4 Changes in version 1.875d:
6 * NUL bytes are no longer allowed in Bison string literals, unfortunately.
8 * %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
9 This is a GNU extension.
11 * Experimental %destructor support has been added to lalr1.cc.
13 Changes in version 1.875c, 2003-08-25:
17 Changes in version 1.875b, 2003-06-17:
19 * GLR grammars now support locations.
22 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
23 - Stray semicolons are no longer allowed at the start of a grammar.
24 - Semicolons are now required after in-grammar declarations.
26 Changes in version 1.875a, 2003-02-01:
28 * For now, %expect-count violations are now just warnings, not errors.
29 This is for compatibility with Bison 1.75 and earlier (when there are
30 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
31 are too many or too few shift/reduce conflicts). However, in future
32 versions of Bison we plan to improve the %expect machinery so that
33 these violations will become errors again.
35 * New configure option --disable-yacc, to disable installation of the
36 yacc command and -ly library introduced in 1.875 for POSIX conformance.
38 Changes in version 1.875, 2003-01-01:
40 * The documentation license has been upgraded to version 1.2
41 of the GNU Free Documentation License.
43 * syntax error processing
45 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
46 locations too. This fixes bugs in error-location computation.
49 It is now possible to reclaim the memory associated to symbols
50 discarded during error recovery. This feature is still experimental.
53 This new directive is preferred over YYERROR_VERBOSE.
55 - #defining yyerror to steal internal variables is discouraged.
56 It is not guaranteed to work forever.
60 - Semicolons are once again optional at the end of grammar rules.
61 This reverts to the behavior of Bison 1.33 and earlier, and improves
62 compatibility with Yacc.
64 - `parse error' -> `syntax error'
65 Bison now uniformly uses the term `syntax error'; formerly, the code
66 and manual sometimes used the term `parse error' instead. POSIX
67 requires `syntax error' in diagnostics, and it was thought better to
70 - The documentation now emphasizes that yylex and yyerror must be
71 declared before use. C99 requires this.
73 - Bison now parses C99 lexical constructs like UCNs and
74 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
76 - File names are properly escaped in C output. E.g., foo\bar.y is
77 output as "foo\\bar.y".
79 - Yacc command and library now available
80 The Bison distribution now installs a `yacc' command, as POSIX requires.
81 Also, Bison now installs a small library liby.a containing
82 implementations of Yacc-compatible yyerror and main functions.
83 This library is normally not useful, but POSIX requires it.
85 - Type clashes now generate warnings, not errors.
87 - If the user does not define YYSTYPE as a macro, Bison now declares it
88 using typedef instead of defining it as a macro.
89 For consistency, YYLTYPE is also declared instead of defined.
91 * Other compatibility issues
93 - %union directives can now have a tag before the `{', e.g., the
94 directive `%union foo {...}' now generates the C code
95 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
96 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
97 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
98 This is for compatibility with both Yacc and Bison 1.35.
100 - `;' is output before the terminating `}' of an action, for
101 compatibility with Bison 1.35.
103 - Bison now uses a Yacc-style format for conflict reports, e.g.,
104 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
106 - `yystype' and `yyltype' are now obsolescent macros instead of being
107 typedefs or tags; they are no longer documented and are planned to be
108 withdrawn in a future release.
113 Users of Bison have to decide how they handle the portability of the
116 - `parsing stack overflow...' -> `parser stack overflow'
117 GLR parsers now report `parser stack overflow' as per the Bison manual.
119 * Bison now warns if it detects conflicting outputs to the same file,
120 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
121 that command outputs both code and header to foo.h.
123 * #line in output files
124 - --no-line works properly.
126 * Bison can no longer be built by a K&R C compiler; it requires C89 or
127 later to be built. This change originally took place a few versions
128 ago, but nobody noticed until we recently asked someone to try
129 building Bison with a K&R C compiler.
131 Changes in version 1.75, 2002-10-14:
133 * Bison should now work on 64-bit hosts.
135 * Indonesian translation thanks to Tedi Heriyanto.
138 Fix spurious parse errors.
141 Some people redefine yyerror to steal yyparse' private variables.
142 Reenable this trick until an official feature replaces it.
145 In agreement with POSIX and with other Yaccs, leaving a default
146 action is valid when $$ is untyped, and $1 typed:
150 but the converse remains an error:
154 * Values of mid-rule actions
157 foo: { ... } { $$ = $1; } ...
159 was incorrectly rejected: $1 is defined in the second mid-rule
160 action, and is equal to the $$ of the first mid-rule action.
162 Changes in version 1.50, 2002-10-04:
167 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
168 almost any context-free grammar, ambiguous or not. The new declarations
169 %dprec and %merge on grammar rules allow parse-time resolution of
170 ambiguities. Contributed by Paul Hilfinger.
172 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
173 like the Alpha, so please stick to 32-bit hosts for now.
176 When not in Yacc compatibility mode, when the output file was not
177 specified, running `bison foo/bar.y' created `foo/bar.c'. It
181 The undefined token was systematically mapped to 2 which prevented
182 the use of 2 by the user. This is no longer the case.
184 * Unknown token numbers
185 If yylex returned an out of range value, yyparse could die. This is
189 According to POSIX, the error token must be 256.
190 Bison extends this requirement by making it a preference: *if* the
191 user specified that one of her tokens is numbered 256, then error
192 will be mapped onto another number.
194 * Verbose error messages
195 They no longer report `..., expecting error or...' for states where
196 error recovery is possible.
199 Defaults to `$end' instead of `$'.
201 * Error recovery now conforms to documentation and to POSIX
202 When a Bison-generated parser encounters a syntax error, it now pops
203 the stack until it finds a state that allows shifting the error
204 token. Formerly, it popped the stack until it found a state that
205 allowed some non-error action other than a default reduction on the
206 error token. The new behavior has long been the documented behavior,
207 and has long been required by POSIX. For more details, please see
208 <http://mail.gnu.org/pipermail/bug-bison/2002-May/001452.html>.
211 Popped tokens and nonterminals are now reported.
214 Larger grammars are now supported (larger token numbers, larger grammar
215 size (= sum of the LHS and RHS lengths), larger LALR tables).
216 Formerly, many of these numbers ran afoul of 16-bit limits;
217 now these limits are 32 bits on most hosts.
219 * Explicit initial rule
220 Bison used to play hacks with the initial rule, which the user does
221 not write. It is now explicit, and visible in the reports and
225 Before, Bison reported the useless rules, but, although not used,
226 included them in the parsers. They are now actually removed.
228 * Useless rules, useless nonterminals
229 They are now reported, as a warning, with their locations.
231 * Rules never reduced
232 Rules that can never be reduced because of conflicts are now
235 * Incorrect `Token not used'
238 %token useless useful
240 exp: '0' %prec useful;
242 where a token was used to set the precedence of the last rule,
243 bison reported both `useful' and `useless' as useless tokens.
245 * Revert the C++ namespace changes introduced in 1.31
246 as they caused too many portability hassles.
249 By an accident of design, the default computation of @$ was
250 performed after another default computation was performed: @$ = @1.
251 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
252 the computation of @$.
255 The token end of file may be specified by the user, in which case,
256 the user symbol is used in the reports, the graphs, and the verbose
257 error messages instead of `$end', which remains being the default.
261 %token MYEOF 0 "end of file"
264 This old option, which has been broken for ages, is removed.
267 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
268 Croatian, thanks to Denis Lackovic.
270 * Incorrect token definitions
271 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
273 * Token definitions as enums
274 Tokens are output both as the traditional #define's, and, provided
275 the compiler supports ANSI C or is a C++ compiler, as enums.
276 This lets debuggers display names instead of integers.
279 In addition to --verbose, bison supports --report=THINGS, which
280 produces additional information:
282 complete the core item sets with their closure
284 explicitly associate lookaheads to items
286 describe shift/reduce conflicts solving.
287 Bison used to systematically output this information on top of
288 the report. Solved conflicts are now attached to their states.
291 Previous versions don't complain when there is a type clash on
292 the default action if the rule has a mid-rule action, such as in:
300 * GNU M4 is now required when using Bison.
302 Changes in version 1.35, 2002-03-25:
305 Some projects use Bison's C parser with C++ compilers, and define
306 YYSTYPE as a class. The recent adjustment of C parsers for data
307 alignment and 64 bit architectures made this impossible.
309 Because for the time being no real solution for C++ parser
310 generation exists, kludges were implemented in the parser to
311 maintain this use. In the future, when Bison has C++ parsers, this
312 kludge will be disabled.
314 This kludge also addresses some C++ problems when the stack was
317 Changes in version 1.34, 2002-03-12:
319 * File name clashes are detected
320 $ bison foo.y -d -o foo.x
321 fatal error: header and parser would both be named `foo.x'
323 * A missing `;' at the end of a rule triggers a warning
324 In accordance with POSIX, and in agreement with other
325 Yacc implementations, Bison will mandate this semicolon in the near
326 future. This eases the implementation of a Bison parser of Bison
327 grammars by making this grammar LALR(1) instead of LR(2). To
328 facilitate the transition, this release introduces a warning.
330 * Revert the C++ namespace changes introduced in 1.31, as they caused too
331 many portability hassles.
333 * DJGPP support added.
335 * Fix test suite portability problems.
337 Changes in version 1.33, 2002-02-07:
340 Groff could not be compiled for the definition of size_t was lacking
341 under some conditions.
346 Changes in version 1.32, 2002-01-23:
348 * Fix Yacc output file names
352 * Italian, Dutch translations
354 Changes in version 1.31, 2002-01-14:
358 * GNU Gettext and %expect
359 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
360 Bison dies on incorrect %expectations, we fear there will be
361 too many bug reports for Gettext, so _for the time being_, %expect
362 does not trigger an error when the input file is named `plural.y'.
364 * Use of alloca in parsers
365 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
366 malloc exclusively. Since 1.29, but was not NEWS'ed.
368 alloca is used only when compiled with GCC, to avoid portability
371 * When the generated parser lacks debugging code, YYDEBUG is now 0
372 (as POSIX requires) instead of being undefined.
375 Bison has always permitted actions such as { $$ = $1 }: it adds the
376 ending semicolon. Now if in Yacc compatibility mode, the semicolon
377 is no longer output: one has to write { $$ = $1; }.
379 * Better C++ compliance
380 The output parsers try to respect C++ namespaces.
381 [This turned out to be a failed experiment, and it was reverted later.]
384 Fixed bugs when reporting useless nonterminals.
387 The parsers work properly on 64 bit hosts.
390 Some calls to strerror resulted in scrambled or missing error messages.
393 When the number of shift/reduce conflicts is correct, don't issue
396 * The verbose report includes the rule line numbers.
398 * Rule line numbers are fixed in traces.
400 * Swedish translation
403 Verbose parse error messages from the parsers are better looking.
404 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
405 Now: parse error: unexpected '/', expecting "number" or '-' or '('
407 * Fixed parser memory leaks.
408 When the generated parser was using malloc to extend its stacks, the
409 previous allocations were not freed.
411 * Fixed verbose output file.
412 Some newlines were missing.
413 Some conflicts in state descriptions were missing.
415 * Fixed conflict report.
416 Option -v was needed to get the result.
420 Mismatches are errors, not warnings.
422 * Fixed incorrect processing of some invalid input.
424 * Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
426 * Fixed some typos in the documentation.
428 * %token MY_EOF 0 is supported.
429 Before, MY_EOF was silently renumbered as 257.
431 * doc/refcard.tex is updated.
433 * %output, %file-prefix, %name-prefix.
437 New, aliasing `--output-file'.
439 Changes in version 1.30, 2001-10-26:
441 * `--defines' and `--graph' have now an optional argument which is the
442 output file name. `-d' and `-g' do not change; they do not take any
445 * `%source_extension' and `%header_extension' are removed, failed
450 Changes in version 1.29, 2001-09-07:
452 * The output file does not define const, as this caused problems when used
453 with common autoconfiguration schemes. If you still use ancient compilers
454 that lack const, compile with the equivalent of the C compiler option
455 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
457 * Added `-g' and `--graph'.
459 * The Bison manual is now distributed under the terms of the GNU FDL.
461 * The input and the output files has automatically a similar extension.
463 * Russian translation added.
465 * NLS support updated; should hopefully be less troublesome.
467 * Added the old Bison reference card.
469 * Added `--locations' and `%locations'.
471 * Added `-S' and `--skeleton'.
473 * `%raw', `-r', `--raw' is disabled.
475 * Special characters are escaped when output. This solves the problems
476 of the #line lines with path names including backslashes.
479 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
480 `%debug', `%source_extension' and `%header_extension'.
483 Automatic location tracking.
485 Changes in version 1.28, 1999-07-06:
487 * Should compile better now with K&R compilers.
491 * Fixed a problem with escaping the double quote character.
493 * There is now a FAQ.
495 Changes in version 1.27:
497 * The make rule which prevented bison.simple from being created on
498 some systems has been fixed.
500 Changes in version 1.26:
502 * Bison now uses automake.
504 * New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
506 * Token numbers now start at 257 as previously documented, not 258.
508 * Bison honors the TMPDIR environment variable.
510 * A couple of buffer overruns have been fixed.
512 * Problems when closing files should now be reported.
514 * Generated parsers should now work even on operating systems which do
515 not provide alloca().
517 Changes in version 1.25, 1995-10-16:
519 * Errors in the input grammar are not fatal; Bison keeps reading
520 the grammar file, and reports all the errors found in it.
522 * Tokens can now be specified as multiple-character strings: for
523 example, you could use "<=" for a token which looks like <=, instead
524 of chosing a name like LESSEQ.
526 * The %token_table declaration says to write a table of tokens (names
527 and numbers) into the parser file. The yylex function can use this
528 table to recognize multiple-character string tokens, or for other
531 * The %no_lines declaration says not to generate any #line preprocessor
532 directives in the parser file.
534 * The %raw declaration says to use internal Bison token numbers, not
535 Yacc-compatible token numbers, when token names are defined as macros.
537 * The --no-parser option produces the parser tables without including
538 the parser engine; a project can now use its own parser engine.
539 The actions go into a separate file called NAME.act, in the form of
540 a switch statement body.
542 Changes in version 1.23:
544 The user can define YYPARSE_PARAM as the name of an argument to be
545 passed into yyparse. The argument should have type void *. It should
546 actually point to an object. Grammar actions can access the variable
547 by casting it to the proper pointer type.
549 Line numbers in output file corrected.
551 Changes in version 1.22:
555 Changes in version 1.20:
557 Output file does not redefine const for C++.
565 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003, 2004
566 Free Software Foundation, Inc.
568 This file is part of Bison, the GNU Compiler Compiler.
570 Bison is free software; you can redistribute it and/or modify
571 it under the terms of the GNU General Public License as published by
572 the Free Software Foundation; either version 2, or (at your option)
575 Bison is distributed in the hope that it will be useful,
576 but WITHOUT ANY WARRANTY; without even the implied warranty of
577 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
578 GNU General Public License for more details.
580 You should have received a copy of the GNU General Public License
581 along with autoconf; see the file COPYING. If not, write to
582 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
583 Boston, MA 02111-1307, USA.