3 * Noteworthy changes in release ?.? (????-??-??) [?]
7 *** Typed midrule actions
9 Because their type is unknown to Bison, the values of midrule actions are
10 not treated like the others: they don't have %printer and %destructor
11 support. It also prevents C++ (Bison) variants to handle them properly.
13 Typed midrule actions address these issues. Instead of:
15 exp: { $<ival>$ = 1; } { $<ival>$ = 2; } { $$ = $<ival>1 + $<ival>2; }
19 exp: <ival>{ $$ = 1; } <ival>{ $$ = 2; } { $$ = $1 + $2; }
21 *** Reports include the type of the symbols
23 The sections about terminal and nonterminal symbols of the '*.output' file
24 now specify their declared type. For instance, for:
28 the report now shows '<ival>':
30 Terminals, with rules where they appear
34 *** Diagnostics about useless rules
36 In the following grammar, the 'exp' nonterminal is trivially useless. So,
37 of course, its rules are useless too.
41 exp: exp '+' exp | exp '-' exp | '(' exp ')'
43 Previously all the useless rules were reported, including those whose
44 left-hand side is the 'exp' nonterminal:
46 warning: 1 nonterminal useless in grammar [-Wother]
47 warning: 4 rules useless in grammar [-Wother]
48 2.14-16: warning: nonterminal useless in grammar: exp [-Wother]
51 2.14-16: warning: rule useless in grammar [-Wother]
54 3.6-16: warning: rule useless in grammar [-Wother]
55 exp: exp '+' exp | exp '-' exp | '(' exp ')'
57 3.20-30: warning: rule useless in grammar [-Wother]
58 exp: exp '+' exp | exp '-' exp | '(' exp ')'
60 3.34-44: warning: rule useless in grammar [-Wother]
61 exp: exp '+' exp | exp '-' exp | '(' exp ')'
64 Now, rules whose left-hand side symbol is useless are no longer reported
65 as useless. The locations of the errors have also been adjusted to point
66 to the first use of the nonterminal as a left-hand side of a rule:
68 warning: 1 nonterminal useless in grammar [-Wother]
69 warning: 4 rules useless in grammar [-Wother]
70 3.1-3: warning: nonterminal useless in grammar: exp [-Wother]
71 exp: exp '+' exp | exp '-' exp | '(' exp ')'
73 2.14-16: warning: rule useless in grammar [-Wother]
79 *** GLR: Predicates support broken by #line directives
81 Predicates (%?) in GLR such as
84 %? {new_syntax} 'w' id new_args
85 | %?{!new_syntax} 'w' id old_args
87 were issued with #lines in the middle of C code.
89 *** Printer and destructor with broken #line directives
91 The #line directives were not properly escaped when emitting the code for
92 %printer/%destructor, which resulted in compiler errors if there are
93 backslashes or double-quotes in the grammar file name.
95 *** Portability on ICC
97 The Intel compiler claims compatibility with GCC, yet rejects its _Pragma.
98 Generated parsers now work around this.
100 * Noteworthy changes in release 3.0.5 (2018-05-27) [stable]
104 *** C++: Fix support of 'syntax_error'
106 One incorrect 'inline' resulted in linking errors about the constructor of
107 the syntax_error exception.
109 *** C++: Fix warnings
111 GCC 7.3 (with -O1 or -O2 but not -O0 or -O3) issued null-dereference
112 warnings about yyformat being possibly null. It also warned about the
113 deprecated implicit definition of copy constructors when there's a
114 user-defined (copy) assignment operator.
116 *** Location of errors
118 In C++ parsers, out-of-bounds errors can happen when a rule with an empty
119 ride-hand side raises a syntax error. The behavior of the default parser
120 (yacc.c) in such a condition was undefined.
122 Now all the parsers match the behavior of glr.c: @$ is used as the
123 location of the error. This handles gracefully rules with and without
126 *** Portability fixes in the test suite
128 On some platforms, some Java and/or C++ tests were failing.
130 * Noteworthy changes in release 3.0.4 (2015-01-23) [stable]
134 *** C++ with Variants (lalr1.cc)
136 Fix a compiler warning when no %destructor use $$.
140 Several portability issues in tests were fixed.
142 * Noteworthy changes in release 3.0.3 (2015-01-15) [stable]
146 *** C++ with Variants (lalr1.cc)
148 Problems with %destructor and '%define parse.assert' have been fixed.
150 *** Named %union support (yacc.c, glr.c)
152 Bison 3.0 introduced a regression on named %union such as
154 %union foo { int ival; };
156 The possibility to use a name was introduced "for Yacc compatibility".
157 It is however not required by POSIX Yacc, and its usefulness is not clear.
159 *** %define api.value.type union with %defines (yacc.c, glr.c)
161 The C parsers were broken when %defines was used together with "%define
162 api.value.type union".
164 *** Redeclarations are reported in proper order
172 bison used to report:
174 /tmp/foo.yy:2.10-11: error: %printer redeclaration for FOO
177 /tmp/foo.yy:3.10-11: previous declaration
181 Now, the "previous" declaration is always the first one.
186 Bison now installs various files in its docdir (which defaults to
187 '/usr/local/share/doc/bison'), including the three fully blown examples
188 extracted from the documentation:
191 Reverse Polish Calculator, a simple introductory example.
193 Multi-function Calc, a calculator with memory and functions and located
196 a calculator in C++ using variant support and token constructors.
198 * Noteworthy changes in release 3.0.2 (2013-12-05) [stable]
202 *** Generated source files when errors are reported
204 When warnings are issued and -Werror is set, bison would still generate
205 the source files (*.c, *.h...). As a consequence, some runs of "make"
206 could fail the first time, but not the second (as the files were generated
209 This is fixed: bison no longer generates this source files, but, of
210 course, still produces the various reports (*.output, *.xml, etc.).
212 *** %empty is used in reports
214 Empty right-hand sides are denoted by '%empty' in all the reports (text,
215 dot, XML and formats derived from it).
217 *** YYERROR and variants
219 When C++ variant support is enabled, an error triggered via YYERROR, but
220 not caught via error recovery, resulted in a double deletion.
222 * Noteworthy changes in release 3.0.1 (2013-11-12) [stable]
226 *** Errors in caret diagnostics
228 On some platforms, some errors could result in endless diagnostics.
230 *** Fixes of the -Werror option
232 Options such as "-Werror -Wno-error=foo" were still turning "foo"
233 diagnostics into errors instead of warnings. This is fixed.
235 Actually, for consistency with GCC, "-Wno-error=foo -Werror" now also
236 leaves "foo" diagnostics as warnings. Similarly, with "-Werror=foo
237 -Wno-error", "foo" diagnostics are now errors.
241 As demonstrated in the documentation, one can now leave spaces between
246 The yacc.1 man page is no longer installed if --disable-yacc was
249 *** Fixes in the test suite
251 Bugs and portability issues.
253 * Noteworthy changes in release 3.0 (2013-07-25) [stable]
255 ** WARNING: Future backward-incompatibilities!
257 Like other GNU packages, Bison will start using some of the C99 features
258 for its own code, especially the definition of variables after statements.
259 The generated C parsers still aim at C90.
261 ** Backward incompatible changes
263 *** Obsolete features
265 Support for YYFAIL is removed (deprecated in Bison 2.4.2): use YYERROR.
267 Support for yystype and yyltype is removed (deprecated in Bison 1.875):
268 use YYSTYPE and YYLTYPE.
270 Support for YYLEX_PARAM and YYPARSE_PARAM is removed (deprecated in Bison
271 1.875): use %lex-param, %parse-param, or %param.
273 Missing semicolons at the end of actions are no longer added (as announced
276 *** Use of YACC='bison -y'
278 TL;DR: With Autoconf <= 2.69, pass -Wno-yacc to (AM_)YFLAGS if you use
281 Traditional Yacc generates 'y.tab.c' whatever the name of the input file.
282 Therefore Makefiles written for Yacc expect 'y.tab.c' (and possibly
283 'y.tab.h' and 'y.outout') to be generated from 'foo.y'.
285 To this end, for ages, AC_PROG_YACC, Autoconf's macro to look for an
286 implementation of Yacc, was using Bison as 'bison -y'. While it does
287 ensure compatible output file names, it also enables warnings for
288 incompatibilities with POSIX Yacc. In other words, 'bison -y' triggers
289 warnings for Bison extensions.
291 Autoconf 2.70+ fixes this incompatibility by using YACC='bison -o y.tab.c'
292 (which also generates 'y.tab.h' and 'y.output' when needed).
293 Alternatively, disable Yacc warnings by passing '-Wno-yacc' to your Yacc
294 flags (YFLAGS, or AM_YFLAGS with Automake).
298 *** The epilogue is no longer affected by internal #defines (glr.c)
300 The glr.c skeleton uses defines such as #define yylval (yystackp->yyval) in
301 generated code. These weren't properly undefined before the inclusion of
302 the user epilogue, so functions such as the following were butchered by the
303 preprocessor expansion:
305 int yylex (YYSTYPE *yylval);
307 This is fixed: yylval, yynerrs, yychar, and yylloc are now valid
308 identifiers for user-provided variables.
310 *** stdio.h is no longer needed when locations are enabled (yacc.c)
312 Changes in Bison 2.7 introduced a dependency on FILE and fprintf when
313 locations are enabled. This is fixed.
315 *** Warnings about useless %pure-parser/%define api.pure are restored
317 ** Diagnostics reported by Bison
319 Most of these features were contributed by Théophile Ranquet and Victor
324 Version 2.7 introduced caret errors, for a prettier output. These are now
325 activated by default. The old format can still be used by invoking Bison
326 with -fno-caret (or -fnone).
328 Some error messages that reproduced excerpts of the grammar are now using
329 the caret information only. For instance on:
336 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
337 in.y:2.12-14: warning: rule useless in parser due to conflicts: exp: 'a' [-Wother]
341 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
342 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
346 and "bison -fno-caret" reports:
348 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
349 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
351 *** Enhancements of the -Werror option
353 The -Werror=CATEGORY option is now recognized, and will treat specified
354 warnings as errors. The warnings need not have been explicitly activated
355 using the -W option, this is similar to what GCC 4.7 does.
357 For example, given the following command line, Bison will treat both
358 warnings related to POSIX Yacc incompatibilities and S/R conflicts as
359 errors (and only those):
361 $ bison -Werror=yacc,error=conflicts-sr input.y
363 If no categories are specified, -Werror will make all active warnings into
364 errors. For example, the following line does the same the previous example:
366 $ bison -Werror -Wnone -Wyacc -Wconflicts-sr input.y
368 (By default -Wconflicts-sr,conflicts-rr,deprecated,other is enabled.)
370 Note that the categories in this -Werror option may not be prefixed with
371 "no-". However, -Wno-error[=CATEGORY] is valid.
373 Note that -y enables -Werror=yacc. Therefore it is now possible to require
374 Yacc-like behavior (e.g., always generate y.tab.c), but to report
375 incompatibilities as warnings: "-y -Wno-error=yacc".
377 *** The display of warnings is now richer
379 The option that controls a given warning is now displayed:
381 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
383 In the case of warnings treated as errors, the prefix is changed from
384 "warning: " to "error: ", and the suffix is displayed, in a manner similar
385 to GCC, as [-Werror=CATEGORY].
387 For instance, where the previous version of Bison would report (and exit
390 bison: warnings being treated as errors
391 input.y:1.1: warning: stray ',' treated as white space
395 input.y:1.1: error: stray ',' treated as white space [-Werror=other]
397 *** Deprecated constructs
399 The new 'deprecated' warning category flags obsolete constructs whose
400 support will be discontinued. It is enabled by default. These warnings
401 used to be reported as 'other' warnings.
403 *** Useless semantic types
405 Bison now warns about useless (uninhabited) semantic types. Since
406 semantic types are not declared to Bison (they are defined in the opaque
407 %union structure), it is %printer/%destructor directives about useless
408 types that trigger the warning:
412 %printer {} <type1> <type3>
413 %destructor {} <type2> <type4>
415 nterm: term { $$ = $1; };
417 3.28-34: warning: type <type3> is used, but is not associated to any symbol
418 4.28-34: warning: type <type4> is used, but is not associated to any symbol
420 *** Undefined but unused symbols
422 Bison used to raise an error for undefined symbols that are not used in
423 the grammar. This is now only a warning.
426 %destructor {} symbol2
431 *** Useless destructors or printers
433 Bison now warns about useless destructors or printers. In the following
434 example, the printer for <type1>, and the destructor for <type2> are
435 useless: all symbols of <type1> (token1) already have a printer, and all
436 symbols of type <type2> (token2) already have a destructor.
438 %token <type1> token1
442 %printer {} token1 <type1> <type3>
443 %destructor {} token2 <type2> <type4>
447 The warnings and error messages about shift/reduce and reduce/reduce
448 conflicts have been normalized. For instance on the following foo.y file:
452 exp: exp '+' exp | '0' | '0';
454 compare the previous version of bison:
457 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
458 $ bison -Werror foo.y
459 bison: warnings being treated as errors
460 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
462 with the new behavior:
465 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
466 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
467 $ bison -Werror foo.y
468 foo.y: error: 1 shift/reduce conflict [-Werror=conflicts-sr]
469 foo.y: error: 2 reduce/reduce conflicts [-Werror=conflicts-rr]
471 When %expect or %expect-rr is used, such as with bar.y:
476 exp: exp '+' exp | '0' | '0';
481 bar.y: conflicts: 1 shift/reduce, 2 reduce/reduce
482 bar.y: expected 0 shift/reduce conflicts
483 bar.y: expected 0 reduce/reduce conflicts
488 bar.y: error: shift/reduce conflicts: 1 found, 0 expected
489 bar.y: error: reduce/reduce conflicts: 2 found, 0 expected
491 ** Incompatibilities with POSIX Yacc
493 The 'yacc' category is no longer part of '-Wall', enable it explicitly
496 ** Additional yylex/yyparse arguments
498 The new directive %param declares additional arguments to both yylex and
499 yyparse. The %lex-param, %parse-param, and %param directives support one
500 or more arguments. Instead of
502 %lex-param {arg1_type *arg1}
503 %lex-param {arg2_type *arg2}
504 %parse-param {arg1_type *arg1}
505 %parse-param {arg2_type *arg2}
509 %param {arg1_type *arg1} {arg2_type *arg2}
511 ** Types of values for %define variables
513 Bison used to make no difference between '%define foo bar' and '%define
514 foo "bar"'. The former is now called a 'keyword value', and the latter a
515 'string value'. A third kind was added: 'code values', such as '%define
518 Keyword variables are used for fixed value sets, e.g.,
522 Code variables are used for value in the target language, e.g.,
524 %define api.value.type {struct semantic_type}
526 String variables are used remaining cases, e.g. file names.
528 ** Variable api.token.prefix
530 The variable api.token.prefix changes the way tokens are identified in
531 the generated files. This is especially useful to avoid collisions
532 with identifiers in the target language. For instance
534 %token FILE for ERROR
535 %define api.token.prefix {TOK_}
537 start: FILE for ERROR;
539 will generate the definition of the symbols TOK_FILE, TOK_for, and
540 TOK_ERROR in the generated sources. In particular, the scanner must
541 use these prefixed token names, although the grammar itself still
542 uses the short names (as in the sample rule given above).
544 ** Variable api.value.type
546 This new %define variable supersedes the #define macro YYSTYPE. The use
547 of YYSTYPE is discouraged. In particular, #defining YYSTYPE *and* either
548 using %union or %defining api.value.type results in undefined behavior.
550 Either define api.value.type, or use "%union":
557 %token <ival> INT "integer"
558 %token <sval> STRING "string"
559 %printer { fprintf (yyo, "%d", $$); } <ival>
560 %destructor { free ($$); } <sval>
563 yylval.ival = 42; return INT;
564 yylval.sval = "42"; return STRING;
566 The %define variable api.value.type supports both keyword and code values.
568 The keyword value 'union' means that the user provides genuine types, not
569 union member names such as "ival" and "sval" above (WARNING: will fail if
570 -y/--yacc/%yacc is enabled).
572 %define api.value.type union
573 %token <int> INT "integer"
574 %token <char *> STRING "string"
575 %printer { fprintf (yyo, "%d", $$); } <int>
576 %destructor { free ($$); } <char *>
579 yylval.INT = 42; return INT;
580 yylval.STRING = "42"; return STRING;
582 The keyword value variant is somewhat equivalent, but for C++ special
583 provision is made to allow classes to be used (more about this below).
585 %define api.value.type variant
586 %token <int> INT "integer"
587 %token <std::string> STRING "string"
589 Code values (in braces) denote user defined types. This is where YYSTYPE
607 %define api.value.type {struct my_value}
608 %token <u.ival> INT "integer"
609 %token <u.sval> STRING "string"
610 %printer { fprintf (yyo, "%d", $$); } <u.ival>
611 %destructor { free ($$); } <u.sval>
614 yylval.u.ival = 42; return INT;
615 yylval.u.sval = "42"; return STRING;
617 ** Variable parse.error
619 This variable controls the verbosity of error messages. The use of the
620 %error-verbose directive is deprecated in favor of "%define parse.error
623 ** Renamed %define variables
625 The following variables have been renamed for consistency. Backward
626 compatibility is ensured, but upgrading is recommended.
628 lr.default-reductions -> lr.default-reduction
629 lr.keep-unreachable-states -> lr.keep-unreachable-state
630 namespace -> api.namespace
631 stype -> api.value.type
633 ** Semantic predicates
635 Contributed by Paul Hilfinger.
637 The new, experimental, semantic-predicate feature allows actions of the
638 form "%?{ BOOLEAN-EXPRESSION }", which cause syntax errors (as for
639 YYERROR) if the expression evaluates to 0, and are evaluated immediately
640 in GLR parsers, rather than being deferred. The result is that they allow
641 the programmer to prune possible parses based on the values of run-time
644 ** The directive %expect-rr is now an error in non GLR mode
646 It used to be an error only if used in non GLR mode, _and_ if there are
647 reduce/reduce conflicts.
649 ** Tokens are numbered in their order of appearance
651 Contributed by Valentin Tolmer.
653 With '%token A B', A had a number less than the one of B. However,
654 precedence declarations used to generate a reversed order. This is now
655 fixed, and introducing tokens with any of %token, %left, %right,
656 %precedence, or %nonassoc yields the same result.
658 When mixing declarations of tokens with a litteral character (e.g., 'a')
659 or with an identifier (e.g., B) in a precedence declaration, Bison
660 numbered the litteral characters first. For example
664 would lead to the tokens declared in this order: 'c' 'd' A B. Again, the
665 input order is now preserved.
667 These changes were made so that one can remove useless precedence and
668 associativity declarations (i.e., map %nonassoc, %left or %right to
669 %precedence, or to %token) and get exactly the same output.
671 ** Useless precedence and associativity
673 Contributed by Valentin Tolmer.
675 When developing and maintaining a grammar, useless associativity and
676 precedence directives are common. They can be a nuisance: new ambiguities
677 arising are sometimes masked because their conflicts are resolved due to
678 the extra precedence or associativity information. Furthermore, it can
679 hinder the comprehension of a new grammar: one will wonder about the role
680 of a precedence, where in fact it is useless. The following changes aim
681 at detecting and reporting these extra directives.
683 *** Precedence warning category
685 A new category of warning, -Wprecedence, was introduced. It flags the
686 useless precedence and associativity directives.
688 *** Useless associativity
690 Bison now warns about symbols with a declared associativity that is never
691 used to resolve conflicts. In that case, using %precedence is sufficient;
692 the parsing tables will remain unchanged. Solving these warnings may raise
693 useless precedence warnings, as the symbols no longer have associativity.
707 warning: useless associativity for '+', use %precedence [-Wprecedence]
711 *** Useless precedence
713 Bison now warns about symbols with a declared precedence and no declared
714 associativity (i.e., declared with %precedence), and whose precedence is
715 never used. In that case, the symbol can be safely declared with %token
716 instead, without modifying the parsing tables. For example:
720 exp: "var" '=' "number";
724 warning: useless precedence for '=' [-Wprecedence]
728 *** Useless precedence and associativity
730 In case of both useless precedence and associativity, the issue is flagged
735 exp: "var" '=' "number";
739 warning: useless precedence and associativity for '=' [-Wprecedence]
745 With help from Joel E. Denny and Gabriel Rassoul.
747 Empty rules (i.e., with an empty right-hand side) can now be explicitly
748 marked by the new %empty directive. Using %empty on a non-empty rule is
749 an error. The new -Wempty-rule warning reports empty rules without
750 %empty. On the following grammar:
760 3.4-5: warning: empty rule without %empty [-Wempty-rule]
763 5.8-13: error: %empty on non-empty rule
767 ** Java skeleton improvements
769 The constants for token names were moved to the Lexer interface. Also, it
770 is possible to add code to the parser's constructors using "%code init"
771 and "%define init_throws".
772 Contributed by Paolo Bonzini.
774 The Java skeleton now supports push parsing.
775 Contributed by Dennis Heimbigner.
777 ** C++ skeletons improvements
779 *** The parser header is no longer mandatory (lalr1.cc, glr.cc)
781 Using %defines is now optional. Without it, the needed support classes
782 are defined in the generated parser, instead of additional files (such as
783 location.hh, position.hh and stack.hh).
785 *** Locations are no longer mandatory (lalr1.cc, glr.cc)
787 Both lalr1.cc and glr.cc no longer require %location.
789 *** syntax_error exception (lalr1.cc)
791 The C++ parser features a syntax_error exception, which can be
792 thrown from the scanner or from user rules to raise syntax errors.
793 This facilitates reporting errors caught in sub-functions (e.g.,
794 rejecting too large integral literals from a conversion function
795 used by the scanner, or rejecting invalid combinations from a
796 factory invoked by the user actions).
798 *** %define api.value.type variant
800 This is based on a submission from Michiel De Wilde. With help
801 from Théophile Ranquet.
803 In this mode, complex C++ objects can be used as semantic values. For
806 %token <::std::string> TEXT;
809 %type <::std::string> item;
810 %type <::std::list<std::string>> list;
813 list { std::cout << $1 << std::endl; }
817 %empty { /* Generates an empty string list. */ }
818 | list item ";" { std::swap ($$, $1); $$.push_back ($2); }
822 TEXT { std::swap ($$, $1); }
823 | NUMBER { $$ = string_cast ($1); }
826 *** %define api.token.constructor
828 When variants are enabled, Bison can generate functions to build the
829 tokens. This guarantees that the token type (e.g., NUMBER) is consistent
830 with the semantic value (e.g., int):
832 parser::symbol_type yylex ()
834 parser::location_type loc = ...;
836 return parser::make_TEXT ("Hello, world!", loc);
838 return parser::make_NUMBER (42, loc);
840 return parser::make_SEMICOLON (loc);
846 There are operator- and operator-= for 'location'. Negative line/column
847 increments can no longer underflow the resulting value.
849 * Noteworthy changes in release 2.7.1 (2013-04-15) [stable]
853 *** Fix compiler attribute portability (yacc.c)
855 With locations enabled, __attribute__ was used unprotected.
857 *** Fix some compiler warnings (lalr1.cc)
859 * Noteworthy changes in release 2.7 (2012-12-12) [stable]
863 Warnings about uninitialized yylloc in yyparse have been fixed.
865 Restored C90 compliance (yet no report was ever made).
867 ** Diagnostics are improved
869 Contributed by Théophile Ranquet.
871 *** Changes in the format of error messages
873 This used to be the format of many error reports:
875 input.y:2.7-12: %type redeclaration for exp
876 input.y:1.7-12: previous declaration
880 input.y:2.7-12: error: %type redeclaration for exp
881 input.y:1.7-12: previous declaration
883 *** New format for error reports: carets
885 Caret errors have been added to Bison:
887 input.y:2.7-12: error: %type redeclaration for exp
890 input.y:1.7-12: previous declaration
896 input.y:3.20-23: error: ambiguous reference: '$exp'
897 exp: exp '+' exp { $exp = $1 + $3; };
899 input.y:3.1-3: refers to: $exp at $$
900 exp: exp '+' exp { $exp = $1 + $3; };
902 input.y:3.6-8: refers to: $exp at $1
903 exp: exp '+' exp { $exp = $1 + $3; };
905 input.y:3.14-16: refers to: $exp at $3
906 exp: exp '+' exp { $exp = $1 + $3; };
909 The default behavior for now is still not to display these unless
910 explicitly asked with -fcaret (or -fall). However, in a later release, it
911 will be made the default behavior (but may still be deactivated with
914 ** New value for %define variable: api.pure full
916 The %define variable api.pure requests a pure (reentrant) parser. However,
917 for historical reasons, using it in a location-tracking Yacc parser
918 resulted in a yyerror function that did not take a location as a
919 parameter. With this new value, the user may request a better pure parser,
920 where yyerror does take a location as a parameter (in location-tracking
923 The use of "%define api.pure true" is deprecated in favor of this new
924 "%define api.pure full".
926 ** New %define variable: api.location.type (glr.cc, lalr1.cc, lalr1.java)
928 The %define variable api.location.type defines the name of the type to use
929 for locations. When defined, Bison no longer generates the position.hh
930 and location.hh files, nor does the parser will include them: the user is
931 then responsible to define her type.
933 This can be used in programs with several parsers to factor their location
934 and position files: let one of them generate them, and the others just use
937 This feature was actually introduced, but not documented, in Bison 2.5,
938 under the name "location_type" (which is maintained for backward
941 For consistency, lalr1.java's %define variables location_type and
942 position_type are deprecated in favor of api.location.type and
945 ** Exception safety (lalr1.cc)
947 The parse function now catches exceptions, uses the %destructors to
948 release memory (the lookahead symbol and the symbols pushed on the stack)
949 before re-throwing the exception.
951 This feature is somewhat experimental. User feedback would be
954 ** Graph improvements in DOT and XSLT
956 Contributed by Théophile Ranquet.
958 The graphical presentation of the states is more readable: their shape is
959 now rectangular, the state number is clearly displayed, and the items are
960 numbered and left-justified.
962 The reductions are now explicitly represented as transitions to other
963 diamond shaped nodes.
965 These changes are present in both --graph output and xml2dot.xsl XSLT
966 processing, with minor (documented) differences.
968 ** %language is no longer an experimental feature.
970 The introduction of this feature, in 2.4, was four years ago. The
971 --language option and the %language directive are no longer experimental.
975 The sections about shift/reduce and reduce/reduce conflicts resolution
976 have been fixed and extended.
978 Although introduced more than four years ago, XML and Graphviz reports
979 were not properly documented.
981 The translation of mid-rule actions is now described.
983 * Noteworthy changes in release 2.6.5 (2012-11-07) [stable]
985 We consider compiler warnings about Bison generated parsers to be bugs.
986 Rather than working around them in your own project, please consider
987 reporting them to us.
991 Warnings about uninitialized yylval and/or yylloc for push parsers with a
992 pure interface have been fixed for GCC 4.0 up to 4.8, and Clang 2.9 to
995 Other issues in the test suite have been addressed.
997 Null characters are correctly displayed in error messages.
999 When possible, yylloc is correctly initialized before calling yylex. It
1000 is no longer necessary to initialize it in the %initial-action.
1002 * Noteworthy changes in release 2.6.4 (2012-10-23) [stable]
1004 Bison 2.6.3's --version was incorrect. This release fixes this issue.
1006 * Noteworthy changes in release 2.6.3 (2012-10-22) [stable]
1010 Bugs and portability issues in the test suite have been fixed.
1012 Some errors in translations have been addressed, and --help now directs
1013 users to the appropriate place to report them.
1015 Stray Info files shipped by accident are removed.
1017 Incorrect definitions of YY_, issued by yacc.c when no parser header is
1018 generated, are removed.
1020 All the generated headers are self-contained.
1022 ** Header guards (yacc.c, glr.c, glr.cc)
1024 In order to avoid collisions, the header guards are now
1025 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
1026 For instance the header generated from
1028 %define api.prefix "calc"
1029 %defines "lib/parse.h"
1031 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
1033 ** Fix compiler warnings in the generated parser (yacc.c, glr.c)
1035 The compilation of pure parsers (%define api.pure) can trigger GCC
1038 input.c: In function 'yyparse':
1039 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
1040 function [-Wmaybe-uninitialized]
1044 This is now fixed; pragmas to avoid these warnings are no longer needed.
1046 Warnings from clang ("equality comparison with extraneous parentheses" and
1047 "function declared 'noreturn' should not return") have also been
1050 * Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
1054 Buffer overruns, complaints from Flex, and portability issues in the test
1055 suite have been fixed.
1057 ** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
1059 Trailing end-of-lines in %parse-param or %lex-param would result in
1060 invalid C++. This is fixed.
1062 ** Spurious spaces and end-of-lines
1064 The generated files no longer end (nor start) with empty lines.
1066 * Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
1068 Bison no longer executes user-specified M4 code when processing a grammar.
1072 In addition to the removal of the features announced in Bison 2.6, the
1073 next major release will remove the "Temporary hack for adding a semicolon
1074 to the user action", as announced in the release 2.5. Instead of:
1076 exp: exp "+" exp { $$ = $1 + $3 };
1080 exp: exp "+" exp { $$ = $1 + $3; };
1084 *** Type names are now properly escaped.
1086 *** glr.cc: set_debug_level and debug_level work as expected.
1088 *** Stray @ or $ in actions
1090 While Bison used to warn about stray $ or @ in action rules, it did not
1091 for other actions such as printers, destructors, or initial actions. It
1094 ** Type names in actions
1096 For consistency with rule actions, it is now possible to qualify $$ by a
1097 type-name in destructors, printers, and initial actions. For instance:
1099 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
1101 will display two values for each typed and untyped symbol (provided
1102 that YYSTYPE has both "ival" and "fval" fields).
1104 * Noteworthy changes in release 2.6 (2012-07-19) [stable]
1108 The next major release of Bison will drop support for the following
1109 deprecated features. Please report disagreements to bug-bison@gnu.org.
1113 Support for generating parsers in K&R C will be removed. Parsers
1114 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
1117 *** Features deprecated since Bison 1.875
1119 The definitions of yystype and yyltype will be removed; use YYSTYPE and
1122 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
1123 %lex-param, will no longer be supported.
1125 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
1128 *** The generated header will be included (yacc.c)
1130 Instead of duplicating the content of the generated header (definition of
1131 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
1132 as is already the case for GLR or C++ parsers. This change is deferred
1133 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
1136 ** Generated Parser Headers
1138 *** Guards (yacc.c, glr.c, glr.cc)
1140 The generated headers are now guarded, as is already the case for C++
1141 parsers (lalr1.cc). For instance, with --defines=foo.h:
1146 #endif /* !YY_FOO_H */
1148 *** New declarations (yacc.c, glr.c)
1150 The generated header now declares yydebug and yyparse. Both honor
1151 --name-prefix=bar_, and yield
1153 int bar_parse (void);
1157 #define yyparse bar_parse
1160 in order to facilitate the inclusion of several parser headers inside a
1161 single compilation unit.
1163 *** Exported symbols in C++
1165 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
1166 header, are removed, as they prevent the possibility of including several
1167 generated headers from a single compilation unit.
1171 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
1174 ** New %define variable: api.prefix
1176 Now that the generated headers are more complete and properly protected
1177 against multiple inclusions, constant names, such as YYSTYPE are a
1178 problem. While yyparse and others are properly renamed by %name-prefix,
1179 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
1180 would introduce backward compatibility issues in projects not expecting
1181 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
1182 it is deprecated in favor of a new %define variable: api.prefix.
1184 The following examples compares both:
1186 %name-prefix "bar_" | %define api.prefix "bar_"
1187 %token <ival> FOO %token <ival> FOO
1188 %union { int ival; } %union { int ival; }
1194 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
1195 # define BAR_FOO_H # define BAR_FOO_H
1197 /* Enabling traces. */ /* Enabling traces. */
1198 # ifndef YYDEBUG | # ifndef BAR_DEBUG
1199 > # if defined YYDEBUG
1201 > # define BAR_DEBUG 1
1203 > # define BAR_DEBUG 0
1206 # define YYDEBUG 0 | # define BAR_DEBUG 0
1210 # if YYDEBUG | # if BAR_DEBUG
1211 extern int bar_debug; extern int bar_debug;
1214 /* Tokens. */ /* Tokens. */
1215 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
1216 # define YYTOKENTYPE | # define BAR_TOKENTYPE
1217 enum yytokentype { | enum bar_tokentype {
1222 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
1223 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
1224 typedef union YYSTYPE | typedef union BAR_STYPE
1227 } YYSTYPE; | } BAR_STYPE;
1228 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
1231 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
1233 int bar_parse (void); int bar_parse (void);
1235 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
1237 * Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
1241 The next major release will drop support for generating parsers in K&R C.
1243 ** yacc.c: YYBACKUP works as expected.
1245 ** glr.c improvements:
1247 *** Location support is eliminated when not requested:
1249 GLR parsers used to include location-related code even when locations were
1250 not requested, and therefore not even usable.
1252 *** __attribute__ is preserved:
1254 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
1255 when -std is passed to GCC).
1257 ** lalr1.java: several fixes:
1259 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
1260 first token leads to a syntax error. Some minor clean ups.
1264 *** C++11 compatibility:
1266 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
1271 The header files such as "parser.hh", "location.hh", etc. used a constant
1272 name for preprocessor guards, for instance:
1274 #ifndef BISON_LOCATION_HH
1275 # define BISON_LOCATION_HH
1277 #endif // !BISON_LOCATION_HH
1279 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
1280 case characters are converted to upper case, and series of
1281 non-alphanumerical characters are converted to an underscore.
1283 With "bison -o lang++/parser.cc", "location.hh" would now include:
1285 #ifndef YY_LANG_LOCATION_HH
1286 # define YY_LANG_LOCATION_HH
1288 #endif // !YY_LANG_LOCATION_HH
1292 The position and location constructors (and their initialize methods)
1293 accept new arguments for line and column. Several issues in the
1294 documentation were fixed.
1296 ** liby is no longer asking for "rpl_fprintf" on some platforms.
1298 ** Changes in the manual:
1300 *** %printer is documented
1302 The "%printer" directive, supported since at least Bison 1.50, is finally
1303 documented. The "mfcalc" example is extended to demonstrate it.
1305 For consistency with the C skeletons, the C++ parsers now also support
1306 "yyoutput" (as an alias to "debug_stream ()").
1308 *** Several improvements have been made:
1310 The layout for grammar excerpts was changed to a more compact scheme.
1311 Named references are motivated. The description of the automaton
1312 description file (*.output) is updated to the current format. Incorrect
1313 index entries were fixed. Some other errors were fixed.
1317 *** Conflicting prototypes with recent/modified Flex.
1319 Fixed build problems with the current, unreleased, version of Flex, and
1320 some modified versions of 2.5.35, which have modified function prototypes.
1322 *** Warnings during the build procedure have been eliminated.
1324 *** Several portability problems in the test suite have been fixed:
1326 This includes warnings with some compilers, unexpected behavior of tools
1327 such as diff, warning messages from the test suite itself, etc.
1329 *** The install-pdf target works properly:
1331 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
1332 halts in the middle of its course.
1334 * Changes in version 2.5 (2011-05-14):
1336 ** Grammar symbol names can now contain non-initial dashes:
1338 Consistently with directives (such as %error-verbose) and with
1339 %define variables (e.g. push-pull), grammar symbol names may contain
1340 dashes in any position except the beginning. This is a GNU
1341 extension over POSIX Yacc. Thus, use of this extension is reported
1342 by -Wyacc and rejected in Yacc mode (--yacc).
1344 ** Named references:
1346 Historically, Yacc and Bison have supported positional references
1347 ($n, $$) to allow access to symbol values from inside of semantic
1350 Starting from this version, Bison can also accept named references.
1351 When no ambiguity is possible, original symbol names may be used
1352 as named references:
1354 if_stmt : "if" cond_expr "then" then_stmt ';'
1355 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
1357 In the more common case, explicit names may be declared:
1359 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
1360 { $res = mk_if_stmt($cond, $then, $else); }
1362 Location information is also accessible using @name syntax. When
1363 accessing symbol names containing dots or dashes, explicit bracketing
1364 ($[sym.1]) must be used.
1366 These features are experimental in this version. More user feedback
1367 will help to stabilize them.
1368 Contributed by Alex Rozenman.
1370 ** IELR(1) and canonical LR(1):
1372 IELR(1) is a minimal LR(1) parser table generation algorithm. That
1373 is, given any context-free grammar, IELR(1) generates parser tables
1374 with the full language-recognition power of canonical LR(1) but with
1375 nearly the same number of parser states as LALR(1). This reduction
1376 in parser states is often an order of magnitude. More importantly,
1377 because canonical LR(1)'s extra parser states may contain duplicate
1378 conflicts in the case of non-LR(1) grammars, the number of conflicts
1379 for IELR(1) is often an order of magnitude less as well. This can
1380 significantly reduce the complexity of developing of a grammar.
1382 Bison can now generate IELR(1) and canonical LR(1) parser tables in
1383 place of its traditional LALR(1) parser tables, which remain the
1384 default. You can specify the type of parser tables in the grammar
1385 file with these directives:
1387 %define lr.type lalr
1388 %define lr.type ielr
1389 %define lr.type canonical-lr
1391 The default-reduction optimization in the parser tables can also be
1392 adjusted using "%define lr.default-reductions". For details on both
1393 of these features, see the new section "Tuning LR" in the Bison
1396 These features are experimental. More user feedback will help to
1399 ** LAC (Lookahead Correction) for syntax error handling
1401 Contributed by Joel E. Denny.
1403 Canonical LR, IELR, and LALR can suffer from a couple of problems
1404 upon encountering a syntax error. First, the parser might perform
1405 additional parser stack reductions before discovering the syntax
1406 error. Such reductions can perform user semantic actions that are
1407 unexpected because they are based on an invalid token, and they
1408 cause error recovery to begin in a different syntactic context than
1409 the one in which the invalid token was encountered. Second, when
1410 verbose error messages are enabled (with %error-verbose or the
1411 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
1412 syntax error message can both contain invalid tokens and omit valid
1415 The culprits for the above problems are %nonassoc, default
1416 reductions in inconsistent states, and parser state merging. Thus,
1417 IELR and LALR suffer the most. Canonical LR can suffer only if
1418 %nonassoc is used or if default reductions are enabled for
1419 inconsistent states.
1421 LAC is a new mechanism within the parsing algorithm that solves
1422 these problems for canonical LR, IELR, and LALR without sacrificing
1423 %nonassoc, default reductions, or state merging. When LAC is in
1424 use, canonical LR and IELR behave almost exactly the same for both
1425 syntactically acceptable and syntactically unacceptable input.
1426 While LALR still does not support the full language-recognition
1427 power of canonical LR and IELR, LAC at least enables LALR's syntax
1428 error handling to correctly reflect LALR's language-recognition
1431 Currently, LAC is only supported for deterministic parsers in C.
1432 You can enable LAC with the following directive:
1434 %define parse.lac full
1436 See the new section "LAC" in the Bison manual for additional
1437 details including a few caveats.
1439 LAC is an experimental feature. More user feedback will help to
1442 ** %define improvements:
1444 *** Can now be invoked via the command line:
1446 Each of these command-line options
1449 --define=NAME[=VALUE]
1452 --force-define=NAME[=VALUE]
1454 is equivalent to this grammar file declaration
1456 %define NAME ["VALUE"]
1458 except that the manner in which Bison processes multiple definitions
1459 for the same NAME differs. Most importantly, -F and --force-define
1460 quietly override %define, but -D and --define do not. For further
1461 details, see the section "Bison Options" in the Bison manual.
1463 *** Variables renamed:
1465 The following %define variables
1468 lr.keep_unreachable_states
1470 have been renamed to
1473 lr.keep-unreachable-states
1475 The old names are now deprecated but will be maintained indefinitely
1476 for backward compatibility.
1478 *** Values no longer need to be quoted in the grammar file:
1480 If a %define value is an identifier, it no longer needs to be placed
1481 within quotations marks. For example,
1483 %define api.push-pull "push"
1487 %define api.push-pull push
1489 *** Unrecognized variables are now errors not warnings.
1491 *** Multiple invocations for any variable is now an error not a warning.
1493 ** Unrecognized %code qualifiers are now errors not warnings.
1495 ** Character literals not of length one:
1497 Previously, Bison quietly converted all character literals to length
1498 one. For example, without warning, Bison interpreted the operators in
1499 the following grammar to be the same token:
1505 Bison now warns when a character literal is not of length one. In
1506 some future release, Bison will start reporting an error instead.
1508 ** Destructor calls fixed for lookaheads altered in semantic actions:
1510 Previously for deterministic parsers in C, if a user semantic action
1511 altered yychar, the parser in some cases used the old yychar value to
1512 determine which destructor to call for the lookahead upon a syntax
1513 error or upon parser return. This bug has been fixed.
1515 ** C++ parsers use YYRHSLOC:
1517 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
1518 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
1519 to use it. If, for instance, your location structure has "first"
1520 and "last" members, instead of
1522 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1526 (Current).first = (Rhs)[1].location.first; \
1527 (Current).last = (Rhs)[N].location.last; \
1531 (Current).first = (Current).last = (Rhs)[0].location.last; \
1537 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1541 (Current).first = YYRHSLOC (Rhs, 1).first; \
1542 (Current).last = YYRHSLOC (Rhs, N).last; \
1546 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
1550 ** YYLLOC_DEFAULT in C++:
1552 The default implementation of YYLLOC_DEFAULT used to be issued in
1553 the header file. It is now output in the implementation file, after
1554 the user %code sections so that its #ifndef guard does not try to
1555 override the user's YYLLOC_DEFAULT if provided.
1557 ** YYFAIL now produces warnings and Java parsers no longer implement it:
1559 YYFAIL has existed for many years as an undocumented feature of
1560 deterministic parsers in C generated by Bison. More recently, it was
1561 a documented feature of Bison's experimental Java parsers. As
1562 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
1563 semantic action now produces a deprecation warning, and Java parsers
1564 no longer implement YYFAIL at all. For further details, including a
1565 discussion of how to suppress C preprocessor warnings about YYFAIL
1566 being unused, see the Bison 2.4.2 NEWS entry.
1568 ** Temporary hack for adding a semicolon to the user action:
1570 Previously, Bison appended a semicolon to every user action for
1571 reductions when the output language defaulted to C (specifically, when
1572 neither %yacc, %language, %skeleton, or equivalent command-line
1573 options were specified). This allowed actions such as
1575 exp: exp "+" exp { $$ = $1 + $3 };
1579 exp: exp "+" exp { $$ = $1 + $3; };
1581 As a first step in removing this misfeature, Bison now issues a
1582 warning when it appends a semicolon. Moreover, in cases where Bison
1583 cannot easily determine whether a semicolon is needed (for example, an
1584 action ending with a cpp directive or a braced compound initializer),
1585 it no longer appends one. Thus, the C compiler might now complain
1586 about a missing semicolon where it did not before. Future releases of
1587 Bison will cease to append semicolons entirely.
1589 ** Verbose syntax error message fixes:
1591 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
1592 specified, syntax error messages produced by the generated parser
1593 include the unexpected token as well as a list of expected tokens.
1594 The effect of %nonassoc on these verbose messages has been corrected
1595 in two ways, but a more complete fix requires LAC, described above:
1597 *** When %nonassoc is used, there can exist parser states that accept no
1598 tokens, and so the parser does not always require a lookahead token
1599 in order to detect a syntax error. Because no unexpected token or
1600 expected tokens can then be reported, the verbose syntax error
1601 message described above is suppressed, and the parser instead
1602 reports the simpler message, "syntax error". Previously, this
1603 suppression was sometimes erroneously triggered by %nonassoc when a
1604 lookahead was actually required. Now verbose messages are
1605 suppressed only when all previous lookaheads have already been
1606 shifted or discarded.
1608 *** Previously, the list of expected tokens erroneously included tokens
1609 that would actually induce a syntax error because conflicts for them
1610 were resolved with %nonassoc in the current parser state. Such
1611 tokens are now properly omitted from the list.
1613 *** Expected token lists are still often wrong due to state merging
1614 (from LALR or IELR) and default reductions, which can both add
1615 invalid tokens and subtract valid tokens. Canonical LR almost
1616 completely fixes this problem by eliminating state merging and
1617 default reductions. However, there is one minor problem left even
1618 when using canonical LR and even after the fixes above. That is,
1619 if the resolution of a conflict with %nonassoc appears in a later
1620 parser state than the one at which some syntax error is
1621 discovered, the conflicted token is still erroneously included in
1622 the expected token list. Bison's new LAC implementation,
1623 described above, eliminates this problem and the need for
1624 canonical LR. However, LAC is still experimental and is disabled
1627 ** Java skeleton fixes:
1629 *** A location handling bug has been fixed.
1631 *** The top element of each of the value stack and location stack is now
1632 cleared when popped so that it can be garbage collected.
1634 *** Parser traces now print the top element of the stack.
1636 ** -W/--warnings fixes:
1638 *** Bison now properly recognizes the "no-" versions of categories:
1640 For example, given the following command line, Bison now enables all
1641 warnings except warnings for incompatibilities with POSIX Yacc:
1643 bison -Wall,no-yacc gram.y
1645 *** Bison now treats S/R and R/R conflicts like other warnings:
1647 Previously, conflict reports were independent of Bison's normal
1648 warning system. Now, Bison recognizes the warning categories
1649 "conflicts-sr" and "conflicts-rr". This change has important
1650 consequences for the -W and --warnings command-line options. For
1653 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
1654 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
1655 bison -Wnone gram.y # no conflicts are reported
1656 bison -Werror gram.y # any conflict is an error
1658 However, as before, if the %expect or %expect-rr directive is
1659 specified, an unexpected number of conflicts is an error, and an
1660 expected number of conflicts is not reported, so -W and --warning
1661 then have no effect on the conflict report.
1663 *** The "none" category no longer disables a preceding "error":
1665 For example, for the following command line, Bison now reports
1666 errors instead of warnings for incompatibilities with POSIX Yacc:
1668 bison -Werror,none,yacc gram.y
1670 *** The "none" category now disables all Bison warnings:
1672 Previously, the "none" category disabled only Bison warnings for
1673 which there existed a specific -W/--warning category. However,
1674 given the following command line, Bison is now guaranteed to
1675 suppress all warnings:
1679 ** Precedence directives can now assign token number 0:
1681 Since Bison 2.3b, which restored the ability of precedence
1682 directives to assign token numbers, doing so for token number 0 has
1683 produced an assertion failure. For example:
1687 This bug has been fixed.
1689 * Changes in version 2.4.3 (2010-08-05):
1691 ** Bison now obeys -Werror and --warnings=error for warnings about
1692 grammar rules that are useless in the parser due to conflicts.
1694 ** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
1697 ** Failures in the test suite for GCC 4.5 have been fixed.
1699 ** Failures in the test suite for some versions of Sun Studio C++ have
1702 ** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
1703 warnings about undefined %prec identifiers will not be converted to
1704 errors in Bison 2.5. They will remain warnings, which should be
1705 sufficient for POSIX while avoiding backward compatibility issues.
1707 ** Minor documentation fixes.
1709 * Changes in version 2.4.2 (2010-03-20):
1711 ** Some portability problems that resulted in failures and livelocks
1712 in the test suite on some versions of at least Solaris, AIX, HP-UX,
1713 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
1714 errors should no longer cause M4 to report a broken pipe on the
1717 ** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
1719 POSIX specifies that an error be reported for any identifier that does
1720 not appear on the LHS of a grammar rule and that is not defined by
1721 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
1722 error report for the case when an identifier appears only after a
1723 %prec directive. It is now restored. However, for backward
1724 compatibility with recent Bison releases, it is only a warning for
1725 now. In Bison 2.5 and later, it will return to being an error.
1726 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
1727 warning will not be converted to an error in Bison 2.5.]
1729 ** Detection of GNU M4 1.4.6 or newer during configure is improved.
1731 ** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
1732 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
1735 ** %code is now a permanent feature.
1737 A traditional Yacc prologue directive is written in the form:
1741 To provide a more flexible alternative, Bison 2.3b introduced the
1742 %code directive with the following forms for C/C++:
1745 %code requires {CODE}
1746 %code provides {CODE}
1749 These forms are now considered permanent features of Bison. See the
1750 %code entries in the section "Bison Declaration Summary" in the Bison
1751 manual for a summary of their functionality. See the section
1752 "Prologue Alternatives" for a detailed discussion including the
1753 advantages of %code over the traditional Yacc prologue directive.
1755 Bison's Java feature as a whole including its current usage of %code
1756 is still considered experimental.
1758 ** YYFAIL is deprecated and will eventually be removed.
1760 YYFAIL has existed for many years as an undocumented feature of
1761 deterministic parsers in C generated by Bison. Previously, it was
1762 documented for Bison's experimental Java parsers. YYFAIL is no longer
1763 documented for Java parsers and is formally deprecated in both cases.
1764 Users are strongly encouraged to migrate to YYERROR, which is
1767 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
1768 induce a syntax error. The most obvious difference from YYERROR is
1769 that YYFAIL will automatically invoke yyerror to report the syntax
1770 error so that you don't have to. However, there are several other
1771 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
1772 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1773 used. For a more detailed discussion, see:
1775 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
1777 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
1778 deterministic parsers in C will continue to implement it. However,
1779 because YYFAIL is already flawed, it seems futile to try to make new
1780 Bison features compatible with it. Thus, during parser generation,
1781 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
1782 rule action. In a later release, YYFAIL will be disabled for
1783 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1784 be removed altogether.
1786 There exists at least one case where Bison 2.5's YYFAIL warning will
1787 be a false positive. Some projects add phony uses of YYFAIL and other
1788 Bison-defined macros for the sole purpose of suppressing C
1789 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
1790 To avoid Bison's future warning, such YYFAIL uses can be moved to the
1791 epilogue (that is, after the second "%%") in the Bison input file. In
1792 this release (2.4.2), Bison already generates its own code to suppress
1793 C preprocessor warnings for YYFAIL, so projects can remove their own
1794 phony uses of YYFAIL if compatibility with Bison releases prior to
1795 2.4.2 is not necessary.
1797 ** Internationalization.
1799 Fix a regression introduced in Bison 2.4: Under some circumstances,
1800 message translations were not installed although supported by the
1803 * Changes in version 2.4.1 (2008-12-11):
1805 ** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
1806 declarations have been fixed.
1808 ** Temporary hack for adding a semicolon to the user action.
1810 Bison used to prepend a trailing semicolon at the end of the user
1811 action for reductions. This allowed actions such as
1813 exp: exp "+" exp { $$ = $1 + $3 };
1817 exp: exp "+" exp { $$ = $1 + $3; };
1819 Some grammars still depend on this "feature". Bison 2.4.1 restores
1820 the previous behavior in the case of C output (specifically, when
1821 neither %language or %skeleton or equivalent command-line options
1822 are used) to leave more time for grammars depending on the old
1823 behavior to be adjusted. Future releases of Bison will disable this
1826 ** A few minor improvements to the Bison manual.
1828 * Changes in version 2.4 (2008-11-02):
1830 ** %language is an experimental feature.
1832 We first introduced this feature in test release 2.3b as a cleaner
1833 alternative to %skeleton. Since then, we have discussed the possibility of
1834 modifying its effect on Bison's output file names. Thus, in this release,
1835 we consider %language to be an experimental feature that will likely evolve
1838 ** Forward compatibility with GNU M4 has been improved.
1840 ** Several bugs in the C++ skeleton and the experimental Java skeleton have been
1843 * Changes in version 2.3b (2008-05-27):
1845 ** The quotes around NAME that used to be required in the following directive
1848 %define NAME "VALUE"
1850 ** The directive "%pure-parser" is now deprecated in favor of:
1854 which has the same effect except that Bison is more careful to warn about
1855 unreasonable usage in the latter case.
1859 Bison can now generate an LALR(1) parser in C with a push interface. That
1860 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
1861 push one token at a time to the parser using "yypush_parse", which will
1862 return to the caller after processing each token. By default, the push
1863 interface is disabled. Either of the following directives will enable it:
1865 %define api.push_pull "push" // Just push; does not require yylex.
1866 %define api.push_pull "both" // Push and pull; requires yylex.
1868 See the new section "A Push Parser" in the Bison manual for details.
1870 The current push parsing interface is experimental and may evolve. More user
1871 feedback will help to stabilize it.
1873 ** The -g and --graph options now output graphs in Graphviz DOT format,
1874 not VCG format. Like --graph, -g now also takes an optional FILE argument
1875 and thus cannot be bundled with other short options.
1879 Bison can now generate an LALR(1) parser in Java. The skeleton is
1880 "data/lalr1.java". Consider using the new %language directive instead of
1881 %skeleton to select it.
1883 See the new section "Java Parsers" in the Bison manual for details.
1885 The current Java interface is experimental and may evolve. More user
1886 feedback will help to stabilize it.
1887 Contributed by Paolo Bonzini.
1891 This new directive specifies the programming language of the generated
1892 parser, which can be C (the default), C++, or Java. Besides the skeleton
1893 that Bison uses, the directive affects the names of the generated files if
1894 the grammar file's name ends in ".y".
1896 ** XML Automaton Report
1898 Bison can now generate an XML report of the LALR(1) automaton using the new
1899 "--xml" option. The current XML schema is experimental and may evolve. More
1900 user feedback will help to stabilize it.
1901 Contributed by Wojciech Polak.
1903 ** The grammar file may now specify the name of the parser header file using
1904 %defines. For example:
1908 ** When reporting useless rules, useless nonterminals, and unused terminals,
1909 Bison now employs the terms "useless in grammar" instead of "useless",
1910 "useless in parser" instead of "never reduced", and "unused in grammar"
1911 instead of "unused".
1913 ** Unreachable State Removal
1915 Previously, Bison sometimes generated parser tables containing unreachable
1916 states. A state can become unreachable during conflict resolution if Bison
1917 disables a shift action leading to it from a predecessor state. Bison now:
1919 1. Removes unreachable states.
1921 2. Does not report any conflicts that appeared in unreachable states.
1922 WARNING: As a result, you may need to update %expect and %expect-rr
1923 directives in existing grammar files.
1925 3. For any rule used only in such states, Bison now reports the rule as
1926 "useless in parser due to conflicts".
1928 This feature can be disabled with the following directive:
1930 %define lr.keep_unreachable_states
1932 See the %define entry in the "Bison Declaration Summary" in the Bison manual
1933 for further discussion.
1935 ** Lookahead Set Correction in the ".output" Report
1937 When instructed to generate a ".output" file including lookahead sets
1938 (using "--report=lookahead", for example), Bison now prints each reduction's
1939 lookahead set only next to the associated state's one item that (1) is
1940 associated with the same rule as the reduction and (2) has its dot at the end
1941 of its RHS. Previously, Bison also erroneously printed the lookahead set
1942 next to all of the state's other items associated with the same rule. This
1943 bug affected only the ".output" file and not the generated parser source
1946 ** --report-file=FILE is a new option to override the default ".output" file
1949 ** The "=" that used to be required in the following directives is now
1952 %file-prefix "parser"
1956 ** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
1958 Bison 2.3a provided a new set of directives as a more flexible alternative to
1959 the traditional Yacc prologue blocks. Those have now been consolidated into
1960 a single %code directive with an optional qualifier field, which identifies
1961 the purpose of the code and thus the location(s) where Bison should generate
1964 1. "%code {CODE}" replaces "%after-header {CODE}"
1965 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1966 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1967 4. "%code top {CODE}" replaces "%before-header {CODE}"
1969 See the %code entries in section "Bison Declaration Summary" in the Bison
1970 manual for a summary of the new functionality. See the new section "Prologue
1971 Alternatives" for a detailed discussion including the advantages of %code
1972 over the traditional Yacc prologues.
1974 The prologue alternatives are experimental. More user feedback will help to
1975 determine whether they should become permanent features.
1977 ** Revised warning: unset or unused mid-rule values
1979 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1980 used within any of the actions of the parent rule. For example, Bison warns
1983 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1985 Now, Bison also warns about mid-rule values that are used but not set. For
1986 example, Bison warns about unset $$ in the mid-rule action in:
1988 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1990 However, Bison now disables both of these warnings by default since they
1991 sometimes prove to be false alarms in existing grammars employing the Yacc
1992 constructs $0 or $-N (where N is some positive integer).
1994 To enable these warnings, specify the option "--warnings=midrule-values" or
1995 "-W", which is a synonym for "--warnings=all".
1997 ** Default %destructor or %printer with "<*>" or "<>"
1999 Bison now recognizes two separate kinds of default %destructor's and
2002 1. Place "<*>" in a %destructor/%printer symbol list to define a default
2003 %destructor/%printer for all grammar symbols for which you have formally
2004 declared semantic type tags.
2006 2. Place "<>" in a %destructor/%printer symbol list to define a default
2007 %destructor/%printer for all grammar symbols without declared semantic
2010 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
2011 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
2012 longer applies any %destructor to a mid-rule value if that mid-rule value is
2013 not actually ever referenced using either $$ or $n in a semantic action.
2015 The default %destructor's and %printer's are experimental. More user
2016 feedback will help to determine whether they should become permanent
2019 See the section "Freeing Discarded Symbols" in the Bison manual for further
2022 ** %left, %right, and %nonassoc can now declare token numbers. This is required
2023 by POSIX. However, see the end of section "Operator Precedence" in the Bison
2024 manual for a caveat concerning the treatment of literal strings.
2026 ** The nonfunctional --no-parser, -n, and %no-parser options have been
2027 completely removed from Bison.
2029 * Changes in version 2.3a, 2006-09-13:
2031 ** Instead of %union, you can define and use your own union type
2032 YYSTYPE if your grammar contains at least one <type> tag.
2033 Your YYSTYPE need not be a macro; it can be a typedef.
2034 This change is for compatibility with other Yacc implementations,
2035 and is required by POSIX.
2037 ** Locations columns and lines start at 1.
2038 In accordance with the GNU Coding Standards and Emacs.
2040 ** You may now declare per-type and default %destructor's and %printer's:
2044 %union { char *string; }
2045 %token <string> STRING1
2046 %token <string> STRING2
2047 %type <string> string1
2048 %type <string> string2
2049 %union { char character; }
2050 %token <character> CHR
2051 %type <character> chr
2052 %destructor { free ($$); } %symbol-default
2053 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
2054 %destructor { } <character>
2056 guarantees that, when the parser discards any user-defined symbol that has a
2057 semantic type tag other than "<character>", it passes its semantic value to
2058 "free". However, when the parser discards a "STRING1" or a "string1", it
2059 also prints its line number to "stdout". It performs only the second
2060 "%destructor" in this case, so it invokes "free" only once.
2062 [Although we failed to mention this here in the 2.3a release, the default
2063 %destructor's and %printer's were experimental, and they were rewritten in
2066 ** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
2067 "--yacc", or "%yacc"), Bison no longer generates #define statements for
2068 associating token numbers with token names. Removing the #define statements
2069 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
2070 requires them. Bison still generates an enum for token names in all cases.
2072 ** Handling of traditional Yacc prologue blocks is now more consistent but
2073 potentially incompatible with previous releases of Bison.
2075 As before, you declare prologue blocks in your grammar file with the
2076 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
2077 prologue blocks that you've declared before the first %union. To generate
2078 the post-prologue, Bison concatenates all prologue blocks that you've
2079 declared after the first %union.
2081 Previous releases of Bison inserted the pre-prologue into both the header
2082 file and the code file in all cases except for LALR(1) parsers in C. In the
2083 latter case, Bison inserted it only into the code file. For parsers in C++,
2084 the point of insertion was before any token definitions (which associate
2085 token numbers with names). For parsers in C, the point of insertion was
2086 after the token definitions.
2088 Now, Bison never inserts the pre-prologue into the header file. In the code
2089 file, it always inserts it before the token definitions.
2091 ** Bison now provides a more flexible alternative to the traditional Yacc
2092 prologue blocks: %before-header, %start-header, %end-header, and
2095 For example, the following declaration order in the grammar file reflects the
2096 order in which Bison will output these code blocks. However, you are free to
2097 declare these code blocks in your grammar file in whatever order is most
2101 /* Bison treats this block like a pre-prologue block: it inserts it into
2102 * the code file before the contents of the header file. It does *not*
2103 * insert it into the header file. This is a good place to put
2104 * #include's that you want at the top of your code file. A common
2105 * example is '#include "system.h"'. */
2108 /* Bison inserts this block into both the header file and the code file.
2109 * In both files, the point of insertion is before any Bison-generated
2110 * token, semantic type, location type, and class definitions. This is a
2111 * good place to define %union dependencies, for example. */
2114 /* Unlike the traditional Yacc prologue blocks, the output order for the
2115 * new %*-header blocks is not affected by their declaration position
2116 * relative to any %union in the grammar file. */
2119 /* Bison inserts this block into both the header file and the code file.
2120 * In both files, the point of insertion is after the Bison-generated
2121 * definitions. This is a good place to declare or define public
2122 * functions or data structures that depend on the Bison-generated
2126 /* Bison treats this block like a post-prologue block: it inserts it into
2127 * the code file after the contents of the header file. It does *not*
2128 * insert it into the header file. This is a good place to declare or
2129 * define internal functions or data structures that depend on the
2130 * Bison-generated definitions. */
2133 If you have multiple occurrences of any one of the above declarations, Bison
2134 will concatenate the contents in declaration order.
2136 [Although we failed to mention this here in the 2.3a release, the prologue
2137 alternatives were experimental, and they were rewritten in future versions.]
2139 ** The option "--report=look-ahead" has been changed to "--report=lookahead".
2140 The old spelling still works, but is not documented and may be removed
2141 in a future release.
2143 * Changes in version 2.3, 2006-06-05:
2145 ** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
2146 for compatibility with LALR(1) grammars.
2148 ** It is now documented that any definition of YYSTYPE or YYLTYPE should
2149 be to a type name that does not contain parentheses or brackets.
2151 * Changes in version 2.2, 2006-05-19:
2153 ** The distribution terms for all Bison-generated parsers now permit
2154 using the parsers in nonfree programs. Previously, this permission
2155 was granted only for Bison-generated LALR(1) parsers in C.
2157 ** %name-prefix changes the namespace name in C++ outputs.
2159 ** The C++ parsers export their token_type.
2161 ** Bison now allows multiple %union declarations, and concatenates
2162 their contents together.
2164 ** New warning: unused values
2165 Right-hand side symbols whose values are not used are reported,
2166 if the symbols have destructors. For instance:
2168 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
2172 will trigger a warning about $$ and $5 in the first rule, and $3 in
2173 the second ($1 is copied to $$ by the default rule). This example
2174 most likely contains three errors, and could be rewritten as:
2176 exp: exp "?" exp ":" exp
2177 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
2179 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
2182 However, if the original actions were really intended, memory leaks
2183 and all, the warnings can be suppressed by letting Bison believe the
2184 values are used, e.g.:
2186 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
2187 | exp "+" exp { $$ = $1; (void) $3; }
2190 If there are mid-rule actions, the warning is issued if no action
2191 uses it. The following triggers no warning: $1 and $3 are used.
2193 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
2195 The warning is intended to help catching lost values and memory leaks.
2196 If a value is ignored, its associated memory typically is not reclaimed.
2198 ** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
2199 Destructors are now called when user code invokes YYABORT, YYACCEPT,
2200 and YYERROR, for all objects on the stack, other than objects
2201 corresponding to the right-hand side of the current rule.
2203 ** %expect, %expect-rr
2204 Incorrect numbers of expected conflicts are now actual errors,
2205 instead of warnings.
2207 ** GLR, YACC parsers.
2208 The %parse-params are available in the destructors (and the
2209 experimental printers) as per the documentation.
2211 ** Bison now warns if it finds a stray "$" or "@" in an action.
2213 ** %require "VERSION"
2214 This specifies that the grammar file depends on features implemented
2215 in Bison version VERSION or higher.
2217 ** lalr1.cc: The token and value types are now class members.
2218 The tokens were defined as free form enums and cpp macros. YYSTYPE
2219 was defined as a free form union. They are now class members:
2220 tokens are enumerations of the "yy::parser::token" struct, and the
2221 semantic values have the "yy::parser::semantic_type" type.
2223 If you do not want or can update to this scheme, the directive
2224 '%define "global_tokens_and_yystype" "1"' triggers the global
2225 definition of tokens and YYSTYPE. This change is suitable both
2226 for previous releases of Bison, and this one.
2228 If you wish to update, then make sure older version of Bison will
2229 fail using '%require "2.2"'.
2231 ** DJGPP support added.
2233 * Changes in version 2.1, 2005-09-16:
2235 ** The C++ lalr1.cc skeleton supports %lex-param.
2237 ** Bison-generated parsers now support the translation of diagnostics like
2238 "syntax error" into languages other than English. The default
2239 language is still English. For details, please see the new
2240 Internationalization section of the Bison manual. Software
2241 distributors should also see the new PACKAGING file. Thanks to
2242 Bruno Haible for this new feature.
2244 ** Wording in the Bison-generated parsers has been changed slightly to
2245 simplify translation. In particular, the message "memory exhausted"
2246 has replaced "parser stack overflow", as the old message was not
2247 always accurate for modern Bison-generated parsers.
2249 ** Destructors are now called when the parser aborts, for all symbols left
2250 behind on the stack. Also, the start symbol is now destroyed after a
2251 successful parse. In both cases, the behavior was formerly inconsistent.
2253 ** When generating verbose diagnostics, Bison-generated parsers no longer
2254 quote the literal strings associated with tokens. For example, for
2255 a syntax error associated with '%token NUM "number"' they might
2256 print 'syntax error, unexpected number' instead of 'syntax error,
2257 unexpected "number"'.
2259 * Changes in version 2.0, 2004-12-25:
2261 ** Possibly-incompatible changes
2263 - Bison-generated parsers no longer default to using the alloca function
2264 (when available) to extend the parser stack, due to widespread
2265 problems in unchecked stack-overflow detection. You can "#define
2266 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
2267 the manual to determine safe values for YYMAXDEPTH in that case.
2269 - Error token location.
2270 During error recovery, the location of the syntax error is updated
2271 to cover the whole sequence covered by the error token: it includes
2272 the shifted symbols thrown away during the first part of the error
2273 recovery, and the lookahead rejected during the second part.
2275 - Semicolon changes:
2276 . Stray semicolons are no longer allowed at the start of a grammar.
2277 . Semicolons are now required after in-grammar declarations.
2279 - Unescaped newlines are no longer allowed in character constants or
2280 string literals. They were never portable, and GCC 3.4.0 has
2281 dropped support for them. Better diagnostics are now generated if
2282 forget a closing quote.
2284 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
2288 - GLR grammars now support locations.
2290 - New directive: %initial-action.
2291 This directive allows the user to run arbitrary code (including
2292 initializing @$) from yyparse before parsing starts.
2294 - A new directive "%expect-rr N" specifies the expected number of
2295 reduce/reduce conflicts in GLR parsers.
2297 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
2298 This is a GNU extension.
2300 - The option "--report=lookahead" was changed to "--report=look-ahead".
2301 [However, this was changed back after 2.3.]
2303 - Experimental %destructor support has been added to lalr1.cc.
2305 - New configure option --disable-yacc, to disable installation of the
2306 yacc command and -ly library introduced in 1.875 for POSIX conformance.
2310 - For now, %expect-count violations are now just warnings, not errors.
2311 This is for compatibility with Bison 1.75 and earlier (when there are
2312 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
2313 are too many or too few shift/reduce conflicts). However, in future
2314 versions of Bison we plan to improve the %expect machinery so that
2315 these violations will become errors again.
2317 - Within Bison itself, numbers (e.g., goto numbers) are no longer
2318 arbitrarily limited to 16-bit counts.
2320 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
2322 * Changes in version 1.875, 2003-01-01:
2324 ** The documentation license has been upgraded to version 1.2
2325 of the GNU Free Documentation License.
2327 ** syntax error processing
2329 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
2330 locations too. This fixes bugs in error-location computation.
2333 It is now possible to reclaim the memory associated to symbols
2334 discarded during error recovery. This feature is still experimental.
2337 This new directive is preferred over YYERROR_VERBOSE.
2339 - #defining yyerror to steal internal variables is discouraged.
2340 It is not guaranteed to work forever.
2342 ** POSIX conformance
2344 - Semicolons are once again optional at the end of grammar rules.
2345 This reverts to the behavior of Bison 1.33 and earlier, and improves
2346 compatibility with Yacc.
2348 - "parse error" -> "syntax error"
2349 Bison now uniformly uses the term "syntax error"; formerly, the code
2350 and manual sometimes used the term "parse error" instead. POSIX
2351 requires "syntax error" in diagnostics, and it was thought better to
2354 - The documentation now emphasizes that yylex and yyerror must be
2355 declared before use. C99 requires this.
2357 - Bison now parses C99 lexical constructs like UCNs and
2358 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
2360 - File names are properly escaped in C output. E.g., foo\bar.y is
2361 output as "foo\\bar.y".
2363 - Yacc command and library now available
2364 The Bison distribution now installs a "yacc" command, as POSIX requires.
2365 Also, Bison now installs a small library liby.a containing
2366 implementations of Yacc-compatible yyerror and main functions.
2367 This library is normally not useful, but POSIX requires it.
2369 - Type clashes now generate warnings, not errors.
2371 - If the user does not define YYSTYPE as a macro, Bison now declares it
2372 using typedef instead of defining it as a macro.
2373 For consistency, YYLTYPE is also declared instead of defined.
2375 ** Other compatibility issues
2377 - %union directives can now have a tag before the "{", e.g., the
2378 directive "%union foo {...}" now generates the C code
2379 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
2380 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
2381 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
2382 This is for compatibility with both Yacc and Bison 1.35.
2384 - ";" is output before the terminating "}" of an action, for
2385 compatibility with Bison 1.35.
2387 - Bison now uses a Yacc-style format for conflict reports, e.g.,
2388 "conflicts: 2 shift/reduce, 1 reduce/reduce".
2390 - "yystype" and "yyltype" are now obsolescent macros instead of being
2391 typedefs or tags; they are no longer documented and are planned to be
2392 withdrawn in a future release.
2397 Users of Bison have to decide how they handle the portability of the
2400 - "parsing stack overflow..." -> "parser stack overflow"
2401 GLR parsers now report "parser stack overflow" as per the Bison manual.
2403 ** %parse-param and %lex-param
2404 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
2405 additional context to yyparse and yylex. They suffer from several
2408 - a single argument only can be added,
2409 - their types are weak (void *),
2410 - this context is not passed to ancillary functions such as yyerror,
2411 - only yacc.c parsers support them.
2413 The new %parse-param/%lex-param directives provide a more precise control.
2416 %parse-param {int *nastiness}
2417 %lex-param {int *nastiness}
2418 %parse-param {int *randomness}
2420 results in the following signatures:
2422 int yylex (int *nastiness);
2423 int yyparse (int *nastiness, int *randomness);
2425 or, if both %pure-parser and %locations are used:
2427 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
2428 int yyparse (int *nastiness, int *randomness);
2430 ** Bison now warns if it detects conflicting outputs to the same file,
2431 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
2432 that command outputs both code and header to foo.h.
2434 ** #line in output files
2435 - --no-line works properly.
2437 ** Bison can no longer be built by a K&R C compiler; it requires C89 or
2438 later to be built. This change originally took place a few versions
2439 ago, but nobody noticed until we recently asked someone to try
2440 building Bison with a K&R C compiler.
2442 * Changes in version 1.75, 2002-10-14:
2444 ** Bison should now work on 64-bit hosts.
2446 ** Indonesian translation thanks to Tedi Heriyanto.
2449 Fix spurious parse errors.
2452 Some people redefine yyerror to steal yyparse' private variables.
2453 Reenable this trick until an official feature replaces it.
2456 In agreement with POSIX and with other Yaccs, leaving a default
2457 action is valid when $$ is untyped, and $1 typed:
2461 but the converse remains an error:
2465 ** Values of mid-rule actions
2468 foo: { ... } { $$ = $1; } ...
2470 was incorrectly rejected: $1 is defined in the second mid-rule
2471 action, and is equal to the $$ of the first mid-rule action.
2473 * Changes in version 1.50, 2002-10-04:
2478 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
2479 almost any context-free grammar, ambiguous or not. The new declarations
2480 %dprec and %merge on grammar rules allow parse-time resolution of
2481 ambiguities. Contributed by Paul Hilfinger.
2483 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
2484 like the Alpha, so please stick to 32-bit hosts for now.
2487 When not in Yacc compatibility mode, when the output file was not
2488 specified, running "bison foo/bar.y" created "foo/bar.c". It
2489 now creates "bar.c".
2492 The undefined token was systematically mapped to 2 which prevented
2493 the use of 2 by the user. This is no longer the case.
2495 ** Unknown token numbers
2496 If yylex returned an out of range value, yyparse could die. This is
2500 According to POSIX, the error token must be 256.
2501 Bison extends this requirement by making it a preference: *if* the
2502 user specified that one of her tokens is numbered 256, then error
2503 will be mapped onto another number.
2505 ** Verbose error messages
2506 They no longer report "..., expecting error or..." for states where
2507 error recovery is possible.
2510 Defaults to "$end" instead of "$".
2512 ** Error recovery now conforms to documentation and to POSIX
2513 When a Bison-generated parser encounters a syntax error, it now pops
2514 the stack until it finds a state that allows shifting the error
2515 token. Formerly, it popped the stack until it found a state that
2516 allowed some non-error action other than a default reduction on the
2517 error token. The new behavior has long been the documented behavior,
2518 and has long been required by POSIX. For more details, please see
2519 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
2520 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
2523 Popped tokens and nonterminals are now reported.
2526 Larger grammars are now supported (larger token numbers, larger grammar
2527 size (= sum of the LHS and RHS lengths), larger LALR tables).
2528 Formerly, many of these numbers ran afoul of 16-bit limits;
2529 now these limits are 32 bits on most hosts.
2531 ** Explicit initial rule
2532 Bison used to play hacks with the initial rule, which the user does
2533 not write. It is now explicit, and visible in the reports and
2537 Before, Bison reported the useless rules, but, although not used,
2538 included them in the parsers. They are now actually removed.
2540 ** Useless rules, useless nonterminals
2541 They are now reported, as a warning, with their locations.
2543 ** Rules never reduced
2544 Rules that can never be reduced because of conflicts are now
2547 ** Incorrect "Token not used"
2548 On a grammar such as
2550 %token useless useful
2552 exp: '0' %prec useful;
2554 where a token was used to set the precedence of the last rule,
2555 bison reported both "useful" and "useless" as useless tokens.
2557 ** Revert the C++ namespace changes introduced in 1.31
2558 as they caused too many portability hassles.
2560 ** Default locations
2561 By an accident of design, the default computation of @$ was
2562 performed after another default computation was performed: @$ = @1.
2563 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
2564 the computation of @$.
2566 ** Token end-of-file
2567 The token end of file may be specified by the user, in which case,
2568 the user symbol is used in the reports, the graphs, and the verbose
2569 error messages instead of "$end", which remains being the default.
2573 %token MYEOF 0 "end of file"
2576 This old option, which has been broken for ages, is removed.
2579 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
2580 Croatian, thanks to Denis Lackovic.
2582 ** Incorrect token definitions
2585 bison used to output
2588 ** Token definitions as enums
2589 Tokens are output both as the traditional #define's, and, provided
2590 the compiler supports ANSI C or is a C++ compiler, as enums.
2591 This lets debuggers display names instead of integers.
2594 In addition to --verbose, bison supports --report=THINGS, which
2595 produces additional information:
2597 complete the core item sets with their closure
2598 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
2599 explicitly associate lookahead tokens to items
2601 describe shift/reduce conflicts solving.
2602 Bison used to systematically output this information on top of
2603 the report. Solved conflicts are now attached to their states.
2606 Previous versions don't complain when there is a type clash on
2607 the default action if the rule has a mid-rule action, such as in:
2615 ** GNU M4 is now required when using Bison.
2617 * Changes in version 1.35, 2002-03-25:
2620 Some projects use Bison's C parser with C++ compilers, and define
2621 YYSTYPE as a class. The recent adjustment of C parsers for data
2622 alignment and 64 bit architectures made this impossible.
2624 Because for the time being no real solution for C++ parser
2625 generation exists, kludges were implemented in the parser to
2626 maintain this use. In the future, when Bison has C++ parsers, this
2627 kludge will be disabled.
2629 This kludge also addresses some C++ problems when the stack was
2632 * Changes in version 1.34, 2002-03-12:
2634 ** File name clashes are detected
2635 $ bison foo.y -d -o foo.x
2636 fatal error: header and parser would both be named "foo.x"
2638 ** A missing ";" at the end of a rule triggers a warning
2639 In accordance with POSIX, and in agreement with other
2640 Yacc implementations, Bison will mandate this semicolon in the near
2641 future. This eases the implementation of a Bison parser of Bison
2642 grammars by making this grammar LALR(1) instead of LR(2). To
2643 facilitate the transition, this release introduces a warning.
2645 ** Revert the C++ namespace changes introduced in 1.31, as they caused too
2646 many portability hassles.
2648 ** DJGPP support added.
2650 ** Fix test suite portability problems.
2652 * Changes in version 1.33, 2002-02-07:
2655 Groff could not be compiled for the definition of size_t was lacking
2656 under some conditions.
2661 * Changes in version 1.32, 2002-01-23:
2663 ** Fix Yacc output file names
2665 ** Portability fixes
2667 ** Italian, Dutch translations
2669 * Changes in version 1.31, 2002-01-14:
2673 ** GNU Gettext and %expect
2674 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
2675 Bison dies on incorrect %expectations, we fear there will be
2676 too many bug reports for Gettext, so _for the time being_, %expect
2677 does not trigger an error when the input file is named "plural.y".
2679 ** Use of alloca in parsers
2680 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
2681 malloc exclusively. Since 1.29, but was not NEWS'ed.
2683 alloca is used only when compiled with GCC, to avoid portability
2686 ** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
2688 ** When the generated parser lacks debugging code, YYDEBUG is now 0
2689 (as POSIX requires) instead of being undefined.
2692 Bison has always permitted actions such as { $$ = $1 }: it adds the
2693 ending semicolon. Now if in Yacc compatibility mode, the semicolon
2694 is no longer output: one has to write { $$ = $1; }.
2696 ** Better C++ compliance
2697 The output parsers try to respect C++ namespaces.
2698 [This turned out to be a failed experiment, and it was reverted later.]
2701 Fixed bugs when reporting useless nonterminals.
2704 The parsers work properly on 64 bit hosts.
2707 Some calls to strerror resulted in scrambled or missing error messages.
2710 When the number of shift/reduce conflicts is correct, don't issue
2713 ** The verbose report includes the rule line numbers.
2715 ** Rule line numbers are fixed in traces.
2717 ** Swedish translation
2720 Verbose parse error messages from the parsers are better looking.
2721 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
2722 Now: parse error: unexpected '/', expecting "number" or '-' or '('
2724 ** Fixed parser memory leaks.
2725 When the generated parser was using malloc to extend its stacks, the
2726 previous allocations were not freed.
2728 ** Fixed verbose output file.
2729 Some newlines were missing.
2730 Some conflicts in state descriptions were missing.
2732 ** Fixed conflict report.
2733 Option -v was needed to get the result.
2737 Mismatches are errors, not warnings.
2739 ** Fixed incorrect processing of some invalid input.
2741 ** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
2743 ** Fixed some typos in the documentation.
2745 ** %token MY_EOF 0 is supported.
2746 Before, MY_EOF was silently renumbered as 257.
2748 ** doc/refcard.tex is updated.
2750 ** %output, %file-prefix, %name-prefix.
2754 New, aliasing "--output-file".
2756 * Changes in version 1.30, 2001-10-26:
2758 ** "--defines" and "--graph" have now an optional argument which is the
2759 output file name. "-d" and "-g" do not change; they do not take any
2762 ** "%source_extension" and "%header_extension" are removed, failed
2765 ** Portability fixes.
2767 * Changes in version 1.29, 2001-09-07:
2769 ** The output file does not define const, as this caused problems when used
2770 with common autoconfiguration schemes. If you still use ancient compilers
2771 that lack const, compile with the equivalent of the C compiler option
2772 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
2774 ** Added "-g" and "--graph".
2776 ** The Bison manual is now distributed under the terms of the GNU FDL.
2778 ** The input and the output files has automatically a similar extension.
2780 ** Russian translation added.
2782 ** NLS support updated; should hopefully be less troublesome.
2784 ** Added the old Bison reference card.
2786 ** Added "--locations" and "%locations".
2788 ** Added "-S" and "--skeleton".
2790 ** "%raw", "-r", "--raw" is disabled.
2792 ** Special characters are escaped when output. This solves the problems
2793 of the #line lines with path names including backslashes.
2796 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
2797 "%debug", "%source_extension" and "%header_extension".
2800 Automatic location tracking.
2802 * Changes in version 1.28, 1999-07-06:
2804 ** Should compile better now with K&R compilers.
2808 ** Fixed a problem with escaping the double quote character.
2810 ** There is now a FAQ.
2812 * Changes in version 1.27:
2814 ** The make rule which prevented bison.simple from being created on
2815 some systems has been fixed.
2817 * Changes in version 1.26:
2819 ** Bison now uses Automake.
2821 ** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
2823 ** Token numbers now start at 257 as previously documented, not 258.
2825 ** Bison honors the TMPDIR environment variable.
2827 ** A couple of buffer overruns have been fixed.
2829 ** Problems when closing files should now be reported.
2831 ** Generated parsers should now work even on operating systems which do
2832 not provide alloca().
2834 * Changes in version 1.25, 1995-10-16:
2836 ** Errors in the input grammar are not fatal; Bison keeps reading
2837 the grammar file, and reports all the errors found in it.
2839 ** Tokens can now be specified as multiple-character strings: for
2840 example, you could use "<=" for a token which looks like <=, instead
2841 of choosing a name like LESSEQ.
2843 ** The %token_table declaration says to write a table of tokens (names
2844 and numbers) into the parser file. The yylex function can use this
2845 table to recognize multiple-character string tokens, or for other
2848 ** The %no_lines declaration says not to generate any #line preprocessor
2849 directives in the parser file.
2851 ** The %raw declaration says to use internal Bison token numbers, not
2852 Yacc-compatible token numbers, when token names are defined as macros.
2854 ** The --no-parser option produces the parser tables without including
2855 the parser engine; a project can now use its own parser engine.
2856 The actions go into a separate file called NAME.act, in the form of
2857 a switch statement body.
2859 * Changes in version 1.23:
2861 The user can define YYPARSE_PARAM as the name of an argument to be
2862 passed into yyparse. The argument should have type void *. It should
2863 actually point to an object. Grammar actions can access the variable
2864 by casting it to the proper pointer type.
2866 Line numbers in output file corrected.
2868 * Changes in version 1.22:
2870 --help option added.
2872 * Changes in version 1.20:
2874 Output file does not redefine const for C++.
2878 Copyright (C) 1995-2015, 2018 Free Software Foundation, Inc.
2880 This file is part of Bison, the GNU Parser Generator.
2882 This program is free software: you can redistribute it and/or modify
2883 it under the terms of the GNU General Public License as published by
2884 the Free Software Foundation, either version 3 of the License, or
2885 (at your option) any later version.
2887 This program is distributed in the hope that it will be useful,
2888 but WITHOUT ANY WARRANTY; without even the implied warranty of
2889 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2890 GNU General Public License for more details.
2892 You should have received a copy of the GNU General Public License
2893 along with this program. If not, see <http://www.gnu.org/licenses/>.
2895 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2896 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2897 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2898 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2899 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2900 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2901 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2902 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2903 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2904 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2905 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2906 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2907 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
2908 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
2909 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
2910 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2911 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
2912 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
2913 LocalWords: TOK calc yyo fval Wconflicts parsers yystackp yyval yynerrs
2914 LocalWords: Théophile Ranquet Santet fno fnone stype associativity Tolmer
2915 LocalWords: Wprecedence Rassoul Wempty Paolo Bonzini parser's Michiel loc
2916 LocalWords: redeclaration sval fcaret reentrant XSLT xsl Wmaybe yyvsp Tedi
2917 LocalWords: pragmas noreturn untyped Rozenman unexpanded Wojciech Polak
2918 LocalWords: Alexandre MERCHANTABILITY yytype