4 * Changes in version ?.? (????-??-??):
6 ** Additional yylex/yyparse arguments
8 The new directive %param declare additional argument to both yylex
9 and yyparse. The %lex-param, %parse-param, and %param directive
10 support one or more arguments. Instead of
12 %lex-param {arg1_type *arg1}
13 %lex-param {arg2_type *arg2}
14 %parse-param {arg1_type *arg1}
15 %parse-param {arg2_type *arg2}
19 %param {arg1_type *arg1} {arg2_type *arg2}
21 ** Java skeleton improvements
23 The constants for token names were moved to the Lexer interface.
24 Also, it is possible to add code to the parser's constructors using
25 "%code init" and "%define init_throws".
27 ** Variable api.tokens.prefix
29 The variable api.tokens.prefix changes the way tokens are identified in
30 the generated files. This is especially useful to avoid collisions
31 with identifiers in the target language. For instance
34 %define api.tokens.prefix "TOK_"
36 start: FILE for ERROR;
38 will generate the definition of the symbols TOK_FILE, TOK_for, and
39 TOK_ERROR in the generated sources. In particular, the scanner must
40 use these prefixed token names, although the grammar itself still
41 uses the short names (as in the sample rule given above).
43 ** Variable api.namespace
45 The "namespace" variable is renamed "api.namespace". Backward
46 compatibility is ensured, but upgrading is recommended.
48 ** Variable parse.error
50 The variable error controls the verbosity of error messages. The
51 use of the %error-verbose directive is deprecated in favor of
52 %define parse.error "verbose".
54 * Changes in version 2.5 (????-??-??):
56 ** Named References Support
58 Historically, Yacc and Bison have supported positional references
59 ($n, $$) to allow access to symbol values from inside of semantic
62 Starting from this version, Bison can also accept named references.
63 When no ambiguity is possible, original symbol names may be used
66 if_stmt : 'if' cond_expr 'then' then_stmt ';'
67 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
69 In the more common case, explicit names may be declared:
71 stmt[res] : 'if' expr[cond] 'then' stmt[then] 'else' stmt[else] ';'
72 { $res = mk_if_stmt($cond, $then, $else); }
74 Location information is also accessible using @name syntax. When
75 accessing symbol names containing dots or dashes, explicit bracketing
76 ($[sym.1]) must be used.
78 These features are experimental in this version. More user feedback
79 will help to stabilize them.
81 ** IELR(1) and Canonical LR(1) Support
83 IELR(1) is a minimal LR(1) parser table generation algorithm. That
84 is, given any context-free grammar, IELR(1) generates parser tables
85 with the full language recognition power of canonical LR(1) but with
86 nearly the same number of parser states as LALR(1). This reduction in
87 parser states is often an order of magnitude. More importantly,
88 because canonical LR(1)'s extra parser states may contain duplicate
89 conflicts in the case of non-LR(1) grammars, the number of conflicts
90 for IELR(1) is often an order of magnitude less as well. This can
91 significantly reduce the complexity of developing of a grammar.
93 Bison can now generate IELR(1) and canonical LR(1) parser tables in
94 place of its traditional LALR(1) parser tables, which remain the
95 default. You can specify the type of parser tables in the grammar
96 file with these directives:
100 %define lr.type canonical-lr
102 The default reduction optimization in the parser tables can also be
103 adjusted using `%define lr.default-reductions'. See the documentation
104 for `%define lr.type' and `%define lr.default-reductions' in the
105 section `Bison Declaration Summary' in the Bison manual for the
108 These features are experimental. More user feedback will help to
111 ** Unrecognized %code qualifiers are now an error not a warning.
113 ** %define improvements.
115 *** Unrecognized variables are now an error not a warning.
117 *** Multiple invocations for any variable is now an error not a warning.
119 *** Can now be invoked via the command line.
121 Each of these command-line options
124 --define=NAME[=VALUE]
127 --force-define=NAME[=VALUE]
129 is equivalent to this grammar file declaration
131 %define NAME ["VALUE"]
133 except that the manner in which Bison processes multiple definitions
134 for the same NAME differs. Most importantly, -F and --force-define
135 quietly override %define, but -D and --define do not. For further
136 details, see the section "Bison Options" in the Bison manual.
138 *** Variables renamed.
140 The following %define variables
143 lr.keep_unreachable_states
148 lr.keep-unreachable-states
150 The old names are now deprecated but will be maintained indefinitely
151 for backward compatibility.
153 *** Values no longer need to be quoted in grammar file.
155 If a %define value is an identifier, it no longer needs to be placed
156 within quotations marks. For example,
158 %define api.push-pull "push"
162 %define api.push-pull push
166 Consistently with directives (such as %error-verbose) and variables
167 (e.g. push-pull), symbol names may include dashes in any position,
168 similarly to periods and underscores. This is GNU extension over
169 POSIX Yacc whose use is reported by -Wyacc, and rejected in Yacc
172 ** YYFAIL now produces warnings and Java parsers no longer implement it.
174 YYFAIL has existed for many years as an undocumented feature of
175 deterministic parsers in C generated by Bison. More recently, it was
176 a documented feature of Bison's experimental Java parsers. As
177 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
178 semantic action now produces a deprecation warning, and Java parsers
179 no longer implement YYFAIL at all. For further details, including a
180 discussion of how to suppress C preprocessor warnings about YYFAIL
181 being unused, see the Bison 2.4.2 NEWS entry.
183 ** Temporary hack for adding a semicolon to the user action.
185 Previously, Bison appended a semicolon to every user action for
186 reductions when the output language defaulted to C (specifically, when
187 neither %yacc, %language, %skeleton, or equivalent command-line
188 options were specified). This allowed actions such as
190 exp: exp "+" exp { $$ = $1 + $3 };
194 exp: exp "+" exp { $$ = $1 + $3; };
196 As a first step in removing this misfeature, Bison now issues a
197 warning when it appends a semicolon. Moreover, in cases where Bison
198 cannot easily determine whether a semicolon is needed (for example, an
199 action ending with a cpp directive or a braced compound initializer),
200 it no longer appends one. Thus, the C compiler might now complain
201 about a missing semicolon where it did not before. Future releases of
202 Bison will cease to append semicolons entirely.
204 ** Character literals not of length one.
206 Previously, Bison quietly converted all character literals to length
207 one. For example, without warning, Bison interpreted the operators in
208 the following grammar to be the same token:
214 Bison now warns when a character literal is not of length one. In
215 some future release, Bison will report an error instead.
217 ** Verbose error messages fixed for nonassociative tokens.
219 When %error-verbose is specified, syntax error messages produced by
220 the generated parser include the unexpected token as well as a list of
221 expected tokens. Previously, this list erroneously included tokens
222 that would actually induce a syntax error because conflicts for them
223 were resolved with %nonassoc. Such tokens are now properly omitted
226 * Changes in version 2.4.2 (????-??-??):
228 ** Detection of GNU M4 1.4.6 or newer during configure is improved.
230 ** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
231 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
234 ** %code is now a permanent feature.
236 A traditional Yacc prologue directive is written in the form:
240 To provide a more flexible alternative, Bison 2.3b introduced the
241 %code directive with the following forms for C/C++:
244 %code requires {CODE}
245 %code provides {CODE}
248 These forms are now considered permanent features of Bison. See the
249 %code entries in the section "Bison Declaration Summary" in the Bison
250 manual for a summary of their functionality. See the section
251 "Prologue Alternatives" for a detailed discussion including the
252 advantages of %code over the traditional Yacc prologue directive.
254 Bison's Java feature as a whole including its current usage of %code
255 is still considered experimental.
257 ** YYFAIL is deprecated and will eventually be removed.
259 YYFAIL has existed for many years as an undocumented feature of
260 deterministic parsers in C generated by Bison. Previously, it was
261 documented for Bison's experimental Java parsers. YYFAIL is no longer
262 documented for Java parsers and is formally deprecated in both cases.
263 Users are strongly encouraged to migrate to YYERROR, which is
266 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
267 induce a syntax error. The most obvious difference from YYERROR is
268 that YYFAIL will automatically invoke yyerror to report the syntax
269 error so that you don't have to. However, there are several other
270 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
271 inherent flaws when %error-verbose or `#define YYERROR_VERBOSE' is
272 used. For a more detailed discussion, see:
274 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
276 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
277 deterministic parsers in C will continue to implement it. However,
278 because YYFAIL is already flawed, it seems futile to try to make new
279 Bison features compatible with it. Thus, during parser generation,
280 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
281 rule action. In a later release, YYFAIL will be disabled for
282 %error-verbose and `#define YYERROR_VERBOSE'. Eventually, YYFAIL will
283 be removed altogether.
285 There exists at least one case where Bison 2.5's YYFAIL warning will
286 be a false positive. Some projects add phony uses of YYFAIL and other
287 Bison-defined macros for the sole purpose of suppressing C
288 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
289 To avoid Bison's future warning, such YYFAIL uses can be moved to the
290 epilogue (that is, after the second `%%') in the Bison input file. In
291 this release (2.4.2), Bison already generates its own code to suppress
292 C preprocessor warnings for YYFAIL, so projects can remove their own
293 phony uses of YYFAIL if compatibility with Bison releases prior to
294 2.4.2 is not necessary.
296 ** Internationalization.
298 Fix a regression introduced in Bison 2.4: Under some circumstances,
299 message translations were not installed although supported by the
302 * Changes in version 2.4.1 (2008-12-11):
304 ** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
305 declarations have been fixed.
307 ** Temporary hack for adding a semicolon to the user action.
309 Bison used to prepend a trailing semicolon at the end of the user
310 action for reductions. This allowed actions such as
312 exp: exp "+" exp { $$ = $1 + $3 };
316 exp: exp "+" exp { $$ = $1 + $3; };
318 Some grammars still depend on this `feature'. Bison 2.4.1 restores
319 the previous behavior in the case of C output (specifically, when
320 neither %language or %skeleton or equivalent command-line options
321 are used) to leave more time for grammars depending on the old
322 behavior to be adjusted. Future releases of Bison will disable this
325 ** A few minor improvements to the Bison manual.
327 * Changes in version 2.4 (2008-11-02):
329 ** %language is an experimental feature.
331 We first introduced this feature in test release 2.3b as a cleaner
332 alternative to %skeleton. Since then, we have discussed the possibility of
333 modifying its effect on Bison's output file names. Thus, in this release,
334 we consider %language to be an experimental feature that will likely evolve
337 ** Forward compatibility with GNU M4 has been improved.
339 ** Several bugs in the C++ skeleton and the experimental Java skeleton have been
342 * Changes in version 2.3b (2008-05-27):
344 ** The quotes around NAME that used to be required in the following directive
349 ** The directive `%pure-parser' is now deprecated in favor of:
353 which has the same effect except that Bison is more careful to warn about
354 unreasonable usage in the latter case.
358 Bison can now generate an LALR(1) parser in C with a push interface. That
359 is, instead of invoking `yyparse', which pulls tokens from `yylex', you can
360 push one token at a time to the parser using `yypush_parse', which will
361 return to the caller after processing each token. By default, the push
362 interface is disabled. Either of the following directives will enable it:
364 %define api.push_pull "push" // Just push; does not require yylex.
365 %define api.push_pull "both" // Push and pull; requires yylex.
367 See the new section `A Push Parser' in the Bison manual for details.
369 The current push parsing interface is experimental and may evolve. More user
370 feedback will help to stabilize it.
372 ** The -g and --graph options now output graphs in Graphviz DOT format,
373 not VCG format. Like --graph, -g now also takes an optional FILE argument
374 and thus cannot be bundled with other short options.
378 Bison can now generate an LALR(1) parser in Java. The skeleton is
379 `data/lalr1.java'. Consider using the new %language directive instead of
380 %skeleton to select it.
382 See the new section `Java Parsers' in the Bison manual for details.
384 The current Java interface is experimental and may evolve. More user
385 feedback will help to stabilize it.
389 This new directive specifies the programming language of the generated
390 parser, which can be C (the default), C++, or Java. Besides the skeleton
391 that Bison uses, the directive affects the names of the generated files if
392 the grammar file's name ends in ".y".
394 ** XML Automaton Report
396 Bison can now generate an XML report of the LALR(1) automaton using the new
397 `--xml' option. The current XML schema is experimental and may evolve. More
398 user feedback will help to stabilize it.
400 ** The grammar file may now specify the name of the parser header file using
401 %defines. For example:
405 ** When reporting useless rules, useless nonterminals, and unused terminals,
406 Bison now employs the terms "useless in grammar" instead of "useless",
407 "useless in parser" instead of "never reduced", and "unused in grammar"
410 ** Unreachable State Removal
412 Previously, Bison sometimes generated parser tables containing unreachable
413 states. A state can become unreachable during conflict resolution if Bison
414 disables a shift action leading to it from a predecessor state. Bison now:
416 1. Removes unreachable states.
418 2. Does not report any conflicts that appeared in unreachable states.
419 WARNING: As a result, you may need to update %expect and %expect-rr
420 directives in existing grammar files.
422 3. For any rule used only in such states, Bison now reports the rule as
423 "useless in parser due to conflicts".
425 This feature can be disabled with the following directive:
427 %define lr.keep_unreachable_states
429 See the %define entry in the `Bison Declaration Summary' in the Bison manual
430 for further discussion.
432 ** Lookahead Set Correction in the `.output' Report
434 When instructed to generate a `.output' file including lookahead sets
435 (using `--report=lookahead', for example), Bison now prints each reduction's
436 lookahead set only next to the associated state's one item that (1) is
437 associated with the same rule as the reduction and (2) has its dot at the end
438 of its RHS. Previously, Bison also erroneously printed the lookahead set
439 next to all of the state's other items associated with the same rule. This
440 bug affected only the `.output' file and not the generated parser source
443 ** --report-file=FILE is a new option to override the default `.output' file
446 ** The `=' that used to be required in the following directives is now
449 %file-prefix "parser"
453 ** An Alternative to `%{...%}' -- `%code QUALIFIER {CODE}'
455 Bison 2.3a provided a new set of directives as a more flexible alternative to
456 the traditional Yacc prologue blocks. Those have now been consolidated into
457 a single %code directive with an optional qualifier field, which identifies
458 the purpose of the code and thus the location(s) where Bison should generate
461 1. `%code {CODE}' replaces `%after-header {CODE}'
462 2. `%code requires {CODE}' replaces `%start-header {CODE}'
463 3. `%code provides {CODE}' replaces `%end-header {CODE}'
464 4. `%code top {CODE}' replaces `%before-header {CODE}'
466 See the %code entries in section `Bison Declaration Summary' in the Bison
467 manual for a summary of the new functionality. See the new section `Prologue
468 Alternatives' for a detailed discussion including the advantages of %code
469 over the traditional Yacc prologues.
471 The prologue alternatives are experimental. More user feedback will help to
472 determine whether they should become permanent features.
474 ** Revised warning: unset or unused mid-rule values
476 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
477 used within any of the actions of the parent rule. For example, Bison warns
480 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
482 Now, Bison also warns about mid-rule values that are used but not set. For
483 example, Bison warns about unset $$ in the mid-rule action in:
485 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
487 However, Bison now disables both of these warnings by default since they
488 sometimes prove to be false alarms in existing grammars employing the Yacc
489 constructs $0 or $-N (where N is some positive integer).
491 To enable these warnings, specify the option `--warnings=midrule-values' or
492 `-W', which is a synonym for `--warnings=all'.
494 ** Default %destructor or %printer with `<*>' or `<>'
496 Bison now recognizes two separate kinds of default %destructor's and
499 1. Place `<*>' in a %destructor/%printer symbol list to define a default
500 %destructor/%printer for all grammar symbols for which you have formally
501 declared semantic type tags.
503 2. Place `<>' in a %destructor/%printer symbol list to define a default
504 %destructor/%printer for all grammar symbols without declared semantic
507 Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
508 `<*>' and `<>' combined achieve the same effect with one exception: Bison no
509 longer applies any %destructor to a mid-rule value if that mid-rule value is
510 not actually ever referenced using either $$ or $n in a semantic action.
512 The default %destructor's and %printer's are experimental. More user
513 feedback will help to determine whether they should become permanent
516 See the section `Freeing Discarded Symbols' in the Bison manual for further
519 ** %left, %right, and %nonassoc can now declare token numbers. This is required
520 by POSIX. However, see the end of section `Operator Precedence' in the Bison
521 manual for a caveat concerning the treatment of literal strings.
523 ** The nonfunctional --no-parser, -n, and %no-parser options have been
524 completely removed from Bison.
526 * Changes in version 2.3a, 2006-09-13:
528 ** Instead of %union, you can define and use your own union type
529 YYSTYPE if your grammar contains at least one <type> tag.
530 Your YYSTYPE need not be a macro; it can be a typedef.
531 This change is for compatibility with other Yacc implementations,
532 and is required by POSIX.
534 ** Locations columns and lines start at 1.
535 In accordance with the GNU Coding Standards and Emacs.
537 ** You may now declare per-type and default %destructor's and %printer's:
541 %union { char *string; }
542 %token <string> STRING1
543 %token <string> STRING2
544 %type <string> string1
545 %type <string> string2
546 %union { char character; }
547 %token <character> CHR
548 %type <character> chr
549 %destructor { free ($$); } %symbol-default
550 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
551 %destructor { } <character>
553 guarantees that, when the parser discards any user-defined symbol that has a
554 semantic type tag other than `<character>', it passes its semantic value to
555 `free'. However, when the parser discards a `STRING1' or a `string1', it
556 also prints its line number to `stdout'. It performs only the second
557 `%destructor' in this case, so it invokes `free' only once.
559 [Although we failed to mention this here in the 2.3a release, the default
560 %destructor's and %printer's were experimental, and they were rewritten in
563 ** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
564 `--yacc', or `%yacc'), Bison no longer generates #define statements for
565 associating token numbers with token names. Removing the #define statements
566 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
567 requires them. Bison still generates an enum for token names in all cases.
569 ** Handling of traditional Yacc prologue blocks is now more consistent but
570 potentially incompatible with previous releases of Bison.
572 As before, you declare prologue blocks in your grammar file with the
573 `%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all
574 prologue blocks that you've declared before the first %union. To generate
575 the post-prologue, Bison concatenates all prologue blocks that you've
576 declared after the first %union.
578 Previous releases of Bison inserted the pre-prologue into both the header
579 file and the code file in all cases except for LALR(1) parsers in C. In the
580 latter case, Bison inserted it only into the code file. For parsers in C++,
581 the point of insertion was before any token definitions (which associate
582 token numbers with names). For parsers in C, the point of insertion was
583 after the token definitions.
585 Now, Bison never inserts the pre-prologue into the header file. In the code
586 file, it always inserts it before the token definitions.
588 ** Bison now provides a more flexible alternative to the traditional Yacc
589 prologue blocks: %before-header, %start-header, %end-header, and
592 For example, the following declaration order in the grammar file reflects the
593 order in which Bison will output these code blocks. However, you are free to
594 declare these code blocks in your grammar file in whatever order is most
598 /* Bison treats this block like a pre-prologue block: it inserts it into
599 * the code file before the contents of the header file. It does *not*
600 * insert it into the header file. This is a good place to put
601 * #include's that you want at the top of your code file. A common
602 * example is `#include "system.h"'. */
605 /* Bison inserts this block into both the header file and the code file.
606 * In both files, the point of insertion is before any Bison-generated
607 * token, semantic type, location type, and class definitions. This is a
608 * good place to define %union dependencies, for example. */
611 /* Unlike the traditional Yacc prologue blocks, the output order for the
612 * new %*-header blocks is not affected by their declaration position
613 * relative to any %union in the grammar file. */
616 /* Bison inserts this block into both the header file and the code file.
617 * In both files, the point of insertion is after the Bison-generated
618 * definitions. This is a good place to declare or define public
619 * functions or data structures that depend on the Bison-generated
623 /* Bison treats this block like a post-prologue block: it inserts it into
624 * the code file after the contents of the header file. It does *not*
625 * insert it into the header file. This is a good place to declare or
626 * define internal functions or data structures that depend on the
627 * Bison-generated definitions. */
630 If you have multiple occurrences of any one of the above declarations, Bison
631 will concatenate the contents in declaration order.
633 [Although we failed to mention this here in the 2.3a release, the prologue
634 alternatives were experimental, and they were rewritten in future versions.]
636 ** The option `--report=look-ahead' has been changed to `--report=lookahead'.
637 The old spelling still works, but is not documented and may be removed
640 * Changes in version 2.3, 2006-06-05:
642 ** GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
643 for compatibility with LALR(1) grammars.
645 ** It is now documented that any definition of YYSTYPE or YYLTYPE should
646 be to a type name that does not contain parentheses or brackets.
648 * Changes in version 2.2, 2006-05-19:
650 ** The distribution terms for all Bison-generated parsers now permit
651 using the parsers in nonfree programs. Previously, this permission
652 was granted only for Bison-generated LALR(1) parsers in C.
654 ** %name-prefix changes the namespace name in C++ outputs.
656 ** The C++ parsers export their token_type.
658 ** Bison now allows multiple %union declarations, and concatenates
659 their contents together.
661 ** New warning: unused values
662 Right-hand side symbols whose values are not used are reported,
663 if the symbols have destructors. For instance:
665 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
669 will trigger a warning about $$ and $5 in the first rule, and $3 in
670 the second ($1 is copied to $$ by the default rule). This example
671 most likely contains three errors, and could be rewritten as:
673 exp: exp "?" exp ":" exp
674 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
676 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
679 However, if the original actions were really intended, memory leaks
680 and all, the warnings can be suppressed by letting Bison believe the
681 values are used, e.g.:
683 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
684 | exp "+" exp { $$ = $1; (void) $3; }
687 If there are mid-rule actions, the warning is issued if no action
688 uses it. The following triggers no warning: $1 and $3 are used.
690 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
692 The warning is intended to help catching lost values and memory leaks.
693 If a value is ignored, its associated memory typically is not reclaimed.
695 ** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
696 Destructors are now called when user code invokes YYABORT, YYACCEPT,
697 and YYERROR, for all objects on the stack, other than objects
698 corresponding to the right-hand side of the current rule.
700 ** %expect, %expect-rr
701 Incorrect numbers of expected conflicts are now actual errors,
704 ** GLR, YACC parsers.
705 The %parse-params are available in the destructors (and the
706 experimental printers) as per the documentation.
708 ** Bison now warns if it finds a stray `$' or `@' in an action.
710 ** %require "VERSION"
711 This specifies that the grammar file depends on features implemented
712 in Bison version VERSION or higher.
714 ** lalr1.cc: The token and value types are now class members.
715 The tokens were defined as free form enums and cpp macros. YYSTYPE
716 was defined as a free form union. They are now class members:
717 tokens are enumerations of the `yy::parser::token' struct, and the
718 semantic values have the `yy::parser::semantic_type' type.
720 If you do not want or can update to this scheme, the directive
721 `%define "global_tokens_and_yystype" "1"' triggers the global
722 definition of tokens and YYSTYPE. This change is suitable both
723 for previous releases of Bison, and this one.
725 If you wish to update, then make sure older version of Bison will
726 fail using `%require "2.2"'.
728 ** DJGPP support added.
730 * Changes in version 2.1, 2005-09-16:
732 ** The C++ lalr1.cc skeleton supports %lex-param.
734 ** Bison-generated parsers now support the translation of diagnostics like
735 "syntax error" into languages other than English. The default
736 language is still English. For details, please see the new
737 Internationalization section of the Bison manual. Software
738 distributors should also see the new PACKAGING file. Thanks to
739 Bruno Haible for this new feature.
741 ** Wording in the Bison-generated parsers has been changed slightly to
742 simplify translation. In particular, the message "memory exhausted"
743 has replaced "parser stack overflow", as the old message was not
744 always accurate for modern Bison-generated parsers.
746 ** Destructors are now called when the parser aborts, for all symbols left
747 behind on the stack. Also, the start symbol is now destroyed after a
748 successful parse. In both cases, the behavior was formerly inconsistent.
750 ** When generating verbose diagnostics, Bison-generated parsers no longer
751 quote the literal strings associated with tokens. For example, for
752 a syntax error associated with '%token NUM "number"' they might
753 print 'syntax error, unexpected number' instead of 'syntax error,
754 unexpected "number"'.
756 * Changes in version 2.0, 2004-12-25:
758 ** Possibly-incompatible changes
760 - Bison-generated parsers no longer default to using the alloca function
761 (when available) to extend the parser stack, due to widespread
762 problems in unchecked stack-overflow detection. You can "#define
763 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
764 the manual to determine safe values for YYMAXDEPTH in that case.
766 - Error token location.
767 During error recovery, the location of the syntax error is updated
768 to cover the whole sequence covered by the error token: it includes
769 the shifted symbols thrown away during the first part of the error
770 recovery, and the lookahead rejected during the second part.
773 . Stray semicolons are no longer allowed at the start of a grammar.
774 . Semicolons are now required after in-grammar declarations.
776 - Unescaped newlines are no longer allowed in character constants or
777 string literals. They were never portable, and GCC 3.4.0 has
778 dropped support for them. Better diagnostics are now generated if
779 forget a closing quote.
781 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
785 - GLR grammars now support locations.
787 - New directive: %initial-action.
788 This directive allows the user to run arbitrary code (including
789 initializing @$) from yyparse before parsing starts.
791 - A new directive "%expect-rr N" specifies the expected number of
792 reduce/reduce conflicts in GLR parsers.
794 - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
795 This is a GNU extension.
797 - The option `--report=lookahead' was changed to `--report=look-ahead'.
798 [However, this was changed back after 2.3.]
800 - Experimental %destructor support has been added to lalr1.cc.
802 - New configure option --disable-yacc, to disable installation of the
803 yacc command and -ly library introduced in 1.875 for POSIX conformance.
807 - For now, %expect-count violations are now just warnings, not errors.
808 This is for compatibility with Bison 1.75 and earlier (when there are
809 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
810 are too many or too few shift/reduce conflicts). However, in future
811 versions of Bison we plan to improve the %expect machinery so that
812 these violations will become errors again.
814 - Within Bison itself, numbers (e.g., goto numbers) are no longer
815 arbitrarily limited to 16-bit counts.
817 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
819 * Changes in version 1.875, 2003-01-01:
821 ** The documentation license has been upgraded to version 1.2
822 of the GNU Free Documentation License.
824 ** syntax error processing
826 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
827 locations too. This fixes bugs in error-location computation.
830 It is now possible to reclaim the memory associated to symbols
831 discarded during error recovery. This feature is still experimental.
834 This new directive is preferred over YYERROR_VERBOSE.
836 - #defining yyerror to steal internal variables is discouraged.
837 It is not guaranteed to work forever.
841 - Semicolons are once again optional at the end of grammar rules.
842 This reverts to the behavior of Bison 1.33 and earlier, and improves
843 compatibility with Yacc.
845 - `parse error' -> `syntax error'
846 Bison now uniformly uses the term `syntax error'; formerly, the code
847 and manual sometimes used the term `parse error' instead. POSIX
848 requires `syntax error' in diagnostics, and it was thought better to
851 - The documentation now emphasizes that yylex and yyerror must be
852 declared before use. C99 requires this.
854 - Bison now parses C99 lexical constructs like UCNs and
855 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
857 - File names are properly escaped in C output. E.g., foo\bar.y is
858 output as "foo\\bar.y".
860 - Yacc command and library now available
861 The Bison distribution now installs a `yacc' command, as POSIX requires.
862 Also, Bison now installs a small library liby.a containing
863 implementations of Yacc-compatible yyerror and main functions.
864 This library is normally not useful, but POSIX requires it.
866 - Type clashes now generate warnings, not errors.
868 - If the user does not define YYSTYPE as a macro, Bison now declares it
869 using typedef instead of defining it as a macro.
870 For consistency, YYLTYPE is also declared instead of defined.
872 ** Other compatibility issues
874 - %union directives can now have a tag before the `{', e.g., the
875 directive `%union foo {...}' now generates the C code
876 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
877 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
878 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
879 This is for compatibility with both Yacc and Bison 1.35.
881 - `;' is output before the terminating `}' of an action, for
882 compatibility with Bison 1.35.
884 - Bison now uses a Yacc-style format for conflict reports, e.g.,
885 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
887 - `yystype' and `yyltype' are now obsolescent macros instead of being
888 typedefs or tags; they are no longer documented and are planned to be
889 withdrawn in a future release.
894 Users of Bison have to decide how they handle the portability of the
897 - `parsing stack overflow...' -> `parser stack overflow'
898 GLR parsers now report `parser stack overflow' as per the Bison manual.
900 ** Bison now warns if it detects conflicting outputs to the same file,
901 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
902 that command outputs both code and header to foo.h.
904 ** #line in output files
905 - --no-line works properly.
907 ** Bison can no longer be built by a K&R C compiler; it requires C89 or
908 later to be built. This change originally took place a few versions
909 ago, but nobody noticed until we recently asked someone to try
910 building Bison with a K&R C compiler.
912 * Changes in version 1.75, 2002-10-14:
914 ** Bison should now work on 64-bit hosts.
916 ** Indonesian translation thanks to Tedi Heriyanto.
919 Fix spurious parse errors.
922 Some people redefine yyerror to steal yyparse' private variables.
923 Reenable this trick until an official feature replaces it.
926 In agreement with POSIX and with other Yaccs, leaving a default
927 action is valid when $$ is untyped, and $1 typed:
931 but the converse remains an error:
935 ** Values of mid-rule actions
938 foo: { ... } { $$ = $1; } ...
940 was incorrectly rejected: $1 is defined in the second mid-rule
941 action, and is equal to the $$ of the first mid-rule action.
943 * Changes in version 1.50, 2002-10-04:
948 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
949 almost any context-free grammar, ambiguous or not. The new declarations
950 %dprec and %merge on grammar rules allow parse-time resolution of
951 ambiguities. Contributed by Paul Hilfinger.
953 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
954 like the Alpha, so please stick to 32-bit hosts for now.
957 When not in Yacc compatibility mode, when the output file was not
958 specified, running `bison foo/bar.y' created `foo/bar.c'. It
962 The undefined token was systematically mapped to 2 which prevented
963 the use of 2 by the user. This is no longer the case.
965 ** Unknown token numbers
966 If yylex returned an out of range value, yyparse could die. This is
970 According to POSIX, the error token must be 256.
971 Bison extends this requirement by making it a preference: *if* the
972 user specified that one of her tokens is numbered 256, then error
973 will be mapped onto another number.
975 ** Verbose error messages
976 They no longer report `..., expecting error or...' for states where
977 error recovery is possible.
980 Defaults to `$end' instead of `$'.
982 ** Error recovery now conforms to documentation and to POSIX
983 When a Bison-generated parser encounters a syntax error, it now pops
984 the stack until it finds a state that allows shifting the error
985 token. Formerly, it popped the stack until it found a state that
986 allowed some non-error action other than a default reduction on the
987 error token. The new behavior has long been the documented behavior,
988 and has long been required by POSIX. For more details, please see
989 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
990 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
993 Popped tokens and nonterminals are now reported.
996 Larger grammars are now supported (larger token numbers, larger grammar
997 size (= sum of the LHS and RHS lengths), larger LALR tables).
998 Formerly, many of these numbers ran afoul of 16-bit limits;
999 now these limits are 32 bits on most hosts.
1001 ** Explicit initial rule
1002 Bison used to play hacks with the initial rule, which the user does
1003 not write. It is now explicit, and visible in the reports and
1007 Before, Bison reported the useless rules, but, although not used,
1008 included them in the parsers. They are now actually removed.
1010 ** Useless rules, useless nonterminals
1011 They are now reported, as a warning, with their locations.
1013 ** Rules never reduced
1014 Rules that can never be reduced because of conflicts are now
1017 ** Incorrect `Token not used'
1018 On a grammar such as
1020 %token useless useful
1022 exp: '0' %prec useful;
1024 where a token was used to set the precedence of the last rule,
1025 bison reported both `useful' and `useless' as useless tokens.
1027 ** Revert the C++ namespace changes introduced in 1.31
1028 as they caused too many portability hassles.
1030 ** Default locations
1031 By an accident of design, the default computation of @$ was
1032 performed after another default computation was performed: @$ = @1.
1033 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1034 the computation of @$.
1036 ** Token end-of-file
1037 The token end of file may be specified by the user, in which case,
1038 the user symbol is used in the reports, the graphs, and the verbose
1039 error messages instead of `$end', which remains being the default.
1043 %token MYEOF 0 "end of file"
1046 This old option, which has been broken for ages, is removed.
1049 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
1050 Croatian, thanks to Denis Lackovic.
1052 ** Incorrect token definitions
1053 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
1055 ** Token definitions as enums
1056 Tokens are output both as the traditional #define's, and, provided
1057 the compiler supports ANSI C or is a C++ compiler, as enums.
1058 This lets debuggers display names instead of integers.
1061 In addition to --verbose, bison supports --report=THINGS, which
1062 produces additional information:
1064 complete the core item sets with their closure
1065 - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back]
1066 explicitly associate lookahead tokens to items
1068 describe shift/reduce conflicts solving.
1069 Bison used to systematically output this information on top of
1070 the report. Solved conflicts are now attached to their states.
1073 Previous versions don't complain when there is a type clash on
1074 the default action if the rule has a mid-rule action, such as in:
1082 ** GNU M4 is now required when using Bison.
1084 * Changes in version 1.35, 2002-03-25:
1087 Some projects use Bison's C parser with C++ compilers, and define
1088 YYSTYPE as a class. The recent adjustment of C parsers for data
1089 alignment and 64 bit architectures made this impossible.
1091 Because for the time being no real solution for C++ parser
1092 generation exists, kludges were implemented in the parser to
1093 maintain this use. In the future, when Bison has C++ parsers, this
1094 kludge will be disabled.
1096 This kludge also addresses some C++ problems when the stack was
1099 * Changes in version 1.34, 2002-03-12:
1101 ** File name clashes are detected
1102 $ bison foo.y -d -o foo.x
1103 fatal error: header and parser would both be named `foo.x'
1105 ** A missing `;' at the end of a rule triggers a warning
1106 In accordance with POSIX, and in agreement with other
1107 Yacc implementations, Bison will mandate this semicolon in the near
1108 future. This eases the implementation of a Bison parser of Bison
1109 grammars by making this grammar LALR(1) instead of LR(2). To
1110 facilitate the transition, this release introduces a warning.
1112 ** Revert the C++ namespace changes introduced in 1.31, as they caused too
1113 many portability hassles.
1115 ** DJGPP support added.
1117 ** Fix test suite portability problems.
1119 * Changes in version 1.33, 2002-02-07:
1122 Groff could not be compiled for the definition of size_t was lacking
1123 under some conditions.
1128 * Changes in version 1.32, 2002-01-23:
1130 ** Fix Yacc output file names
1132 ** Portability fixes
1134 ** Italian, Dutch translations
1136 * Changes in version 1.31, 2002-01-14:
1140 ** GNU Gettext and %expect
1141 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1142 Bison dies on incorrect %expectations, we fear there will be
1143 too many bug reports for Gettext, so _for the time being_, %expect
1144 does not trigger an error when the input file is named `plural.y'.
1146 ** Use of alloca in parsers
1147 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1148 malloc exclusively. Since 1.29, but was not NEWS'ed.
1150 alloca is used only when compiled with GCC, to avoid portability
1153 ** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
1155 ** When the generated parser lacks debugging code, YYDEBUG is now 0
1156 (as POSIX requires) instead of being undefined.
1159 Bison has always permitted actions such as { $$ = $1 }: it adds the
1160 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1161 is no longer output: one has to write { $$ = $1; }.
1163 ** Better C++ compliance
1164 The output parsers try to respect C++ namespaces.
1165 [This turned out to be a failed experiment, and it was reverted later.]
1168 Fixed bugs when reporting useless nonterminals.
1171 The parsers work properly on 64 bit hosts.
1174 Some calls to strerror resulted in scrambled or missing error messages.
1177 When the number of shift/reduce conflicts is correct, don't issue
1180 ** The verbose report includes the rule line numbers.
1182 ** Rule line numbers are fixed in traces.
1184 ** Swedish translation
1187 Verbose parse error messages from the parsers are better looking.
1188 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1189 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1191 ** Fixed parser memory leaks.
1192 When the generated parser was using malloc to extend its stacks, the
1193 previous allocations were not freed.
1195 ** Fixed verbose output file.
1196 Some newlines were missing.
1197 Some conflicts in state descriptions were missing.
1199 ** Fixed conflict report.
1200 Option -v was needed to get the result.
1204 Mismatches are errors, not warnings.
1206 ** Fixed incorrect processing of some invalid input.
1208 ** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
1210 ** Fixed some typos in the documentation.
1212 ** %token MY_EOF 0 is supported.
1213 Before, MY_EOF was silently renumbered as 257.
1215 ** doc/refcard.tex is updated.
1217 ** %output, %file-prefix, %name-prefix.
1221 New, aliasing `--output-file'.
1223 * Changes in version 1.30, 2001-10-26:
1225 ** `--defines' and `--graph' have now an optional argument which is the
1226 output file name. `-d' and `-g' do not change; they do not take any
1229 ** `%source_extension' and `%header_extension' are removed, failed
1232 ** Portability fixes.
1234 * Changes in version 1.29, 2001-09-07:
1236 ** The output file does not define const, as this caused problems when used
1237 with common autoconfiguration schemes. If you still use ancient compilers
1238 that lack const, compile with the equivalent of the C compiler option
1239 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
1241 ** Added `-g' and `--graph'.
1243 ** The Bison manual is now distributed under the terms of the GNU FDL.
1245 ** The input and the output files has automatically a similar extension.
1247 ** Russian translation added.
1249 ** NLS support updated; should hopefully be less troublesome.
1251 ** Added the old Bison reference card.
1253 ** Added `--locations' and `%locations'.
1255 ** Added `-S' and `--skeleton'.
1257 ** `%raw', `-r', `--raw' is disabled.
1259 ** Special characters are escaped when output. This solves the problems
1260 of the #line lines with path names including backslashes.
1263 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
1264 `%debug', `%source_extension' and `%header_extension'.
1267 Automatic location tracking.
1269 * Changes in version 1.28, 1999-07-06:
1271 ** Should compile better now with K&R compilers.
1275 ** Fixed a problem with escaping the double quote character.
1277 ** There is now a FAQ.
1279 * Changes in version 1.27:
1281 ** The make rule which prevented bison.simple from being created on
1282 some systems has been fixed.
1284 * Changes in version 1.26:
1286 ** Bison now uses automake.
1288 ** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
1290 ** Token numbers now start at 257 as previously documented, not 258.
1292 ** Bison honors the TMPDIR environment variable.
1294 ** A couple of buffer overruns have been fixed.
1296 ** Problems when closing files should now be reported.
1298 ** Generated parsers should now work even on operating systems which do
1299 not provide alloca().
1301 * Changes in version 1.25, 1995-10-16:
1303 ** Errors in the input grammar are not fatal; Bison keeps reading
1304 the grammar file, and reports all the errors found in it.
1306 ** Tokens can now be specified as multiple-character strings: for
1307 example, you could use "<=" for a token which looks like <=, instead
1308 of chosing a name like LESSEQ.
1310 ** The %token_table declaration says to write a table of tokens (names
1311 and numbers) into the parser file. The yylex function can use this
1312 table to recognize multiple-character string tokens, or for other
1315 ** The %no_lines declaration says not to generate any #line preprocessor
1316 directives in the parser file.
1318 ** The %raw declaration says to use internal Bison token numbers, not
1319 Yacc-compatible token numbers, when token names are defined as macros.
1321 ** The --no-parser option produces the parser tables without including
1322 the parser engine; a project can now use its own parser engine.
1323 The actions go into a separate file called NAME.act, in the form of
1324 a switch statement body.
1326 * Changes in version 1.23:
1328 The user can define YYPARSE_PARAM as the name of an argument to be
1329 passed into yyparse. The argument should have type void *. It should
1330 actually point to an object. Grammar actions can access the variable
1331 by casting it to the proper pointer type.
1333 Line numbers in output file corrected.
1335 * Changes in version 1.22:
1337 --help option added.
1339 * Changes in version 1.20:
1341 Output file does not redefine const for C++.
1349 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
1350 2004, 2005, 2006, 2007, 2008, 2009 Free Software Foundation, Inc.
1352 This file is part of Bison, the GNU Parser Generator.
1354 This program is free software: you can redistribute it and/or modify
1355 it under the terms of the GNU General Public License as published by
1356 the Free Software Foundation, either version 3 of the License, or
1357 (at your option) any later version.
1359 This program is distributed in the hope that it will be useful,
1360 but WITHOUT ANY WARRANTY; without even the implied warranty of
1361 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1362 GNU General Public License for more details.
1364 You should have received a copy of the GNU General Public License
1365 along with this program. If not, see <http://www.gnu.org/licenses/>.