3 * Noteworthy changes in release ?.? (????-??-??) [?]
7 *** C++ with Variants (lalr1.cc)
9 Problems with %destructor and '%define parse.assert' have been fixed.
11 *** Named %union support (yacc.c, glr.c)
13 Bison 3.0 introduced a regression on named %union such as
15 %union foo { int ival; };
17 The possibility to use a name was introduced "for Yacc compatibility".
18 It is however not required by POSIX Yacc, and its usefulness is not clear.
20 *** %define api.value.type union with %defines (yacc.c, glr.c)
22 The C parsers were broken when %defines was used together with "%define
23 api.value.type union".
25 *** Redeclarations are reported in proper order
35 /tmp/foo.yy:2.10-11: error: %printer redeclaration for FOO
38 /tmp/foo.yy:3.10-11: previous declaration
42 Now, the "previous" declaration is always the first one.
47 Bison now installs various files in its docdir (which defaults to
48 '/usr/local/share/doc/bison'), including the three fully blown examples
49 extracted from the documentation:
52 Reverse polish calculator, a simple introductory example.
54 Multi-function Calc, a calculator with memory and functions and located
57 a calculator in C++ using variant support and token constructors.
59 * Noteworthy changes in release 3.0.2 (2013-12-05) [stable]
63 *** Generated source files when errors are reported
65 When warnings are issued and -Werror is set, bison would still generate
66 the source files (*.c, *.h...). As a consequence, some runs of "make"
67 could fail the first time, but not the second (as the files were generated
70 This is fixed: bison no longer generates this source files, but, of
71 course, still produces the various reports (*.output, *.xml, etc.).
73 *** %empty is used in reports
75 Empty right-hand sides are denoted by '%empty' in all the reports (text,
76 dot, XML and formats derived from it).
78 *** YYERROR and variants
80 When C++ variant support is enabled, an error triggered via YYERROR, but
81 not caught via error recovery, resulted in a double deletion.
83 * Noteworthy changes in release 3.0.1 (2013-11-12) [stable]
87 *** Errors in caret diagnostics
89 On some platforms, some errors could result in endless diagnostics.
91 *** Fixes of the -Werror option
93 Options such as "-Werror -Wno-error=foo" were still turning "foo"
94 diagnostics into errors instead of warnings. This is fixed.
96 Actually, for consistency with GCC, "-Wno-error=foo -Werror" now also
97 leaves "foo" diagnostics as warnings. Similarly, with "-Werror=foo
98 -Wno-error", "foo" diagnostics are now errors.
102 As demonstrated in the documentation, one can now leave spaces between
107 The yacc.1 man page is no longer installed if --disable-yacc was
110 *** Fixes in the test suite
112 Bugs and portability issues.
114 * Noteworthy changes in release 3.0 (2013-07-25) [stable]
116 ** WARNING: Future backward-incompatibilities!
118 Like other GNU packages, Bison will start using some of the C99 features
119 for its own code, especially the definition of variables after statements.
120 The generated C parsers still aim at C90.
122 ** Backward incompatible changes
124 *** Obsolete features
126 Support for YYFAIL is removed (deprecated in Bison 2.4.2): use YYERROR.
128 Support for yystype and yyltype is removed (deprecated in Bison 1.875):
129 use YYSTYPE and YYLTYPE.
131 Support for YYLEX_PARAM and YYPARSE_PARAM is removed (deprecated in Bison
132 1.875): use %lex-param, %parse-param, or %param.
134 Missing semicolons at the end of actions are no longer added (as announced
137 *** Use of YACC='bison -y'
139 TL;DR: With Autoconf <= 2.69, pass -Wno-yacc to (AM_)YFLAGS if you use
142 Traditional Yacc generates 'y.tab.c' whatever the name of the input file.
143 Therefore Makefiles written for Yacc expect 'y.tab.c' (and possibly
144 'y.tab.h' and 'y.outout') to be generated from 'foo.y'.
146 To this end, for ages, AC_PROG_YACC, Autoconf's macro to look for an
147 implementation of Yacc, was using Bison as 'bison -y'. While it does
148 ensure compatible output file names, it also enables warnings for
149 incompatibilities with POSIX Yacc. In other words, 'bison -y' triggers
150 warnings for Bison extensions.
152 Autoconf 2.70+ fixes this incompatibility by using YACC='bison -o y.tab.c'
153 (which also generates 'y.tab.h' and 'y.output' when needed).
154 Alternatively, disable Yacc warnings by passing '-Wno-yacc' to your Yacc
155 flags (YFLAGS, or AM_YFLAGS with Automake).
159 *** The epilogue is no longer affected by internal #defines (glr.c)
161 The glr.c skeleton uses defines such as #define yylval (yystackp->yyval) in
162 generated code. These weren't properly undefined before the inclusion of
163 the user epilogue, so functions such as the following were butchered by the
164 preprocessor expansion:
166 int yylex (YYSTYPE *yylval);
168 This is fixed: yylval, yynerrs, yychar, and yylloc are now valid
169 identifiers for user-provided variables.
171 *** stdio.h is no longer needed when locations are enabled (yacc.c)
173 Changes in Bison 2.7 introduced a dependency on FILE and fprintf when
174 locations are enabled. This is fixed.
176 *** Warnings about useless %pure-parser/%define api.pure are restored
178 ** Diagnostics reported by Bison
180 Most of these features were contributed by Théophile Ranquet and Victor
185 Version 2.7 introduced caret errors, for a prettier output. These are now
186 activated by default. The old format can still be used by invoking Bison
187 with -fno-caret (or -fnone).
189 Some error messages that reproduced excerpts of the grammar are now using
190 the caret information only. For instance on:
197 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
198 in.y:2.12-14: warning: rule useless in parser due to conflicts: exp: 'a' [-Wother]
202 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
203 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
207 and "bison -fno-caret" reports:
209 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
210 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
212 *** Enhancements of the -Werror option
214 The -Werror=CATEGORY option is now recognized, and will treat specified
215 warnings as errors. The warnings need not have been explicitly activated
216 using the -W option, this is similar to what GCC 4.7 does.
218 For example, given the following command line, Bison will treat both
219 warnings related to POSIX Yacc incompatibilities and S/R conflicts as
220 errors (and only those):
222 $ bison -Werror=yacc,error=conflicts-sr input.y
224 If no categories are specified, -Werror will make all active warnings into
225 errors. For example, the following line does the same the previous example:
227 $ bison -Werror -Wnone -Wyacc -Wconflicts-sr input.y
229 (By default -Wconflicts-sr,conflicts-rr,deprecated,other is enabled.)
231 Note that the categories in this -Werror option may not be prefixed with
232 "no-". However, -Wno-error[=CATEGORY] is valid.
234 Note that -y enables -Werror=yacc. Therefore it is now possible to require
235 Yacc-like behavior (e.g., always generate y.tab.c), but to report
236 incompatibilities as warnings: "-y -Wno-error=yacc".
238 *** The display of warnings is now richer
240 The option that controls a given warning is now displayed:
242 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
244 In the case of warnings treated as errors, the prefix is changed from
245 "warning: " to "error: ", and the suffix is displayed, in a manner similar
246 to GCC, as [-Werror=CATEGORY].
248 For instance, where the previous version of Bison would report (and exit
251 bison: warnings being treated as errors
252 input.y:1.1: warning: stray ',' treated as white space
256 input.y:1.1: error: stray ',' treated as white space [-Werror=other]
258 *** Deprecated constructs
260 The new 'deprecated' warning category flags obsolete constructs whose
261 support will be discontinued. It is enabled by default. These warnings
262 used to be reported as 'other' warnings.
264 *** Useless semantic types
266 Bison now warns about useless (uninhabited) semantic types. Since
267 semantic types are not declared to Bison (they are defined in the opaque
268 %union structure), it is %printer/%destructor directives about useless
269 types that trigger the warning:
273 %printer {} <type1> <type3>
274 %destructor {} <type2> <type4>
276 nterm: term { $$ = $1; };
278 3.28-34: warning: type <type3> is used, but is not associated to any symbol
279 4.28-34: warning: type <type4> is used, but is not associated to any symbol
281 *** Undefined but unused symbols
283 Bison used to raise an error for undefined symbols that are not used in
284 the grammar. This is now only a warning.
287 %destructor {} symbol2
292 *** Useless destructors or printers
294 Bison now warns about useless destructors or printers. In the following
295 example, the printer for <type1>, and the destructor for <type2> are
296 useless: all symbols of <type1> (token1) already have a printer, and all
297 symbols of type <type2> (token2) already have a destructor.
299 %token <type1> token1
303 %printer {} token1 <type1> <type3>
304 %destructor {} token2 <type2> <type4>
308 The warnings and error messages about shift/reduce and reduce/reduce
309 conflicts have been normalized. For instance on the following foo.y file:
313 exp: exp '+' exp | '0' | '0';
315 compare the previous version of bison:
318 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
319 $ bison -Werror foo.y
320 bison: warnings being treated as errors
321 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
323 with the new behavior:
326 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
327 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
328 $ bison -Werror foo.y
329 foo.y: error: 1 shift/reduce conflict [-Werror=conflicts-sr]
330 foo.y: error: 2 reduce/reduce conflicts [-Werror=conflicts-rr]
332 When %expect or %expect-rr is used, such as with bar.y:
337 exp: exp '+' exp | '0' | '0';
342 bar.y: conflicts: 1 shift/reduce, 2 reduce/reduce
343 bar.y: expected 0 shift/reduce conflicts
344 bar.y: expected 0 reduce/reduce conflicts
349 bar.y: error: shift/reduce conflicts: 1 found, 0 expected
350 bar.y: error: reduce/reduce conflicts: 2 found, 0 expected
352 ** Incompatibilities with POSIX Yacc
354 The 'yacc' category is no longer part of '-Wall', enable it explicitly
357 ** Additional yylex/yyparse arguments
359 The new directive %param declares additional arguments to both yylex and
360 yyparse. The %lex-param, %parse-param, and %param directives support one
361 or more arguments. Instead of
363 %lex-param {arg1_type *arg1}
364 %lex-param {arg2_type *arg2}
365 %parse-param {arg1_type *arg1}
366 %parse-param {arg2_type *arg2}
370 %param {arg1_type *arg1} {arg2_type *arg2}
372 ** Types of values for %define variables
374 Bison used to make no difference between '%define foo bar' and '%define
375 foo "bar"'. The former is now called a 'keyword value', and the latter a
376 'string value'. A third kind was added: 'code values', such as '%define
379 Keyword variables are used for fixed value sets, e.g.,
383 Code variables are used for value in the target language, e.g.,
385 %define api.value.type {struct semantic_type}
387 String variables are used remaining cases, e.g. file names.
389 ** Variable api.token.prefix
391 The variable api.token.prefix changes the way tokens are identified in
392 the generated files. This is especially useful to avoid collisions
393 with identifiers in the target language. For instance
395 %token FILE for ERROR
396 %define api.token.prefix {TOK_}
398 start: FILE for ERROR;
400 will generate the definition of the symbols TOK_FILE, TOK_for, and
401 TOK_ERROR in the generated sources. In particular, the scanner must
402 use these prefixed token names, although the grammar itself still
403 uses the short names (as in the sample rule given above).
405 ** Variable api.value.type
407 This new %define variable supersedes the #define macro YYSTYPE. The use
408 of YYSTYPE is discouraged. In particular, #defining YYSTYPE *and* either
409 using %union or %defining api.value.type results in undefined behavior.
411 Either define api.value.type, or use "%union":
418 %token <ival> INT "integer"
419 %token <sval> STRING "string"
420 %printer { fprintf (yyo, "%d", $$); } <ival>
421 %destructor { free ($$); } <sval>
424 yylval.ival = 42; return INT;
425 yylval.sval = "42"; return STRING;
427 The %define variable api.value.type supports both keyword and code values.
429 The keyword value 'union' means that the user provides genuine types, not
430 union member names such as "ival" and "sval" above (WARNING: will fail if
431 -y/--yacc/%yacc is enabled).
433 %define api.value.type union
434 %token <int> INT "integer"
435 %token <char *> STRING "string"
436 %printer { fprintf (yyo, "%d", $$); } <int>
437 %destructor { free ($$); } <char *>
440 yylval.INT = 42; return INT;
441 yylval.STRING = "42"; return STRING;
443 The keyword value variant is somewhat equivalent, but for C++ special
444 provision is made to allow classes to be used (more about this below).
446 %define api.value.type variant
447 %token <int> INT "integer"
448 %token <std::string> STRING "string"
450 Code values (in braces) denote user defined types. This is where YYSTYPE
468 %define api.value.type {struct my_value}
469 %token <u.ival> INT "integer"
470 %token <u.sval> STRING "string"
471 %printer { fprintf (yyo, "%d", $$); } <u.ival>
472 %destructor { free ($$); } <u.sval>
475 yylval.u.ival = 42; return INT;
476 yylval.u.sval = "42"; return STRING;
478 ** Variable parse.error
480 This variable controls the verbosity of error messages. The use of the
481 %error-verbose directive is deprecated in favor of "%define parse.error
484 ** Renamed %define variables
486 The following variables have been renamed for consistency. Backward
487 compatibility is ensured, but upgrading is recommended.
489 lr.default-reductions -> lr.default-reduction
490 lr.keep-unreachable-states -> lr.keep-unreachable-state
491 namespace -> api.namespace
492 stype -> api.value.type
494 ** Semantic predicates
496 Contributed by Paul Hilfinger.
498 The new, experimental, semantic-predicate feature allows actions of the
499 form "%?{ BOOLEAN-EXPRESSION }", which cause syntax errors (as for
500 YYERROR) if the expression evaluates to 0, and are evaluated immediately
501 in GLR parsers, rather than being deferred. The result is that they allow
502 the programmer to prune possible parses based on the values of run-time
505 ** The directive %expect-rr is now an error in non GLR mode
507 It used to be an error only if used in non GLR mode, _and_ if there are
508 reduce/reduce conflicts.
510 ** Tokens are numbered in their order of appearance
512 Contributed by Valentin Tolmer.
514 With '%token A B', A had a number less than the one of B. However,
515 precedence declarations used to generate a reversed order. This is now
516 fixed, and introducing tokens with any of %token, %left, %right,
517 %precedence, or %nonassoc yields the same result.
519 When mixing declarations of tokens with a litteral character (e.g., 'a')
520 or with an identifier (e.g., B) in a precedence declaration, Bison
521 numbered the litteral characters first. For example
525 would lead to the tokens declared in this order: 'c' 'd' A B. Again, the
526 input order is now preserved.
528 These changes were made so that one can remove useless precedence and
529 associativity declarations (i.e., map %nonassoc, %left or %right to
530 %precedence, or to %token) and get exactly the same output.
532 ** Useless precedence and associativity
534 Contributed by Valentin Tolmer.
536 When developing and maintaining a grammar, useless associativity and
537 precedence directives are common. They can be a nuisance: new ambiguities
538 arising are sometimes masked because their conflicts are resolved due to
539 the extra precedence or associativity information. Furthermore, it can
540 hinder the comprehension of a new grammar: one will wonder about the role
541 of a precedence, where in fact it is useless. The following changes aim
542 at detecting and reporting these extra directives.
544 *** Precedence warning category
546 A new category of warning, -Wprecedence, was introduced. It flags the
547 useless precedence and associativity directives.
549 *** Useless associativity
551 Bison now warns about symbols with a declared associativity that is never
552 used to resolve conflicts. In that case, using %precedence is sufficient;
553 the parsing tables will remain unchanged. Solving these warnings may raise
554 useless precedence warnings, as the symbols no longer have associativity.
568 warning: useless associativity for '+', use %precedence [-Wprecedence]
572 *** Useless precedence
574 Bison now warns about symbols with a declared precedence and no declared
575 associativity (i.e., declared with %precedence), and whose precedence is
576 never used. In that case, the symbol can be safely declared with %token
577 instead, without modifying the parsing tables. For example:
581 exp: "var" '=' "number";
585 warning: useless precedence for '=' [-Wprecedence]
589 *** Useless precedence and associativity
591 In case of both useless precedence and associativity, the issue is flagged
596 exp: "var" '=' "number";
600 warning: useless precedence and associativity for '=' [-Wprecedence]
606 With help from Joel E. Denny and Gabriel Rassoul.
608 Empty rules (i.e., with an empty right-hand side) can now be explicitly
609 marked by the new %empty directive. Using %empty on a non-empty rule is
610 an error. The new -Wempty-rule warning reports empty rules without
611 %empty. On the following grammar:
621 3.4-5: warning: empty rule without %empty [-Wempty-rule]
624 5.8-13: error: %empty on non-empty rule
628 ** Java skeleton improvements
630 The constants for token names were moved to the Lexer interface. Also, it
631 is possible to add code to the parser's constructors using "%code init"
632 and "%define init_throws".
633 Contributed by Paolo Bonzini.
635 The Java skeleton now supports push parsing.
636 Contributed by Dennis Heimbigner.
638 ** C++ skeletons improvements
640 *** The parser header is no longer mandatory (lalr1.cc, glr.cc)
642 Using %defines is now optional. Without it, the needed support classes
643 are defined in the generated parser, instead of additional files (such as
644 location.hh, position.hh and stack.hh).
646 *** Locations are no longer mandatory (lalr1.cc, glr.cc)
648 Both lalr1.cc and glr.cc no longer require %location.
650 *** syntax_error exception (lalr1.cc)
652 The C++ parser features a syntax_error exception, which can be
653 thrown from the scanner or from user rules to raise syntax errors.
654 This facilitates reporting errors caught in sub-functions (e.g.,
655 rejecting too large integral literals from a conversion function
656 used by the scanner, or rejecting invalid combinations from a
657 factory invoked by the user actions).
659 *** %define api.value.type variant
661 This is based on a submission from Michiel De Wilde. With help
662 from Théophile Ranquet.
664 In this mode, complex C++ objects can be used as semantic values. For
667 %token <::std::string> TEXT;
670 %type <::std::string> item;
671 %type <::std::list<std::string>> list;
674 list { std::cout << $1 << std::endl; }
678 %empty { /* Generates an empty string list. */ }
679 | list item ";" { std::swap ($$, $1); $$.push_back ($2); }
683 TEXT { std::swap ($$, $1); }
684 | NUMBER { $$ = string_cast ($1); }
687 *** %define api.token.constructor
689 When variants are enabled, Bison can generate functions to build the
690 tokens. This guarantees that the token type (e.g., NUMBER) is consistent
691 with the semantic value (e.g., int):
693 parser::symbol_type yylex ()
695 parser::location_type loc = ...;
697 return parser::make_TEXT ("Hello, world!", loc);
699 return parser::make_NUMBER (42, loc);
701 return parser::make_SEMICOLON (loc);
707 There are operator- and operator-= for 'location'. Negative line/column
708 increments can no longer underflow the resulting value.
710 * Noteworthy changes in release 2.7.1 (2013-04-15) [stable]
714 *** Fix compiler attribute portability (yacc.c)
716 With locations enabled, __attribute__ was used unprotected.
718 *** Fix some compiler warnings (lalr1.cc)
720 * Noteworthy changes in release 2.7 (2012-12-12) [stable]
724 Warnings about uninitialized yylloc in yyparse have been fixed.
726 Restored C90 compliance (yet no report was ever made).
728 ** Diagnostics are improved
730 Contributed by Théophile Ranquet.
732 *** Changes in the format of error messages
734 This used to be the format of many error reports:
736 input.y:2.7-12: %type redeclaration for exp
737 input.y:1.7-12: previous declaration
741 input.y:2.7-12: error: %type redeclaration for exp
742 input.y:1.7-12: previous declaration
744 *** New format for error reports: carets
746 Caret errors have been added to Bison:
748 input.y:2.7-12: error: %type redeclaration for exp
751 input.y:1.7-12: previous declaration
757 input.y:3.20-23: error: ambiguous reference: '$exp'
758 exp: exp '+' exp { $exp = $1 + $3; };
760 input.y:3.1-3: refers to: $exp at $$
761 exp: exp '+' exp { $exp = $1 + $3; };
763 input.y:3.6-8: refers to: $exp at $1
764 exp: exp '+' exp { $exp = $1 + $3; };
766 input.y:3.14-16: refers to: $exp at $3
767 exp: exp '+' exp { $exp = $1 + $3; };
770 The default behavior for now is still not to display these unless
771 explicitly asked with -fcaret (or -fall). However, in a later release, it
772 will be made the default behavior (but may still be deactivated with
775 ** New value for %define variable: api.pure full
777 The %define variable api.pure requests a pure (reentrant) parser. However,
778 for historical reasons, using it in a location-tracking Yacc parser
779 resulted in a yyerror function that did not take a location as a
780 parameter. With this new value, the user may request a better pure parser,
781 where yyerror does take a location as a parameter (in location-tracking
784 The use of "%define api.pure true" is deprecated in favor of this new
785 "%define api.pure full".
787 ** New %define variable: api.location.type (glr.cc, lalr1.cc, lalr1.java)
789 The %define variable api.location.type defines the name of the type to use
790 for locations. When defined, Bison no longer generates the position.hh
791 and location.hh files, nor does the parser will include them: the user is
792 then responsible to define her type.
794 This can be used in programs with several parsers to factor their location
795 and position files: let one of them generate them, and the others just use
798 This feature was actually introduced, but not documented, in Bison 2.5,
799 under the name "location_type" (which is maintained for backward
802 For consistency, lalr1.java's %define variables location_type and
803 position_type are deprecated in favor of api.location.type and
806 ** Exception safety (lalr1.cc)
808 The parse function now catches exceptions, uses the %destructors to
809 release memory (the lookahead symbol and the symbols pushed on the stack)
810 before re-throwing the exception.
812 This feature is somewhat experimental. User feedback would be
815 ** Graph improvements in DOT and XSLT
817 Contributed by Théophile Ranquet.
819 The graphical presentation of the states is more readable: their shape is
820 now rectangular, the state number is clearly displayed, and the items are
821 numbered and left-justified.
823 The reductions are now explicitly represented as transitions to other
824 diamond shaped nodes.
826 These changes are present in both --graph output and xml2dot.xsl XSLT
827 processing, with minor (documented) differences.
829 ** %language is no longer an experimental feature.
831 The introduction of this feature, in 2.4, was four years ago. The
832 --language option and the %language directive are no longer experimental.
836 The sections about shift/reduce and reduce/reduce conflicts resolution
837 have been fixed and extended.
839 Although introduced more than four years ago, XML and Graphviz reports
840 were not properly documented.
842 The translation of mid-rule actions is now described.
844 * Noteworthy changes in release 2.6.5 (2012-11-07) [stable]
846 We consider compiler warnings about Bison generated parsers to be bugs.
847 Rather than working around them in your own project, please consider
848 reporting them to us.
852 Warnings about uninitialized yylval and/or yylloc for push parsers with a
853 pure interface have been fixed for GCC 4.0 up to 4.8, and Clang 2.9 to
856 Other issues in the test suite have been addressed.
858 Null characters are correctly displayed in error messages.
860 When possible, yylloc is correctly initialized before calling yylex. It
861 is no longer necessary to initialize it in the %initial-action.
863 * Noteworthy changes in release 2.6.4 (2012-10-23) [stable]
865 Bison 2.6.3's --version was incorrect. This release fixes this issue.
867 * Noteworthy changes in release 2.6.3 (2012-10-22) [stable]
871 Bugs and portability issues in the test suite have been fixed.
873 Some errors in translations have been addressed, and --help now directs
874 users to the appropriate place to report them.
876 Stray Info files shipped by accident are removed.
878 Incorrect definitions of YY_, issued by yacc.c when no parser header is
879 generated, are removed.
881 All the generated headers are self-contained.
883 ** Header guards (yacc.c, glr.c, glr.cc)
885 In order to avoid collisions, the header guards are now
886 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
887 For instance the header generated from
889 %define api.prefix "calc"
890 %defines "lib/parse.h"
892 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
894 ** Fix compiler warnings in the generated parser (yacc.c, glr.c)
896 The compilation of pure parsers (%define api.pure) can trigger GCC
899 input.c: In function 'yyparse':
900 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
901 function [-Wmaybe-uninitialized]
905 This is now fixed; pragmas to avoid these warnings are no longer needed.
907 Warnings from clang ("equality comparison with extraneous parentheses" and
908 "function declared 'noreturn' should not return") have also been
911 * Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
915 Buffer overruns, complaints from Flex, and portability issues in the test
916 suite have been fixed.
918 ** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
920 Trailing end-of-lines in %parse-param or %lex-param would result in
921 invalid C++. This is fixed.
923 ** Spurious spaces and end-of-lines
925 The generated files no longer end (nor start) with empty lines.
927 * Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
929 Bison no longer executes user-specified M4 code when processing a grammar.
933 In addition to the removal of the features announced in Bison 2.6, the
934 next major release will remove the "Temporary hack for adding a semicolon
935 to the user action", as announced in the release 2.5. Instead of:
937 exp: exp "+" exp { $$ = $1 + $3 };
941 exp: exp "+" exp { $$ = $1 + $3; };
945 *** Type names are now properly escaped.
947 *** glr.cc: set_debug_level and debug_level work as expected.
949 *** Stray @ or $ in actions
951 While Bison used to warn about stray $ or @ in action rules, it did not
952 for other actions such as printers, destructors, or initial actions. It
955 ** Type names in actions
957 For consistency with rule actions, it is now possible to qualify $$ by a
958 type-name in destructors, printers, and initial actions. For instance:
960 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
962 will display two values for each typed and untyped symbol (provided
963 that YYSTYPE has both "ival" and "fval" fields).
965 * Noteworthy changes in release 2.6 (2012-07-19) [stable]
969 The next major release of Bison will drop support for the following
970 deprecated features. Please report disagreements to bug-bison@gnu.org.
974 Support for generating parsers in K&R C will be removed. Parsers
975 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
978 *** Features deprecated since Bison 1.875
980 The definitions of yystype and yyltype will be removed; use YYSTYPE and
983 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
984 %lex-param, will no longer be supported.
986 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
989 *** The generated header will be included (yacc.c)
991 Instead of duplicating the content of the generated header (definition of
992 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
993 as is already the case for GLR or C++ parsers. This change is deferred
994 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
997 ** Generated Parser Headers
999 *** Guards (yacc.c, glr.c, glr.cc)
1001 The generated headers are now guarded, as is already the case for C++
1002 parsers (lalr1.cc). For instance, with --defines=foo.h:
1007 #endif /* !YY_FOO_H */
1009 *** New declarations (yacc.c, glr.c)
1011 The generated header now declares yydebug and yyparse. Both honor
1012 --name-prefix=bar_, and yield
1014 int bar_parse (void);
1018 #define yyparse bar_parse
1021 in order to facilitate the inclusion of several parser headers inside a
1022 single compilation unit.
1024 *** Exported symbols in C++
1026 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
1027 header, are removed, as they prevent the possibility of including several
1028 generated headers from a single compilation unit.
1032 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
1035 ** New %define variable: api.prefix
1037 Now that the generated headers are more complete and properly protected
1038 against multiple inclusions, constant names, such as YYSTYPE are a
1039 problem. While yyparse and others are properly renamed by %name-prefix,
1040 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
1041 would introduce backward compatibility issues in projects not expecting
1042 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
1043 it is deprecated in favor of a new %define variable: api.prefix.
1045 The following examples compares both:
1047 %name-prefix "bar_" | %define api.prefix "bar_"
1048 %token <ival> FOO %token <ival> FOO
1049 %union { int ival; } %union { int ival; }
1055 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
1056 # define BAR_FOO_H # define BAR_FOO_H
1058 /* Enabling traces. */ /* Enabling traces. */
1059 # ifndef YYDEBUG | # ifndef BAR_DEBUG
1060 > # if defined YYDEBUG
1062 > # define BAR_DEBUG 1
1064 > # define BAR_DEBUG 0
1067 # define YYDEBUG 0 | # define BAR_DEBUG 0
1071 # if YYDEBUG | # if BAR_DEBUG
1072 extern int bar_debug; extern int bar_debug;
1075 /* Tokens. */ /* Tokens. */
1076 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
1077 # define YYTOKENTYPE | # define BAR_TOKENTYPE
1078 enum yytokentype { | enum bar_tokentype {
1083 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
1084 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
1085 typedef union YYSTYPE | typedef union BAR_STYPE
1088 } YYSTYPE; | } BAR_STYPE;
1089 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
1092 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
1094 int bar_parse (void); int bar_parse (void);
1096 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
1098 * Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
1102 The next major release will drop support for generating parsers in K&R C.
1104 ** yacc.c: YYBACKUP works as expected.
1106 ** glr.c improvements:
1108 *** Location support is eliminated when not requested:
1110 GLR parsers used to include location-related code even when locations were
1111 not requested, and therefore not even usable.
1113 *** __attribute__ is preserved:
1115 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
1116 when -std is passed to GCC).
1118 ** lalr1.java: several fixes:
1120 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
1121 first token leads to a syntax error. Some minor clean ups.
1125 *** C++11 compatibility:
1127 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
1132 The header files such as "parser.hh", "location.hh", etc. used a constant
1133 name for preprocessor guards, for instance:
1135 #ifndef BISON_LOCATION_HH
1136 # define BISON_LOCATION_HH
1138 #endif // !BISON_LOCATION_HH
1140 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
1141 case characters are converted to upper case, and series of
1142 non-alphanumerical characters are converted to an underscore.
1144 With "bison -o lang++/parser.cc", "location.hh" would now include:
1146 #ifndef YY_LANG_LOCATION_HH
1147 # define YY_LANG_LOCATION_HH
1149 #endif // !YY_LANG_LOCATION_HH
1153 The position and location constructors (and their initialize methods)
1154 accept new arguments for line and column. Several issues in the
1155 documentation were fixed.
1157 ** liby is no longer asking for "rpl_fprintf" on some platforms.
1159 ** Changes in the manual:
1161 *** %printer is documented
1163 The "%printer" directive, supported since at least Bison 1.50, is finally
1164 documented. The "mfcalc" example is extended to demonstrate it.
1166 For consistency with the C skeletons, the C++ parsers now also support
1167 "yyoutput" (as an alias to "debug_stream ()").
1169 *** Several improvements have been made:
1171 The layout for grammar excerpts was changed to a more compact scheme.
1172 Named references are motivated. The description of the automaton
1173 description file (*.output) is updated to the current format. Incorrect
1174 index entries were fixed. Some other errors were fixed.
1178 *** Conflicting prototypes with recent/modified Flex.
1180 Fixed build problems with the current, unreleased, version of Flex, and
1181 some modified versions of 2.5.35, which have modified function prototypes.
1183 *** Warnings during the build procedure have been eliminated.
1185 *** Several portability problems in the test suite have been fixed:
1187 This includes warnings with some compilers, unexpected behavior of tools
1188 such as diff, warning messages from the test suite itself, etc.
1190 *** The install-pdf target works properly:
1192 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
1193 halts in the middle of its course.
1195 * Changes in version 2.5 (2011-05-14):
1197 ** Grammar symbol names can now contain non-initial dashes:
1199 Consistently with directives (such as %error-verbose) and with
1200 %define variables (e.g. push-pull), grammar symbol names may contain
1201 dashes in any position except the beginning. This is a GNU
1202 extension over POSIX Yacc. Thus, use of this extension is reported
1203 by -Wyacc and rejected in Yacc mode (--yacc).
1205 ** Named references:
1207 Historically, Yacc and Bison have supported positional references
1208 ($n, $$) to allow access to symbol values from inside of semantic
1211 Starting from this version, Bison can also accept named references.
1212 When no ambiguity is possible, original symbol names may be used
1213 as named references:
1215 if_stmt : "if" cond_expr "then" then_stmt ';'
1216 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
1218 In the more common case, explicit names may be declared:
1220 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
1221 { $res = mk_if_stmt($cond, $then, $else); }
1223 Location information is also accessible using @name syntax. When
1224 accessing symbol names containing dots or dashes, explicit bracketing
1225 ($[sym.1]) must be used.
1227 These features are experimental in this version. More user feedback
1228 will help to stabilize them.
1229 Contributed by Alex Rozenman.
1231 ** IELR(1) and canonical LR(1):
1233 IELR(1) is a minimal LR(1) parser table generation algorithm. That
1234 is, given any context-free grammar, IELR(1) generates parser tables
1235 with the full language-recognition power of canonical LR(1) but with
1236 nearly the same number of parser states as LALR(1). This reduction
1237 in parser states is often an order of magnitude. More importantly,
1238 because canonical LR(1)'s extra parser states may contain duplicate
1239 conflicts in the case of non-LR(1) grammars, the number of conflicts
1240 for IELR(1) is often an order of magnitude less as well. This can
1241 significantly reduce the complexity of developing of a grammar.
1243 Bison can now generate IELR(1) and canonical LR(1) parser tables in
1244 place of its traditional LALR(1) parser tables, which remain the
1245 default. You can specify the type of parser tables in the grammar
1246 file with these directives:
1248 %define lr.type lalr
1249 %define lr.type ielr
1250 %define lr.type canonical-lr
1252 The default-reduction optimization in the parser tables can also be
1253 adjusted using "%define lr.default-reductions". For details on both
1254 of these features, see the new section "Tuning LR" in the Bison
1257 These features are experimental. More user feedback will help to
1260 ** LAC (Lookahead Correction) for syntax error handling
1262 Contributed by Joel E. Denny.
1264 Canonical LR, IELR, and LALR can suffer from a couple of problems
1265 upon encountering a syntax error. First, the parser might perform
1266 additional parser stack reductions before discovering the syntax
1267 error. Such reductions can perform user semantic actions that are
1268 unexpected because they are based on an invalid token, and they
1269 cause error recovery to begin in a different syntactic context than
1270 the one in which the invalid token was encountered. Second, when
1271 verbose error messages are enabled (with %error-verbose or the
1272 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
1273 syntax error message can both contain invalid tokens and omit valid
1276 The culprits for the above problems are %nonassoc, default
1277 reductions in inconsistent states, and parser state merging. Thus,
1278 IELR and LALR suffer the most. Canonical LR can suffer only if
1279 %nonassoc is used or if default reductions are enabled for
1280 inconsistent states.
1282 LAC is a new mechanism within the parsing algorithm that solves
1283 these problems for canonical LR, IELR, and LALR without sacrificing
1284 %nonassoc, default reductions, or state merging. When LAC is in
1285 use, canonical LR and IELR behave almost exactly the same for both
1286 syntactically acceptable and syntactically unacceptable input.
1287 While LALR still does not support the full language-recognition
1288 power of canonical LR and IELR, LAC at least enables LALR's syntax
1289 error handling to correctly reflect LALR's language-recognition
1292 Currently, LAC is only supported for deterministic parsers in C.
1293 You can enable LAC with the following directive:
1295 %define parse.lac full
1297 See the new section "LAC" in the Bison manual for additional
1298 details including a few caveats.
1300 LAC is an experimental feature. More user feedback will help to
1303 ** %define improvements:
1305 *** Can now be invoked via the command line:
1307 Each of these command-line options
1310 --define=NAME[=VALUE]
1313 --force-define=NAME[=VALUE]
1315 is equivalent to this grammar file declaration
1317 %define NAME ["VALUE"]
1319 except that the manner in which Bison processes multiple definitions
1320 for the same NAME differs. Most importantly, -F and --force-define
1321 quietly override %define, but -D and --define do not. For further
1322 details, see the section "Bison Options" in the Bison manual.
1324 *** Variables renamed:
1326 The following %define variables
1329 lr.keep_unreachable_states
1331 have been renamed to
1334 lr.keep-unreachable-states
1336 The old names are now deprecated but will be maintained indefinitely
1337 for backward compatibility.
1339 *** Values no longer need to be quoted in the grammar file:
1341 If a %define value is an identifier, it no longer needs to be placed
1342 within quotations marks. For example,
1344 %define api.push-pull "push"
1348 %define api.push-pull push
1350 *** Unrecognized variables are now errors not warnings.
1352 *** Multiple invocations for any variable is now an error not a warning.
1354 ** Unrecognized %code qualifiers are now errors not warnings.
1356 ** Character literals not of length one:
1358 Previously, Bison quietly converted all character literals to length
1359 one. For example, without warning, Bison interpreted the operators in
1360 the following grammar to be the same token:
1366 Bison now warns when a character literal is not of length one. In
1367 some future release, Bison will start reporting an error instead.
1369 ** Destructor calls fixed for lookaheads altered in semantic actions:
1371 Previously for deterministic parsers in C, if a user semantic action
1372 altered yychar, the parser in some cases used the old yychar value to
1373 determine which destructor to call for the lookahead upon a syntax
1374 error or upon parser return. This bug has been fixed.
1376 ** C++ parsers use YYRHSLOC:
1378 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
1379 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
1380 to use it. If, for instance, your location structure has "first"
1381 and "last" members, instead of
1383 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1387 (Current).first = (Rhs)[1].location.first; \
1388 (Current).last = (Rhs)[N].location.last; \
1392 (Current).first = (Current).last = (Rhs)[0].location.last; \
1398 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1402 (Current).first = YYRHSLOC (Rhs, 1).first; \
1403 (Current).last = YYRHSLOC (Rhs, N).last; \
1407 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
1411 ** YYLLOC_DEFAULT in C++:
1413 The default implementation of YYLLOC_DEFAULT used to be issued in
1414 the header file. It is now output in the implementation file, after
1415 the user %code sections so that its #ifndef guard does not try to
1416 override the user's YYLLOC_DEFAULT if provided.
1418 ** YYFAIL now produces warnings and Java parsers no longer implement it:
1420 YYFAIL has existed for many years as an undocumented feature of
1421 deterministic parsers in C generated by Bison. More recently, it was
1422 a documented feature of Bison's experimental Java parsers. As
1423 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
1424 semantic action now produces a deprecation warning, and Java parsers
1425 no longer implement YYFAIL at all. For further details, including a
1426 discussion of how to suppress C preprocessor warnings about YYFAIL
1427 being unused, see the Bison 2.4.2 NEWS entry.
1429 ** Temporary hack for adding a semicolon to the user action:
1431 Previously, Bison appended a semicolon to every user action for
1432 reductions when the output language defaulted to C (specifically, when
1433 neither %yacc, %language, %skeleton, or equivalent command-line
1434 options were specified). This allowed actions such as
1436 exp: exp "+" exp { $$ = $1 + $3 };
1440 exp: exp "+" exp { $$ = $1 + $3; };
1442 As a first step in removing this misfeature, Bison now issues a
1443 warning when it appends a semicolon. Moreover, in cases where Bison
1444 cannot easily determine whether a semicolon is needed (for example, an
1445 action ending with a cpp directive or a braced compound initializer),
1446 it no longer appends one. Thus, the C compiler might now complain
1447 about a missing semicolon where it did not before. Future releases of
1448 Bison will cease to append semicolons entirely.
1450 ** Verbose syntax error message fixes:
1452 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
1453 specified, syntax error messages produced by the generated parser
1454 include the unexpected token as well as a list of expected tokens.
1455 The effect of %nonassoc on these verbose messages has been corrected
1456 in two ways, but a more complete fix requires LAC, described above:
1458 *** When %nonassoc is used, there can exist parser states that accept no
1459 tokens, and so the parser does not always require a lookahead token
1460 in order to detect a syntax error. Because no unexpected token or
1461 expected tokens can then be reported, the verbose syntax error
1462 message described above is suppressed, and the parser instead
1463 reports the simpler message, "syntax error". Previously, this
1464 suppression was sometimes erroneously triggered by %nonassoc when a
1465 lookahead was actually required. Now verbose messages are
1466 suppressed only when all previous lookaheads have already been
1467 shifted or discarded.
1469 *** Previously, the list of expected tokens erroneously included tokens
1470 that would actually induce a syntax error because conflicts for them
1471 were resolved with %nonassoc in the current parser state. Such
1472 tokens are now properly omitted from the list.
1474 *** Expected token lists are still often wrong due to state merging
1475 (from LALR or IELR) and default reductions, which can both add
1476 invalid tokens and subtract valid tokens. Canonical LR almost
1477 completely fixes this problem by eliminating state merging and
1478 default reductions. However, there is one minor problem left even
1479 when using canonical LR and even after the fixes above. That is,
1480 if the resolution of a conflict with %nonassoc appears in a later
1481 parser state than the one at which some syntax error is
1482 discovered, the conflicted token is still erroneously included in
1483 the expected token list. Bison's new LAC implementation,
1484 described above, eliminates this problem and the need for
1485 canonical LR. However, LAC is still experimental and is disabled
1488 ** Java skeleton fixes:
1490 *** A location handling bug has been fixed.
1492 *** The top element of each of the value stack and location stack is now
1493 cleared when popped so that it can be garbage collected.
1495 *** Parser traces now print the top element of the stack.
1497 ** -W/--warnings fixes:
1499 *** Bison now properly recognizes the "no-" versions of categories:
1501 For example, given the following command line, Bison now enables all
1502 warnings except warnings for incompatibilities with POSIX Yacc:
1504 bison -Wall,no-yacc gram.y
1506 *** Bison now treats S/R and R/R conflicts like other warnings:
1508 Previously, conflict reports were independent of Bison's normal
1509 warning system. Now, Bison recognizes the warning categories
1510 "conflicts-sr" and "conflicts-rr". This change has important
1511 consequences for the -W and --warnings command-line options. For
1514 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
1515 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
1516 bison -Wnone gram.y # no conflicts are reported
1517 bison -Werror gram.y # any conflict is an error
1519 However, as before, if the %expect or %expect-rr directive is
1520 specified, an unexpected number of conflicts is an error, and an
1521 expected number of conflicts is not reported, so -W and --warning
1522 then have no effect on the conflict report.
1524 *** The "none" category no longer disables a preceding "error":
1526 For example, for the following command line, Bison now reports
1527 errors instead of warnings for incompatibilities with POSIX Yacc:
1529 bison -Werror,none,yacc gram.y
1531 *** The "none" category now disables all Bison warnings:
1533 Previously, the "none" category disabled only Bison warnings for
1534 which there existed a specific -W/--warning category. However,
1535 given the following command line, Bison is now guaranteed to
1536 suppress all warnings:
1540 ** Precedence directives can now assign token number 0:
1542 Since Bison 2.3b, which restored the ability of precedence
1543 directives to assign token numbers, doing so for token number 0 has
1544 produced an assertion failure. For example:
1548 This bug has been fixed.
1550 * Changes in version 2.4.3 (2010-08-05):
1552 ** Bison now obeys -Werror and --warnings=error for warnings about
1553 grammar rules that are useless in the parser due to conflicts.
1555 ** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
1558 ** Failures in the test suite for GCC 4.5 have been fixed.
1560 ** Failures in the test suite for some versions of Sun Studio C++ have
1563 ** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
1564 warnings about undefined %prec identifiers will not be converted to
1565 errors in Bison 2.5. They will remain warnings, which should be
1566 sufficient for POSIX while avoiding backward compatibility issues.
1568 ** Minor documentation fixes.
1570 * Changes in version 2.4.2 (2010-03-20):
1572 ** Some portability problems that resulted in failures and livelocks
1573 in the test suite on some versions of at least Solaris, AIX, HP-UX,
1574 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
1575 errors should no longer cause M4 to report a broken pipe on the
1578 ** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
1580 POSIX specifies that an error be reported for any identifier that does
1581 not appear on the LHS of a grammar rule and that is not defined by
1582 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
1583 error report for the case when an identifier appears only after a
1584 %prec directive. It is now restored. However, for backward
1585 compatibility with recent Bison releases, it is only a warning for
1586 now. In Bison 2.5 and later, it will return to being an error.
1587 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
1588 warning will not be converted to an error in Bison 2.5.]
1590 ** Detection of GNU M4 1.4.6 or newer during configure is improved.
1592 ** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
1593 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
1596 ** %code is now a permanent feature.
1598 A traditional Yacc prologue directive is written in the form:
1602 To provide a more flexible alternative, Bison 2.3b introduced the
1603 %code directive with the following forms for C/C++:
1606 %code requires {CODE}
1607 %code provides {CODE}
1610 These forms are now considered permanent features of Bison. See the
1611 %code entries in the section "Bison Declaration Summary" in the Bison
1612 manual for a summary of their functionality. See the section
1613 "Prologue Alternatives" for a detailed discussion including the
1614 advantages of %code over the traditional Yacc prologue directive.
1616 Bison's Java feature as a whole including its current usage of %code
1617 is still considered experimental.
1619 ** YYFAIL is deprecated and will eventually be removed.
1621 YYFAIL has existed for many years as an undocumented feature of
1622 deterministic parsers in C generated by Bison. Previously, it was
1623 documented for Bison's experimental Java parsers. YYFAIL is no longer
1624 documented for Java parsers and is formally deprecated in both cases.
1625 Users are strongly encouraged to migrate to YYERROR, which is
1628 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
1629 induce a syntax error. The most obvious difference from YYERROR is
1630 that YYFAIL will automatically invoke yyerror to report the syntax
1631 error so that you don't have to. However, there are several other
1632 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
1633 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1634 used. For a more detailed discussion, see:
1636 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
1638 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
1639 deterministic parsers in C will continue to implement it. However,
1640 because YYFAIL is already flawed, it seems futile to try to make new
1641 Bison features compatible with it. Thus, during parser generation,
1642 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
1643 rule action. In a later release, YYFAIL will be disabled for
1644 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1645 be removed altogether.
1647 There exists at least one case where Bison 2.5's YYFAIL warning will
1648 be a false positive. Some projects add phony uses of YYFAIL and other
1649 Bison-defined macros for the sole purpose of suppressing C
1650 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
1651 To avoid Bison's future warning, such YYFAIL uses can be moved to the
1652 epilogue (that is, after the second "%%") in the Bison input file. In
1653 this release (2.4.2), Bison already generates its own code to suppress
1654 C preprocessor warnings for YYFAIL, so projects can remove their own
1655 phony uses of YYFAIL if compatibility with Bison releases prior to
1656 2.4.2 is not necessary.
1658 ** Internationalization.
1660 Fix a regression introduced in Bison 2.4: Under some circumstances,
1661 message translations were not installed although supported by the
1664 * Changes in version 2.4.1 (2008-12-11):
1666 ** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
1667 declarations have been fixed.
1669 ** Temporary hack for adding a semicolon to the user action.
1671 Bison used to prepend a trailing semicolon at the end of the user
1672 action for reductions. This allowed actions such as
1674 exp: exp "+" exp { $$ = $1 + $3 };
1678 exp: exp "+" exp { $$ = $1 + $3; };
1680 Some grammars still depend on this "feature". Bison 2.4.1 restores
1681 the previous behavior in the case of C output (specifically, when
1682 neither %language or %skeleton or equivalent command-line options
1683 are used) to leave more time for grammars depending on the old
1684 behavior to be adjusted. Future releases of Bison will disable this
1687 ** A few minor improvements to the Bison manual.
1689 * Changes in version 2.4 (2008-11-02):
1691 ** %language is an experimental feature.
1693 We first introduced this feature in test release 2.3b as a cleaner
1694 alternative to %skeleton. Since then, we have discussed the possibility of
1695 modifying its effect on Bison's output file names. Thus, in this release,
1696 we consider %language to be an experimental feature that will likely evolve
1699 ** Forward compatibility with GNU M4 has been improved.
1701 ** Several bugs in the C++ skeleton and the experimental Java skeleton have been
1704 * Changes in version 2.3b (2008-05-27):
1706 ** The quotes around NAME that used to be required in the following directive
1709 %define NAME "VALUE"
1711 ** The directive "%pure-parser" is now deprecated in favor of:
1715 which has the same effect except that Bison is more careful to warn about
1716 unreasonable usage in the latter case.
1720 Bison can now generate an LALR(1) parser in C with a push interface. That
1721 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
1722 push one token at a time to the parser using "yypush_parse", which will
1723 return to the caller after processing each token. By default, the push
1724 interface is disabled. Either of the following directives will enable it:
1726 %define api.push_pull "push" // Just push; does not require yylex.
1727 %define api.push_pull "both" // Push and pull; requires yylex.
1729 See the new section "A Push Parser" in the Bison manual for details.
1731 The current push parsing interface is experimental and may evolve. More user
1732 feedback will help to stabilize it.
1734 ** The -g and --graph options now output graphs in Graphviz DOT format,
1735 not VCG format. Like --graph, -g now also takes an optional FILE argument
1736 and thus cannot be bundled with other short options.
1740 Bison can now generate an LALR(1) parser in Java. The skeleton is
1741 "data/lalr1.java". Consider using the new %language directive instead of
1742 %skeleton to select it.
1744 See the new section "Java Parsers" in the Bison manual for details.
1746 The current Java interface is experimental and may evolve. More user
1747 feedback will help to stabilize it.
1748 Contributed by Paolo Bonzini.
1752 This new directive specifies the programming language of the generated
1753 parser, which can be C (the default), C++, or Java. Besides the skeleton
1754 that Bison uses, the directive affects the names of the generated files if
1755 the grammar file's name ends in ".y".
1757 ** XML Automaton Report
1759 Bison can now generate an XML report of the LALR(1) automaton using the new
1760 "--xml" option. The current XML schema is experimental and may evolve. More
1761 user feedback will help to stabilize it.
1762 Contributed by Wojciech Polak.
1764 ** The grammar file may now specify the name of the parser header file using
1765 %defines. For example:
1769 ** When reporting useless rules, useless nonterminals, and unused terminals,
1770 Bison now employs the terms "useless in grammar" instead of "useless",
1771 "useless in parser" instead of "never reduced", and "unused in grammar"
1772 instead of "unused".
1774 ** Unreachable State Removal
1776 Previously, Bison sometimes generated parser tables containing unreachable
1777 states. A state can become unreachable during conflict resolution if Bison
1778 disables a shift action leading to it from a predecessor state. Bison now:
1780 1. Removes unreachable states.
1782 2. Does not report any conflicts that appeared in unreachable states.
1783 WARNING: As a result, you may need to update %expect and %expect-rr
1784 directives in existing grammar files.
1786 3. For any rule used only in such states, Bison now reports the rule as
1787 "useless in parser due to conflicts".
1789 This feature can be disabled with the following directive:
1791 %define lr.keep_unreachable_states
1793 See the %define entry in the "Bison Declaration Summary" in the Bison manual
1794 for further discussion.
1796 ** Lookahead Set Correction in the ".output" Report
1798 When instructed to generate a ".output" file including lookahead sets
1799 (using "--report=lookahead", for example), Bison now prints each reduction's
1800 lookahead set only next to the associated state's one item that (1) is
1801 associated with the same rule as the reduction and (2) has its dot at the end
1802 of its RHS. Previously, Bison also erroneously printed the lookahead set
1803 next to all of the state's other items associated with the same rule. This
1804 bug affected only the ".output" file and not the generated parser source
1807 ** --report-file=FILE is a new option to override the default ".output" file
1810 ** The "=" that used to be required in the following directives is now
1813 %file-prefix "parser"
1817 ** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
1819 Bison 2.3a provided a new set of directives as a more flexible alternative to
1820 the traditional Yacc prologue blocks. Those have now been consolidated into
1821 a single %code directive with an optional qualifier field, which identifies
1822 the purpose of the code and thus the location(s) where Bison should generate
1825 1. "%code {CODE}" replaces "%after-header {CODE}"
1826 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1827 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1828 4. "%code top {CODE}" replaces "%before-header {CODE}"
1830 See the %code entries in section "Bison Declaration Summary" in the Bison
1831 manual for a summary of the new functionality. See the new section "Prologue
1832 Alternatives" for a detailed discussion including the advantages of %code
1833 over the traditional Yacc prologues.
1835 The prologue alternatives are experimental. More user feedback will help to
1836 determine whether they should become permanent features.
1838 ** Revised warning: unset or unused mid-rule values
1840 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1841 used within any of the actions of the parent rule. For example, Bison warns
1844 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1846 Now, Bison also warns about mid-rule values that are used but not set. For
1847 example, Bison warns about unset $$ in the mid-rule action in:
1849 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1851 However, Bison now disables both of these warnings by default since they
1852 sometimes prove to be false alarms in existing grammars employing the Yacc
1853 constructs $0 or $-N (where N is some positive integer).
1855 To enable these warnings, specify the option "--warnings=midrule-values" or
1856 "-W", which is a synonym for "--warnings=all".
1858 ** Default %destructor or %printer with "<*>" or "<>"
1860 Bison now recognizes two separate kinds of default %destructor's and
1863 1. Place "<*>" in a %destructor/%printer symbol list to define a default
1864 %destructor/%printer for all grammar symbols for which you have formally
1865 declared semantic type tags.
1867 2. Place "<>" in a %destructor/%printer symbol list to define a default
1868 %destructor/%printer for all grammar symbols without declared semantic
1871 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1872 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
1873 longer applies any %destructor to a mid-rule value if that mid-rule value is
1874 not actually ever referenced using either $$ or $n in a semantic action.
1876 The default %destructor's and %printer's are experimental. More user
1877 feedback will help to determine whether they should become permanent
1880 See the section "Freeing Discarded Symbols" in the Bison manual for further
1883 ** %left, %right, and %nonassoc can now declare token numbers. This is required
1884 by POSIX. However, see the end of section "Operator Precedence" in the Bison
1885 manual for a caveat concerning the treatment of literal strings.
1887 ** The nonfunctional --no-parser, -n, and %no-parser options have been
1888 completely removed from Bison.
1890 * Changes in version 2.3a, 2006-09-13:
1892 ** Instead of %union, you can define and use your own union type
1893 YYSTYPE if your grammar contains at least one <type> tag.
1894 Your YYSTYPE need not be a macro; it can be a typedef.
1895 This change is for compatibility with other Yacc implementations,
1896 and is required by POSIX.
1898 ** Locations columns and lines start at 1.
1899 In accordance with the GNU Coding Standards and Emacs.
1901 ** You may now declare per-type and default %destructor's and %printer's:
1905 %union { char *string; }
1906 %token <string> STRING1
1907 %token <string> STRING2
1908 %type <string> string1
1909 %type <string> string2
1910 %union { char character; }
1911 %token <character> CHR
1912 %type <character> chr
1913 %destructor { free ($$); } %symbol-default
1914 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1915 %destructor { } <character>
1917 guarantees that, when the parser discards any user-defined symbol that has a
1918 semantic type tag other than "<character>", it passes its semantic value to
1919 "free". However, when the parser discards a "STRING1" or a "string1", it
1920 also prints its line number to "stdout". It performs only the second
1921 "%destructor" in this case, so it invokes "free" only once.
1923 [Although we failed to mention this here in the 2.3a release, the default
1924 %destructor's and %printer's were experimental, and they were rewritten in
1927 ** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1928 "--yacc", or "%yacc"), Bison no longer generates #define statements for
1929 associating token numbers with token names. Removing the #define statements
1930 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1931 requires them. Bison still generates an enum for token names in all cases.
1933 ** Handling of traditional Yacc prologue blocks is now more consistent but
1934 potentially incompatible with previous releases of Bison.
1936 As before, you declare prologue blocks in your grammar file with the
1937 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
1938 prologue blocks that you've declared before the first %union. To generate
1939 the post-prologue, Bison concatenates all prologue blocks that you've
1940 declared after the first %union.
1942 Previous releases of Bison inserted the pre-prologue into both the header
1943 file and the code file in all cases except for LALR(1) parsers in C. In the
1944 latter case, Bison inserted it only into the code file. For parsers in C++,
1945 the point of insertion was before any token definitions (which associate
1946 token numbers with names). For parsers in C, the point of insertion was
1947 after the token definitions.
1949 Now, Bison never inserts the pre-prologue into the header file. In the code
1950 file, it always inserts it before the token definitions.
1952 ** Bison now provides a more flexible alternative to the traditional Yacc
1953 prologue blocks: %before-header, %start-header, %end-header, and
1956 For example, the following declaration order in the grammar file reflects the
1957 order in which Bison will output these code blocks. However, you are free to
1958 declare these code blocks in your grammar file in whatever order is most
1962 /* Bison treats this block like a pre-prologue block: it inserts it into
1963 * the code file before the contents of the header file. It does *not*
1964 * insert it into the header file. This is a good place to put
1965 * #include's that you want at the top of your code file. A common
1966 * example is '#include "system.h"'. */
1969 /* Bison inserts this block into both the header file and the code file.
1970 * In both files, the point of insertion is before any Bison-generated
1971 * token, semantic type, location type, and class definitions. This is a
1972 * good place to define %union dependencies, for example. */
1975 /* Unlike the traditional Yacc prologue blocks, the output order for the
1976 * new %*-header blocks is not affected by their declaration position
1977 * relative to any %union in the grammar file. */
1980 /* Bison inserts this block into both the header file and the code file.
1981 * In both files, the point of insertion is after the Bison-generated
1982 * definitions. This is a good place to declare or define public
1983 * functions or data structures that depend on the Bison-generated
1987 /* Bison treats this block like a post-prologue block: it inserts it into
1988 * the code file after the contents of the header file. It does *not*
1989 * insert it into the header file. This is a good place to declare or
1990 * define internal functions or data structures that depend on the
1991 * Bison-generated definitions. */
1994 If you have multiple occurrences of any one of the above declarations, Bison
1995 will concatenate the contents in declaration order.
1997 [Although we failed to mention this here in the 2.3a release, the prologue
1998 alternatives were experimental, and they were rewritten in future versions.]
2000 ** The option "--report=look-ahead" has been changed to "--report=lookahead".
2001 The old spelling still works, but is not documented and may be removed
2002 in a future release.
2004 * Changes in version 2.3, 2006-06-05:
2006 ** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
2007 for compatibility with LALR(1) grammars.
2009 ** It is now documented that any definition of YYSTYPE or YYLTYPE should
2010 be to a type name that does not contain parentheses or brackets.
2012 * Changes in version 2.2, 2006-05-19:
2014 ** The distribution terms for all Bison-generated parsers now permit
2015 using the parsers in nonfree programs. Previously, this permission
2016 was granted only for Bison-generated LALR(1) parsers in C.
2018 ** %name-prefix changes the namespace name in C++ outputs.
2020 ** The C++ parsers export their token_type.
2022 ** Bison now allows multiple %union declarations, and concatenates
2023 their contents together.
2025 ** New warning: unused values
2026 Right-hand side symbols whose values are not used are reported,
2027 if the symbols have destructors. For instance:
2029 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
2033 will trigger a warning about $$ and $5 in the first rule, and $3 in
2034 the second ($1 is copied to $$ by the default rule). This example
2035 most likely contains three errors, and could be rewritten as:
2037 exp: exp "?" exp ":" exp
2038 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
2040 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
2043 However, if the original actions were really intended, memory leaks
2044 and all, the warnings can be suppressed by letting Bison believe the
2045 values are used, e.g.:
2047 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
2048 | exp "+" exp { $$ = $1; (void) $3; }
2051 If there are mid-rule actions, the warning is issued if no action
2052 uses it. The following triggers no warning: $1 and $3 are used.
2054 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
2056 The warning is intended to help catching lost values and memory leaks.
2057 If a value is ignored, its associated memory typically is not reclaimed.
2059 ** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
2060 Destructors are now called when user code invokes YYABORT, YYACCEPT,
2061 and YYERROR, for all objects on the stack, other than objects
2062 corresponding to the right-hand side of the current rule.
2064 ** %expect, %expect-rr
2065 Incorrect numbers of expected conflicts are now actual errors,
2066 instead of warnings.
2068 ** GLR, YACC parsers.
2069 The %parse-params are available in the destructors (and the
2070 experimental printers) as per the documentation.
2072 ** Bison now warns if it finds a stray "$" or "@" in an action.
2074 ** %require "VERSION"
2075 This specifies that the grammar file depends on features implemented
2076 in Bison version VERSION or higher.
2078 ** lalr1.cc: The token and value types are now class members.
2079 The tokens were defined as free form enums and cpp macros. YYSTYPE
2080 was defined as a free form union. They are now class members:
2081 tokens are enumerations of the "yy::parser::token" struct, and the
2082 semantic values have the "yy::parser::semantic_type" type.
2084 If you do not want or can update to this scheme, the directive
2085 '%define "global_tokens_and_yystype" "1"' triggers the global
2086 definition of tokens and YYSTYPE. This change is suitable both
2087 for previous releases of Bison, and this one.
2089 If you wish to update, then make sure older version of Bison will
2090 fail using '%require "2.2"'.
2092 ** DJGPP support added.
2094 * Changes in version 2.1, 2005-09-16:
2096 ** The C++ lalr1.cc skeleton supports %lex-param.
2098 ** Bison-generated parsers now support the translation of diagnostics like
2099 "syntax error" into languages other than English. The default
2100 language is still English. For details, please see the new
2101 Internationalization section of the Bison manual. Software
2102 distributors should also see the new PACKAGING file. Thanks to
2103 Bruno Haible for this new feature.
2105 ** Wording in the Bison-generated parsers has been changed slightly to
2106 simplify translation. In particular, the message "memory exhausted"
2107 has replaced "parser stack overflow", as the old message was not
2108 always accurate for modern Bison-generated parsers.
2110 ** Destructors are now called when the parser aborts, for all symbols left
2111 behind on the stack. Also, the start symbol is now destroyed after a
2112 successful parse. In both cases, the behavior was formerly inconsistent.
2114 ** When generating verbose diagnostics, Bison-generated parsers no longer
2115 quote the literal strings associated with tokens. For example, for
2116 a syntax error associated with '%token NUM "number"' they might
2117 print 'syntax error, unexpected number' instead of 'syntax error,
2118 unexpected "number"'.
2120 * Changes in version 2.0, 2004-12-25:
2122 ** Possibly-incompatible changes
2124 - Bison-generated parsers no longer default to using the alloca function
2125 (when available) to extend the parser stack, due to widespread
2126 problems in unchecked stack-overflow detection. You can "#define
2127 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
2128 the manual to determine safe values for YYMAXDEPTH in that case.
2130 - Error token location.
2131 During error recovery, the location of the syntax error is updated
2132 to cover the whole sequence covered by the error token: it includes
2133 the shifted symbols thrown away during the first part of the error
2134 recovery, and the lookahead rejected during the second part.
2136 - Semicolon changes:
2137 . Stray semicolons are no longer allowed at the start of a grammar.
2138 . Semicolons are now required after in-grammar declarations.
2140 - Unescaped newlines are no longer allowed in character constants or
2141 string literals. They were never portable, and GCC 3.4.0 has
2142 dropped support for them. Better diagnostics are now generated if
2143 forget a closing quote.
2145 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
2149 - GLR grammars now support locations.
2151 - New directive: %initial-action.
2152 This directive allows the user to run arbitrary code (including
2153 initializing @$) from yyparse before parsing starts.
2155 - A new directive "%expect-rr N" specifies the expected number of
2156 reduce/reduce conflicts in GLR parsers.
2158 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
2159 This is a GNU extension.
2161 - The option "--report=lookahead" was changed to "--report=look-ahead".
2162 [However, this was changed back after 2.3.]
2164 - Experimental %destructor support has been added to lalr1.cc.
2166 - New configure option --disable-yacc, to disable installation of the
2167 yacc command and -ly library introduced in 1.875 for POSIX conformance.
2171 - For now, %expect-count violations are now just warnings, not errors.
2172 This is for compatibility with Bison 1.75 and earlier (when there are
2173 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
2174 are too many or too few shift/reduce conflicts). However, in future
2175 versions of Bison we plan to improve the %expect machinery so that
2176 these violations will become errors again.
2178 - Within Bison itself, numbers (e.g., goto numbers) are no longer
2179 arbitrarily limited to 16-bit counts.
2181 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
2183 * Changes in version 1.875, 2003-01-01:
2185 ** The documentation license has been upgraded to version 1.2
2186 of the GNU Free Documentation License.
2188 ** syntax error processing
2190 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
2191 locations too. This fixes bugs in error-location computation.
2194 It is now possible to reclaim the memory associated to symbols
2195 discarded during error recovery. This feature is still experimental.
2198 This new directive is preferred over YYERROR_VERBOSE.
2200 - #defining yyerror to steal internal variables is discouraged.
2201 It is not guaranteed to work forever.
2203 ** POSIX conformance
2205 - Semicolons are once again optional at the end of grammar rules.
2206 This reverts to the behavior of Bison 1.33 and earlier, and improves
2207 compatibility with Yacc.
2209 - "parse error" -> "syntax error"
2210 Bison now uniformly uses the term "syntax error"; formerly, the code
2211 and manual sometimes used the term "parse error" instead. POSIX
2212 requires "syntax error" in diagnostics, and it was thought better to
2215 - The documentation now emphasizes that yylex and yyerror must be
2216 declared before use. C99 requires this.
2218 - Bison now parses C99 lexical constructs like UCNs and
2219 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
2221 - File names are properly escaped in C output. E.g., foo\bar.y is
2222 output as "foo\\bar.y".
2224 - Yacc command and library now available
2225 The Bison distribution now installs a "yacc" command, as POSIX requires.
2226 Also, Bison now installs a small library liby.a containing
2227 implementations of Yacc-compatible yyerror and main functions.
2228 This library is normally not useful, but POSIX requires it.
2230 - Type clashes now generate warnings, not errors.
2232 - If the user does not define YYSTYPE as a macro, Bison now declares it
2233 using typedef instead of defining it as a macro.
2234 For consistency, YYLTYPE is also declared instead of defined.
2236 ** Other compatibility issues
2238 - %union directives can now have a tag before the "{", e.g., the
2239 directive "%union foo {...}" now generates the C code
2240 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
2241 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
2242 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
2243 This is for compatibility with both Yacc and Bison 1.35.
2245 - ";" is output before the terminating "}" of an action, for
2246 compatibility with Bison 1.35.
2248 - Bison now uses a Yacc-style format for conflict reports, e.g.,
2249 "conflicts: 2 shift/reduce, 1 reduce/reduce".
2251 - "yystype" and "yyltype" are now obsolescent macros instead of being
2252 typedefs or tags; they are no longer documented and are planned to be
2253 withdrawn in a future release.
2258 Users of Bison have to decide how they handle the portability of the
2261 - "parsing stack overflow..." -> "parser stack overflow"
2262 GLR parsers now report "parser stack overflow" as per the Bison manual.
2264 ** %parse-param and %lex-param
2265 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
2266 additional context to yyparse and yylex. They suffer from several
2269 - a single argument only can be added,
2270 - their types are weak (void *),
2271 - this context is not passed to ancillary functions such as yyerror,
2272 - only yacc.c parsers support them.
2274 The new %parse-param/%lex-param directives provide a more precise control.
2277 %parse-param {int *nastiness}
2278 %lex-param {int *nastiness}
2279 %parse-param {int *randomness}
2281 results in the following signatures:
2283 int yylex (int *nastiness);
2284 int yyparse (int *nastiness, int *randomness);
2286 or, if both %pure-parser and %locations are used:
2288 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
2289 int yyparse (int *nastiness, int *randomness);
2291 ** Bison now warns if it detects conflicting outputs to the same file,
2292 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
2293 that command outputs both code and header to foo.h.
2295 ** #line in output files
2296 - --no-line works properly.
2298 ** Bison can no longer be built by a K&R C compiler; it requires C89 or
2299 later to be built. This change originally took place a few versions
2300 ago, but nobody noticed until we recently asked someone to try
2301 building Bison with a K&R C compiler.
2303 * Changes in version 1.75, 2002-10-14:
2305 ** Bison should now work on 64-bit hosts.
2307 ** Indonesian translation thanks to Tedi Heriyanto.
2310 Fix spurious parse errors.
2313 Some people redefine yyerror to steal yyparse' private variables.
2314 Reenable this trick until an official feature replaces it.
2317 In agreement with POSIX and with other Yaccs, leaving a default
2318 action is valid when $$ is untyped, and $1 typed:
2322 but the converse remains an error:
2326 ** Values of mid-rule actions
2329 foo: { ... } { $$ = $1; } ...
2331 was incorrectly rejected: $1 is defined in the second mid-rule
2332 action, and is equal to the $$ of the first mid-rule action.
2334 * Changes in version 1.50, 2002-10-04:
2339 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
2340 almost any context-free grammar, ambiguous or not. The new declarations
2341 %dprec and %merge on grammar rules allow parse-time resolution of
2342 ambiguities. Contributed by Paul Hilfinger.
2344 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
2345 like the Alpha, so please stick to 32-bit hosts for now.
2348 When not in Yacc compatibility mode, when the output file was not
2349 specified, running "bison foo/bar.y" created "foo/bar.c". It
2350 now creates "bar.c".
2353 The undefined token was systematically mapped to 2 which prevented
2354 the use of 2 by the user. This is no longer the case.
2356 ** Unknown token numbers
2357 If yylex returned an out of range value, yyparse could die. This is
2361 According to POSIX, the error token must be 256.
2362 Bison extends this requirement by making it a preference: *if* the
2363 user specified that one of her tokens is numbered 256, then error
2364 will be mapped onto another number.
2366 ** Verbose error messages
2367 They no longer report "..., expecting error or..." for states where
2368 error recovery is possible.
2371 Defaults to "$end" instead of "$".
2373 ** Error recovery now conforms to documentation and to POSIX
2374 When a Bison-generated parser encounters a syntax error, it now pops
2375 the stack until it finds a state that allows shifting the error
2376 token. Formerly, it popped the stack until it found a state that
2377 allowed some non-error action other than a default reduction on the
2378 error token. The new behavior has long been the documented behavior,
2379 and has long been required by POSIX. For more details, please see
2380 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
2381 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
2384 Popped tokens and nonterminals are now reported.
2387 Larger grammars are now supported (larger token numbers, larger grammar
2388 size (= sum of the LHS and RHS lengths), larger LALR tables).
2389 Formerly, many of these numbers ran afoul of 16-bit limits;
2390 now these limits are 32 bits on most hosts.
2392 ** Explicit initial rule
2393 Bison used to play hacks with the initial rule, which the user does
2394 not write. It is now explicit, and visible in the reports and
2398 Before, Bison reported the useless rules, but, although not used,
2399 included them in the parsers. They are now actually removed.
2401 ** Useless rules, useless nonterminals
2402 They are now reported, as a warning, with their locations.
2404 ** Rules never reduced
2405 Rules that can never be reduced because of conflicts are now
2408 ** Incorrect "Token not used"
2409 On a grammar such as
2411 %token useless useful
2413 exp: '0' %prec useful;
2415 where a token was used to set the precedence of the last rule,
2416 bison reported both "useful" and "useless" as useless tokens.
2418 ** Revert the C++ namespace changes introduced in 1.31
2419 as they caused too many portability hassles.
2421 ** Default locations
2422 By an accident of design, the default computation of @$ was
2423 performed after another default computation was performed: @$ = @1.
2424 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
2425 the computation of @$.
2427 ** Token end-of-file
2428 The token end of file may be specified by the user, in which case,
2429 the user symbol is used in the reports, the graphs, and the verbose
2430 error messages instead of "$end", which remains being the default.
2434 %token MYEOF 0 "end of file"
2437 This old option, which has been broken for ages, is removed.
2440 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
2441 Croatian, thanks to Denis Lackovic.
2443 ** Incorrect token definitions
2446 bison used to output
2449 ** Token definitions as enums
2450 Tokens are output both as the traditional #define's, and, provided
2451 the compiler supports ANSI C or is a C++ compiler, as enums.
2452 This lets debuggers display names instead of integers.
2455 In addition to --verbose, bison supports --report=THINGS, which
2456 produces additional information:
2458 complete the core item sets with their closure
2459 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
2460 explicitly associate lookahead tokens to items
2462 describe shift/reduce conflicts solving.
2463 Bison used to systematically output this information on top of
2464 the report. Solved conflicts are now attached to their states.
2467 Previous versions don't complain when there is a type clash on
2468 the default action if the rule has a mid-rule action, such as in:
2476 ** GNU M4 is now required when using Bison.
2478 * Changes in version 1.35, 2002-03-25:
2481 Some projects use Bison's C parser with C++ compilers, and define
2482 YYSTYPE as a class. The recent adjustment of C parsers for data
2483 alignment and 64 bit architectures made this impossible.
2485 Because for the time being no real solution for C++ parser
2486 generation exists, kludges were implemented in the parser to
2487 maintain this use. In the future, when Bison has C++ parsers, this
2488 kludge will be disabled.
2490 This kludge also addresses some C++ problems when the stack was
2493 * Changes in version 1.34, 2002-03-12:
2495 ** File name clashes are detected
2496 $ bison foo.y -d -o foo.x
2497 fatal error: header and parser would both be named "foo.x"
2499 ** A missing ";" at the end of a rule triggers a warning
2500 In accordance with POSIX, and in agreement with other
2501 Yacc implementations, Bison will mandate this semicolon in the near
2502 future. This eases the implementation of a Bison parser of Bison
2503 grammars by making this grammar LALR(1) instead of LR(2). To
2504 facilitate the transition, this release introduces a warning.
2506 ** Revert the C++ namespace changes introduced in 1.31, as they caused too
2507 many portability hassles.
2509 ** DJGPP support added.
2511 ** Fix test suite portability problems.
2513 * Changes in version 1.33, 2002-02-07:
2516 Groff could not be compiled for the definition of size_t was lacking
2517 under some conditions.
2522 * Changes in version 1.32, 2002-01-23:
2524 ** Fix Yacc output file names
2526 ** Portability fixes
2528 ** Italian, Dutch translations
2530 * Changes in version 1.31, 2002-01-14:
2534 ** GNU Gettext and %expect
2535 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
2536 Bison dies on incorrect %expectations, we fear there will be
2537 too many bug reports for Gettext, so _for the time being_, %expect
2538 does not trigger an error when the input file is named "plural.y".
2540 ** Use of alloca in parsers
2541 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
2542 malloc exclusively. Since 1.29, but was not NEWS'ed.
2544 alloca is used only when compiled with GCC, to avoid portability
2547 ** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
2549 ** When the generated parser lacks debugging code, YYDEBUG is now 0
2550 (as POSIX requires) instead of being undefined.
2553 Bison has always permitted actions such as { $$ = $1 }: it adds the
2554 ending semicolon. Now if in Yacc compatibility mode, the semicolon
2555 is no longer output: one has to write { $$ = $1; }.
2557 ** Better C++ compliance
2558 The output parsers try to respect C++ namespaces.
2559 [This turned out to be a failed experiment, and it was reverted later.]
2562 Fixed bugs when reporting useless nonterminals.
2565 The parsers work properly on 64 bit hosts.
2568 Some calls to strerror resulted in scrambled or missing error messages.
2571 When the number of shift/reduce conflicts is correct, don't issue
2574 ** The verbose report includes the rule line numbers.
2576 ** Rule line numbers are fixed in traces.
2578 ** Swedish translation
2581 Verbose parse error messages from the parsers are better looking.
2582 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
2583 Now: parse error: unexpected '/', expecting "number" or '-' or '('
2585 ** Fixed parser memory leaks.
2586 When the generated parser was using malloc to extend its stacks, the
2587 previous allocations were not freed.
2589 ** Fixed verbose output file.
2590 Some newlines were missing.
2591 Some conflicts in state descriptions were missing.
2593 ** Fixed conflict report.
2594 Option -v was needed to get the result.
2598 Mismatches are errors, not warnings.
2600 ** Fixed incorrect processing of some invalid input.
2602 ** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
2604 ** Fixed some typos in the documentation.
2606 ** %token MY_EOF 0 is supported.
2607 Before, MY_EOF was silently renumbered as 257.
2609 ** doc/refcard.tex is updated.
2611 ** %output, %file-prefix, %name-prefix.
2615 New, aliasing "--output-file".
2617 * Changes in version 1.30, 2001-10-26:
2619 ** "--defines" and "--graph" have now an optional argument which is the
2620 output file name. "-d" and "-g" do not change; they do not take any
2623 ** "%source_extension" and "%header_extension" are removed, failed
2626 ** Portability fixes.
2628 * Changes in version 1.29, 2001-09-07:
2630 ** The output file does not define const, as this caused problems when used
2631 with common autoconfiguration schemes. If you still use ancient compilers
2632 that lack const, compile with the equivalent of the C compiler option
2633 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
2635 ** Added "-g" and "--graph".
2637 ** The Bison manual is now distributed under the terms of the GNU FDL.
2639 ** The input and the output files has automatically a similar extension.
2641 ** Russian translation added.
2643 ** NLS support updated; should hopefully be less troublesome.
2645 ** Added the old Bison reference card.
2647 ** Added "--locations" and "%locations".
2649 ** Added "-S" and "--skeleton".
2651 ** "%raw", "-r", "--raw" is disabled.
2653 ** Special characters are escaped when output. This solves the problems
2654 of the #line lines with path names including backslashes.
2657 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
2658 "%debug", "%source_extension" and "%header_extension".
2661 Automatic location tracking.
2663 * Changes in version 1.28, 1999-07-06:
2665 ** Should compile better now with K&R compilers.
2669 ** Fixed a problem with escaping the double quote character.
2671 ** There is now a FAQ.
2673 * Changes in version 1.27:
2675 ** The make rule which prevented bison.simple from being created on
2676 some systems has been fixed.
2678 * Changes in version 1.26:
2680 ** Bison now uses Automake.
2682 ** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
2684 ** Token numbers now start at 257 as previously documented, not 258.
2686 ** Bison honors the TMPDIR environment variable.
2688 ** A couple of buffer overruns have been fixed.
2690 ** Problems when closing files should now be reported.
2692 ** Generated parsers should now work even on operating systems which do
2693 not provide alloca().
2695 * Changes in version 1.25, 1995-10-16:
2697 ** Errors in the input grammar are not fatal; Bison keeps reading
2698 the grammar file, and reports all the errors found in it.
2700 ** Tokens can now be specified as multiple-character strings: for
2701 example, you could use "<=" for a token which looks like <=, instead
2702 of choosing a name like LESSEQ.
2704 ** The %token_table declaration says to write a table of tokens (names
2705 and numbers) into the parser file. The yylex function can use this
2706 table to recognize multiple-character string tokens, or for other
2709 ** The %no_lines declaration says not to generate any #line preprocessor
2710 directives in the parser file.
2712 ** The %raw declaration says to use internal Bison token numbers, not
2713 Yacc-compatible token numbers, when token names are defined as macros.
2715 ** The --no-parser option produces the parser tables without including
2716 the parser engine; a project can now use its own parser engine.
2717 The actions go into a separate file called NAME.act, in the form of
2718 a switch statement body.
2720 * Changes in version 1.23:
2722 The user can define YYPARSE_PARAM as the name of an argument to be
2723 passed into yyparse. The argument should have type void *. It should
2724 actually point to an object. Grammar actions can access the variable
2725 by casting it to the proper pointer type.
2727 Line numbers in output file corrected.
2729 * Changes in version 1.22:
2731 --help option added.
2733 * Changes in version 1.20:
2735 Output file does not redefine const for C++.
2739 Copyright (C) 1995-2015 Free Software Foundation, Inc.
2741 This file is part of Bison, the GNU Parser Generator.
2743 This program is free software: you can redistribute it and/or modify
2744 it under the terms of the GNU General Public License as published by
2745 the Free Software Foundation, either version 3 of the License, or
2746 (at your option) any later version.
2748 This program is distributed in the hope that it will be useful,
2749 but WITHOUT ANY WARRANTY; without even the implied warranty of
2750 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2751 GNU General Public License for more details.
2753 You should have received a copy of the GNU General Public License
2754 along with this program. If not, see <http://www.gnu.org/licenses/>.
2756 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2757 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2758 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2759 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2760 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2761 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2762 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2763 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2764 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2765 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2766 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2767 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2768 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
2769 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
2770 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
2771 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2772 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
2773 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
2774 LocalWords: TOK calc yyo fval Wconflicts parsers yystackp yyval yynerrs
2775 LocalWords: Théophile Ranquet Santet fno fnone stype associativity Tolmer
2776 LocalWords: Wprecedence Rassoul Wempty Paolo Bonzini parser's Michiel loc
2777 LocalWords: redeclaration sval fcaret reentrant XSLT xsl Wmaybe yyvsp Tedi
2778 LocalWords: pragmas noreturn untyped Rozenman unexpanded Wojciech Polak
2779 LocalWords: Alexandre MERCHANTABILITY yytype