4 Changes in version 1.75a, 2002-10-24:
6 * Bison can no longer be built by a K&R C compiler; it requires C89 or
7 later to be built. This change originally took place a few versions
8 ago, but nobody noticed until we recently asked someone to try
9 building Bison with a K&R C compiler.
11 Changes in version 1.75, 2002-10-14:
13 * Bison should now work on 64-bit hosts.
15 * Indonesian translation thanks to Tedi Heriyanto.
18 Fix spurious parse errors.
21 Some people redefine yyerror to steal yyparse' private variables.
22 Reenable this trick until an official feature replaces it.
25 In agreement with POSIX and with other Yaccs, leaving a default
26 action is valid when $$ is untyped, and $1 typed:
30 but the converse remains an error:
34 * Values of mid-rule actions
37 foo: { ... } { $$ = $1; } ...
39 was incorrectly rejected: $1 is defined in the second mid-rule
40 action, and is equal to the $$ of the first mid-rule action.
42 Changes in version 1.50, 2002-10-04:
47 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
48 almost any context-free grammar, ambiguous or not. The new declarations
49 %dprec and %merge on grammar rules allow parse-time resolution of
50 ambiguities. Contributed by Paul Hilfinger.
52 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
53 like the Alpha, so please stick to 32-bit hosts for now.
56 When not in Yacc compatibility mode, when the output file was not
57 specified, running `bison foo/bar.y' created `foo/bar.c'. It
61 The undefined token was systematically mapped to 2 which prevented
62 the use of 2 by the user. This is no longer the case.
64 * Unknown token numbers
65 If yylex returned an out of range value, yyparse could die. This is
69 According to POSIX, the error token must be 256.
70 Bison extends this requirement by making it a preference: *if* the
71 user specified that one of her tokens is numbered 256, then error
72 will be mapped onto another number.
74 * Verbose error messages
75 They no longer report `..., expecting error or...' for states where
76 error recovery is possible.
79 Defaults to `$end' instead of `$'.
81 * Error recovery now conforms to documentation and to POSIX
82 When a Bison-generated parser encounters a syntax error, it now pops
83 the stack until it finds a state that allows shifting the error
84 token. Formerly, it popped the stack until it found a state that
85 allowed some non-error action other than a default reduction on the
86 error token. The new behavior has long been the documented behavior,
87 and has long been required by POSIX. For more details, please see
88 <http://mail.gnu.org/pipermail/bug-bison/2002-May/001452.html>.
91 Popped tokens and nonterminals are now reported.
94 Larger grammars are now supported (larger token numbers, larger grammar
95 size (= sum of the LHS and RHS lengths), larger LALR tables).
96 Formerly, many of these numbers ran afoul of 16-bit limits;
97 now these limits are 32 bits on most hosts.
99 * Explicit initial rule
100 Bison used to play hacks with the initial rule, which the user does
101 not write. It is now explicit, and visible in the reports and
105 Before, Bison reported the useless rules, but, although not used,
106 included them in the parsers. They are now actually removed.
108 * Useless rules, useless nonterminals
109 They are now reported, as a warning, with their locations.
111 * Rules never reduced
112 Rules that can never be reduced because of conflicts are now
115 * Incorrect `Token not used'
118 %token useless useful
120 exp: '0' %prec useful;
122 where a token was used to set the precedence of the last rule,
123 bison reported both `useful' and `useless' as useless tokens.
125 * Revert the C++ namespace changes introduced in 1.31
126 as they caused too many portability hassles.
129 By an accident of design, the default computation of @$ was
130 performed after another default computation was performed: @$ = @1.
131 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
132 the computation of @$.
135 The token end of file may be specified by the user, in which case,
136 the user symbol is used in the reports, the graphs, and the verbose
137 error messages instead of `$end', which remains being the default.
141 %token YYEOF 0 "end of file"
144 This old option, which has been broken for ages, is removed.
147 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
148 Croatian, thanks to Denis Lackovic.
150 * Incorrect token definitions
151 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
153 * Token definitions as enums
154 Tokens are output both as the traditional #define's, and, provided
155 the compiler supports ANSI C or is a C++ compiler, as enums.
156 This lets debuggers display names instead of integers.
159 In addition to --verbose, bison supports --report=THINGS, which
160 produces additional information:
162 complete the core item sets with their closure
164 explicitly associate lookaheads to items
166 describe shift/reduce conflicts solving.
167 Bison used to systematically output this information on top of
168 the report. Solved conflicts are now attached to their states.
171 Previous versions don't complain when there is a type clash on
172 the default action if the rule has a mid-rule action, such as in:
180 * GNU M4 is now required when using Bison.
182 Changes in version 1.35, 2002-03-25:
185 Some projects use Bison's C parser with C++ compilers, and define
186 YYSTYPE as a class. The recent adjustment of C parsers for data
187 alignment and 64 bit architectures made this impossible.
189 Because for the time being no real solution for C++ parser
190 generation exists, kludges were implemented in the parser to
191 maintain this use. In the future, when Bison has C++ parsers, this
192 kludge will be disabled.
194 This kludge also addresses some C++ problems when the stack was
197 Changes in version 1.34, 2002-03-12:
199 * File name clashes are detected
200 $ bison foo.y -d -o foo.x
201 fatal error: header and parser would both be named `foo.x'
203 * A missing `;' at the end of a rule triggers a warning
204 In accordance with POSIX, and in agreement with other
205 Yacc implementations, Bison will mandate this semicolon in the near
206 future. This eases the implementation of a Bison parser of Bison
207 grammars by making this grammar LALR(1) instead of LR(2). To
208 facilitate the transition, this release introduces a warning.
210 * Revert the C++ namespace changes introduced in 1.31, as they caused too
211 many portability hassles.
213 * DJGPP support added.
215 * Fix test suite portability problems.
217 Changes in version 1.33, 2002-02-07:
220 Groff could not be compiled for the definition of size_t was lacking
221 under some conditions.
226 Changes in version 1.32, 2002-01-23:
228 * Fix Yacc output file names
232 * Italian, Dutch translations
234 Changes in version 1.31, 2002-01-14:
238 * GNU Gettext and %expect
239 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
240 Bison dies on incorrect %expectations, we fear there will be
241 too many bug reports for Gettext, so _for the time being_, %expect
242 does not trigger an error when the input file is named `plural.y'.
244 * Use of alloca in parsers
245 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
246 malloc exclusively. Since 1.29, but was not NEWS'ed.
248 alloca is used only when compiled with GCC, to avoid portability
251 * When the generated parser lacks debugging code, YYDEBUG is now 0
252 (as POSIX requires) instead of being undefined.
255 Bison has always permitted actions such as { $$ = $1 }: it adds the
256 ending semicolon. Now if in Yacc compatibility mode, the semicolon
257 is no longer output: one has to write { $$ = $1; }.
259 * Better C++ compliance
260 The output parsers try to respect C++ namespaces.
261 [This turned out to be a failed experiment, and it was reverted later.]
264 Fixed bugs when reporting useless nonterminals.
267 The parsers work properly on 64 bit hosts.
270 Some calls to strerror resulted in scrambled or missing error messages.
273 When the number of shift/reduce conflicts is correct, don't issue
276 * The verbose report includes the rule line numbers.
278 * Rule line numbers are fixed in traces.
280 * Swedish translation
283 Verbose parse error messages from the parsers are better looking.
284 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
285 Now: parse error: unexpected '/', expecting "number" or '-' or '('
287 * Fixed parser memory leaks.
288 When the generated parser was using malloc to extend its stacks, the
289 previous allocations were not freed.
291 * Fixed verbose output file.
292 Some newlines were missing.
293 Some conflicts in state descriptions were missing.
295 * Fixed conflict report.
296 Option -v was needed to get the result.
300 Mismatches are errors, not warnings.
302 * Fixed incorrect processing of some invalid input.
304 * Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
306 * Fixed some typos in the documentation.
308 * %token MY_EOF 0 is supported.
309 Before, MY_EOF was silently renumbered as 257.
311 * doc/refcard.tex is updated.
313 * %output, %file-prefix, %name-prefix.
317 New, aliasing `--output-file'.
319 Changes in version 1.30, 2001-10-26:
321 * `--defines' and `--graph' have now an optionnal argument which is the
322 output file name. `-d' and `-g' do not change, they do not take any
325 * `%source_extension' and `%header_extension' are removed, failed
330 Changes in version 1.29, 2001-09-07:
332 * The output file does not define const, as this caused problems when used
333 with common autoconfiguration schemes. If you still use ancient compilers
334 that lack const, compile with the equivalent of the C compiler option
335 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
337 * Added `-g' and `--graph'.
339 * The Bison manual is now distributed under the terms of the GNU FDL.
341 * The input and the output files has automatically a similar extension.
343 * Russian translation added.
345 * NLS support updated; should hopefully be less troublesome.
347 * Added the old Bison reference card.
349 * Added `--locations' and `%locations'.
351 * Added `-S' and `--skeleton'.
353 * `%raw', `-r', `--raw' is disabled.
355 * Special characters are escaped when output. This solves the problems
356 of the #line lines with path names including backslashes.
359 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
360 `%debug', `%source_extension' and `%header_extension'.
363 Automatic location tracking.
365 Changes in version 1.28, 1999-07-06:
367 * Should compile better now with K&R compilers.
371 * Fixed a problem with escaping the double quote character.
373 * There is now a FAQ.
375 Changes in version 1.27:
377 * The make rule which prevented bison.simple from being created on
378 some systems has been fixed.
380 Changes in version 1.26:
382 * Bison now uses automake.
384 * New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
386 * Token numbers now start at 257 as previously documented, not 258.
388 * Bison honors the TMPDIR environment variable.
390 * A couple of buffer overruns have been fixed.
392 * Problems when closing files should now be reported.
394 * Generated parsers should now work even on operating systems which do
395 not provide alloca().
397 Changes in version 1.25, 1995-10-16:
399 * Errors in the input grammar are not fatal; Bison keeps reading
400 the grammar file, and reports all the errors found in it.
402 * Tokens can now be specified as multiple-character strings: for
403 example, you could use "<=" for a token which looks like <=, instead
404 of chosing a name like LESSEQ.
406 * The %token_table declaration says to write a table of tokens (names
407 and numbers) into the parser file. The yylex function can use this
408 table to recognize multiple-character string tokens, or for other
411 * The %no_lines declaration says not to generate any #line preprocessor
412 directives in the parser file.
414 * The %raw declaration says to use internal Bison token numbers, not
415 Yacc-compatible token numbers, when token names are defined as macros.
417 * The --no-parser option produces the parser tables without including
418 the parser engine; a project can now use its own parser engine.
419 The actions go into a separate file called NAME.act, in the form of
420 a switch statement body.
422 Changes in version 1.23:
424 The user can define YYPARSE_PARAM as the name of an argument to be
425 passed into yyparse. The argument should have type void *. It should
426 actually point to an object. Grammar actions can access the variable
427 by casting it to the proper pointer type.
429 Line numbers in output file corrected.
431 Changes in version 1.22:
435 Changes in version 1.20:
437 Output file does not redefine const for C++.
445 Copyright (C) 2001, 2002 Free Software Foundation, Inc.
447 This file is part of GNU Autoconf.
449 GNU Autoconf is free software; you can redistribute it and/or modify
450 it under the terms of the GNU General Public License as published by
451 the Free Software Foundation; either version 2, or (at your option)
454 GNU Autoconf is distributed in the hope that it will be useful,
455 but WITHOUT ANY WARRANTY; without even the implied warranty of
456 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
457 GNU General Public License for more details.
459 You should have received a copy of the GNU General Public License
460 along with autoconf; see the file COPYING. If not, write to
461 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
462 Boston, MA 02111-1307, USA.