Stack Manipulation
1.1 anton 972: @subsection Stack pointer manipulation
973: doc-sp@
974: doc-sp!
975: doc-fp@
976: doc-fp!
977: doc-rp@
978: doc-rp!
979: doc-lp@
980: doc-lp!
981:
1.4 anton 982: @node Memory access, Control Structures, Stack Manipulation, Words
1.1 anton 983: @section Memory access
984:
1.4 anton 985: @menu
986: * Stack-Memory transfers::
987: * Address arithmetic::
988: * Memory block access::
989: @end menu
990:
991: @node Stack-Memory transfers, Address arithmetic, Memory access, Memory access
1.1 anton 992: @subsection Stack-Memory transfers
993:
994: doc-@
995: doc-!
996: doc-+!
997: doc-c@
998: doc-c!
999: doc-2@
1000: doc-2!
1001: doc-f@
1002: doc-f!
1003: doc-sf@
1004: doc-sf!
1005: doc-df@
1006: doc-df!
1007:
1.4 anton 1008: @node Address arithmetic, Memory block access, Stack-Memory transfers, Memory access
1.1 anton 1009: @subsection Address arithmetic
1010:
1011: ANS Forth does not specify the sizes of the data types. Instead, it
1012: offers a number of words for computing sizes and doing address
1013: arithmetic. Basically, address arithmetic is performed in terms of
1014: address units (aus); on most systems the address unit is one byte. Note
1015: that a character may have more than one au, so @code{chars} is no noop
1016: (on systems where it is a noop, it compiles to nothing).
1017:
1018: ANS Forth also defines words for aligning addresses for specific
1019: addresses. Many computers require that accesses to specific data types
1020: must only occur at specific addresses; e.g., that cells may only be
1021: accessed at addresses divisible by 4. Even if a machine allows unaligned
1022: accesses, it can usually perform aligned accesses faster.
1023:
1.17 anton 1024: For the performance-conscious: alignment operations are usually only
1.1 anton 1025: necessary during the definition of a data structure, not during the
1026: (more frequent) accesses to it.
1027:
1028: ANS Forth defines no words for character-aligning addresses. This is not
1029: an oversight, but reflects the fact that addresses that are not
1030: char-aligned have no use in the standard and therefore will not be
1031: created.
1032:
1033: The standard guarantees that addresses returned by @code{CREATE}d words
1.17 anton 1034: are cell-aligned; in addition, Gforth guarantees that these addresses
1.1 anton 1035: are aligned for all purposes.
1036:
1.9 anton 1037: Note that the standard defines a word @code{char}, which has nothing to
1038: do with address arithmetic.
1039:
1.1 anton 1040: doc-chars
1041: doc-char+
1042: doc-cells
1043: doc-cell+
1044: doc-align
1045: doc-aligned
1046: doc-floats
1047: doc-float+
1048: doc-falign
1049: doc-faligned
1050: doc-sfloats
1051: doc-sfloat+
1052: doc-sfalign
1053: doc-sfaligned
1054: doc-dfloats
1055: doc-dfloat+
1056: doc-dfalign
1057: doc-dfaligned
1.10 anton 1058: doc-maxalign
1059: doc-maxaligned
1060: doc-cfalign
1061: doc-cfaligned
1.1 anton 1062: doc-address-unit-bits
1063:
1.4 anton 1064: @node Memory block access, , Address arithmetic, Memory access
1.1 anton 1065: @subsection Memory block access
1066:
1067: doc-move
1068: doc-erase
1069:
1070: While the previous words work on address units, the rest works on
1071: characters.
1072:
1073: doc-cmove
1074: doc-cmove>
1075: doc-fill
1076: doc-blank
1077:
1.4 anton 1078: @node Control Structures, Locals, Memory access, Words
1.1 anton 1079: @section Control Structures
1080:
1081: Control structures in Forth cannot be used in interpret state, only in
1082: compile state, i.e., in a colon definition. We do not like this
1083: limitation, but have not seen a satisfying way around it yet, although
1084: many schemes have been proposed.
1085:
1.4 anton 1086: @menu
1087: * Selection::
1088: * Simple Loops::
1089: * Counted Loops::
1090: * Arbitrary control structures::
1091: * Calls and returns::
1092: * Exception Handling::
1093: @end menu
1094:
1095: @node Selection, Simple Loops, Control Structures, Control Structures
1.1 anton 1096: @subsection Selection
1097:
1098: @example
1099: @var{flag}
1100: IF
1101: @var{code}
1102: ENDIF
1103: @end example
1104: or
1105: @example
1106: @var{flag}
1107: IF
1108: @var{code1}
1109: ELSE
1110: @var{code2}
1111: ENDIF
1112: @end example
1113:
1.4 anton 1114: You can use @code{THEN} instead of @code{ENDIF}. Indeed, @code{THEN} is
1.1 anton 1115: standard, and @code{ENDIF} is not, although it is quite popular. We
1116: recommend using @code{ENDIF}, because it is less confusing for people
1117: who also know other languages (and is not prone to reinforcing negative
1118: prejudices against Forth in these people). Adding @code{ENDIF} to a
1119: system that only supplies @code{THEN} is simple:
1120: @example
1121: : endif POSTPONE then ; immediate
1122: @end example
1123:
1124: [According to @cite{Webster's New Encyclopedic Dictionary}, @dfn{then
1125: (adv.)} has the following meanings:
1126: @quotation
1127: ... 2b: following next after in order ... 3d: as a necessary consequence
1128: (if you were there, then you saw them).
1129: @end quotation
1130: Forth's @code{THEN} has the meaning 2b, whereas @code{THEN} in Pascal
1131: and many other programming languages has the meaning 3d.]
1132:
1133: We also provide the words @code{?dup-if} and @code{?dup-0=-if}, so you
1134: can avoid using @code{?dup}.
1135:
1136: @example
1137: @var{n}
1138: CASE
1139: @var{n1} OF @var{code1} ENDOF
1140: @var{n2} OF @var{code2} ENDOF
1.4 anton 1141: @dots{}
1.1 anton 1142: ENDCASE
1143: @end example
1144:
1145: Executes the first @var{codei}, where the @var{ni} is equal to
1146: @var{n}. A default case can be added by simply writing the code after
1147: the last @code{ENDOF}. It may use @var{n}, which is on top of the stack,
1148: but must not consume it.
1149:
1.4 anton 1150: @node Simple Loops, Counted Loops, Selection, Control Structures
1.1 anton 1151: @subsection Simple Loops
1152:
1153: @example
1154: BEGIN
1155: @var{code1}
1156: @var{flag}
1157: WHILE
1158: @var{code2}
1159: REPEAT
1160: @end example
1161:
1162: @var{code1} is executed and @var{flag} is computed. If it is true,
1163: @var{code2} is executed and the loop is restarted; If @var{flag} is false, execution continues after the @code{REPEAT}.
1164:
1165: @example
1166: BEGIN
1167: @var{code}
1168: @var{flag}
1169: UNTIL
1170: @end example
1171:
1172: @var{code} is executed. The loop is restarted if @code{flag} is false.
1173:
1174: @example
1175: BEGIN
1176: @var{code}
1177: AGAIN
1178: @end example
1179:
1180: This is an endless loop.
1181:
1.4 anton 1182: @node Counted Loops, Arbitrary control structures, Simple Loops, Control Structures
1.1 anton 1183: @subsection Counted Loops
1184:
1185: The basic counted loop is:
1186: @example
1187: @var{limit} @var{start}
1188: ?DO
1189: @var{body}
1190: LOOP
1191: @end example
1192:
1193: This performs one iteration for every integer, starting from @var{start}
1194: and up to, but excluding @var{limit}. The counter, aka index, can be
1195: accessed with @code{i}. E.g., the loop
1196: @example
1197: 10 0 ?DO
1198: i .
1199: LOOP
1200: @end example
1201: prints
1202: @example
1203: 0 1 2 3 4 5 6 7 8 9
1204: @end example
1205: The index of the innermost loop can be accessed with @code{i}, the index
1206: of the next loop with @code{j}, and the index of the third loop with
1207: @code{k}.
1208:
1209: The loop control data are kept on the return stack, so there are some
1210: restrictions on mixing return stack accesses and counted loop
1211: words. E.g., if you put values on the return stack outside the loop, you
1212: cannot read them inside the loop. If you put values on the return stack
1213: within a loop, you have to remove them before the end of the loop and
1214: before accessing the index of the loop.
1215:
1216: There are several variations on the counted loop:
1217:
1218: @code{LEAVE} leaves the innermost counted loop immediately.
1219:
1.18 anton 1220: If @var{start} is greater than @var{limit}, a @code{?DO} loop is entered
1221: (and @code{LOOP} iterates until they become equal by wrap-around
1222: arithmetic). This behaviour is usually not what you want. Therefore,
1223: Gforth offers @code{+DO} and @code{U+DO} (as replacements for
1224: @code{?DO}), which do not enter the loop if @var{start} is greater than
1225: @var{limit}; @code{+DO} is for signed loop parameters, @code{U+DO} for
1226: unsigned loop parameters. These words can be implemented easily on
1227: standard systems, so using them does not make your programs hard to
1228: port; e.g.:
1229: @example
1230: : +DO ( compile-time: -- do-sys; run-time: n1 n2 -- )
1231: POSTPONE over POSTPONE min POSTPONE ?DO ; immediate
1232: @end example
1233:
1.1 anton 1234: @code{LOOP} can be replaced with @code{@var{n} +LOOP}; this updates the
1235: index by @var{n} instead of by 1. The loop is terminated when the border
1236: between @var{limit-1} and @var{limit} is crossed. E.g.:
1237:
1.18 anton 1238: @code{4 0 +DO i . 2 +LOOP} prints @code{0 2}
1.1 anton 1239:
1.18 anton 1240: @code{4 1 +DO i . 2 +LOOP} prints @code{1 3}
1.1 anton 1241:
1242: The behaviour of @code{@var{n} +LOOP} is peculiar when @var{n} is negative:
1243:
1.2 anton 1244: @code{-1 0 ?DO i . -1 +LOOP} prints @code{0 -1}
1.1 anton 1245:
1.2 anton 1246: @code{ 0 0 ?DO i . -1 +LOOP} prints nothing
1.1 anton 1247:
1.18 anton 1248: Therefore we recommend avoiding @code{@var{n} +LOOP} with negative
1249: @var{n}. One alternative is @code{@var{u} -LOOP}, which reduces the
1250: index by @var{u} each iteration. The loop is terminated when the border
1251: between @var{limit+1} and @var{limit} is crossed. Gforth also provides
1252: @code{-DO} and @code{U-DO} for down-counting loops. E.g.:
1.1 anton 1253:
1.18 anton 1254: @code{-2 0 -DO i . 1 -LOOP} prints @code{0 -1}
1.1 anton 1255:
1.18 anton 1256: @code{-1 0 -DO i . 1 -LOOP} prints @code{0}
1.1 anton 1257:
1.18 anton 1258: @code{ 0 0 -DO i . 1 -LOOP} prints nothing
1.1 anton 1259:
1.18 anton 1260: Another alternative is @code{@var{n} S+LOOP}, where the negative
1261: case behaves symmetrical to the positive case:
1.1 anton 1262:
1.18 anton 1263: @code{-2 0 -DO i . -1 S+LOOP} prints @code{0 -1}
1264:
1265: The loop is terminated when the border between @var{limit@minus{}sgn(n)}
1266: and @var{limit} is crossed. Unfortunately, neither @code{-LOOP} nor
1267: @code{S+LOOP} are part of the ANS Forth standard, and they are not easy
1268: to implement using standard words. If you want to write standard
1269: programs, just avoid counting down.
1270:
1271: @code{?DO} can also be replaced by @code{DO}. @code{DO} always enters
1272: the loop, independent of the loop parameters. Do not use @code{DO}, even
1273: if you know that the loop is entered in any case. Such knowledge tends
1274: to become invalid during maintenance of a program, and then the
1275: @code{DO} will make trouble.
1.1 anton 1276:
1277: @code{UNLOOP} is used to prepare for an abnormal loop exit, e.g., via
1278: @code{EXIT}. @code{UNLOOP} removes the loop control parameters from the
1279: return stack so @code{EXIT} can get to its return address.
1280:
1281: Another counted loop is
1282: @example
1283: @var{n}
1284: FOR
1285: @var{body}
1286: NEXT
1287: @end example
1288: This is the preferred loop of native code compiler writers who are too
1.17 anton 1289: lazy to optimize @code{?DO} loops properly. In Gforth, this loop
1.1 anton 1290: iterates @var{n+1} times; @code{i} produces values starting with @var{n}
1291: and ending with 0. Other Forth systems may behave differently, even if
1292: they support @code{FOR} loops.
1293:
1.4 anton 1294: @node Arbitrary control structures, Calls and returns, Counted Loops, Control Structures
1.2 anton 1295: @subsection Arbitrary control structures
1296:
1297: ANS Forth permits and supports using control structures in a non-nested
1298: way. Information about incomplete control structures is stored on the
1299: control-flow stack. This stack may be implemented on the Forth data
1.17 anton 1300: stack, and this is what we have done in Gforth.
1.2 anton 1301:
1302: An @i{orig} entry represents an unresolved forward branch, a @i{dest}
1303: entry represents a backward branch target. A few words are the basis for
1304: building any control structure possible (except control structures that
1305: need storage, like calls, coroutines, and backtracking).
1306:
1.3 anton 1307: doc-if
1308: doc-ahead
1309: doc-then
1310: doc-begin
1311: doc-until
1312: doc-again
1313: doc-cs-pick
1314: doc-cs-roll
1.2 anton 1315:
1.17 anton 1316: On many systems control-flow stack items take one word, in Gforth they
1.2 anton 1317: currently take three (this may change in the future). Therefore it is a
1318: really good idea to manipulate the control flow stack with
1319: @code{cs-pick} and @code{cs-roll}, not with data stack manipulation
1320: words.
1321:
1322: Some standard control structure words are built from these words:
1323:
1.3 anton 1324: doc-else
1325: doc-while
1326: doc-repeat
1.2 anton 1327:
1328: Counted loop words constitute a separate group of words:
1329:
1.3 anton 1330: doc-?do
1.18 anton 1331: doc-+do
1332: doc-u+do
1333: doc--do
1334: doc-u-do
1.3 anton 1335: doc-do
1336: doc-for
1337: doc-loop
1338: doc-s+loop
1339: doc-+loop
1.18 anton 1340: doc--loop
1.3 anton 1341: doc-next
1342: doc-leave
1343: doc-?leave
1344: doc-unloop
1.10 anton 1345: doc-done
1.2 anton 1346:
1347: The standard does not allow using @code{cs-pick} and @code{cs-roll} on
1348: @i{do-sys}. Our system allows it, but it's your job to ensure that for
1349: every @code{?DO} etc. there is exactly one @code{UNLOOP} on any path
1.3 anton 1350: through the definition (@code{LOOP} etc. compile an @code{UNLOOP} on the
1351: fall-through path). Also, you have to ensure that all @code{LEAVE}s are
1.7 pazsan 1352: resolved (by using one of the loop-ending words or @code{DONE}).
1.2 anton 1353:
1354: Another group of control structure words are
1355:
1.3 anton 1356: doc-case
1357: doc-endcase
1358: doc-of
1359: doc-endof
1.2 anton 1360:
1361: @i{case-sys} and @i{of-sys} cannot be processed using @code{cs-pick} and
1362: @code{cs-roll}.
1363:
1.3 anton 1364: @subsubsection Programming Style
1365:
1366: In order to ensure readability we recommend that you do not create
1367: arbitrary control structures directly, but define new control structure
1368: words for the control structure you want and use these words in your
1369: program.
1370:
1371: E.g., instead of writing
1372:
1373: @example
1374: begin
1375: ...
1376: if [ 1 cs-roll ]
1377: ...
1378: again then
1379: @end example
1380:
1381: we recommend defining control structure words, e.g.,
1382:
1383: @example
1384: : while ( dest -- orig dest )
1385: POSTPONE if
1386: 1 cs-roll ; immediate
1387:
1388: : repeat ( orig dest -- )
1389: POSTPONE again
1390: POSTPONE then ; immediate
1391: @end example
1392:
1393: and then using these to create the control structure:
1394:
1395: @example
1396: begin
1397: ...
1398: while
1399: ...
1400: repeat
1401: @end example
1402:
1403: That's much easier to read, isn't it? Of course, @code{BEGIN} and
1404: @code{WHILE} are predefined, so in this example it would not be
1405: necessary to define them.
1406:
1.4 anton 1407: @node Calls and returns, Exception Handling, Arbitrary control structures, Control Structures
1.3 anton 1408: @subsection Calls and returns
1409:
1410: A definition can be called simply be writing the name of the
1.17 anton 1411: definition. When the end of the definition is reached, it returns. An
1412: earlier return can be forced using
1.3 anton 1413:
1414: doc-exit
1415:
1416: Don't forget to clean up the return stack and @code{UNLOOP} any
1417: outstanding @code{?DO}...@code{LOOP}s before @code{EXIT}ing. The
1418: primitive compiled by @code{EXIT} is
1419:
1420: doc-;s
1421:
1.4 anton 1422: @node Exception Handling, , Calls and returns, Control Structures
1.3 anton 1423: @subsection Exception Handling
1424:
1425: doc-catch
1426: doc-throw
1427:
1.4 anton 1428: @node Locals, Defining Words, Control Structures, Words
1.1 anton 1429: @section Locals
1430:
1.2 anton 1431: Local variables can make Forth programming more enjoyable and Forth
1432: programs easier to read. Unfortunately, the locals of ANS Forth are
1433: laden with restrictions. Therefore, we provide not only the ANS Forth
1434: locals wordset, but also our own, more powerful locals wordset (we
1435: implemented the ANS Forth locals wordset through our locals wordset).
1436:
1437: @menu
1.17 anton 1438: * Gforth locals::
1.4 anton 1439: * ANS Forth locals::
1.2 anton 1440: @end menu
1441:
1.17 anton 1442: @node Gforth locals, ANS Forth locals, Locals, Locals
1443: @subsection Gforth locals
1.2 anton 1444:
1445: Locals can be defined with
1446:
1447: @example
1448: @{ local1 local2 ... -- comment @}
1449: @end example
1450: or
1451: @example
1452: @{ local1 local2 ... @}
1453: @end example
1454:
1455: E.g.,
1456: @example
1457: : max @{ n1 n2 -- n3 @}
1458: n1 n2 > if
1459: n1
1460: else
1461: n2
1462: endif ;
1463: @end example
1464:
1465: The similarity of locals definitions with stack comments is intended. A
1466: locals definition often replaces the stack comment of a word. The order
1467: of the locals corresponds to the order in a stack comment and everything
1468: after the @code{--} is really a comment.
1469:
1470: This similarity has one disadvantage: It is too easy to confuse locals
1471: declarations with stack comments, causing bugs and making them hard to
1472: find. However, this problem can be avoided by appropriate coding
1473: conventions: Do not use both notations in the same program. If you do,
1474: they should be distinguished using additional means, e.g. by position.
1475:
1476: The name of the local may be preceded by a type specifier, e.g.,
1477: @code{F:} for a floating point value:
1478:
1479: @example
1480: : CX* @{ F: Ar F: Ai F: Br F: Bi -- Cr Ci @}
1481: \ complex multiplication
1482: Ar Br f* Ai Bi f* f-
1483: Ar Bi f* Ai Br f* f+ ;
1484: @end example
1485:
1.17 anton 1486: Gforth currently supports cells (@code{W:}, @code{W^}), doubles
1.2 anton 1487: (@code{D:}, @code{D^}), floats (@code{F:}, @code{F^}) and characters
1488: (@code{C:}, @code{C^}) in two flavours: a value-flavoured local (defined
1489: with @code{W:}, @code{D:} etc.) produces its value and can be changed
1490: with @code{TO}. A variable-flavoured local (defined with @code{W^} etc.)
1491: produces its address (which becomes invalid when the variable's scope is
1492: left). E.g., the standard word @code{emit} can be defined in therms of
1493: @code{type} like this:
1494:
1495: @example
1496: : emit @{ C^ char* -- @}
1497: char* 1 type ;
1498: @end example
1499:
1500: A local without type specifier is a @code{W:} local. Both flavours of
1501: locals are initialized with values from the data or FP stack.
1502:
1503: Currently there is no way to define locals with user-defined data
1504: structures, but we are working on it.
1505:
1.17 anton 1506: Gforth allows defining locals everywhere in a colon definition. This
1.7 pazsan 1507: poses the following questions:
1.2 anton 1508:
1.4 anton 1509: @menu
1510: * Where are locals visible by name?::
1.14 anton 1511: * How long do locals live?::
1.4 anton 1512: * Programming Style::
1513: * Implementation::
1514: @end menu
1515:
1.17 anton 1516: @node Where are locals visible by name?, How long do locals live?, Gforth locals, Gforth locals
1.2 anton 1517: @subsubsection Where are locals visible by name?
1518:
1519: Basically, the answer is that locals are visible where you would expect
1520: it in block-structured languages, and sometimes a little longer. If you
1521: want to restrict the scope of a local, enclose its definition in
1522: @code{SCOPE}...@code{ENDSCOPE}.
1523:
1524: doc-scope
1525: doc-endscope
1526:
1527: These words behave like control structure words, so you can use them
1528: with @code{CS-PICK} and @code{CS-ROLL} to restrict the scope in
1529: arbitrary ways.
1530:
1531: If you want a more exact answer to the visibility question, here's the
1532: basic principle: A local is visible in all places that can only be
1533: reached through the definition of the local@footnote{In compiler
1534: construction terminology, all places dominated by the definition of the
1535: local.}. In other words, it is not visible in places that can be reached
1536: without going through the definition of the local. E.g., locals defined
1537: in @code{IF}...@code{ENDIF} are visible until the @code{ENDIF}, locals
1538: defined in @code{BEGIN}...@code{UNTIL} are visible after the
1539: @code{UNTIL} (until, e.g., a subsequent @code{ENDSCOPE}).
1540:
1541: The reasoning behind this solution is: We want to have the locals
1542: visible as long as it is meaningful. The user can always make the
1543: visibility shorter by using explicit scoping. In a place that can
1544: only be reached through the definition of a local, the meaning of a
1545: local name is clear. In other places it is not: How is the local
1546: initialized at the control flow path that does not contain the
1547: definition? Which local is meant, if the same name is defined twice in
1548: two independent control flow paths?
1549:
1550: This should be enough detail for nearly all users, so you can skip the
1551: rest of this section. If you relly must know all the gory details and
1552: options, read on.
1553:
1554: In order to implement this rule, the compiler has to know which places
1555: are unreachable. It knows this automatically after @code{AHEAD},
1556: @code{AGAIN}, @code{EXIT} and @code{LEAVE}; in other cases (e.g., after
1557: most @code{THROW}s), you can use the word @code{UNREACHABLE} to tell the
1558: compiler that the control flow never reaches that place. If
1559: @code{UNREACHABLE} is not used where it could, the only consequence is
1560: that the visibility of some locals is more limited than the rule above
1561: says. If @code{UNREACHABLE} is used where it should not (i.e., if you
1562: lie to the compiler), buggy code will be produced.
1563:
1564: Another problem with this rule is that at @code{BEGIN}, the compiler
1.3 anton 1565: does not know which locals will be visible on the incoming
1566: back-edge. All problems discussed in the following are due to this
1567: ignorance of the compiler (we discuss the problems using @code{BEGIN}
1568: loops as examples; the discussion also applies to @code{?DO} and other
1.2 anton 1569: loops). Perhaps the most insidious example is:
1570: @example
1571: AHEAD
1572: BEGIN
1573: x
1574: [ 1 CS-ROLL ] THEN
1.4 anton 1575: @{ x @}
1.2 anton 1576: ...
1577: UNTIL
1578: @end example
1579:
1580: This should be legal according to the visibility rule. The use of
1581: @code{x} can only be reached through the definition; but that appears
1582: textually below the use.
1583:
1584: From this example it is clear that the visibility rules cannot be fully
1585: implemented without major headaches. Our implementation treats common
1586: cases as advertised and the exceptions are treated in a safe way: The
1587: compiler makes a reasonable guess about the locals visible after a
1588: @code{BEGIN}; if it is too pessimistic, the
1589: user will get a spurious error about the local not being defined; if the
1590: compiler is too optimistic, it will notice this later and issue a
1591: warning. In the case above the compiler would complain about @code{x}
1592: being undefined at its use. You can see from the obscure examples in
1593: this section that it takes quite unusual control structures to get the
1594: compiler into trouble, and even then it will often do fine.
1595:
1596: If the @code{BEGIN} is reachable from above, the most optimistic guess
1597: is that all locals visible before the @code{BEGIN} will also be
1598: visible after the @code{BEGIN}. This guess is valid for all loops that
1599: are entered only through the @code{BEGIN}, in particular, for normal
1600: @code{BEGIN}...@code{WHILE}...@code{REPEAT} and
1601: @code{BEGIN}...@code{UNTIL} loops and it is implemented in our
1602: compiler. When the branch to the @code{BEGIN} is finally generated by
1603: @code{AGAIN} or @code{UNTIL}, the compiler checks the guess and
1604: warns the user if it was too optimisitic:
1605: @example
1606: IF
1.4 anton 1607: @{ x @}
1.2 anton 1608: BEGIN
1609: \ x ?
1610: [ 1 cs-roll ] THEN
1611: ...
1612: UNTIL
1613: @end example
1614:
1615: Here, @code{x} lives only until the @code{BEGIN}, but the compiler
1616: optimistically assumes that it lives until the @code{THEN}. It notices
1617: this difference when it compiles the @code{UNTIL} and issues a
1618: warning. The user can avoid the warning, and make sure that @code{x}
1619: is not used in the wrong area by using explicit scoping:
1620: @example
1621: IF
1622: SCOPE
1.4 anton 1623: @{ x @}
1.2 anton 1624: ENDSCOPE
1625: BEGIN
1626: [ 1 cs-roll ] THEN
1627: ...
1628: UNTIL
1629: @end example
1630:
1631: Since the guess is optimistic, there will be no spurious error messages
1632: about undefined locals.
1633:
1634: If the @code{BEGIN} is not reachable from above (e.g., after
1635: @code{AHEAD} or @code{EXIT}), the compiler cannot even make an
1636: optimistic guess, as the locals visible after the @code{BEGIN} may be
1637: defined later. Therefore, the compiler assumes that no locals are
1.17 anton 1638: visible after the @code{BEGIN}. However, the user can use
1.2 anton 1639: @code{ASSUME-LIVE} to make the compiler assume that the same locals are
1.17 anton 1640: visible at the BEGIN as at the point where the top control-flow stack
1641: item was created.
1.2 anton 1642:
1643: doc-assume-live
1644:
1645: E.g.,
1646: @example
1.4 anton 1647: @{ x @}
1.2 anton 1648: AHEAD
1649: ASSUME-LIVE
1650: BEGIN
1651: x
1652: [ 1 CS-ROLL ] THEN
1653: ...
1654: UNTIL
1655: @end example
1656:
1657: Other cases where the locals are defined before the @code{BEGIN} can be
1658: handled by inserting an appropriate @code{CS-ROLL} before the
1659: @code{ASSUME-LIVE} (and changing the control-flow stack manipulation
1660: behind the @code{ASSUME-LIVE}).
1661:
1662: Cases where locals are defined after the @code{BEGIN} (but should be
1663: visible immediately after the @code{BEGIN}) can only be handled by
1664: rearranging the loop. E.g., the ``most insidious'' example above can be
1665: arranged into:
1666: @example
1667: BEGIN
1.4 anton 1668: @{ x @}
1.2 anton 1669: ... 0=
1670: WHILE
1671: x
1672: REPEAT
1673: @end example
1674:
1.17 anton 1675: @node How long do locals live?, Programming Style, Where are locals visible by name?, Gforth locals
1.2 anton 1676: @subsubsection How long do locals live?
1677:
1678: The right answer for the lifetime question would be: A local lives at
1679: least as long as it can be accessed. For a value-flavoured local this
1680: means: until the end of its visibility. However, a variable-flavoured
1681: local could be accessed through its address far beyond its visibility
1682: scope. Ultimately, this would mean that such locals would have to be
1683: garbage collected. Since this entails un-Forth-like implementation
1684: complexities, I adopted the same cowardly solution as some other
1685: languages (e.g., C): The local lives only as long as it is visible;
1686: afterwards its address is invalid (and programs that access it
1687: afterwards are erroneous).
1688:
1.17 anton 1689: @node Programming Style, Implementation, How long do locals live?, Gforth locals
1.2 anton 1690: @subsubsection Programming Style
1691:
1692: The freedom to define locals anywhere has the potential to change
1693: programming styles dramatically. In particular, the need to use the
1694: return stack for intermediate storage vanishes. Moreover, all stack
1695: manipulations (except @code{PICK}s and @code{ROLL}s with run-time
1696: determined arguments) can be eliminated: If the stack items are in the
1697: wrong order, just write a locals definition for all of them; then
1698: write the items in the order you want.
1699:
1700: This seems a little far-fetched and eliminating stack manipulations is
1.4 anton 1701: unlikely to become a conscious programming objective. Still, the number
1702: of stack manipulations will be reduced dramatically if local variables
1.17 anton 1703: are used liberally (e.g., compare @code{max} in @ref{Gforth locals} with
1.4 anton 1704: a traditional implementation of @code{max}).
1.2 anton 1705:
1706: This shows one potential benefit of locals: making Forth programs more
1707: readable. Of course, this benefit will only be realized if the
1708: programmers continue to honour the principle of factoring instead of
1709: using the added latitude to make the words longer.
1710:
1711: Using @code{TO} can and should be avoided. Without @code{TO},
1712: every value-flavoured local has only a single assignment and many
1713: advantages of functional languages apply to Forth. I.e., programs are
1714: easier to analyse, to optimize and to read: It is clear from the
1715: definition what the local stands for, it does not turn into something
1716: different later.
1717:
1718: E.g., a definition using @code{TO} might look like this:
1719: @example
1720: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1721: u1 u2 min 0
1722: ?do
1723: addr1 c@ addr2 c@ - ?dup
1724: if
1725: unloop exit
1726: then
1727: addr1 char+ TO addr1
1728: addr2 char+ TO addr2
1729: loop
1730: u1 u2 - ;
1731: @end example
1732: Here, @code{TO} is used to update @code{addr1} and @code{addr2} at
1733: every loop iteration. @code{strcmp} is a typical example of the
1734: readability problems of using @code{TO}. When you start reading
1735: @code{strcmp}, you think that @code{addr1} refers to the start of the
1736: string. Only near the end of the loop you realize that it is something
1737: else.
1738:
1739: This can be avoided by defining two locals at the start of the loop that
1740: are initialized with the right value for the current iteration.
1741: @example
1742: : strcmp @{ addr1 u1 addr2 u2 -- n @}
1743: addr1 addr2
1744: u1 u2 min 0
1745: ?do @{ s1 s2 @}
1746: s1 c@ s2 c@ - ?dup
1747: if
1748: unloop exit
1749: then
1750: s1 char+ s2 char+
1751: loop
1752: 2drop
1753: u1 u2 - ;
1754: @end example
1755: Here it is clear from the start that @code{s1} has a different value
1756: in every loop iteration.
1757:
1.17 anton 1758: @node Implementation, , Programming Style, Gforth locals
1.2 anton 1759: @subsubsection Implementation
1760:
1.17 anton 1761: Gforth uses an extra locals stack. The most compelling reason for
1.2 anton 1762: this is that the return stack is not float-aligned; using an extra stack
1763: also eliminates the problems and restrictions of using the return stack
1764: as locals stack. Like the other stacks, the locals stack grows toward
1765: lower addresses. A few primitives allow an efficient implementation:
1766:
1767: doc-@local#
1768: doc-f@local#
1769: doc-laddr#
1770: doc-lp+!#
1771: doc-lp!
1772: doc->l
1773: doc-f>l
1774:
1775: In addition to these primitives, some specializations of these
1776: primitives for commonly occurring inline arguments are provided for
1777: efficiency reasons, e.g., @code{@@local0} as specialization of
1778: @code{@@local#} for the inline argument 0. The following compiling words
1779: compile the right specialized version, or the general version, as
1780: appropriate:
1781:
1.12 anton 1782: doc-compile-@local
1783: doc-compile-f@local
1.2 anton 1784: doc-compile-lp+!
1785:
1786: Combinations of conditional branches and @code{lp+!#} like
1787: @code{?branch-lp+!#} (the locals pointer is only changed if the branch
1788: is taken) are provided for efficiency and correctness in loops.
1789:
1790: A special area in the dictionary space is reserved for keeping the
1791: local variable names. @code{@{} switches the dictionary pointer to this
1792: area and @code{@}} switches it back and generates the locals
1793: initializing code. @code{W:} etc.@ are normal defining words. This
1794: special area is cleared at the start of every colon definition.
1795:
1.17 anton 1796: A special feature of Gforth's dictionary is used to implement the
1.2 anton 1797: definition of locals without type specifiers: every wordlist (aka
1798: vocabulary) has its own methods for searching
1.4 anton 1799: etc. (@pxref{Wordlists}). For the present purpose we defined a wordlist
1.2 anton 1800: with a special search method: When it is searched for a word, it
1801: actually creates that word using @code{W:}. @code{@{} changes the search
1802: order to first search the wordlist containing @code{@}}, @code{W:} etc.,
1803: and then the wordlist for defining locals without type specifiers.
1804:
1805: The lifetime rules support a stack discipline within a colon
1806: definition: The lifetime of a local is either nested with other locals
1807: lifetimes or it does not overlap them.
1808:
1809: At @code{BEGIN}, @code{IF}, and @code{AHEAD} no code for locals stack
1810: pointer manipulation is generated. Between control structure words
1811: locals definitions can push locals onto the locals stack. @code{AGAIN}
1812: is the simplest of the other three control flow words. It has to
1813: restore the locals stack depth of the corresponding @code{BEGIN}
1814: before branching. The code looks like this:
1815: @format
1816: @code{lp+!#} current-locals-size @minus{} dest-locals-size
1817: @code{branch} <begin>
1818: @end format
1819:
1820: @code{UNTIL} is a little more complicated: If it branches back, it
1821: must adjust the stack just like @code{AGAIN}. But if it falls through,
1822: the locals stack must not be changed. The compiler generates the
1823: following code:
1824: @format
1825: @code{?branch-lp+!#} <begin> current-locals-size @minus{} dest-locals-size
1826: @end format
1827: The locals stack pointer is only adjusted if the branch is taken.
1828:
1829: @code{THEN} can produce somewhat inefficient code:
1830: @format
1831: @code{lp+!#} current-locals-size @minus{} orig-locals-size
1832: <orig target>:
1833: @code{lp+!#} orig-locals-size @minus{} new-locals-size
1834: @end format
1835: The second @code{lp+!#} adjusts the locals stack pointer from the
1.4 anton 1836: level at the @var{orig} point to the level after the @code{THEN}. The
1.2 anton 1837: first @code{lp+!#} adjusts the locals stack pointer from the current
1838: level to the level at the orig point, so the complete effect is an
1839: adjustment from the current level to the right level after the
1840: @code{THEN}.
1841:
1842: In a conventional Forth implementation a dest control-flow stack entry
1843: is just the target address and an orig entry is just the address to be
1844: patched. Our locals implementation adds a wordlist to every orig or dest
1845: item. It is the list of locals visible (or assumed visible) at the point
1846: described by the entry. Our implementation also adds a tag to identify
1847: the kind of entry, in particular to differentiate between live and dead
1848: (reachable and unreachable) orig entries.
1849:
1850: A few unusual operations have to be performed on locals wordlists:
1851:
1852: doc-common-list
1853: doc-sub-list?
1854: doc-list-size
1855:
1856: Several features of our locals wordlist implementation make these
1857: operations easy to implement: The locals wordlists are organised as
1858: linked lists; the tails of these lists are shared, if the lists
1859: contain some of the same locals; and the address of a name is greater
1860: than the address of the names behind it in the list.
1861:
1862: Another important implementation detail is the variable
1863: @code{dead-code}. It is used by @code{BEGIN} and @code{THEN} to
1864: determine if they can be reached directly or only through the branch
1865: that they resolve. @code{dead-code} is set by @code{UNREACHABLE},
1866: @code{AHEAD}, @code{EXIT} etc., and cleared at the start of a colon
1867: definition, by @code{BEGIN} and usually by @code{THEN}.
1868:
1869: Counted loops are similar to other loops in most respects, but
1870: @code{LEAVE} requires special attention: It performs basically the same
1871: service as @code{AHEAD}, but it does not create a control-flow stack
1872: entry. Therefore the information has to be stored elsewhere;
1873: traditionally, the information was stored in the target fields of the
1874: branches created by the @code{LEAVE}s, by organizing these fields into a
1875: linked list. Unfortunately, this clever trick does not provide enough
1876: space for storing our extended control flow information. Therefore, we
1877: introduce another stack, the leave stack. It contains the control-flow
1878: stack entries for all unresolved @code{LEAVE}s.
1879:
1880: Local names are kept until the end of the colon definition, even if
1881: they are no longer visible in any control-flow path. In a few cases
1882: this may lead to increased space needs for the locals name area, but
1883: usually less than reclaiming this space would cost in code size.
1884:
1885:
1.17 anton 1886: @node ANS Forth locals, , Gforth locals, Locals
1.2 anton 1887: @subsection ANS Forth locals
1888:
1889: The ANS Forth locals wordset does not define a syntax for locals, but
1890: words that make it possible to define various syntaxes. One of the
1.17 anton 1891: possible syntaxes is a subset of the syntax we used in the Gforth locals
1.2 anton 1892: wordset, i.e.:
1893:
1894: @example
1895: @{ local1 local2 ... -- comment @}
1896: @end example
1897: or
1898: @example
1899: @{ local1 local2 ... @}
1900: @end example
1901:
1902: The order of the locals corresponds to the order in a stack comment. The
1903: restrictions are:
1.1 anton 1904:
1.2 anton 1905: @itemize @bullet
1906: @item
1.17 anton 1907: Locals can only be cell-sized values (no type specifiers are allowed).
1.2 anton 1908: @item
1909: Locals can be defined only outside control structures.
1910: @item
1911: Locals can interfere with explicit usage of the return stack. For the
1912: exact (and long) rules, see the standard. If you don't use return stack
1.17 anton 1913: accessing words in a definition using locals, you will be all right. The
1.2 anton 1914: purpose of this rule is to make locals implementation on the return
1915: stack easier.
1916: @item
1917: The whole definition must be in one line.
1918: @end itemize
1919:
1920: Locals defined in this way behave like @code{VALUE}s
1.4 anton 1921: (@xref{Values}). I.e., they are initialized from the stack. Using their
1.2 anton 1922: name produces their value. Their value can be changed using @code{TO}.
1923:
1.17 anton 1924: Since this syntax is supported by Gforth directly, you need not do
1.2 anton 1925: anything to use it. If you want to port a program using this syntax to
1926: another ANS Forth system, use @file{anslocal.fs} to implement the syntax
1927: on the other system.
1928:
1929: Note that a syntax shown in the standard, section A.13 looks
1930: similar, but is quite different in having the order of locals
1931: reversed. Beware!
1932:
1933: The ANS Forth locals wordset itself consists of the following word
1934:
1935: doc-(local)
1936:
1937: The ANS Forth locals extension wordset defines a syntax, but it is so
1938: awful that we strongly recommend not to use it. We have implemented this
1.17 anton 1939: syntax to make porting to Gforth easy, but do not document it here. The
1.2 anton 1940: problem with this syntax is that the locals are defined in an order
1941: reversed with respect to the standard stack comment notation, making
1942: programs harder to read, and easier to misread and miswrite. The only
1943: merit of this syntax is that it is easy to implement using the ANS Forth
1944: locals wordset.
1.3 anton 1945:
1.4 anton 1946: @node Defining Words, Wordlists, Locals, Words
1947: @section Defining Words
1948:
1.14 anton 1949: @menu
1950: * Values::
1951: @end menu
1952:
1.4 anton 1953: @node Values, , Defining Words, Defining Words
1954: @subsection Values
1955:
1956: @node Wordlists, Files, Defining Words, Words
1957: @section Wordlists
1958:
1959: @node Files, Blocks, Wordlists, Words
1960: @section Files
1961:
1962: @node Blocks, Other I/O, Files, Words
1963: @section Blocks
1964:
1965: @node Other I/O, Programming Tools, Blocks, Words
1966: @section Other I/O
1967:
1.18 anton 1968: @node Programming Tools, Assembler and Code words, Other I/O, Words
1.4 anton 1969: @section Programming Tools
1970:
1.5 anton 1971: @menu
1972: * Debugging:: Simple and quick.
1973: * Assertions:: Making your programs self-checking.
1974: @end menu
1975:
1976: @node Debugging, Assertions, Programming Tools, Programming Tools
1.4 anton 1977: @subsection Debugging
1978:
1979: The simple debugging aids provided in @file{debugging.fs}
1980: are meant to support a different style of debugging than the
1981: tracing/stepping debuggers used in languages with long turn-around
1982: times.
1983:
1984: A much better (faster) way in fast-compilig languages is to add
1985: printing code at well-selected places, let the program run, look at
1986: the output, see where things went wrong, add more printing code, etc.,
1987: until the bug is found.
1988:
1989: The word @code{~~} is easy to insert. It just prints debugging
1990: information (by default the source location and the stack contents). It
1991: is also easy to remove (@kbd{C-x ~} in the Emacs Forth mode to
1992: query-replace them with nothing). The deferred words
1993: @code{printdebugdata} and @code{printdebugline} control the output of
1994: @code{~~}. The default source location output format works well with
1995: Emacs' compilation mode, so you can step through the program at the
1.5 anton 1996: source level using @kbd{C-x `} (the advantage over a stepping debugger
1997: is that you can step in any direction and you know where the crash has
1998: happened or where the strange data has occurred).
1.4 anton 1999:
2000: Note that the default actions clobber the contents of the pictured
2001: numeric output string, so you should not use @code{~~}, e.g., between
2002: @code{<#} and @code{#>}.
2003:
2004: doc-~~
2005: doc-printdebugdata
2006: doc-printdebugline
2007:
1.5 anton 2008: @node Assertions, , Debugging, Programming Tools
1.4 anton 2009: @subsection Assertions
2010:
1.5 anton 2011: It is a good idea to make your programs self-checking, in particular, if
2012: you use an assumption (e.g., that a certain field of a data structure is
1.17 anton 2013: never zero) that may become wrong during maintenance. Gforth supports
1.5 anton 2014: assertions for this purpose. They are used like this:
2015:
2016: @example
2017: assert( @var{flag} )
2018: @end example
2019:
2020: The code between @code{assert(} and @code{)} should compute a flag, that
2021: should be true if everything is alright and false otherwise. It should
2022: not change anything else on the stack. The overall stack effect of the
2023: assertion is @code{( -- )}. E.g.
2024:
2025: @example
2026: assert( 1 1 + 2 = ) \ what we learn in school
2027: assert( dup 0<> ) \ assert that the top of stack is not zero
2028: assert( false ) \ this code should not be reached
2029: @end example
2030:
2031: The need for assertions is different at different times. During
2032: debugging, we want more checking, in production we sometimes care more
2033: for speed. Therefore, assertions can be turned off, i.e., the assertion
2034: becomes a comment. Depending on the importance of an assertion and the
2035: time it takes to check it, you may want to turn off some assertions and
1.17 anton 2036: keep others turned on. Gforth provides several levels of assertions for
1.5 anton 2037: this purpose:
2038:
2039: doc-assert0(
2040: doc-assert1(
2041: doc-assert2(
2042: doc-assert3(
2043: doc-assert(
2044: doc-)
2045:
2046: @code{Assert(} is the same as @code{assert1(}. The variable
2047: @code{assert-level} specifies the highest assertions that are turned
2048: on. I.e., at the default @code{assert-level} of one, @code{assert0(} and
2049: @code{assert1(} assertions perform checking, while @code{assert2(} and
2050: @code{assert3(} assertions are treated as comments.
2051:
2052: Note that the @code{assert-level} is evaluated at compile-time, not at
2053: run-time. I.e., you cannot turn assertions on or off at run-time, you
2054: have to set the @code{assert-level} appropriately before compiling a
2055: piece of code. You can compile several pieces of code at several
2056: @code{assert-level}s (e.g., a trusted library at level 1 and newly
2057: written code at level 3).
2058:
2059: doc-assert-level
2060:
2061: If an assertion fails, a message compatible with Emacs' compilation mode
2062: is produced and the execution is aborted (currently with @code{ABORT"}.
2063: If there is interest, we will introduce a special throw code. But if you
2064: intend to @code{catch} a specific condition, using @code{throw} is
2065: probably more appropriate than an assertion).
2066:
1.18 anton 2067: @node Assembler and Code words, Threading Words, Programming Tools, Words
2068: @section Assembler and Code words
2069:
2070: Gforth provides some words for defining primitives (words written in
2071: machine code), and for defining the the machine-code equivalent of
2072: @code{DOES>}-based defining words. However, the machine-independent
2073: nature of Gforth poses a few problems: First of all. Gforth runs on
2074: several architectures, so it can provide no standard assembler. What's
2075: worse is that the register allocation not only depends on the processor,
2076: but also on the gcc version and options used.
2077:
2078: The words Gforth offers encapsulate some system dependences (e.g., the
2079: header structure), so a system-independent assembler may be used in
2080: Gforth. If you do not have an assembler, you can compile machine code
2081: directly with @code{,} and @code{c,}.
2082:
2083: doc-assembler
2084: doc-code
2085: doc-end-code
2086: doc-;code
2087: doc-flush-icache
2088:
2089: If @code{flush-icache} does not work correctly, @code{code} words
2090: etc. will not work (reliably), either.
2091:
2092: These words are rarely used. Therefore they reside in @code{code.fs},
2093: which is usually not loaded (except @code{flush-icache}, which is always
1.19 anton 2094: present). You can load them with @code{require code.fs}.
1.18 anton 2095:
2096: Another option for implementing normal and defining words efficiently
2097: is: adding the wanted functionality to the source of Gforth. For normal
2098: words you just have to edit @file{primitives}, defining words (for fast
2099: defined words) probably require changes in @file{engine.c},
2100: @file{kernal.fs}, @file{prims2x.fs}, and possibly @file{cross.fs}.
2101:
2102:
2103: @node Threading Words, , Assembler and Code words, Words
1.4 anton 2104: @section Threading Words
2105:
2106: These words provide access to code addresses and other threading stuff
1.17 anton 2107: in Gforth (and, possibly, other interpretive Forths). It more or less
1.4 anton 2108: abstracts away the differences between direct and indirect threading
2109: (and, for direct threading, the machine dependences). However, at
2110: present this wordset is still inclomplete. It is also pretty low-level;
2111: some day it will hopefully be made unnecessary by an internals words set
2112: that abstracts implementation details away completely.
2113:
2114: doc->code-address
2115: doc->does-code
2116: doc-code-address!
2117: doc-does-code!
2118: doc-does-handler!
2119: doc-/does-handler
2120:
1.18 anton 2121: The code addresses produced by various defining words are produced by
2122: the following words:
1.14 anton 2123:
1.18 anton 2124: doc-docol:
2125: doc-docon:
2126: doc-dovar:
2127: doc-douser:
2128: doc-dodefer:
2129: doc-dofield:
2130:
2131: Currently there is no installation-independent way for recogizing words
2132: defined by a @code{CREATE}...@code{DOES>} word; however, once you know
2133: that a word is defined by a @code{CREATE}...@code{DOES>} word, you can
2134: use @code{>DOES-CODE}.
1.14 anton 2135:
1.4 anton 2136: @node ANS conformance, Model, Words, Top
2137: @chapter ANS conformance
2138:
1.17 anton 2139: To the best of our knowledge, Gforth is an
1.14 anton 2140:
1.15 anton 2141: ANS Forth System
2142: @itemize
2143: @item providing the Core Extensions word set
2144: @item providing the Block word set
2145: @item providing the Block Extensions word set
2146: @item providing the Double-Number word set
2147: @item providing the Double-Number Extensions word set
2148: @item providing the Exception word set
2149: @item providing the Exception Extensions word set
2150: @item providing the Facility word set
2151: @item providing @code{MS} and @code{TIME&DATE} from the Facility Extensions word set
2152: @item providing the File Access word set
2153: @item providing the File Access Extensions word set
2154: @item providing the Floating-Point word set
2155: @item providing the Floating-Point Extensions word set
2156: @item providing the Locals word set
2157: @item providing the Locals Extensions word set
2158: @item providing the Memory-Allocation word set
2159: @item providing the Memory-Allocation Extensions word set (that one's easy)
2160: @item providing the Programming-Tools word set
1.18 anton 2161: @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 2162: @item providing the Search-Order word set
2163: @item providing the Search-Order Extensions word set
2164: @item providing the String word set
2165: @item providing the String Extensions word set (another easy one)
2166: @end itemize
2167:
2168: In addition, ANS Forth systems are required to document certain
2169: implementation choices. This chapter tries to meet these
2170: requirements. In many cases it gives a way to ask the system for the
2171: information instead of providing the information directly, in
2172: particular, if the information depends on the processor, the operating
2173: system or the installation options chosen, or if they are likely to
1.17 anton 2174: change during the maintenance of Gforth.
1.15 anton 2175:
1.14 anton 2176: @comment The framework for the rest has been taken from pfe.
2177:
2178: @menu
2179: * The Core Words::
2180: * The optional Block word set::
2181: * The optional Double Number word set::
2182: * The optional Exception word set::
2183: * The optional Facility word set::
2184: * The optional File-Access word set::
2185: * The optional Floating-Point word set::
2186: * The optional Locals word set::
2187: * The optional Memory-Allocation word set::
2188: * The optional Programming-Tools word set::
2189: * The optional Search-Order word set::
2190: @end menu
2191:
2192:
2193: @c =====================================================================
2194: @node The Core Words, The optional Block word set, ANS conformance, ANS conformance
2195: @comment node-name, next, previous, up
2196: @section The Core Words
2197: @c =====================================================================
2198:
2199: @menu
1.15 anton 2200: * core-idef:: Implementation Defined Options
2201: * core-ambcond:: Ambiguous Conditions
2202: * core-other:: Other System Documentation
1.14 anton 2203: @end menu
2204:
2205: @c ---------------------------------------------------------------------
2206: @node core-idef, core-ambcond, The Core Words, The Core Words
2207: @subsection Implementation Defined Options
2208: @c ---------------------------------------------------------------------
2209:
2210: @table @i
2211:
2212: @item (Cell) aligned addresses:
1.17 anton 2213: processor-dependent. Gforth's alignment words perform natural alignment
1.14 anton 2214: (e.g., an address aligned for a datum of size 8 is divisible by
2215: 8). Unaligned accesses usually result in a @code{-23 THROW}.
2216:
2217: @item @code{EMIT} and non-graphic characters:
2218: The character is output using the C library function (actually, macro)
2219: @code{putchar}.
2220:
2221: @item character editing of @code{ACCEPT} and @code{EXPECT}:
2222: This is modeled on the GNU readline library (@pxref{Readline
2223: Interaction, , Command Line Editing, readline, The GNU Readline
2224: Library}) with Emacs-like key bindings. @kbd{Tab} deviates a little by
2225: producing a full word completion every time you type it (instead of
2226: producing the common prefix of all completions).
2227:
2228: @item character set:
2229: The character set of your computer and display device. Gforth is
2230: 8-bit-clean (but some other component in your system may make trouble).
2231:
2232: @item Character-aligned address requirements:
2233: installation-dependent. Currently a character is represented by a C
2234: @code{unsigned char}; in the future we might switch to @code{wchar_t}
2235: (Comments on that requested).
2236:
2237: @item character-set extensions and matching of names:
1.17 anton 2238: Any character except the ASCII NUL charcter can be used in a
2239: name. Matching is case-insensitive. The matching is performed using the
2240: C function @code{strncasecmp}, whose function is probably influenced by
2241: the locale. E.g., the @code{C} locale does not know about accents and
1.14 anton 2242: umlauts, so they are matched case-sensitively in that locale. For
2243: portability reasons it is best to write programs such that they work in
2244: the @code{C} locale. Then one can use libraries written by a Polish
2245: programmer (who might use words containing ISO Latin-2 encoded
2246: characters) and by a French programmer (ISO Latin-1) in the same program
2247: (of course, @code{WORDS} will produce funny results for some of the
2248: words (which ones, depends on the font you are using)). Also, the locale
2249: you prefer may not be available in other operating systems. Hopefully,
2250: Unicode will solve these problems one day.
2251:
2252: @item conditions under which control characters match a space delimiter:
2253: If @code{WORD} is called with the space character as a delimiter, all
2254: white-space characters (as identified by the C macro @code{isspace()})
2255: are delimiters. @code{PARSE}, on the other hand, treats space like other
2256: delimiters. @code{PARSE-WORD} treats space like @code{WORD}, but behaves
2257: like @code{PARSE} otherwise. @code{(NAME)}, which is used by the outer
2258: interpreter (aka text interpreter) by default, treats all white-space
2259: characters as delimiters.
2260:
2261: @item format of the control flow stack:
2262: The data stack is used as control flow stack. The size of a control flow
2263: stack item in cells is given by the constant @code{cs-item-size}. At the
2264: time of this writing, an item consists of a (pointer to a) locals list
2265: (third), an address in the code (second), and a tag for identifying the
2266: item (TOS). The following tags are used: @code{defstart},
2267: @code{live-orig}, @code{dead-orig}, @code{dest}, @code{do-dest},
2268: @code{scopestart}.
2269:
2270: @item conversion of digits > 35
2271: The characters @code{[\]^_'} are the digits with the decimal value
2272: 36@minus{}41. There is no way to input many of the larger digits.
2273:
2274: @item display after input terminates in @code{ACCEPT} and @code{EXPECT}:
2275: The cursor is moved to the end of the entered string. If the input is
2276: terminated using the @kbd{Return} key, a space is typed.
2277:
2278: @item exception abort sequence of @code{ABORT"}:
2279: The error string is stored into the variable @code{"error} and a
2280: @code{-2 throw} is performed.
2281:
2282: @item input line terminator:
2283: For interactive input, @kbd{C-m} and @kbd{C-j} terminate lines. One of
2284: these characters is typically produced when you type the @kbd{Enter} or
2285: @kbd{Return} key.
2286:
2287: @item maximum size of a counted string:
2288: @code{s" /counted-string" environment? drop .}. Currently 255 characters
2289: on all ports, but this may change.
2290:
2291: @item maximum size of a parsed string:
2292: Given by the constant @code{/line}. Currently 255 characters.
2293:
2294: @item maximum size of a definition name, in characters:
2295: 31
2296:
2297: @item maximum string length for @code{ENVIRONMENT?}, in characters:
2298: 31
2299:
2300: @item method of selecting the user input device:
1.17 anton 2301: The user input device is the standard input. There is currently no way to
2302: change it from within Gforth. However, the input can typically be
2303: redirected in the command line that starts Gforth.
1.14 anton 2304:
2305: @item method of selecting the user output device:
2306: The user output device is the standard output. It cannot be redirected
1.17 anton 2307: from within Gforth, but typically from the command line that starts
2308: Gforth. Gforth uses buffered output, so output on a terminal does not
1.14 anton 2309: become visible before the next newline or buffer overflow. Output on
2310: non-terminals is invisible until the buffer overflows.
2311:
2312: @item methods of dictionary compilation:
1.17 anton 2313: What are we expected to document here?
1.14 anton 2314:
2315: @item number of bits in one address unit:
2316: @code{s" address-units-bits" environment? drop .}. 8 in all current
2317: ports.
2318:
2319: @item number representation and arithmetic:
2320: Processor-dependent. Binary two's complement on all current ports.
2321:
2322: @item ranges for integer types:
2323: Installation-dependent. Make environmental queries for @code{MAX-N},
2324: @code{MAX-U}, @code{MAX-D} and @code{MAX-UD}. The lower bounds for
2325: unsigned (and positive) types is 0. The lower bound for signed types on
2326: two's complement and one's complement machines machines can be computed
2327: by adding 1 to the upper bound.
2328:
2329: @item read-only data space regions:
2330: The whole Forth data space is writable.
2331:
2332: @item size of buffer at @code{WORD}:
2333: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2334: shared with the pictured numeric output string. If overwriting
2335: @code{PAD} is acceptable, it is as large as the remaining dictionary
2336: space, although only as much can be sensibly used as fits in a counted
2337: string.
2338:
2339: @item size of one cell in address units:
2340: @code{1 cells .}.
2341:
2342: @item size of one character in address units:
2343: @code{1 chars .}. 1 on all current ports.
2344:
2345: @item size of the keyboard terminal buffer:
2346: Varies. You can determine the size at a specific time using @code{lp@
2347: tib - .}. It is shared with the locals stack and TIBs of files that
2348: include the current file. You can change the amount of space for TIBs
1.17 anton 2349: and locals stack at Gforth startup with the command line option
1.14 anton 2350: @code{-l}.
2351:
2352: @item size of the pictured numeric output buffer:
2353: @code{PAD HERE - .}. 104 characters on 32-bit machines. The buffer is
2354: shared with @code{WORD}.
2355:
2356: @item size of the scratch area returned by @code{PAD}:
2357: The remainder of dictionary space. You can even use the unused part of
2358: the data stack space. The current size can be computed with @code{sp@
2359: pad - .}.
2360:
2361: @item system case-sensitivity characteristics:
2362: Dictionary searches are case insensitive. However, as explained above
2363: under @i{character-set extensions}, the matching for non-ASCII
2364: characters is determined by the locale you are using. In the default
2365: @code{C} locale all non-ASCII characters are matched case-sensitively.
2366:
2367: @item system prompt:
2368: @code{ ok} in interpret state, @code{ compiled} in compile state.
2369:
2370: @item division rounding:
2371: installation dependent. @code{s" floored" environment? drop .}. We leave
2372: the choice to gcc (what to use for @code{/}) and to you (whether to use
2373: @code{fm/mod}, @code{sm/rem} or simply @code{/}).
2374:
2375: @item values of @code{STATE} when true:
2376: -1.
2377:
2378: @item values returned after arithmetic overflow:
2379: On two's complement machines, arithmetic is performed modulo
2380: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2381: arithmetic (with appropriate mapping for signed types). Division by zero
2382: typically results in a @code{-55 throw} (floatingpoint unidentified
2383: fault), although a @code{-10 throw} (divide by zero) would be more
2384: appropriate.
2385:
2386: @item whether the current definition can be found after @t{DOES>}:
2387: No.
2388:
2389: @end table
2390:
2391: @c ---------------------------------------------------------------------
2392: @node core-ambcond, core-other, core-idef, The Core Words
2393: @subsection Ambiguous conditions
2394: @c ---------------------------------------------------------------------
2395:
2396: @table @i
2397:
2398: @item a name is neither a word nor a number:
2399: @code{-13 throw} (Undefined word)
2400:
2401: @item a definition name exceeds the maximum length allowed:
2402: @code{-19 throw} (Word name too long)
2403:
2404: @item addressing a region not inside the various data spaces of the forth system:
2405: The stacks, code space and name space are accessible. Machine code space is
2406: typically readable. Accessing other addresses gives results dependent on
2407: the operating system. On decent systems: @code{-9 throw} (Invalid memory
2408: address).
2409:
2410: @item argument type incompatible with parameter:
2411: This is usually not caught. Some words perform checks, e.g., the control
2412: flow words, and issue a @code{ABORT"} or @code{-12 THROW} (Argument type
2413: mismatch).
2414:
2415: @item attempting to obtain the execution token of a word with undefined execution semantics:
2416: You get an execution token representing the compilation semantics
2417: instead.
2418:
2419: @item dividing by zero:
2420: typically results in a @code{-55 throw} (floating point unidentified
2421: fault), although a @code{-10 throw} (divide by zero) would be more
2422: appropriate.
2423:
2424: @item insufficient data stack or return stack space:
2425: Not checked. This typically results in mysterious illegal memory
2426: accesses, producing @code{-9 throw} (Invalid memory address) or
2427: @code{-23 throw} (Address alignment exception).
2428:
2429: @item insufficient space for loop control parameters:
2430: like other return stack overflows.
2431:
2432: @item insufficient space in the dictionary:
2433: Not checked. Similar results as stack overflows. However, typically the
2434: error appears at a different place when one inserts or removes code.
2435:
2436: @item interpreting a word with undefined interpretation semantics:
2437: For some words, we defined interpretation semantics. For the others:
2438: @code{-14 throw} (Interpreting a compile-only word). Note that this is
2439: checked only by the outer (aka text) interpreter; if the word is
2440: @code{execute}d in some other way, it will typically perform it's
2441: compilation semantics even in interpret state. (We could change @code{'}
2442: and relatives not to give the xt of such words, but we think that would
2443: be too restrictive).
2444:
2445: @item modifying the contents of the input buffer or a string literal:
2446: These are located in writable memory and can be modified.
2447:
2448: @item overflow of the pictured numeric output string:
2449: Not checked.
2450:
2451: @item parsed string overflow:
2452: @code{PARSE} cannot overflow. @code{WORD} does not check for overflow.
2453:
2454: @item producing a result out of range:
2455: On two's complement machines, arithmetic is performed modulo
2456: 2**bits-per-cell for single arithmetic and 4**bits-per-cell for double
2457: arithmetic (with appropriate mapping for signed types). Division by zero
2458: typically results in a @code{-55 throw} (floatingpoint unidentified
2459: fault), although a @code{-10 throw} (divide by zero) would be more
2460: appropriate. @code{convert} and @code{>number} currently overflow
2461: silently.
2462:
2463: @item reading from an empty data or return stack:
2464: The data stack is checked by the outer (aka text) interpreter after
2465: every word executed. If it has underflowed, a @code{-4 throw} (Stack
2466: underflow) is performed. Apart from that, the stacks are not checked and
2467: underflows can result in similar behaviour as overflows (of adjacent
2468: stacks).
2469:
2470: @item unexepected end of the input buffer, resulting in an attempt to use a zero-length string as a name:
2471: @code{Create} and its descendants perform a @code{-16 throw} (Attempt to
2472: use zero-length string as a name). Words like @code{'} probably will not
2473: find what they search. Note that it is possible to create zero-length
2474: names with @code{nextname} (should it not?).
2475:
2476: @item @code{>IN} greater than input buffer:
2477: The next invocation of a parsing word returns a string wih length 0.
2478:
2479: @item @code{RECURSE} appears after @code{DOES>}:
2480: Compiles a recursive call to the defining word not to the defined word.
2481:
2482: @item argument input source different than current input source for @code{RESTORE-INPUT}:
2483: !!???If the argument input source is a valid input source then it gets
1.19 anton 2484: restored. Otherwise causes @code{-12 THROW}, which, unless caught, issues
1.14 anton 2485: the message "argument type mismatch" and aborts.
2486:
2487: @item data space containing definitions gets de-allocated:
2488: Deallocation with @code{allot} is not checked. This typically resuls in
2489: memory access faults or execution of illegal instructions.
2490:
2491: @item data space read/write with incorrect alignment:
2492: Processor-dependent. Typically results in a @code{-23 throw} (Address
2493: alignment exception). Under Linux on a 486 or later processor with
2494: alignment turned on, incorrect alignment results in a @code{-9 throw}
2495: (Invalid memory address). There are reportedly some processors with
2496: alignment restrictions that do not report them.
2497:
2498: @item data space pointer not properly aligned, @code{,}, @code{C,}:
2499: Like other alignment errors.
2500:
2501: @item less than u+2 stack items (@code{PICK} and @code{ROLL}):
2502: Not checked. May cause an illegal memory access.
2503:
2504: @item loop control parameters not available:
2505: Not checked. The counted loop words simply assume that the top of return
2506: stack items are loop control parameters and behave accordingly.
2507:
2508: @item most recent definition does not have a name (@code{IMMEDIATE}):
2509: @code{abort" last word was headerless"}.
2510:
2511: @item name not defined by @code{VALUE} used by @code{TO}:
2512: @code{-32 throw} (Invalid name argument)
2513:
1.15 anton 2514: @item name not found (@code{'}, @code{POSTPONE}, @code{[']}, @code{[COMPILE]}):
1.14 anton 2515: @code{-13 throw} (Undefined word)
2516:
2517: @item parameters are not of the same type (@code{DO}, @code{?DO}, @code{WITHIN}):
2518: Gforth behaves as if they were of the same type. I.e., you can predict
2519: the behaviour by interpreting all parameters as, e.g., signed.
2520:
2521: @item @code{POSTPONE} or @code{[COMPILE]} applied to @code{TO}:
2522: Assume @code{: X POSTPONE TO ; IMMEDIATE}. @code{X} is equivalent to
2523: @code{TO}.
2524:
2525: @item String longer than a counted string returned by @code{WORD}:
2526: Not checked. The string will be ok, but the count will, of course,
2527: contain only the least significant bits of the length.
2528:
1.15 anton 2529: @item u greater than or equal to the number of bits in a cell (@code{LSHIFT}, @code{RSHIFT}):
1.14 anton 2530: Processor-dependent. Typical behaviours are returning 0 and using only
2531: the low bits of the shift count.
2532:
2533: @item word not defined via @code{CREATE}:
2534: @code{>BODY} produces the PFA of the word no matter how it was defined.
2535:
2536: @code{DOES>} changes the execution semantics of the last defined word no
2537: matter how it was defined. E.g., @code{CONSTANT DOES>} is equivalent to
2538: @code{CREATE , DOES>}.
2539:
2540: @item words improperly used outside @code{<#} and @code{#>}:
2541: Not checked. As usual, you can expect memory faults.
2542:
2543: @end table
2544:
2545:
2546: @c ---------------------------------------------------------------------
2547: @node core-other, , core-ambcond, The Core Words
2548: @subsection Other system documentation
2549: @c ---------------------------------------------------------------------
2550:
2551: @table @i
2552:
2553: @item nonstandard words using @code{PAD}:
2554: None.
2555:
2556: @item operator's terminal facilities available:
2557: !!??
2558:
2559: @item program data space available:
2560: @code{sp@ here - .} gives the space remaining for dictionary and data
2561: stack together.
2562:
2563: @item return stack space available:
2564: !!??
2565:
2566: @item stack space available:
2567: @code{sp@ here - .} gives the space remaining for dictionary and data
2568: stack together.
2569:
2570: @item system dictionary space required, in address units:
2571: Type @code{here forthstart - .} after startup. At the time of this
2572: writing, this gives 70108 (bytes) on a 32-bit system.
2573: @end table
2574:
2575:
2576: @c =====================================================================
2577: @node The optional Block word set, The optional Double Number word set, The Core Words, ANS conformance
2578: @section The optional Block word set
2579: @c =====================================================================
2580:
2581: @menu
1.15 anton 2582: * block-idef:: Implementation Defined Options
2583: * block-ambcond:: Ambiguous Conditions
2584: * block-other:: Other System Documentation
1.14 anton 2585: @end menu
2586:
2587:
2588: @c ---------------------------------------------------------------------
2589: @node block-idef, block-ambcond, The optional Block word set, The optional Block word set
2590: @subsection Implementation Defined Options
2591: @c ---------------------------------------------------------------------
2592:
2593: @table @i
2594:
2595: @item the format for display by @code{LIST}:
2596: First the screen number is displayed, then 16 lines of 64 characters,
2597: each line preceded by the line number.
2598:
2599: @item the length of a line affected by @code{\}:
2600: 64 characters.
2601: @end table
2602:
2603:
2604: @c ---------------------------------------------------------------------
2605: @node block-ambcond, block-other, block-idef, The optional Block word set
2606: @subsection Ambiguous conditions
2607: @c ---------------------------------------------------------------------
2608:
2609: @table @i
2610:
2611: @item correct block read was not possible:
2612: Typically results in a @code{throw} of some OS-derived value (between
2613: -512 and -2048). If the blocks file was just not long enough, blanks are
2614: supplied for the missing portion.
2615:
2616: @item I/O exception in block transfer:
2617: Typically results in a @code{throw} of some OS-derived value (between
2618: -512 and -2048).
2619:
2620: @item invalid block number:
2621: @code{-35 throw} (Invalid block number)
2622:
2623: @item a program directly alters the contents of @code{BLK}:
2624: The input stream is switched to that other block, at the same
2625: position. If the storing to @code{BLK} happens when interpreting
2626: non-block input, the system will get quite confused when the block ends.
2627:
2628: @item no current block buffer for @code{UPDATE}:
2629: @code{UPDATE} has no effect.
2630:
2631: @end table
2632:
2633:
2634: @c ---------------------------------------------------------------------
2635: @node block-other, , block-ambcond, The optional Block word set
2636: @subsection Other system documentation
2637: @c ---------------------------------------------------------------------
2638:
2639: @table @i
2640:
2641: @item any restrictions a multiprogramming system places on the use of buffer addresses:
2642: No restrictions (yet).
2643:
2644: @item the number of blocks available for source and data:
2645: depends on your disk space.
2646:
2647: @end table
2648:
2649:
2650: @c =====================================================================
2651: @node The optional Double Number word set, The optional Exception word set, The optional Block word set, ANS conformance
2652: @section The optional Double Number word set
2653: @c =====================================================================
2654:
2655: @menu
1.15 anton 2656: * double-ambcond:: Ambiguous Conditions
1.14 anton 2657: @end menu
2658:
2659:
2660: @c ---------------------------------------------------------------------
1.15 anton 2661: @node double-ambcond, , The optional Double Number word set, The optional Double Number word set
1.14 anton 2662: @subsection Ambiguous conditions
2663: @c ---------------------------------------------------------------------
2664:
2665: @table @i
2666:
1.15 anton 2667: @item @var{d} outside of range of @var{n} in @code{D>S}:
1.14 anton 2668: The least significant cell of @var{d} is produced.
2669:
2670: @end table
2671:
2672:
2673: @c =====================================================================
2674: @node The optional Exception word set, The optional Facility word set, The optional Double Number word set, ANS conformance
2675: @section The optional Exception word set
2676: @c =====================================================================
2677:
2678: @menu
1.15 anton 2679: * exception-idef:: Implementation Defined Options
1.14 anton 2680: @end menu
2681:
2682:
2683: @c ---------------------------------------------------------------------
1.15 anton 2684: @node exception-idef, , The optional Exception word set, The optional Exception word set
1.14 anton 2685: @subsection Implementation Defined Options
2686: @c ---------------------------------------------------------------------
2687:
2688: @table @i
2689: @item @code{THROW}-codes used in the system:
2690: The codes -256@minus{}-511 are used for reporting signals (see
2691: @file{errore.fs}). The codes -512@minus{}-2047 are used for OS errors
2692: (for file and memory allocation operations). The mapping from OS error
2693: numbers to throw code is -512@minus{}@var{errno}. One side effect of
2694: this mapping is that undefined OS errors produce a message with a
2695: strange number; e.g., @code{-1000 THROW} results in @code{Unknown error
2696: 488} on my system.
2697: @end table
2698:
2699: @c =====================================================================
2700: @node The optional Facility word set, The optional File-Access word set, The optional Exception word set, ANS conformance
2701: @section The optional Facility word set
2702: @c =====================================================================
2703:
2704: @menu
1.15 anton 2705: * facility-idef:: Implementation Defined Options
2706: * facility-ambcond:: Ambiguous Conditions
1.14 anton 2707: @end menu
2708:
2709:
2710: @c ---------------------------------------------------------------------
2711: @node facility-idef, facility-ambcond, The optional Facility word set, The optional Facility word set
2712: @subsection Implementation Defined Options
2713: @c ---------------------------------------------------------------------
2714:
2715: @table @i
2716:
2717: @item encoding of keyboard events (@code{EKEY}):
2718: Not yet implemeted.
2719:
2720: @item duration of a system clock tick
2721: System dependent. With respect to @code{MS}, the time is specified in
2722: microseconds. How well the OS and the hardware implement this, is
2723: another question.
2724:
2725: @item repeatability to be expected from the execution of @code{MS}:
2726: System dependent. On Unix, a lot depends on load. If the system is
1.17 anton 2727: lightly loaded, and the delay is short enough that Gforth does not get
1.14 anton 2728: swapped out, the performance should be acceptable. Under MS-DOS and
2729: other single-tasking systems, it should be good.
2730:
2731: @end table
2732:
2733:
2734: @c ---------------------------------------------------------------------
1.15 anton 2735: @node facility-ambcond, , facility-idef, The optional Facility word set
1.14 anton 2736: @subsection Ambiguous conditions
2737: @c ---------------------------------------------------------------------
2738:
2739: @table @i
2740:
2741: @item @code{AT-XY} can't be performed on user output device:
2742: Largely terminal dependant. No range checks are done on the arguments.
2743: No errors are reported. You may see some garbage appearing, you may see
2744: simply nothing happen.
2745:
2746: @end table
2747:
2748:
2749: @c =====================================================================
2750: @node The optional File-Access word set, The optional Floating-Point word set, The optional Facility word set, ANS conformance
2751: @section The optional File-Access word set
2752: @c =====================================================================
2753:
2754: @menu
1.15 anton 2755: * file-idef:: Implementation Defined Options
2756: * file-ambcond:: Ambiguous Conditions
1.14 anton 2757: @end menu
2758:
2759:
2760: @c ---------------------------------------------------------------------
2761: @node file-idef, file-ambcond, The optional File-Access word set, The optional File-Access word set
2762: @subsection Implementation Defined Options
2763: @c ---------------------------------------------------------------------
2764:
2765: @table @i
2766:
2767: @item File access methods used:
2768: @code{R/O}, @code{R/W} and @code{BIN} work as you would
2769: expect. @code{W/O} translates into the C file opening mode @code{w} (or
2770: @code{wb}): The file is cleared, if it exists, and created, if it does
1.15 anton 2771: not (both with @code{open-file} and @code{create-file}). Under Unix
1.14 anton 2772: @code{create-file} creates a file with 666 permissions modified by your
2773: umask.
2774:
2775: @item file exceptions:
2776: The file words do not raise exceptions (except, perhaps, memory access
2777: faults when you pass illegal addresses or file-ids).
2778:
2779: @item file line terminator:
2780: System-dependent. Gforth uses C's newline character as line
2781: terminator. What the actual character code(s) of this are is
2782: system-dependent.
2783:
2784: @item file name format
2785: System dependent. Gforth just uses the file name format of your OS.
2786:
2787: @item information returned by @code{FILE-STATUS}:
2788: @code{FILE-STATUS} returns the most powerful file access mode allowed
2789: for the file: Either @code{R/O}, @code{W/O} or @code{R/W}. If the file
2790: cannot be accessed, @code{R/O BIN} is returned. @code{BIN} is applicable
2791: along with the retured mode.
2792:
2793: @item input file state after an exception when including source:
2794: All files that are left via the exception are closed.
2795:
2796: @item @var{ior} values and meaning:
1.15 anton 2797: The @var{ior}s returned by the file and memory allocation words are
2798: intended as throw codes. They typically are in the range
2799: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
2800: @var{ior}s is -512@minus{}@var{errno}.
1.14 anton 2801:
2802: @item maximum depth of file input nesting:
2803: limited by the amount of return stack, locals/TIB stack, and the number
2804: of open files available. This should not give you troubles.
2805:
2806: @item maximum size of input line:
2807: @code{/line}. Currently 255.
2808:
2809: @item methods of mapping block ranges to files:
2810: Currently, the block words automatically access the file
2811: @file{blocks.fb} in the currend working directory. More sophisticated
2812: methods could be implemented if there is demand (and a volunteer).
2813:
2814: @item number of string buffers provided by @code{S"}:
2815: 1
2816:
2817: @item size of string buffer used by @code{S"}:
2818: @code{/line}. currently 255.
2819:
2820: @end table
2821:
2822: @c ---------------------------------------------------------------------
1.15 anton 2823: @node file-ambcond, , file-idef, The optional File-Access word set
1.14 anton 2824: @subsection Ambiguous conditions
2825: @c ---------------------------------------------------------------------
2826:
2827: @table @i
2828:
2829: @item attempting to position a file outside it's boundaries:
2830: @code{REPOSITION-FILE} is performed as usual: Afterwards,
2831: @code{FILE-POSITION} returns the value given to @code{REPOSITION-FILE}.
2832:
2833: @item attempting to read from file positions not yet written:
2834: End-of-file, i.e., zero characters are read and no error is reported.
2835:
2836: @item @var{file-id} is invalid (@code{INCLUDE-FILE}):
2837: An appropriate exception may be thrown, but a memory fault or other
2838: problem is more probable.
2839:
2840: @item I/O exception reading or closing @var{file-id} (@code{include-file}, @code{included}):
2841: The @var{ior} produced by the operation, that discovered the problem, is
2842: thrown.
2843:
2844: @item named file cannot be opened (@code{included}):
2845: The @var{ior} produced by @code{open-file} is thrown.
2846:
2847: @item requesting an unmapped block number:
2848: There are no unmapped legal block numbers. On some operating systems,
2849: writing a block with a large number may overflow the file system and
2850: have an error message as consequence.
2851:
2852: @item using @code{source-id} when @code{blk} is non-zero:
2853: @code{source-id} performs its function. Typically it will give the id of
2854: the source which loaded the block. (Better ideas?)
2855:
2856: @end table
2857:
2858:
2859: @c =====================================================================
2860: @node The optional Floating-Point word set, The optional Locals word set, The optional File-Access word set, ANS conformance
1.15 anton 2861: @section The optional Floating-Point word set
1.14 anton 2862: @c =====================================================================
2863:
2864: @menu
1.15 anton 2865: * floating-idef:: Implementation Defined Options
2866: * floating-ambcond:: Ambiguous Conditions
1.14 anton 2867: @end menu
2868:
2869:
2870: @c ---------------------------------------------------------------------
2871: @node floating-idef, floating-ambcond, The optional Floating-Point word set, The optional Floating-Point word set
2872: @subsection Implementation Defined Options
2873: @c ---------------------------------------------------------------------
2874:
2875: @table @i
2876:
1.15 anton 2877: @item format and range of floating point numbers:
2878: System-dependent; the @code{double} type of C.
1.14 anton 2879:
1.15 anton 2880: @item results of @code{REPRESENT} when @var{float} is out of range:
2881: System dependent; @code{REPRESENT} is implemented using the C library
2882: function @code{ecvt()} and inherits its behaviour in this respect.
1.14 anton 2883:
1.15 anton 2884: @item rounding or truncation of floating-point numbers:
2885: What's the question?!!
1.14 anton 2886:
1.15 anton 2887: @item size of floating-point stack:
2888: @code{s" FLOATING-STACK" environment? drop .}. Can be changed at startup
2889: with the command-line option @code{-f}.
1.14 anton 2890:
1.15 anton 2891: @item width of floating-point stack:
2892: @code{1 floats}.
1.14 anton 2893:
2894: @end table
2895:
2896:
2897: @c ---------------------------------------------------------------------
1.15 anton 2898: @node floating-ambcond, , floating-idef, The optional Floating-Point word set
2899: @subsection Ambiguous conditions
1.14 anton 2900: @c ---------------------------------------------------------------------
2901:
2902: @table @i
2903:
1.15 anton 2904: @item @code{df@@} or @code{df!} used with an address that is not double-float aligned:
2905: System-dependent. Typically results in an alignment fault like other
2906: alignment violations.
1.14 anton 2907:
1.15 anton 2908: @item @code{f@@} or @code{f!} used with an address that is not float aligned:
2909: System-dependent. Typically results in an alignment fault like other
2910: alignment violations.
1.14 anton 2911:
1.15 anton 2912: @item Floating-point result out of range:
2913: System-dependent. Can result in a @code{-55 THROW} (Floating-point
2914: unidentified fault), or can produce a special value representing, e.g.,
2915: Infinity.
1.14 anton 2916:
1.15 anton 2917: @item @code{sf@@} or @code{sf!} used with an address that is not single-float aligned:
2918: System-dependent. Typically results in an alignment fault like other
2919: alignment violations.
1.14 anton 2920:
1.15 anton 2921: @item BASE is not decimal (@code{REPRESENT}, @code{F.}, @code{FE.}, @code{FS.}):
2922: The floating-point number is converted into decimal nonetheless.
1.14 anton 2923:
1.15 anton 2924: @item Both arguments are equal to zero (@code{FATAN2}):
2925: System-dependent. @code{FATAN2} is implemented using the C library
2926: function @code{atan2()}.
1.14 anton 2927:
1.15 anton 2928: @item Using ftan on an argument @var{r1} where cos(@var{r1}) is zero:
2929: System-dependent. Anyway, typically the cos of @var{r1} will not be zero
2930: because of small errors and the tan will be a very large (or very small)
2931: but finite number.
1.14 anton 2932:
1.15 anton 2933: @item @var{d} cannot be presented precisely as a float in @code{D>F}:
2934: The result is rounded to the nearest float.
1.14 anton 2935:
1.15 anton 2936: @item dividing by zero:
2937: @code{-55 throw} (Floating-point unidentified fault)
1.14 anton 2938:
1.15 anton 2939: @item exponent too big for conversion (@code{DF!}, @code{DF@@}, @code{SF!}, @code{SF@@}):
2940: System dependent. On IEEE-FP based systems the number is converted into
2941: an infinity.
1.14 anton 2942:
1.15 anton 2943: @item @var{float}<1 (@code{facosh}):
2944: @code{-55 throw} (Floating-point unidentified fault)
1.14 anton 2945:
1.15 anton 2946: @item @var{float}=<-1 (@code{flnp1}):
2947: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
2948: negative infinity is typically produced for @var{float}=-1.
1.14 anton 2949:
1.15 anton 2950: @item @var{float}=<0 (@code{fln}, @code{flog}):
2951: @code{-55 throw} (Floating-point unidentified fault). On IEEE-FP systems
2952: negative infinity is typically produced for @var{float}=0.
1.14 anton 2953:
1.15 anton 2954: @item @var{float}<0 (@code{fasinh}, @code{fsqrt}):
2955: @code{-55 throw} (Floating-point unidentified fault). @code{fasinh}
2956: produces values for these inputs on my Linux box (Bug in the C library?)
1.14 anton 2957:
1.15 anton 2958: @item |@var{float}|>1 (@code{facos}, @code{fasin}, @code{fatanh}):
2959: @code{-55 throw} (Floating-point unidentified fault).
1.14 anton 2960:
1.15 anton 2961: @item integer part of float cannot be represented by @var{d} in @code{f>d}:
2962: @code{-55 throw} (Floating-point unidentified fault).
1.14 anton 2963:
1.15 anton 2964: @item string larger than pictured numeric output area (@code{f.}, @code{fe.}, @code{fs.}):
2965: This does not happen.
2966: @end table
1.14 anton 2967:
2968:
2969:
2970: @c =====================================================================
1.15 anton 2971: @node The optional Locals word set, The optional Memory-Allocation word set, The optional Floating-Point word set, ANS conformance
2972: @section The optional Locals word set
1.14 anton 2973: @c =====================================================================
2974:
2975: @menu
1.15 anton 2976: * locals-idef:: Implementation Defined Options
2977: * locals-ambcond:: Ambiguous Conditions
1.14 anton 2978: @end menu
2979:
2980:
2981: @c ---------------------------------------------------------------------
1.15 anton 2982: @node locals-idef, locals-ambcond, The optional Locals word set, The optional Locals word set
1.14 anton 2983: @subsection Implementation Defined Options
2984: @c ---------------------------------------------------------------------
2985:
2986: @table @i
2987:
1.15 anton 2988: @item maximum number of locals in a definition:
2989: @code{s" #locals" environment? drop .}. Currently 15. This is a lower
2990: bound, e.g., on a 32-bit machine there can be 41 locals of up to 8
2991: characters. The number of locals in a definition is bounded by the size
2992: of locals-buffer, which contains the names of the locals.
1.14 anton 2993:
2994: @end table
2995:
2996:
2997: @c ---------------------------------------------------------------------
1.15 anton 2998: @node locals-ambcond, , locals-idef, The optional Locals word set
1.14 anton 2999: @subsection Ambiguous conditions
3000: @c ---------------------------------------------------------------------
3001:
3002: @table @i
3003:
1.15 anton 3004: @item executing a named local in interpretation state:
3005: @code{-14 throw} (Interpreting a compile-only word).
1.14 anton 3006:
1.15 anton 3007: @item @var{name} not defined by @code{VALUE} or @code{(LOCAL)} (@code{TO}):
3008: @code{-32 throw} (Invalid name argument)
1.14 anton 3009:
3010: @end table
3011:
3012:
3013: @c =====================================================================
1.15 anton 3014: @node The optional Memory-Allocation word set, The optional Programming-Tools word set, The optional Locals word set, ANS conformance
3015: @section The optional Memory-Allocation word set
1.14 anton 3016: @c =====================================================================
3017:
3018: @menu
1.15 anton 3019: * memory-idef:: Implementation Defined Options
1.14 anton 3020: @end menu
3021:
3022:
3023: @c ---------------------------------------------------------------------
1.15 anton 3024: @node memory-idef, , The optional Memory-Allocation word set, The optional Memory-Allocation word set
1.14 anton 3025: @subsection Implementation Defined Options
3026: @c ---------------------------------------------------------------------
3027:
3028: @table @i
3029:
1.15 anton 3030: @item values and meaning of @var{ior}:
3031: The @var{ior}s returned by the file and memory allocation words are
3032: intended as throw codes. They typically are in the range
3033: -512@minus{}-2047 of OS errors. The mapping from OS error numbers to
3034: @var{ior}s is -512@minus{}@var{errno}.
1.14 anton 3035:
3036: @end table
3037:
3038: @c =====================================================================
1.15 anton 3039: @node The optional Programming-Tools word set, The optional Search-Order word set, The optional Memory-Allocation word set, ANS conformance
3040: @section The optional Programming-Tools word set
1.14 anton 3041: @c =====================================================================
3042:
3043: @menu
1.15 anton 3044: * programming-idef:: Implementation Defined Options
3045: * programming-ambcond:: Ambiguous Conditions
1.14 anton 3046: @end menu
3047:
3048:
3049: @c ---------------------------------------------------------------------
1.15 anton 3050: @node programming-idef, programming-ambcond, The optional Programming-Tools word set, The optional Programming-Tools word set
1.14 anton 3051: @subsection Implementation Defined Options
3052: @c ---------------------------------------------------------------------
3053:
3054: @table @i
3055:
1.15 anton 3056: @item ending sequence for input following @code{;code} and @code{code}:
3057: Not implemented (yet).
1.14 anton 3058:
1.15 anton 3059: @item manner of processing input following @code{;code} and @code{code}:
3060: Not implemented (yet).
3061:
3062: @item search order capability for @code{EDITOR} and @code{ASSEMBLER}:
3063: Not implemented (yet). If they were implemented, they would use the
3064: search order wordset.
3065:
3066: @item source and format of display by @code{SEE}:
3067: The source for @code{see} is the intermediate code used by the inner
3068: interpreter. The current @code{see} tries to output Forth source code
3069: as well as possible.
3070:
1.14 anton 3071: @end table
3072:
3073: @c ---------------------------------------------------------------------
1.15 anton 3074: @node programming-ambcond, , programming-idef, The optional Programming-Tools word set
1.14 anton 3075: @subsection Ambiguous conditions
3076: @c ---------------------------------------------------------------------
3077:
3078: @table @i
3079:
1.15 anton 3080: @item deleting the compilation wordlist (@code{FORGET}):
3081: Not implemented (yet).
1.14 anton 3082:
1.15 anton 3083: @item fewer than @var{u}+1 items on the control flow stack (@code{CS-PICK}, @code{CS-ROLL}):
3084: This typically results in an @code{abort"} with a descriptive error
3085: message (may change into a @code{-22 throw} (Control structure mismatch)
3086: in the future). You may also get a memory access error. If you are
3087: unlucky, this ambiguous condition is not caught.
3088:
3089: @item @var{name} can't be found (@code{forget}):
3090: Not implemented (yet).
1.14 anton 3091:
1.15 anton 3092: @item @var{name} not defined via @code{CREATE}:
3093: @code{;code} is not implemented (yet). If it were, it would behave like
3094: @code{DOES>} in this respect, i.e., change the execution semantics of
3095: the last defined word no matter how it was defined.
1.14 anton 3096:
1.15 anton 3097: @item @code{POSTPONE} applied to @code{[IF]}:
3098: After defining @code{: X POSTPONE [IF] ; IMMEDIATE}. @code{X} is
3099: equivalent to @code{[IF]}.
1.14 anton 3100:
1.15 anton 3101: @item reaching the end of the input source before matching @code{[ELSE]} or @code{[THEN]}:
3102: Continue in the same state of conditional compilation in the next outer
3103: input source. Currently there is no warning to the user about this.
1.14 anton 3104:
1.15 anton 3105: @item removing a needed definition (@code{FORGET}):
3106: Not implemented (yet).
1.14 anton 3107:
3108: @end table
3109:
3110:
3111: @c =====================================================================
1.15 anton 3112: @node The optional Search-Order word set, , The optional Programming-Tools word set, ANS conformance
3113: @section The optional Search-Order word set
1.14 anton 3114: @c =====================================================================
3115:
3116: @menu
1.15 anton 3117: * search-idef:: Implementation Defined Options
3118: * search-ambcond:: Ambiguous Conditions
1.14 anton 3119: @end menu
3120:
3121:
3122: @c ---------------------------------------------------------------------
1.15 anton 3123: @node search-idef, search-ambcond, The optional Search-Order word set, The optional Search-Order word set
1.14 anton 3124: @subsection Implementation Defined Options
3125: @c ---------------------------------------------------------------------
3126:
3127: @table @i
3128:
1.15 anton 3129: @item maximum number of word lists in search order:
3130: @code{s" wordlists" environment? drop .}. Currently 16.
3131:
3132: @item minimum search order:
3133: @code{root root}.
1.14 anton 3134:
3135: @end table
3136:
3137: @c ---------------------------------------------------------------------
1.15 anton 3138: @node search-ambcond, , search-idef, The optional Search-Order word set
1.14 anton 3139: @subsection Ambiguous conditions
3140: @c ---------------------------------------------------------------------
3141:
3142: @table @i
3143:
1.15 anton 3144: @item changing the compilation wordlist (during compilation):
3145: The definition is put into the wordlist that is the compilation wordlist
3146: when @code{REVEAL} is executed (by @code{;}, @code{DOES>},
3147: @code{RECURSIVE}, etc.).
1.14 anton 3148:
1.15 anton 3149: @item search order empty (@code{previous}):
3150: @code{abort" Vocstack empty"}.
1.14 anton 3151:
1.15 anton 3152: @item too many word lists in search order (@code{also}):
3153: @code{abort" Vocstack full"}.
1.14 anton 3154:
3155: @end table
1.13 anton 3156:
3157:
1.17 anton 3158: @node Model, Emacs and Gforth, ANS conformance, Top
1.4 anton 3159: @chapter Model
3160:
1.17 anton 3161: @node Emacs and Gforth, Internals, Model, Top
3162: @chapter Emacs and Gforth
1.4 anton 3163:
1.17 anton 3164: Gforth comes with @file{gforth.el}, an improved version of
1.4 anton 3165: @file{forth.el} by Goran Rydqvist (icluded in the TILE package). The
3166: improvements are a better (but still not perfect) handling of
3167: indentation. I have also added comment paragraph filling (@kbd{M-q}),
1.8 anton 3168: commenting (@kbd{C-x \}) and uncommenting (@kbd{C-u C-x \}) regions and
3169: removing debugging tracers (@kbd{C-x ~}, @pxref{Debugging}). I left the
3170: stuff I do not use alone, even though some of it only makes sense for
3171: TILE. To get a description of these features, enter Forth mode and type
3172: @kbd{C-h m}.
1.4 anton 3173:
1.17 anton 3174: In addition, Gforth supports Emacs quite well: The source code locations
1.4 anton 3175: given in error messages, debugging output (from @code{~~}) and failed
3176: assertion messages are in the right format for Emacs' compilation mode
3177: (@pxref{Compilation, , Running Compilations under Emacs, emacs, Emacs
3178: Manual}) so the source location corresponding to an error or other
3179: message is only a few keystrokes away (@kbd{C-x `} for the next error,
3180: @kbd{C-c C-c} for the error under the cursor).
3181:
3182: Also, if you @code{include} @file{etags.fs}, a new @file{TAGS} file
3183: (@pxref{Tags, , Tags Tables, emacs, Emacs Manual}) will be produced that
3184: contains the definitions of all words defined afterwards. You can then
3185: find the source for a word using @kbd{M-.}. Note that emacs can use
1.17 anton 3186: several tags files at the same time (e.g., one for the Gforth sources
1.4 anton 3187: and one for your program).
3188:
3189: To get all these benefits, add the following lines to your @file{.emacs}
3190: file:
3191:
3192: @example
3193: (autoload 'forth-mode "gforth.el")
3194: (setq auto-mode-alist (cons '("\\.fs\\'" . forth-mode) auto-mode-alist))
3195: @end example
3196:
1.17 anton 3197: @node Internals, Bugs, Emacs and Gforth, Top
1.3 anton 3198: @chapter Internals
3199:
1.17 anton 3200: Reading this section is not necessary for programming with Gforth. It
3201: should be helpful for finding your way in the Gforth sources.
1.3 anton 3202:
1.4 anton 3203: @menu
3204: * Portability::
3205: * Threading::
3206: * Primitives::
3207: * System Architecture::
1.17 anton 3208: * Performance::
1.4 anton 3209: @end menu
3210:
3211: @node Portability, Threading, Internals, Internals
1.3 anton 3212: @section Portability
3213:
3214: One of the main goals of the effort is availability across a wide range
3215: of personal machines. fig-Forth, and, to a lesser extent, F83, achieved
3216: this goal by manually coding the engine in assembly language for several
3217: then-popular processors. This approach is very labor-intensive and the
3218: results are short-lived due to progress in computer architecture.
3219:
3220: Others have avoided this problem by coding in C, e.g., Mitch Bradley
3221: (cforth), Mikael Patel (TILE) and Dirk Zoller (pfe). This approach is
3222: particularly popular for UNIX-based Forths due to the large variety of
3223: architectures of UNIX machines. Unfortunately an implementation in C
3224: does not mix well with the goals of efficiency and with using
3225: traditional techniques: Indirect or direct threading cannot be expressed
3226: in C, and switch threading, the fastest technique available in C, is
3227: significantly slower. Another problem with C is that it's very
3228: cumbersome to express double integer arithmetic.
3229:
3230: Fortunately, there is a portable language that does not have these
3231: limitations: GNU C, the version of C processed by the GNU C compiler
3232: (@pxref{C Extensions, , Extensions to the C Language Family, gcc.info,
3233: GNU C Manual}). Its labels as values feature (@pxref{Labels as Values, ,
3234: Labels as Values, gcc.info, GNU C Manual}) makes direct and indirect
3235: threading possible, its @code{long long} type (@pxref{Long Long, ,
3236: Double-Word Integers, gcc.info, GNU C Manual}) corresponds to Forths
3237: double numbers. GNU C is available for free on all important (and many
3238: unimportant) UNIX machines, VMS, 80386s running MS-DOS, the Amiga, and
3239: the Atari ST, so a Forth written in GNU C can run on all these
1.17 anton 3240: machines.
1.3 anton 3241:
3242: Writing in a portable language has the reputation of producing code that
3243: is slower than assembly. For our Forth engine we repeatedly looked at
3244: the code produced by the compiler and eliminated most compiler-induced
3245: inefficiencies by appropriate changes in the source-code.
3246:
3247: However, register allocation cannot be portably influenced by the
3248: programmer, leading to some inefficiencies on register-starved
3249: machines. We use explicit register declarations (@pxref{Explicit Reg
3250: Vars, , Variables in Specified Registers, gcc.info, GNU C Manual}) to
3251: improve the speed on some machines. They are turned on by using the
3252: @code{gcc} switch @code{-DFORCE_REG}. Unfortunately, this feature not
3253: only depends on the machine, but also on the compiler version: On some
3254: machines some compiler versions produce incorrect code when certain
3255: explicit register declarations are used. So by default
3256: @code{-DFORCE_REG} is not used.
3257:
1.4 anton 3258: @node Threading, Primitives, Portability, Internals
1.3 anton 3259: @section Threading
3260:
3261: GNU C's labels as values extension (available since @code{gcc-2.0},
3262: @pxref{Labels as Values, , Labels as Values, gcc.info, GNU C Manual})
3263: makes it possible to take the address of @var{label} by writing
3264: @code{&&@var{label}}. This address can then be used in a statement like
3265: @code{goto *@var{address}}. I.e., @code{goto *&&x} is the same as
3266: @code{goto x}.
3267:
3268: With this feature an indirect threaded NEXT looks like:
3269: @example
3270: cfa = *ip++;
3271: ca = *cfa;
3272: goto *ca;
3273: @end example
3274: For those unfamiliar with the names: @code{ip} is the Forth instruction
3275: pointer; the @code{cfa} (code-field address) corresponds to ANS Forths
3276: execution token and points to the code field of the next word to be
3277: executed; The @code{ca} (code address) fetched from there points to some
3278: executable code, e.g., a primitive or the colon definition handler
3279: @code{docol}.
3280:
3281: Direct threading is even simpler:
3282: @example
3283: ca = *ip++;
3284: goto *ca;
3285: @end example
3286:
3287: Of course we have packaged the whole thing neatly in macros called
3288: @code{NEXT} and @code{NEXT1} (the part of NEXT after fetching the cfa).
3289:
1.4 anton 3290: @menu
3291: * Scheduling::
3292: * Direct or Indirect Threaded?::
3293: * DOES>::
3294: @end menu
3295:
3296: @node Scheduling, Direct or Indirect Threaded?, Threading, Threading
1.3 anton 3297: @subsection Scheduling
3298:
3299: There is a little complication: Pipelined and superscalar processors,
3300: i.e., RISC and some modern CISC machines can process independent
3301: instructions while waiting for the results of an instruction. The
3302: compiler usually reorders (schedules) the instructions in a way that
3303: achieves good usage of these delay slots. However, on our first tries
3304: the compiler did not do well on scheduling primitives. E.g., for
3305: @code{+} implemented as
3306: @example
3307: n=sp[0]+sp[1];
3308: sp++;
3309: sp[0]=n;
3310: NEXT;
3311: @end example
3312: the NEXT comes strictly after the other code, i.e., there is nearly no
3313: scheduling. After a little thought the problem becomes clear: The
3314: compiler cannot know that sp and ip point to different addresses (and
1.4 anton 3315: the version of @code{gcc} we used would not know it even if it was
3316: possible), so it could not move the load of the cfa above the store to
3317: the TOS. Indeed the pointers could be the same, if code on or very near
3318: the top of stack were executed. In the interest of speed we chose to
3319: forbid this probably unused ``feature'' and helped the compiler in
3320: scheduling: NEXT is divided into the loading part (@code{NEXT_P1}) and
3321: the goto part (@code{NEXT_P2}). @code{+} now looks like:
1.3 anton 3322: @example
3323: n=sp[0]+sp[1];
3324: sp++;
3325: NEXT_P1;
3326: sp[0]=n;
3327: NEXT_P2;
3328: @end example
1.4 anton 3329: This can be scheduled optimally by the compiler.
1.3 anton 3330:
3331: This division can be turned off with the switch @code{-DCISC_NEXT}. This
3332: switch is on by default on machines that do not profit from scheduling
3333: (e.g., the 80386), in order to preserve registers.
3334:
1.4 anton 3335: @node Direct or Indirect Threaded?, DOES>, Scheduling, Threading
1.3 anton 3336: @subsection Direct or Indirect Threaded?
3337:
3338: Both! After packaging the nasty details in macro definitions we
3339: realized that we could switch between direct and indirect threading by
3340: simply setting a compilation flag (@code{-DDIRECT_THREADED}) and
3341: defining a few machine-specific macros for the direct-threading case.
3342: On the Forth level we also offer access words that hide the
3343: differences between the threading methods (@pxref{Threading Words}).
3344:
3345: Indirect threading is implemented completely
3346: machine-independently. Direct threading needs routines for creating
3347: jumps to the executable code (e.g. to docol or dodoes). These routines
3348: are inherently machine-dependent, but they do not amount to many source
3349: lines. I.e., even porting direct threading to a new machine is a small
3350: effort.
3351:
1.4 anton 3352: @node DOES>, , Direct or Indirect Threaded?, Threading
1.3 anton 3353: @subsection DOES>
3354: One of the most complex parts of a Forth engine is @code{dodoes}, i.e.,
3355: the chunk of code executed by every word defined by a
3356: @code{CREATE}...@code{DOES>} pair. The main problem here is: How to find
3357: the Forth code to be executed, i.e. the code after the @code{DOES>} (the
3358: DOES-code)? There are two solutions:
3359:
3360: In fig-Forth the code field points directly to the dodoes and the
3361: DOES-code address is stored in the cell after the code address
3362: (i.e. at cfa cell+). It may seem that this solution is illegal in the
3363: Forth-79 and all later standards, because in fig-Forth this address
3364: lies in the body (which is illegal in these standards). However, by
3365: making the code field larger for all words this solution becomes legal
3366: again. We use this approach for the indirect threaded version. Leaving
3367: a cell unused in most words is a bit wasteful, but on the machines we
3368: are targetting this is hardly a problem. The other reason for having a
3369: code field size of two cells is to avoid having different image files
1.4 anton 3370: for direct and indirect threaded systems (@pxref{System Architecture}).
1.3 anton 3371:
3372: The other approach is that the code field points or jumps to the cell
3373: after @code{DOES}. In this variant there is a jump to @code{dodoes} at
3374: this address. @code{dodoes} can then get the DOES-code address by
3375: computing the code address, i.e., the address of the jump to dodoes,
3376: and add the length of that jump field. A variant of this is to have a
3377: call to @code{dodoes} after the @code{DOES>}; then the return address
3378: (which can be found in the return register on RISCs) is the DOES-code
3379: address. Since the two cells available in the code field are usually
3380: used up by the jump to the code address in direct threading, we use
3381: this approach for direct threading. We did not want to add another
3382: cell to the code field.
3383:
1.4 anton 3384: @node Primitives, System Architecture, Threading, Internals
1.3 anton 3385: @section Primitives
3386:
1.4 anton 3387: @menu
3388: * Automatic Generation::
3389: * TOS Optimization::
3390: * Produced code::
3391: @end menu
3392:
3393: @node Automatic Generation, TOS Optimization, Primitives, Primitives
1.3 anton 3394: @subsection Automatic Generation
3395:
3396: Since the primitives are implemented in a portable language, there is no
3397: longer any need to minimize the number of primitives. On the contrary,
3398: having many primitives is an advantage: speed. In order to reduce the
3399: number of errors in primitives and to make programming them easier, we
3400: provide a tool, the primitive generator (@file{prims2x.fs}), that
3401: automatically generates most (and sometimes all) of the C code for a
3402: primitive from the stack effect notation. The source for a primitive
3403: has the following form:
3404:
3405: @format
3406: @var{Forth-name} @var{stack-effect} @var{category} [@var{pronounc.}]
3407: [@code{""}@var{glossary entry}@code{""}]
3408: @var{C code}
3409: [@code{:}
3410: @var{Forth code}]
3411: @end format
3412:
3413: The items in brackets are optional. The category and glossary fields
3414: are there for generating the documentation, the Forth code is there
3415: for manual implementations on machines without GNU C. E.g., the source
3416: for the primitive @code{+} is:
3417: @example
3418: + n1 n2 -- n core plus
3419: n = n1+n2;
3420: @end example
3421:
3422: This looks like a specification, but in fact @code{n = n1+n2} is C
3423: code. Our primitive generation tool extracts a lot of information from
3424: the stack effect notations@footnote{We use a one-stack notation, even
3425: though we have separate data and floating-point stacks; The separate
3426: notation can be generated easily from the unified notation.}: The number
3427: of items popped from and pushed on the stack, their type, and by what
3428: name they are referred to in the C code. It then generates a C code
3429: prelude and postlude for each primitive. The final C code for @code{+}
3430: looks like this:
3431:
3432: @example
3433: I_plus: /* + ( n1 n2 -- n ) */ /* label, stack effect */
3434: /* */ /* documentation */
1.4 anton 3435: @{
1.3 anton 3436: DEF_CA /* definition of variable ca (indirect threading) */
3437: Cell n1; /* definitions of variables */
3438: Cell n2;
3439: Cell n;
3440: n1 = (Cell) sp[1]; /* input */
3441: n2 = (Cell) TOS;
3442: sp += 1; /* stack adjustment */
3443: NAME("+") /* debugging output (with -DDEBUG) */
1.4 anton 3444: @{
1.3 anton 3445: n = n1+n2; /* C code taken from the source */
1.4 anton 3446: @}
1.3 anton 3447: NEXT_P1; /* NEXT part 1 */
3448: TOS = (Cell)n; /* output */
3449: NEXT_P2; /* NEXT part 2 */
1.4 anton 3450: @}
1.3 anton 3451: @end example
3452:
3453: This looks long and inefficient, but the GNU C compiler optimizes quite
3454: well and produces optimal code for @code{+} on, e.g., the R3000 and the
3455: HP RISC machines: Defining the @code{n}s does not produce any code, and
3456: using them as intermediate storage also adds no cost.
3457:
3458: There are also other optimizations, that are not illustrated by this
3459: example: Assignments between simple variables are usually for free (copy
3460: propagation). If one of the stack items is not used by the primitive
3461: (e.g. in @code{drop}), the compiler eliminates the load from the stack
3462: (dead code elimination). On the other hand, there are some things that
3463: the compiler does not do, therefore they are performed by
3464: @file{prims2x.fs}: The compiler does not optimize code away that stores
3465: a stack item to the place where it just came from (e.g., @code{over}).
3466:
3467: While programming a primitive is usually easy, there are a few cases
3468: where the programmer has to take the actions of the generator into
3469: account, most notably @code{?dup}, but also words that do not (always)
3470: fall through to NEXT.
3471:
1.4 anton 3472: @node TOS Optimization, Produced code, Automatic Generation, Primitives
1.3 anton 3473: @subsection TOS Optimization
3474:
3475: An important optimization for stack machine emulators, e.g., Forth
3476: engines, is keeping one or more of the top stack items in
1.4 anton 3477: registers. If a word has the stack effect @var{in1}...@var{inx} @code{--}
3478: @var{out1}...@var{outy}, keeping the top @var{n} items in registers
1.3 anton 3479: @itemize
3480: @item
3481: is better than keeping @var{n-1} items, if @var{x>=n} and @var{y>=n},
3482: due to fewer loads from and stores to the stack.
3483: @item is slower than keeping @var{n-1} items, if @var{x<>y} and @var{x<n} and
3484: @var{y<n}, due to additional moves between registers.
3485: @end itemize
3486:
3487: In particular, keeping one item in a register is never a disadvantage,
3488: if there are enough registers. Keeping two items in registers is a
3489: disadvantage for frequent words like @code{?branch}, constants,
3490: variables, literals and @code{i}. Therefore our generator only produces
3491: code that keeps zero or one items in registers. The generated C code
3492: covers both cases; the selection between these alternatives is made at
3493: C-compile time using the switch @code{-DUSE_TOS}. @code{TOS} in the C
3494: code for @code{+} is just a simple variable name in the one-item case,
3495: otherwise it is a macro that expands into @code{sp[0]}. Note that the
3496: GNU C compiler tries to keep simple variables like @code{TOS} in
3497: registers, and it usually succeeds, if there are enough registers.
3498:
3499: The primitive generator performs the TOS optimization for the
3500: floating-point stack, too (@code{-DUSE_FTOS}). For floating-point
3501: operations the benefit of this optimization is even larger:
3502: floating-point operations take quite long on most processors, but can be
3503: performed in parallel with other operations as long as their results are
3504: not used. If the FP-TOS is kept in a register, this works. If
3505: it is kept on the stack, i.e., in memory, the store into memory has to
3506: wait for the result of the floating-point operation, lengthening the
3507: execution time of the primitive considerably.
3508:
3509: The TOS optimization makes the automatic generation of primitives a
3510: bit more complicated. Just replacing all occurrences of @code{sp[0]} by
3511: @code{TOS} is not sufficient. There are some special cases to
3512: consider:
3513: @itemize
3514: @item In the case of @code{dup ( w -- w w )} the generator must not
3515: eliminate the store to the original location of the item on the stack,
3516: if the TOS optimization is turned on.
1.4 anton 3517: @item Primitives with stack effects of the form @code{--}
3518: @var{out1}...@var{outy} must store the TOS to the stack at the start.
3519: Likewise, primitives with the stack effect @var{in1}...@var{inx} @code{--}
1.3 anton 3520: must load the TOS from the stack at the end. But for the null stack
3521: effect @code{--} no stores or loads should be generated.
3522: @end itemize
3523:
1.4 anton 3524: @node Produced code, , TOS Optimization, Primitives
1.3 anton 3525: @subsection Produced code
3526:
3527: To see what assembly code is produced for the primitives on your machine
3528: with your compiler and your flag settings, type @code{make engine.s} and
1.4 anton 3529: look at the resulting file @file{engine.s}.
1.3 anton 3530:
1.17 anton 3531: @node System Architecture, Performance, Primitives, Internals
1.3 anton 3532: @section System Architecture
3533:
3534: Our Forth system consists not only of primitives, but also of
3535: definitions written in Forth. Since the Forth compiler itself belongs
3536: to those definitions, it is not possible to start the system with the
3537: primitives and the Forth source alone. Therefore we provide the Forth
3538: code as an image file in nearly executable form. At the start of the
3539: system a C routine loads the image file into memory, sets up the
3540: memory (stacks etc.) according to information in the image file, and
3541: starts executing Forth code.
3542:
3543: The image file format is a compromise between the goals of making it
3544: easy to generate image files and making them portable. The easiest way
3545: to generate an image file is to just generate a memory dump. However,
3546: this kind of image file cannot be used on a different machine, or on
3547: the next version of the engine on the same machine, it even might not
3548: work with the same engine compiled by a different version of the C
3549: compiler. We would like to have as few versions of the image file as
3550: possible, because we do not want to distribute many versions of the
3551: same image file, and to make it easy for the users to use their image
3552: files on many machines. We currently need to create a different image
3553: file for machines with different cell sizes and different byte order
1.17 anton 3554: (little- or big-endian)@footnote{We are considering adding information to the
1.3 anton 3555: image file that enables the loader to change the byte order.}.
3556:
3557: Forth code that is going to end up in a portable image file has to
1.4 anton 3558: comply to some restrictions: addresses have to be stored in memory with
3559: special words (@code{A!}, @code{A,}, etc.) in order to make the code
3560: relocatable. Cells, floats, etc., have to be stored at the natural
3561: alignment boundaries@footnote{E.g., store floats (8 bytes) at an address
3562: dividable by~8. This happens automatically in our system when you use
3563: the ANS Forth alignment words.}, in order to avoid alignment faults on
3564: machines with stricter alignment. The image file is produced by a
3565: metacompiler (@file{cross.fs}).
1.3 anton 3566:
3567: So, unlike the image file of Mitch Bradleys @code{cforth}, our image
3568: file is not directly executable, but has to undergo some manipulations
3569: during loading. Address relocation is performed at image load-time, not
3570: at run-time. The loader also has to replace tokens standing for
3571: primitive calls with the appropriate code-field addresses (or code
3572: addresses in the case of direct threading).
1.4 anton 3573:
1.17 anton 3574: @node Performance, , System Architecture, Internals
3575: @section Performance
3576:
3577: On RISCs the Gforth engine is very close to optimal; i.e., it is usually
3578: impossible to write a significantly faster engine.
3579:
3580: On register-starved machines like the 386 architecture processors
3581: improvements are possible, because @code{gcc} does not utilize the
3582: registers as well as a human, even with explicit register declarations;
3583: e.g., Bernd Beuster wrote a Forth system fragment in assembly language
3584: and hand-tuned it for the 486; this system is 1.19 times faster on the
3585: Sieve benchmark on a 486DX2/66 than Gforth compiled with
3586: @code{gcc-2.6.3} with @code{-DFORCE_REG}.
3587:
3588: However, this potential advantage of assembly language implementations
3589: is not necessarily realized in complete Forth systems: We compared
3590: Gforth (compiled with @code{gcc-2.6.3} and @code{-DFORCE_REG}) with
1.18 anton 3591: Win32Forth 1.2093 and LMI's NT Forth (Beta, May 1994), two systems
3592: written in assembly, and with two systems written in C: PFE-0.9.11
3593: (compiled with @code{gcc-2.6.3} with the default configuration for
3594: Linux: @code{-O2 -fomit-frame-pointer -DUSE_REGS}) and ThisForth Beta
3595: (compiled with gcc-2.6.3 -O3 -fomit-frame-pointer). We benchmarked
3596: Gforth, PFE and ThisForth on a 486DX2/66 under Linux. Kenneth O'Heskin
3597: kindly provided the results for Win32Forth and NT Forth on a 486DX2/66
3598: with similar memory performance under Windows NT.
1.17 anton 3599:
3600: We used four small benchmarks: the ubiquitous Sieve; bubble-sorting and
3601: matrix multiplication come from the Stanford integer benchmarks and have
3602: been translated into Forth by Martin Fraeman; we used the versions
3603: included in the TILE Forth package; and a recursive Fibonacci number
3604: computation for benchmark calling performance. The following table shows
3605: the time taken for the benchmarks scaled by the time taken by Gforth (in
3606: other words, it shows the speedup factor that Gforth achieved over the
3607: other systems).
3608:
3609: @example
3610: relative Win32- NT This-
3611: time Gforth Forth Forth PFE Forth
3612: sieve 1.00 1.30 1.07 1.67 2.98
3613: bubble 1.00 1.30 1.40 1.66
3614: matmul 1.00 1.40 1.29 2.24
3615: fib 1.00 1.44 1.26 1.82 2.82
3616: @end example
3617:
3618: You may find the good performance of Gforth compared with the systems
3619: written in assembly language quite surprising. One important reason for
3620: the disappointing performance of these systems is probably that they are
3621: not written optimally for the 486 (e.g., they use the @code{lods}
3622: instruction). In addition, Win32Forth uses a comfortable, but costly
3623: method for relocating the Forth image: like @code{cforth}, it computes
3624: the actual addresses at run time, resulting in two address computations
3625: per NEXT (@pxref{System Architecture}).
3626:
3627: The speedup of Gforth over PFE and ThisForth can be easily explained
3628: with the self-imposed restriction to standard C (although the measured
3629: implementation of PFE uses a GNU C extension: global register
3630: variables), which makes efficient threading impossible. Moreover,
3631: current C compilers have a hard time optimizing other aspects of the
3632: ThisForth source.
3633:
3634: Note that the performance of Gforth on 386 architecture processors
3635: varies widely with the version of @code{gcc} used. E.g., @code{gcc-2.5.8}
3636: failed to allocate any of the virtual machine registers into real
3637: machine registers by itself and would not work correctly with explicit
3638: register declarations, giving a 1.3 times slower engine (on a 486DX2/66
3639: running the Sieve) than the one measured above.
3640:
1.4 anton 3641: @node Bugs, Pedigree, Internals, Top
3642: @chapter Bugs
3643:
1.17 anton 3644: Known bugs are described in the file BUGS in the Gforth distribution.
3645:
3646: If you find a bug, please send a bug report to !!. A bug report should
3647: describe the Gforth version used (it is announced at the start of an
3648: interactive Gforth session), the machine and operating system (on Unix
3649: systems you can use @code{uname -a} to produce this information), the
3650: installation options (!! a way to find them out), and a complete list of
3651: changes you (or your installer) have made to the Gforth sources (if
3652: any); it should contain a program (or a sequence of keyboard commands)
3653: that reproduces the bug and a description of what you think constitutes
3654: the buggy behaviour.
3655:
3656: For a thorough guide on reporting bugs read @ref{Bug Reporting, , How
3657: to Report Bugs, gcc.info, GNU C Manual}.
3658:
3659:
1.4 anton 3660: @node Pedigree, Word Index, Bugs, Top
3661: @chapter Pedigree
3662:
1.17 anton 3663: Gforth descends from BigForth (1993) and fig-Forth. Gforth and PFE (by
3664: Dirk Zoller) will cross-fertilize each other. Of course, a significant part of the design of Gforth was prescribed by ANS Forth.
3665:
3666: Bernd Paysan wrote BigForth, a child of VolksForth.
3667:
3668: VolksForth descends from F83. !! Authors? When?
3669:
3670: Laxen and Perry wrote F83 as a model implementation of the
3671: Forth-83 standard. !! Pedigree? When?
3672:
3673: A team led by Bill Ragsdale implemented fig-Forth on many processors in
3674: 1979. Dean Sanderson and Bill Ragsdale developed the original
3675: implementation of fig-Forth based on microForth.
3676:
3677: !! microForth pedigree
3678:
3679: A part of the information in this section comes from @cite{The Evolution
3680: of Forth} by Elizabeth D. Rather, Donald R. Colburn and Charles
3681: H. Moore, presented at the HOPL-II conference and preprinted in SIGPLAN
3682: Notices 28(3), 1993. You can find more historical and genealogical
3683: information about Forth there.
3684:
1.4 anton 3685: @node Word Index, Node Index, Pedigree, Top
3686: @chapter Word Index
3687:
1.18 anton 3688: This index is as incomplete as the manual. Each word is listed with
3689: stack effect and wordset.
1.17 anton 3690:
3691: @printindex fn
3692:
1.4 anton 3693: @node Node Index, , Word Index, Top
3694: @chapter Node Index
1.17 anton 3695:
3696: This index is even less complete than the manual.
1.1 anton 3697:
3698: @contents
3699: @bye
3700:
FreeBSD-CVSweb <freebsd-cvsweb@FreeBSD.org>