4 * Changes in version 2.5 (????-??-??):
6 ** Named References Support
8 Historically, Yacc and Bison have supported positional references
9 ($n, $$) to allow access to symbol values from inside of semantic
12 Starting from this version, Bison can also accept named references.
13 When no ambiguity is possible, original symbol names may be used
16 if_stmt : 'if' cond_expr 'then' then_stmt ';'
17 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
19 In the more common case, explicit names may be declared:
21 stmt[res] : 'if' expr[cond] 'then' stmt[then] 'else' stmt[else] ';'
22 { $res = mk_if_stmt($cond, $then, $else); }
24 Location information is also accessible using @name syntax. When
25 accessing symbol names containing dots or dashes, explicit bracketing
26 ($[sym.1]) must be used.
28 These features are experimental in this version. More user feedback
29 will help to stabilize them.
31 ** IELR(1) and Canonical LR(1) Support
33 IELR(1) is a minimal LR(1) parser table generation algorithm. That
34 is, given any context-free grammar, IELR(1) generates parser tables
35 with the full language recognition power of canonical LR(1) but with
36 nearly the same number of parser states as LALR(1). This reduction in
37 parser states is often an order of magnitude. More importantly,
38 because canonical LR(1)'s extra parser states may contain duplicate
39 conflicts in the case of non-LR(1) grammars, the number of conflicts
40 for IELR(1) is often an order of magnitude less as well. This can
41 significantly reduce the complexity of developing of a grammar.
43 Bison can now generate IELR(1) and canonical LR(1) parser tables in
44 place of its traditional LALR(1) parser tables, which remain the
45 default. You can specify the type of parser tables in the grammar
46 file with these directives:
50 %define lr.type canonical-lr
52 The default reduction optimization in the parser tables can also be
53 adjusted using `%define lr.default-reductions'. See the documentation
54 for `%define lr.type' and `%define lr.default-reductions' in the
55 section `Bison Declaration Summary' in the Bison manual for the
58 These features are experimental. More user feedback will help to
61 ** Unrecognized %code qualifiers are now an error not a warning.
63 ** %define improvements.
65 *** Unrecognized variables are now an error not a warning.
67 *** Multiple invocations for any variable is now an error not a warning.
69 *** Can now be invoked via the command line.
71 Each of these command-line options
77 --force-define=NAME[=VALUE]
79 is equivalent to this grammar file declaration
81 %define NAME ["VALUE"]
83 except that the manner in which Bison processes multiple definitions
84 for the same NAME differs. Most importantly, -F and --force-define
85 quietly override %define, but -D and --define do not. For further
86 details, see the section "Bison Options" in the Bison manual.
88 *** Variables renamed.
90 The following %define variables
93 lr.keep_unreachable_states
98 lr.keep-unreachable-states
100 The old names are now deprecated but will be maintained indefinitely
101 for backward compatibility.
103 *** Values no longer need to be quoted in grammar file.
105 If a %define value is an identifier, it no longer needs to be placed
106 within quotations marks. For example,
108 %define api.push-pull "push"
112 %define api.push-pull push
116 Consistently with directives (such as %error-verbose) and variables
117 (e.g. push-pull), symbol names may include dashes in any position,
118 similarly to periods and underscores. This is GNU extension over
119 POSIX Yacc whose use is reported by -Wyacc, and rejected in Yacc
122 ** YYFAIL now produces warnings and Java parsers no longer implement it.
124 YYFAIL has existed for many years as an undocumented feature of
125 deterministic parsers in C generated by Bison. More recently, it was
126 a documented feature of Bison's experimental Java parsers. As
127 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
128 semantic action now produces a deprecation warning, and Java parsers
129 no longer implement YYFAIL at all. For further details, including a
130 discussion of how to suppress C preprocessor warnings about YYFAIL
131 being unused, see the Bison 2.4.2 NEWS entry.
133 ** Temporary hack for adding a semicolon to the user action.
135 Previously, Bison appended a semicolon to every user action for
136 reductions when the output language defaulted to C (specifically, when
137 neither %yacc, %language, %skeleton, or equivalent command-line
138 options were specified). This allowed actions such as
140 exp: exp "+" exp { $$ = $1 + $3 };
144 exp: exp "+" exp { $$ = $1 + $3; };
146 As a first step in removing this misfeature, Bison now issues a
147 warning when it appends a semicolon. Moreover, in cases where Bison
148 cannot easily determine whether a semicolon is needed (for example, an
149 action ending with a cpp directive or a braced compound initializer),
150 it no longer appends one. Thus, the C compiler might now complain
151 about a missing semicolon where it did not before. Future releases of
152 Bison will cease to append semicolons entirely.
154 ** Character literals not of length one.
156 Previously, Bison quietly converted all character literals to length
157 one. For example, without warning, Bison interpreted the operators in
158 the following grammar to be the same token:
164 Bison now warns when a character literal is not of length one. In
165 some future release, Bison will report an error instead.
167 ** Verbose error messages fixed for nonassociative tokens.
169 When %error-verbose is specified, syntax error messages produced by
170 the generated parser include the unexpected token as well as a list of
171 expected tokens. Previously, this list erroneously included tokens
172 that would actually induce a syntax error because conflicts for them
173 were resolved with %nonassoc. Such tokens are now properly omitted
176 ** Destructor calls fixed for lookaheads altered in semantic actions.
178 Previously for deterministic parsers in C, if a user semantic action
179 altered yychar, the parser in some cases used the old yychar value to
180 determine which destructor to call for the lookahead upon a syntax
181 error or upon parser return. This bug has been fixed.
183 * Changes in version 2.4.3 (????-??-??):
185 ** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
188 ** Failures in the test suite for GCC 4.5 have been fixed.
190 ** Failures in the test suite for some versions of Sun Studio C++ have
193 ** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
194 warnings about undefined %prec identifiers will not be converted to
195 errors in Bison 2.5. They will remain warnings, which should be
196 sufficient for POSIX while avoiding backward compatibility issues.
198 * Changes in version 2.4.2 (2010-03-20):
200 ** Some portability problems that resulted in failures and livelocks
201 in the test suite on some versions of at least Solaris, AIX, HP-UX,
202 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
203 errors should no longer cause M4 to report a broken pipe on the
206 ** `%prec IDENTIFIER' requires IDENTIFIER to be defined separately.
208 POSIX specifies that an error be reported for any identifier that does
209 not appear on the LHS of a grammar rule and that is not defined by
210 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
211 error report for the case when an identifier appears only after a
212 %prec directive. It is now restored. However, for backward
213 compatibility with recent Bison releases, it is only a warning for
214 now. In Bison 2.5 and later, it will return to being an error.
215 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
216 warning will not be converted to an error in Bison 2.5.]
218 ** Detection of GNU M4 1.4.6 or newer during configure is improved.
220 ** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
221 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
224 ** %code is now a permanent feature.
226 A traditional Yacc prologue directive is written in the form:
230 To provide a more flexible alternative, Bison 2.3b introduced the
231 %code directive with the following forms for C/C++:
234 %code requires {CODE}
235 %code provides {CODE}
238 These forms are now considered permanent features of Bison. See the
239 %code entries in the section "Bison Declaration Summary" in the Bison
240 manual for a summary of their functionality. See the section
241 "Prologue Alternatives" for a detailed discussion including the
242 advantages of %code over the traditional Yacc prologue directive.
244 Bison's Java feature as a whole including its current usage of %code
245 is still considered experimental.
247 ** YYFAIL is deprecated and will eventually be removed.
249 YYFAIL has existed for many years as an undocumented feature of
250 deterministic parsers in C generated by Bison. Previously, it was
251 documented for Bison's experimental Java parsers. YYFAIL is no longer
252 documented for Java parsers and is formally deprecated in both cases.
253 Users are strongly encouraged to migrate to YYERROR, which is
256 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
257 induce a syntax error. The most obvious difference from YYERROR is
258 that YYFAIL will automatically invoke yyerror to report the syntax
259 error so that you don't have to. However, there are several other
260 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
261 inherent flaws when %error-verbose or `#define YYERROR_VERBOSE' is
262 used. For a more detailed discussion, see:
264 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
266 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
267 deterministic parsers in C will continue to implement it. However,
268 because YYFAIL is already flawed, it seems futile to try to make new
269 Bison features compatible with it. Thus, during parser generation,
270 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
271 rule action. In a later release, YYFAIL will be disabled for
272 %error-verbose and `#define YYERROR_VERBOSE'. Eventually, YYFAIL will
273 be removed altogether.
275 There exists at least one case where Bison 2.5's YYFAIL warning will
276 be a false positive. Some projects add phony uses of YYFAIL and other
277 Bison-defined macros for the sole purpose of suppressing C
278 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
279 To avoid Bison's future warning, such YYFAIL uses can be moved to the
280 epilogue (that is, after the second `%%') in the Bison input file. In
281 this release (2.4.2), Bison already generates its own code to suppress
282 C preprocessor warnings for YYFAIL, so projects can remove their own
283 phony uses of YYFAIL if compatibility with Bison releases prior to
284 2.4.2 is not necessary.
286 ** Internationalization.
288 Fix a regression introduced in Bison 2.4: Under some circumstances,
289 message translations were not installed although supported by the
292 * Changes in version 2.4.1 (2008-12-11):
294 ** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
295 declarations have been fixed.
297 ** Temporary hack for adding a semicolon to the user action.
299 Bison used to prepend a trailing semicolon at the end of the user
300 action for reductions. This allowed actions such as
302 exp: exp "+" exp { $$ = $1 + $3 };
306 exp: exp "+" exp { $$ = $1 + $3; };
308 Some grammars still depend on this `feature'. Bison 2.4.1 restores
309 the previous behavior in the case of C output (specifically, when
310 neither %language or %skeleton or equivalent command-line options
311 are used) to leave more time for grammars depending on the old
312 behavior to be adjusted. Future releases of Bison will disable this
315 ** A few minor improvements to the Bison manual.
317 * Changes in version 2.4 (2008-11-02):
319 ** %language is an experimental feature.
321 We first introduced this feature in test release 2.3b as a cleaner
322 alternative to %skeleton. Since then, we have discussed the possibility of
323 modifying its effect on Bison's output file names. Thus, in this release,
324 we consider %language to be an experimental feature that will likely evolve
327 ** Forward compatibility with GNU M4 has been improved.
329 ** Several bugs in the C++ skeleton and the experimental Java skeleton have been
332 * Changes in version 2.3b (2008-05-27):
334 ** The quotes around NAME that used to be required in the following directive
339 ** The directive `%pure-parser' is now deprecated in favor of:
343 which has the same effect except that Bison is more careful to warn about
344 unreasonable usage in the latter case.
348 Bison can now generate an LALR(1) parser in C with a push interface. That
349 is, instead of invoking `yyparse', which pulls tokens from `yylex', you can
350 push one token at a time to the parser using `yypush_parse', which will
351 return to the caller after processing each token. By default, the push
352 interface is disabled. Either of the following directives will enable it:
354 %define api.push_pull "push" // Just push; does not require yylex.
355 %define api.push_pull "both" // Push and pull; requires yylex.
357 See the new section `A Push Parser' in the Bison manual for details.
359 The current push parsing interface is experimental and may evolve. More user
360 feedback will help to stabilize it.
362 ** The -g and --graph options now output graphs in Graphviz DOT format,
363 not VCG format. Like --graph, -g now also takes an optional FILE argument
364 and thus cannot be bundled with other short options.
368 Bison can now generate an LALR(1) parser in Java. The skeleton is
369 `data/lalr1.java'. Consider using the new %language directive instead of
370 %skeleton to select it.
372 See the new section `Java Parsers' in the Bison manual for details.
374 The current Java interface is experimental and may evolve. More user
375 feedback will help to stabilize it.
379 This new directive specifies the programming language of the generated
380 parser, which can be C (the default), C++, or Java. Besides the skeleton
381 that Bison uses, the directive affects the names of the generated files if
382 the grammar file's name ends in ".y".
384 ** XML Automaton Report
386 Bison can now generate an XML report of the LALR(1) automaton using the new
387 `--xml' option. The current XML schema is experimental and may evolve. More
388 user feedback will help to stabilize it.
390 ** The grammar file may now specify the name of the parser header file using
391 %defines. For example:
395 ** When reporting useless rules, useless nonterminals, and unused terminals,
396 Bison now employs the terms "useless in grammar" instead of "useless",
397 "useless in parser" instead of "never reduced", and "unused in grammar"
400 ** Unreachable State Removal
402 Previously, Bison sometimes generated parser tables containing unreachable
403 states. A state can become unreachable during conflict resolution if Bison
404 disables a shift action leading to it from a predecessor state. Bison now:
406 1. Removes unreachable states.
408 2. Does not report any conflicts that appeared in unreachable states.
409 WARNING: As a result, you may need to update %expect and %expect-rr
410 directives in existing grammar files.
412 3. For any rule used only in such states, Bison now reports the rule as
413 "useless in parser due to conflicts".
415 This feature can be disabled with the following directive:
417 %define lr.keep_unreachable_states
419 See the %define entry in the `Bison Declaration Summary' in the Bison manual
420 for further discussion.
422 ** Lookahead Set Correction in the `.output' Report
424 When instructed to generate a `.output' file including lookahead sets
425 (using `--report=lookahead', for example), Bison now prints each reduction's
426 lookahead set only next to the associated state's one item that (1) is
427 associated with the same rule as the reduction and (2) has its dot at the end
428 of its RHS. Previously, Bison also erroneously printed the lookahead set
429 next to all of the state's other items associated with the same rule. This
430 bug affected only the `.output' file and not the generated parser source
433 ** --report-file=FILE is a new option to override the default `.output' file
436 ** The `=' that used to be required in the following directives is now
439 %file-prefix "parser"
443 ** An Alternative to `%{...%}' -- `%code QUALIFIER {CODE}'
445 Bison 2.3a provided a new set of directives as a more flexible alternative to
446 the traditional Yacc prologue blocks. Those have now been consolidated into
447 a single %code directive with an optional qualifier field, which identifies
448 the purpose of the code and thus the location(s) where Bison should generate
451 1. `%code {CODE}' replaces `%after-header {CODE}'
452 2. `%code requires {CODE}' replaces `%start-header {CODE}'
453 3. `%code provides {CODE}' replaces `%end-header {CODE}'
454 4. `%code top {CODE}' replaces `%before-header {CODE}'
456 See the %code entries in section `Bison Declaration Summary' in the Bison
457 manual for a summary of the new functionality. See the new section `Prologue
458 Alternatives' for a detailed discussion including the advantages of %code
459 over the traditional Yacc prologues.
461 The prologue alternatives are experimental. More user feedback will help to
462 determine whether they should become permanent features.
464 ** Revised warning: unset or unused mid-rule values
466 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
467 used within any of the actions of the parent rule. For example, Bison warns
470 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
472 Now, Bison also warns about mid-rule values that are used but not set. For
473 example, Bison warns about unset $$ in the mid-rule action in:
475 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
477 However, Bison now disables both of these warnings by default since they
478 sometimes prove to be false alarms in existing grammars employing the Yacc
479 constructs $0 or $-N (where N is some positive integer).
481 To enable these warnings, specify the option `--warnings=midrule-values' or
482 `-W', which is a synonym for `--warnings=all'.
484 ** Default %destructor or %printer with `<*>' or `<>'
486 Bison now recognizes two separate kinds of default %destructor's and
489 1. Place `<*>' in a %destructor/%printer symbol list to define a default
490 %destructor/%printer for all grammar symbols for which you have formally
491 declared semantic type tags.
493 2. Place `<>' in a %destructor/%printer symbol list to define a default
494 %destructor/%printer for all grammar symbols without declared semantic
497 Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
498 `<*>' and `<>' combined achieve the same effect with one exception: Bison no
499 longer applies any %destructor to a mid-rule value if that mid-rule value is
500 not actually ever referenced using either $$ or $n in a semantic action.
502 The default %destructor's and %printer's are experimental. More user
503 feedback will help to determine whether they should become permanent
506 See the section `Freeing Discarded Symbols' in the Bison manual for further
509 ** %left, %right, and %nonassoc can now declare token numbers. This is required
510 by POSIX. However, see the end of section `Operator Precedence' in the Bison
511 manual for a caveat concerning the treatment of literal strings.
513 ** The nonfunctional --no-parser, -n, and %no-parser options have been
514 completely removed from Bison.
516 * Changes in version 2.3a, 2006-09-13:
518 ** Instead of %union, you can define and use your own union type
519 YYSTYPE if your grammar contains at least one <type> tag.
520 Your YYSTYPE need not be a macro; it can be a typedef.
521 This change is for compatibility with other Yacc implementations,
522 and is required by POSIX.
524 ** Locations columns and lines start at 1.
525 In accordance with the GNU Coding Standards and Emacs.
527 ** You may now declare per-type and default %destructor's and %printer's:
531 %union { char *string; }
532 %token <string> STRING1
533 %token <string> STRING2
534 %type <string> string1
535 %type <string> string2
536 %union { char character; }
537 %token <character> CHR
538 %type <character> chr
539 %destructor { free ($$); } %symbol-default
540 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
541 %destructor { } <character>
543 guarantees that, when the parser discards any user-defined symbol that has a
544 semantic type tag other than `<character>', it passes its semantic value to
545 `free'. However, when the parser discards a `STRING1' or a `string1', it
546 also prints its line number to `stdout'. It performs only the second
547 `%destructor' in this case, so it invokes `free' only once.
549 [Although we failed to mention this here in the 2.3a release, the default
550 %destructor's and %printer's were experimental, and they were rewritten in
553 ** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
554 `--yacc', or `%yacc'), Bison no longer generates #define statements for
555 associating token numbers with token names. Removing the #define statements
556 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
557 requires them. Bison still generates an enum for token names in all cases.
559 ** Handling of traditional Yacc prologue blocks is now more consistent but
560 potentially incompatible with previous releases of Bison.
562 As before, you declare prologue blocks in your grammar file with the
563 `%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all
564 prologue blocks that you've declared before the first %union. To generate
565 the post-prologue, Bison concatenates all prologue blocks that you've
566 declared after the first %union.
568 Previous releases of Bison inserted the pre-prologue into both the header
569 file and the code file in all cases except for LALR(1) parsers in C. In the
570 latter case, Bison inserted it only into the code file. For parsers in C++,
571 the point of insertion was before any token definitions (which associate
572 token numbers with names). For parsers in C, the point of insertion was
573 after the token definitions.
575 Now, Bison never inserts the pre-prologue into the header file. In the code
576 file, it always inserts it before the token definitions.
578 ** Bison now provides a more flexible alternative to the traditional Yacc
579 prologue blocks: %before-header, %start-header, %end-header, and
582 For example, the following declaration order in the grammar file reflects the
583 order in which Bison will output these code blocks. However, you are free to
584 declare these code blocks in your grammar file in whatever order is most
588 /* Bison treats this block like a pre-prologue block: it inserts it into
589 * the code file before the contents of the header file. It does *not*
590 * insert it into the header file. This is a good place to put
591 * #include's that you want at the top of your code file. A common
592 * example is `#include "system.h"'. */
595 /* Bison inserts this block into both the header file and the code file.
596 * In both files, the point of insertion is before any Bison-generated
597 * token, semantic type, location type, and class definitions. This is a
598 * good place to define %union dependencies, for example. */
601 /* Unlike the traditional Yacc prologue blocks, the output order for the
602 * new %*-header blocks is not affected by their declaration position
603 * relative to any %union in the grammar file. */
606 /* Bison inserts this block into both the header file and the code file.
607 * In both files, the point of insertion is after the Bison-generated
608 * definitions. This is a good place to declare or define public
609 * functions or data structures that depend on the Bison-generated
613 /* Bison treats this block like a post-prologue block: it inserts it into
614 * the code file after the contents of the header file. It does *not*
615 * insert it into the header file. This is a good place to declare or
616 * define internal functions or data structures that depend on the
617 * Bison-generated definitions. */
620 If you have multiple occurrences of any one of the above declarations, Bison
621 will concatenate the contents in declaration order.
623 [Although we failed to mention this here in the 2.3a release, the prologue
624 alternatives were experimental, and they were rewritten in future versions.]
626 ** The option `--report=look-ahead' has been changed to `--report=lookahead'.
627 The old spelling still works, but is not documented and may be removed
630 * Changes in version 2.3, 2006-06-05:
632 ** GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
633 for compatibility with LALR(1) grammars.
635 ** It is now documented that any definition of YYSTYPE or YYLTYPE should
636 be to a type name that does not contain parentheses or brackets.
638 * Changes in version 2.2, 2006-05-19:
640 ** The distribution terms for all Bison-generated parsers now permit
641 using the parsers in nonfree programs. Previously, this permission
642 was granted only for Bison-generated LALR(1) parsers in C.
644 ** %name-prefix changes the namespace name in C++ outputs.
646 ** The C++ parsers export their token_type.
648 ** Bison now allows multiple %union declarations, and concatenates
649 their contents together.
651 ** New warning: unused values
652 Right-hand side symbols whose values are not used are reported,
653 if the symbols have destructors. For instance:
655 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
659 will trigger a warning about $$ and $5 in the first rule, and $3 in
660 the second ($1 is copied to $$ by the default rule). This example
661 most likely contains three errors, and could be rewritten as:
663 exp: exp "?" exp ":" exp
664 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
666 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
669 However, if the original actions were really intended, memory leaks
670 and all, the warnings can be suppressed by letting Bison believe the
671 values are used, e.g.:
673 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
674 | exp "+" exp { $$ = $1; (void) $3; }
677 If there are mid-rule actions, the warning is issued if no action
678 uses it. The following triggers no warning: $1 and $3 are used.
680 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
682 The warning is intended to help catching lost values and memory leaks.
683 If a value is ignored, its associated memory typically is not reclaimed.
685 ** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
686 Destructors are now called when user code invokes YYABORT, YYACCEPT,
687 and YYERROR, for all objects on the stack, other than objects
688 corresponding to the right-hand side of the current rule.
690 ** %expect, %expect-rr
691 Incorrect numbers of expected conflicts are now actual errors,
694 ** GLR, YACC parsers.
695 The %parse-params are available in the destructors (and the
696 experimental printers) as per the documentation.
698 ** Bison now warns if it finds a stray `$' or `@' in an action.
700 ** %require "VERSION"
701 This specifies that the grammar file depends on features implemented
702 in Bison version VERSION or higher.
704 ** lalr1.cc: The token and value types are now class members.
705 The tokens were defined as free form enums and cpp macros. YYSTYPE
706 was defined as a free form union. They are now class members:
707 tokens are enumerations of the `yy::parser::token' struct, and the
708 semantic values have the `yy::parser::semantic_type' type.
710 If you do not want or can update to this scheme, the directive
711 `%define "global_tokens_and_yystype" "1"' triggers the global
712 definition of tokens and YYSTYPE. This change is suitable both
713 for previous releases of Bison, and this one.
715 If you wish to update, then make sure older version of Bison will
716 fail using `%require "2.2"'.
718 ** DJGPP support added.
720 * Changes in version 2.1, 2005-09-16:
722 ** The C++ lalr1.cc skeleton supports %lex-param.
724 ** Bison-generated parsers now support the translation of diagnostics like
725 "syntax error" into languages other than English. The default
726 language is still English. For details, please see the new
727 Internationalization section of the Bison manual. Software
728 distributors should also see the new PACKAGING file. Thanks to
729 Bruno Haible for this new feature.
731 ** Wording in the Bison-generated parsers has been changed slightly to
732 simplify translation. In particular, the message "memory exhausted"
733 has replaced "parser stack overflow", as the old message was not
734 always accurate for modern Bison-generated parsers.
736 ** Destructors are now called when the parser aborts, for all symbols left
737 behind on the stack. Also, the start symbol is now destroyed after a
738 successful parse. In both cases, the behavior was formerly inconsistent.
740 ** When generating verbose diagnostics, Bison-generated parsers no longer
741 quote the literal strings associated with tokens. For example, for
742 a syntax error associated with '%token NUM "number"' they might
743 print 'syntax error, unexpected number' instead of 'syntax error,
744 unexpected "number"'.
746 * Changes in version 2.0, 2004-12-25:
748 ** Possibly-incompatible changes
750 - Bison-generated parsers no longer default to using the alloca function
751 (when available) to extend the parser stack, due to widespread
752 problems in unchecked stack-overflow detection. You can "#define
753 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
754 the manual to determine safe values for YYMAXDEPTH in that case.
756 - Error token location.
757 During error recovery, the location of the syntax error is updated
758 to cover the whole sequence covered by the error token: it includes
759 the shifted symbols thrown away during the first part of the error
760 recovery, and the lookahead rejected during the second part.
763 . Stray semicolons are no longer allowed at the start of a grammar.
764 . Semicolons are now required after in-grammar declarations.
766 - Unescaped newlines are no longer allowed in character constants or
767 string literals. They were never portable, and GCC 3.4.0 has
768 dropped support for them. Better diagnostics are now generated if
769 forget a closing quote.
771 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
775 - GLR grammars now support locations.
777 - New directive: %initial-action.
778 This directive allows the user to run arbitrary code (including
779 initializing @$) from yyparse before parsing starts.
781 - A new directive "%expect-rr N" specifies the expected number of
782 reduce/reduce conflicts in GLR parsers.
784 - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
785 This is a GNU extension.
787 - The option `--report=lookahead' was changed to `--report=look-ahead'.
788 [However, this was changed back after 2.3.]
790 - Experimental %destructor support has been added to lalr1.cc.
792 - New configure option --disable-yacc, to disable installation of the
793 yacc command and -ly library introduced in 1.875 for POSIX conformance.
797 - For now, %expect-count violations are now just warnings, not errors.
798 This is for compatibility with Bison 1.75 and earlier (when there are
799 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
800 are too many or too few shift/reduce conflicts). However, in future
801 versions of Bison we plan to improve the %expect machinery so that
802 these violations will become errors again.
804 - Within Bison itself, numbers (e.g., goto numbers) are no longer
805 arbitrarily limited to 16-bit counts.
807 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
809 * Changes in version 1.875, 2003-01-01:
811 ** The documentation license has been upgraded to version 1.2
812 of the GNU Free Documentation License.
814 ** syntax error processing
816 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
817 locations too. This fixes bugs in error-location computation.
820 It is now possible to reclaim the memory associated to symbols
821 discarded during error recovery. This feature is still experimental.
824 This new directive is preferred over YYERROR_VERBOSE.
826 - #defining yyerror to steal internal variables is discouraged.
827 It is not guaranteed to work forever.
831 - Semicolons are once again optional at the end of grammar rules.
832 This reverts to the behavior of Bison 1.33 and earlier, and improves
833 compatibility with Yacc.
835 - `parse error' -> `syntax error'
836 Bison now uniformly uses the term `syntax error'; formerly, the code
837 and manual sometimes used the term `parse error' instead. POSIX
838 requires `syntax error' in diagnostics, and it was thought better to
841 - The documentation now emphasizes that yylex and yyerror must be
842 declared before use. C99 requires this.
844 - Bison now parses C99 lexical constructs like UCNs and
845 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
847 - File names are properly escaped in C output. E.g., foo\bar.y is
848 output as "foo\\bar.y".
850 - Yacc command and library now available
851 The Bison distribution now installs a `yacc' command, as POSIX requires.
852 Also, Bison now installs a small library liby.a containing
853 implementations of Yacc-compatible yyerror and main functions.
854 This library is normally not useful, but POSIX requires it.
856 - Type clashes now generate warnings, not errors.
858 - If the user does not define YYSTYPE as a macro, Bison now declares it
859 using typedef instead of defining it as a macro.
860 For consistency, YYLTYPE is also declared instead of defined.
862 ** Other compatibility issues
864 - %union directives can now have a tag before the `{', e.g., the
865 directive `%union foo {...}' now generates the C code
866 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
867 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
868 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
869 This is for compatibility with both Yacc and Bison 1.35.
871 - `;' is output before the terminating `}' of an action, for
872 compatibility with Bison 1.35.
874 - Bison now uses a Yacc-style format for conflict reports, e.g.,
875 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
877 - `yystype' and `yyltype' are now obsolescent macros instead of being
878 typedefs or tags; they are no longer documented and are planned to be
879 withdrawn in a future release.
884 Users of Bison have to decide how they handle the portability of the
887 - `parsing stack overflow...' -> `parser stack overflow'
888 GLR parsers now report `parser stack overflow' as per the Bison manual.
890 ** Bison now warns if it detects conflicting outputs to the same file,
891 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
892 that command outputs both code and header to foo.h.
894 ** #line in output files
895 - --no-line works properly.
897 ** Bison can no longer be built by a K&R C compiler; it requires C89 or
898 later to be built. This change originally took place a few versions
899 ago, but nobody noticed until we recently asked someone to try
900 building Bison with a K&R C compiler.
902 * Changes in version 1.75, 2002-10-14:
904 ** Bison should now work on 64-bit hosts.
906 ** Indonesian translation thanks to Tedi Heriyanto.
909 Fix spurious parse errors.
912 Some people redefine yyerror to steal yyparse' private variables.
913 Reenable this trick until an official feature replaces it.
916 In agreement with POSIX and with other Yaccs, leaving a default
917 action is valid when $$ is untyped, and $1 typed:
921 but the converse remains an error:
925 ** Values of mid-rule actions
928 foo: { ... } { $$ = $1; } ...
930 was incorrectly rejected: $1 is defined in the second mid-rule
931 action, and is equal to the $$ of the first mid-rule action.
933 * Changes in version 1.50, 2002-10-04:
938 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
939 almost any context-free grammar, ambiguous or not. The new declarations
940 %dprec and %merge on grammar rules allow parse-time resolution of
941 ambiguities. Contributed by Paul Hilfinger.
943 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
944 like the Alpha, so please stick to 32-bit hosts for now.
947 When not in Yacc compatibility mode, when the output file was not
948 specified, running `bison foo/bar.y' created `foo/bar.c'. It
952 The undefined token was systematically mapped to 2 which prevented
953 the use of 2 by the user. This is no longer the case.
955 ** Unknown token numbers
956 If yylex returned an out of range value, yyparse could die. This is
960 According to POSIX, the error token must be 256.
961 Bison extends this requirement by making it a preference: *if* the
962 user specified that one of her tokens is numbered 256, then error
963 will be mapped onto another number.
965 ** Verbose error messages
966 They no longer report `..., expecting error or...' for states where
967 error recovery is possible.
970 Defaults to `$end' instead of `$'.
972 ** Error recovery now conforms to documentation and to POSIX
973 When a Bison-generated parser encounters a syntax error, it now pops
974 the stack until it finds a state that allows shifting the error
975 token. Formerly, it popped the stack until it found a state that
976 allowed some non-error action other than a default reduction on the
977 error token. The new behavior has long been the documented behavior,
978 and has long been required by POSIX. For more details, please see
979 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
980 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
983 Popped tokens and nonterminals are now reported.
986 Larger grammars are now supported (larger token numbers, larger grammar
987 size (= sum of the LHS and RHS lengths), larger LALR tables).
988 Formerly, many of these numbers ran afoul of 16-bit limits;
989 now these limits are 32 bits on most hosts.
991 ** Explicit initial rule
992 Bison used to play hacks with the initial rule, which the user does
993 not write. It is now explicit, and visible in the reports and
997 Before, Bison reported the useless rules, but, although not used,
998 included them in the parsers. They are now actually removed.
1000 ** Useless rules, useless nonterminals
1001 They are now reported, as a warning, with their locations.
1003 ** Rules never reduced
1004 Rules that can never be reduced because of conflicts are now
1007 ** Incorrect `Token not used'
1008 On a grammar such as
1010 %token useless useful
1012 exp: '0' %prec useful;
1014 where a token was used to set the precedence of the last rule,
1015 bison reported both `useful' and `useless' as useless tokens.
1017 ** Revert the C++ namespace changes introduced in 1.31
1018 as they caused too many portability hassles.
1020 ** Default locations
1021 By an accident of design, the default computation of @$ was
1022 performed after another default computation was performed: @$ = @1.
1023 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1024 the computation of @$.
1026 ** Token end-of-file
1027 The token end of file may be specified by the user, in which case,
1028 the user symbol is used in the reports, the graphs, and the verbose
1029 error messages instead of `$end', which remains being the default.
1033 %token MYEOF 0 "end of file"
1036 This old option, which has been broken for ages, is removed.
1039 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
1040 Croatian, thanks to Denis Lackovic.
1042 ** Incorrect token definitions
1043 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
1045 ** Token definitions as enums
1046 Tokens are output both as the traditional #define's, and, provided
1047 the compiler supports ANSI C or is a C++ compiler, as enums.
1048 This lets debuggers display names instead of integers.
1051 In addition to --verbose, bison supports --report=THINGS, which
1052 produces additional information:
1054 complete the core item sets with their closure
1055 - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back]
1056 explicitly associate lookahead tokens to items
1058 describe shift/reduce conflicts solving.
1059 Bison used to systematically output this information on top of
1060 the report. Solved conflicts are now attached to their states.
1063 Previous versions don't complain when there is a type clash on
1064 the default action if the rule has a mid-rule action, such as in:
1072 ** GNU M4 is now required when using Bison.
1074 * Changes in version 1.35, 2002-03-25:
1077 Some projects use Bison's C parser with C++ compilers, and define
1078 YYSTYPE as a class. The recent adjustment of C parsers for data
1079 alignment and 64 bit architectures made this impossible.
1081 Because for the time being no real solution for C++ parser
1082 generation exists, kludges were implemented in the parser to
1083 maintain this use. In the future, when Bison has C++ parsers, this
1084 kludge will be disabled.
1086 This kludge also addresses some C++ problems when the stack was
1089 * Changes in version 1.34, 2002-03-12:
1091 ** File name clashes are detected
1092 $ bison foo.y -d -o foo.x
1093 fatal error: header and parser would both be named `foo.x'
1095 ** A missing `;' at the end of a rule triggers a warning
1096 In accordance with POSIX, and in agreement with other
1097 Yacc implementations, Bison will mandate this semicolon in the near
1098 future. This eases the implementation of a Bison parser of Bison
1099 grammars by making this grammar LALR(1) instead of LR(2). To
1100 facilitate the transition, this release introduces a warning.
1102 ** Revert the C++ namespace changes introduced in 1.31, as they caused too
1103 many portability hassles.
1105 ** DJGPP support added.
1107 ** Fix test suite portability problems.
1109 * Changes in version 1.33, 2002-02-07:
1112 Groff could not be compiled for the definition of size_t was lacking
1113 under some conditions.
1118 * Changes in version 1.32, 2002-01-23:
1120 ** Fix Yacc output file names
1122 ** Portability fixes
1124 ** Italian, Dutch translations
1126 * Changes in version 1.31, 2002-01-14:
1130 ** GNU Gettext and %expect
1131 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1132 Bison dies on incorrect %expectations, we fear there will be
1133 too many bug reports for Gettext, so _for the time being_, %expect
1134 does not trigger an error when the input file is named `plural.y'.
1136 ** Use of alloca in parsers
1137 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1138 malloc exclusively. Since 1.29, but was not NEWS'ed.
1140 alloca is used only when compiled with GCC, to avoid portability
1143 ** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
1145 ** When the generated parser lacks debugging code, YYDEBUG is now 0
1146 (as POSIX requires) instead of being undefined.
1149 Bison has always permitted actions such as { $$ = $1 }: it adds the
1150 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1151 is no longer output: one has to write { $$ = $1; }.
1153 ** Better C++ compliance
1154 The output parsers try to respect C++ namespaces.
1155 [This turned out to be a failed experiment, and it was reverted later.]
1158 Fixed bugs when reporting useless nonterminals.
1161 The parsers work properly on 64 bit hosts.
1164 Some calls to strerror resulted in scrambled or missing error messages.
1167 When the number of shift/reduce conflicts is correct, don't issue
1170 ** The verbose report includes the rule line numbers.
1172 ** Rule line numbers are fixed in traces.
1174 ** Swedish translation
1177 Verbose parse error messages from the parsers are better looking.
1178 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1179 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1181 ** Fixed parser memory leaks.
1182 When the generated parser was using malloc to extend its stacks, the
1183 previous allocations were not freed.
1185 ** Fixed verbose output file.
1186 Some newlines were missing.
1187 Some conflicts in state descriptions were missing.
1189 ** Fixed conflict report.
1190 Option -v was needed to get the result.
1194 Mismatches are errors, not warnings.
1196 ** Fixed incorrect processing of some invalid input.
1198 ** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
1200 ** Fixed some typos in the documentation.
1202 ** %token MY_EOF 0 is supported.
1203 Before, MY_EOF was silently renumbered as 257.
1205 ** doc/refcard.tex is updated.
1207 ** %output, %file-prefix, %name-prefix.
1211 New, aliasing `--output-file'.
1213 * Changes in version 1.30, 2001-10-26:
1215 ** `--defines' and `--graph' have now an optional argument which is the
1216 output file name. `-d' and `-g' do not change; they do not take any
1219 ** `%source_extension' and `%header_extension' are removed, failed
1222 ** Portability fixes.
1224 * Changes in version 1.29, 2001-09-07:
1226 ** The output file does not define const, as this caused problems when used
1227 with common autoconfiguration schemes. If you still use ancient compilers
1228 that lack const, compile with the equivalent of the C compiler option
1229 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
1231 ** Added `-g' and `--graph'.
1233 ** The Bison manual is now distributed under the terms of the GNU FDL.
1235 ** The input and the output files has automatically a similar extension.
1237 ** Russian translation added.
1239 ** NLS support updated; should hopefully be less troublesome.
1241 ** Added the old Bison reference card.
1243 ** Added `--locations' and `%locations'.
1245 ** Added `-S' and `--skeleton'.
1247 ** `%raw', `-r', `--raw' is disabled.
1249 ** Special characters are escaped when output. This solves the problems
1250 of the #line lines with path names including backslashes.
1253 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
1254 `%debug', `%source_extension' and `%header_extension'.
1257 Automatic location tracking.
1259 * Changes in version 1.28, 1999-07-06:
1261 ** Should compile better now with K&R compilers.
1265 ** Fixed a problem with escaping the double quote character.
1267 ** There is now a FAQ.
1269 * Changes in version 1.27:
1271 ** The make rule which prevented bison.simple from being created on
1272 some systems has been fixed.
1274 * Changes in version 1.26:
1276 ** Bison now uses automake.
1278 ** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
1280 ** Token numbers now start at 257 as previously documented, not 258.
1282 ** Bison honors the TMPDIR environment variable.
1284 ** A couple of buffer overruns have been fixed.
1286 ** Problems when closing files should now be reported.
1288 ** Generated parsers should now work even on operating systems which do
1289 not provide alloca().
1291 * Changes in version 1.25, 1995-10-16:
1293 ** Errors in the input grammar are not fatal; Bison keeps reading
1294 the grammar file, and reports all the errors found in it.
1296 ** Tokens can now be specified as multiple-character strings: for
1297 example, you could use "<=" for a token which looks like <=, instead
1298 of chosing a name like LESSEQ.
1300 ** The %token_table declaration says to write a table of tokens (names
1301 and numbers) into the parser file. The yylex function can use this
1302 table to recognize multiple-character string tokens, or for other
1305 ** The %no_lines declaration says not to generate any #line preprocessor
1306 directives in the parser file.
1308 ** The %raw declaration says to use internal Bison token numbers, not
1309 Yacc-compatible token numbers, when token names are defined as macros.
1311 ** The --no-parser option produces the parser tables without including
1312 the parser engine; a project can now use its own parser engine.
1313 The actions go into a separate file called NAME.act, in the form of
1314 a switch statement body.
1316 * Changes in version 1.23:
1318 The user can define YYPARSE_PARAM as the name of an argument to be
1319 passed into yyparse. The argument should have type void *. It should
1320 actually point to an object. Grammar actions can access the variable
1321 by casting it to the proper pointer type.
1323 Line numbers in output file corrected.
1325 * Changes in version 1.22:
1327 --help option added.
1329 * Changes in version 1.20:
1331 Output file does not redefine const for C++.
1339 Copyright (C) 1995-2010 Free Software Foundation, Inc.
1341 This file is part of Bison, the GNU Parser Generator.
1343 This program is free software: you can redistribute it and/or modify
1344 it under the terms of the GNU General Public License as published by
1345 the Free Software Foundation, either version 3 of the License, or
1346 (at your option) any later version.
1348 This program is distributed in the hope that it will be useful,
1349 but WITHOUT ANY WARRANTY; without even the implied warranty of
1350 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1351 GNU General Public License for more details.
1353 You should have received a copy of the GNU General Public License
1354 along with this program. If not, see <http://www.gnu.org/licenses/>.