4 Changes in version 2.0a, 2005-05-22:
6 * Bison-generated C parsers use the _ macro to translate strings from
7 English to the user's language, e.g., _("syntax error"). By default,
8 _ is defined to be a no-op macro so the strings are not translated.
10 * When generating verbose diagnostics, Bison-generated C parsers no longer
11 quote the literal strings associated with tokens. For example, for
12 a syntax error associated with '%token NUM "number"' they might
13 print 'syntax error, unexpected number' instead of 'syntax error,
16 Changes in version 2.0, 2004-12-25:
18 * Possibly-incompatible changes
20 - Bison-generated parsers no longer default to using the alloca function
21 (when available) to extend the parser stack, due to widespread
22 problems in unchecked stack-overflow detection. You can "#define
23 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
24 the manual to determine safe values for YYMAXDEPTH in that case.
26 - Error token location.
27 During error recovery, the location of the syntax error is updated
28 to cover the whole sequence covered by the error token: it includes
29 the shifted symbols thrown away during the first part of the error
30 recovery, and the lookahead rejected during the second part.
33 . Stray semicolons are no longer allowed at the start of a grammar.
34 . Semicolons are now required after in-grammar declarations.
36 - Unescaped newlines are no longer allowed in character constants or
37 string literals. They were never portable, and GCC 3.4.0 has
38 dropped support for them. Better diagnostics are now generated if
39 forget a closing quote.
41 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
45 - GLR grammars now support locations.
47 - New directive: %initial-action.
48 This directive allows the user to run arbitrary code (including
49 initializing @$) from yyparse before parsing starts.
51 - A new directive "%expect-rr N" specifies the expected number of
52 reduce/reduce conflicts in GLR parsers.
54 - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
55 This is a GNU extension.
57 - The option `--report=lookahead' was changed to `--report=look-ahead'.
58 The old spelling still works, but is not documented and will be
61 - Experimental %destructor support has been added to lalr1.cc.
63 - New configure option --disable-yacc, to disable installation of the
64 yacc command and -ly library introduced in 1.875 for POSIX conformance.
68 - For now, %expect-count violations are now just warnings, not errors.
69 This is for compatibility with Bison 1.75 and earlier (when there are
70 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
71 are too many or too few shift/reduce conflicts). However, in future
72 versions of Bison we plan to improve the %expect machinery so that
73 these violations will become errors again.
75 - Within Bison itself, numbers (e.g., goto numbers) are no longer
76 arbitrarily limited to 16-bit counts.
78 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
80 Changes in version 1.875, 2003-01-01:
82 * The documentation license has been upgraded to version 1.2
83 of the GNU Free Documentation License.
85 * syntax error processing
87 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
88 locations too. This fixes bugs in error-location computation.
91 It is now possible to reclaim the memory associated to symbols
92 discarded during error recovery. This feature is still experimental.
95 This new directive is preferred over YYERROR_VERBOSE.
97 - #defining yyerror to steal internal variables is discouraged.
98 It is not guaranteed to work forever.
102 - Semicolons are once again optional at the end of grammar rules.
103 This reverts to the behavior of Bison 1.33 and earlier, and improves
104 compatibility with Yacc.
106 - `parse error' -> `syntax error'
107 Bison now uniformly uses the term `syntax error'; formerly, the code
108 and manual sometimes used the term `parse error' instead. POSIX
109 requires `syntax error' in diagnostics, and it was thought better to
112 - The documentation now emphasizes that yylex and yyerror must be
113 declared before use. C99 requires this.
115 - Bison now parses C99 lexical constructs like UCNs and
116 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
118 - File names are properly escaped in C output. E.g., foo\bar.y is
119 output as "foo\\bar.y".
121 - Yacc command and library now available
122 The Bison distribution now installs a `yacc' command, as POSIX requires.
123 Also, Bison now installs a small library liby.a containing
124 implementations of Yacc-compatible yyerror and main functions.
125 This library is normally not useful, but POSIX requires it.
127 - Type clashes now generate warnings, not errors.
129 - If the user does not define YYSTYPE as a macro, Bison now declares it
130 using typedef instead of defining it as a macro.
131 For consistency, YYLTYPE is also declared instead of defined.
133 * Other compatibility issues
135 - %union directives can now have a tag before the `{', e.g., the
136 directive `%union foo {...}' now generates the C code
137 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
138 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
139 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
140 This is for compatibility with both Yacc and Bison 1.35.
142 - `;' is output before the terminating `}' of an action, for
143 compatibility with Bison 1.35.
145 - Bison now uses a Yacc-style format for conflict reports, e.g.,
146 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
148 - `yystype' and `yyltype' are now obsolescent macros instead of being
149 typedefs or tags; they are no longer documented and are planned to be
150 withdrawn in a future release.
155 Users of Bison have to decide how they handle the portability of the
158 - `parsing stack overflow...' -> `parser stack overflow'
159 GLR parsers now report `parser stack overflow' as per the Bison manual.
161 * Bison now warns if it detects conflicting outputs to the same file,
162 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
163 that command outputs both code and header to foo.h.
165 * #line in output files
166 - --no-line works properly.
168 * Bison can no longer be built by a K&R C compiler; it requires C89 or
169 later to be built. This change originally took place a few versions
170 ago, but nobody noticed until we recently asked someone to try
171 building Bison with a K&R C compiler.
173 Changes in version 1.75, 2002-10-14:
175 * Bison should now work on 64-bit hosts.
177 * Indonesian translation thanks to Tedi Heriyanto.
180 Fix spurious parse errors.
183 Some people redefine yyerror to steal yyparse' private variables.
184 Reenable this trick until an official feature replaces it.
187 In agreement with POSIX and with other Yaccs, leaving a default
188 action is valid when $$ is untyped, and $1 typed:
192 but the converse remains an error:
196 * Values of mid-rule actions
199 foo: { ... } { $$ = $1; } ...
201 was incorrectly rejected: $1 is defined in the second mid-rule
202 action, and is equal to the $$ of the first mid-rule action.
204 Changes in version 1.50, 2002-10-04:
209 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
210 almost any context-free grammar, ambiguous or not. The new declarations
211 %dprec and %merge on grammar rules allow parse-time resolution of
212 ambiguities. Contributed by Paul Hilfinger.
214 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
215 like the Alpha, so please stick to 32-bit hosts for now.
218 When not in Yacc compatibility mode, when the output file was not
219 specified, running `bison foo/bar.y' created `foo/bar.c'. It
223 The undefined token was systematically mapped to 2 which prevented
224 the use of 2 by the user. This is no longer the case.
226 * Unknown token numbers
227 If yylex returned an out of range value, yyparse could die. This is
231 According to POSIX, the error token must be 256.
232 Bison extends this requirement by making it a preference: *if* the
233 user specified that one of her tokens is numbered 256, then error
234 will be mapped onto another number.
236 * Verbose error messages
237 They no longer report `..., expecting error or...' for states where
238 error recovery is possible.
241 Defaults to `$end' instead of `$'.
243 * Error recovery now conforms to documentation and to POSIX
244 When a Bison-generated parser encounters a syntax error, it now pops
245 the stack until it finds a state that allows shifting the error
246 token. Formerly, it popped the stack until it found a state that
247 allowed some non-error action other than a default reduction on the
248 error token. The new behavior has long been the documented behavior,
249 and has long been required by POSIX. For more details, please see
250 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
251 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
254 Popped tokens and nonterminals are now reported.
257 Larger grammars are now supported (larger token numbers, larger grammar
258 size (= sum of the LHS and RHS lengths), larger LALR tables).
259 Formerly, many of these numbers ran afoul of 16-bit limits;
260 now these limits are 32 bits on most hosts.
262 * Explicit initial rule
263 Bison used to play hacks with the initial rule, which the user does
264 not write. It is now explicit, and visible in the reports and
268 Before, Bison reported the useless rules, but, although not used,
269 included them in the parsers. They are now actually removed.
271 * Useless rules, useless nonterminals
272 They are now reported, as a warning, with their locations.
274 * Rules never reduced
275 Rules that can never be reduced because of conflicts are now
278 * Incorrect `Token not used'
281 %token useless useful
283 exp: '0' %prec useful;
285 where a token was used to set the precedence of the last rule,
286 bison reported both `useful' and `useless' as useless tokens.
288 * Revert the C++ namespace changes introduced in 1.31
289 as they caused too many portability hassles.
292 By an accident of design, the default computation of @$ was
293 performed after another default computation was performed: @$ = @1.
294 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
295 the computation of @$.
298 The token end of file may be specified by the user, in which case,
299 the user symbol is used in the reports, the graphs, and the verbose
300 error messages instead of `$end', which remains being the default.
304 %token MYEOF 0 "end of file"
307 This old option, which has been broken for ages, is removed.
310 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
311 Croatian, thanks to Denis Lackovic.
313 * Incorrect token definitions
314 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
316 * Token definitions as enums
317 Tokens are output both as the traditional #define's, and, provided
318 the compiler supports ANSI C or is a C++ compiler, as enums.
319 This lets debuggers display names instead of integers.
322 In addition to --verbose, bison supports --report=THINGS, which
323 produces additional information:
325 complete the core item sets with their closure
326 - lookahead [changed to `look-ahead' in 1.875e and later]
327 explicitly associate look-ahead tokens to items
329 describe shift/reduce conflicts solving.
330 Bison used to systematically output this information on top of
331 the report. Solved conflicts are now attached to their states.
334 Previous versions don't complain when there is a type clash on
335 the default action if the rule has a mid-rule action, such as in:
343 * GNU M4 is now required when using Bison.
345 Changes in version 1.35, 2002-03-25:
348 Some projects use Bison's C parser with C++ compilers, and define
349 YYSTYPE as a class. The recent adjustment of C parsers for data
350 alignment and 64 bit architectures made this impossible.
352 Because for the time being no real solution for C++ parser
353 generation exists, kludges were implemented in the parser to
354 maintain this use. In the future, when Bison has C++ parsers, this
355 kludge will be disabled.
357 This kludge also addresses some C++ problems when the stack was
360 Changes in version 1.34, 2002-03-12:
362 * File name clashes are detected
363 $ bison foo.y -d -o foo.x
364 fatal error: header and parser would both be named `foo.x'
366 * A missing `;' at the end of a rule triggers a warning
367 In accordance with POSIX, and in agreement with other
368 Yacc implementations, Bison will mandate this semicolon in the near
369 future. This eases the implementation of a Bison parser of Bison
370 grammars by making this grammar LALR(1) instead of LR(2). To
371 facilitate the transition, this release introduces a warning.
373 * Revert the C++ namespace changes introduced in 1.31, as they caused too
374 many portability hassles.
376 * DJGPP support added.
378 * Fix test suite portability problems.
380 Changes in version 1.33, 2002-02-07:
383 Groff could not be compiled for the definition of size_t was lacking
384 under some conditions.
389 Changes in version 1.32, 2002-01-23:
391 * Fix Yacc output file names
395 * Italian, Dutch translations
397 Changes in version 1.31, 2002-01-14:
401 * GNU Gettext and %expect
402 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
403 Bison dies on incorrect %expectations, we fear there will be
404 too many bug reports for Gettext, so _for the time being_, %expect
405 does not trigger an error when the input file is named `plural.y'.
407 * Use of alloca in parsers
408 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
409 malloc exclusively. Since 1.29, but was not NEWS'ed.
411 alloca is used only when compiled with GCC, to avoid portability
414 * When the generated parser lacks debugging code, YYDEBUG is now 0
415 (as POSIX requires) instead of being undefined.
418 Bison has always permitted actions such as { $$ = $1 }: it adds the
419 ending semicolon. Now if in Yacc compatibility mode, the semicolon
420 is no longer output: one has to write { $$ = $1; }.
422 * Better C++ compliance
423 The output parsers try to respect C++ namespaces.
424 [This turned out to be a failed experiment, and it was reverted later.]
427 Fixed bugs when reporting useless nonterminals.
430 The parsers work properly on 64 bit hosts.
433 Some calls to strerror resulted in scrambled or missing error messages.
436 When the number of shift/reduce conflicts is correct, don't issue
439 * The verbose report includes the rule line numbers.
441 * Rule line numbers are fixed in traces.
443 * Swedish translation
446 Verbose parse error messages from the parsers are better looking.
447 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
448 Now: parse error: unexpected '/', expecting "number" or '-' or '('
450 * Fixed parser memory leaks.
451 When the generated parser was using malloc to extend its stacks, the
452 previous allocations were not freed.
454 * Fixed verbose output file.
455 Some newlines were missing.
456 Some conflicts in state descriptions were missing.
458 * Fixed conflict report.
459 Option -v was needed to get the result.
463 Mismatches are errors, not warnings.
465 * Fixed incorrect processing of some invalid input.
467 * Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
469 * Fixed some typos in the documentation.
471 * %token MY_EOF 0 is supported.
472 Before, MY_EOF was silently renumbered as 257.
474 * doc/refcard.tex is updated.
476 * %output, %file-prefix, %name-prefix.
480 New, aliasing `--output-file'.
482 Changes in version 1.30, 2001-10-26:
484 * `--defines' and `--graph' have now an optional argument which is the
485 output file name. `-d' and `-g' do not change; they do not take any
488 * `%source_extension' and `%header_extension' are removed, failed
493 Changes in version 1.29, 2001-09-07:
495 * The output file does not define const, as this caused problems when used
496 with common autoconfiguration schemes. If you still use ancient compilers
497 that lack const, compile with the equivalent of the C compiler option
498 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
500 * Added `-g' and `--graph'.
502 * The Bison manual is now distributed under the terms of the GNU FDL.
504 * The input and the output files has automatically a similar extension.
506 * Russian translation added.
508 * NLS support updated; should hopefully be less troublesome.
510 * Added the old Bison reference card.
512 * Added `--locations' and `%locations'.
514 * Added `-S' and `--skeleton'.
516 * `%raw', `-r', `--raw' is disabled.
518 * Special characters are escaped when output. This solves the problems
519 of the #line lines with path names including backslashes.
522 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
523 `%debug', `%source_extension' and `%header_extension'.
526 Automatic location tracking.
528 Changes in version 1.28, 1999-07-06:
530 * Should compile better now with K&R compilers.
534 * Fixed a problem with escaping the double quote character.
536 * There is now a FAQ.
538 Changes in version 1.27:
540 * The make rule which prevented bison.simple from being created on
541 some systems has been fixed.
543 Changes in version 1.26:
545 * Bison now uses automake.
547 * New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
549 * Token numbers now start at 257 as previously documented, not 258.
551 * Bison honors the TMPDIR environment variable.
553 * A couple of buffer overruns have been fixed.
555 * Problems when closing files should now be reported.
557 * Generated parsers should now work even on operating systems which do
558 not provide alloca().
560 Changes in version 1.25, 1995-10-16:
562 * Errors in the input grammar are not fatal; Bison keeps reading
563 the grammar file, and reports all the errors found in it.
565 * Tokens can now be specified as multiple-character strings: for
566 example, you could use "<=" for a token which looks like <=, instead
567 of chosing a name like LESSEQ.
569 * The %token_table declaration says to write a table of tokens (names
570 and numbers) into the parser file. The yylex function can use this
571 table to recognize multiple-character string tokens, or for other
574 * The %no_lines declaration says not to generate any #line preprocessor
575 directives in the parser file.
577 * The %raw declaration says to use internal Bison token numbers, not
578 Yacc-compatible token numbers, when token names are defined as macros.
580 * The --no-parser option produces the parser tables without including
581 the parser engine; a project can now use its own parser engine.
582 The actions go into a separate file called NAME.act, in the form of
583 a switch statement body.
585 Changes in version 1.23:
587 The user can define YYPARSE_PARAM as the name of an argument to be
588 passed into yyparse. The argument should have type void *. It should
589 actually point to an object. Grammar actions can access the variable
590 by casting it to the proper pointer type.
592 Line numbers in output file corrected.
594 Changes in version 1.22:
598 Changes in version 1.20:
600 Output file does not redefine const for C++.
608 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004, 2005
609 Free Software Foundation, Inc.
611 This file is part of Bison, the GNU Compiler Compiler.
613 Bison is free software; you can redistribute it and/or modify
614 it under the terms of the GNU General Public License as published by
615 the Free Software Foundation; either version 2, or (at your option)
618 Bison is distributed in the hope that it will be useful,
619 but WITHOUT ANY WARRANTY; without even the implied warranty of
620 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
621 GNU General Public License for more details.
623 You should have received a copy of the GNU General Public License
624 along with autoconf; see the file COPYING. If not, write to
625 the Free Software Foundation, Inc., 51 Franklin Street, Fifth Floor,
626 Boston, MA 02110-1301, USA.