3 * Noteworthy changes in release ?.? (????-??-??) [?]
6 * Noteworthy changes in release 3.0.2 (2013-12-05) [stable]
10 *** Generated source files when errors are reported
12 When warnings are issued and -Werror is set, bison would still generate
13 the source files (*.c, *.h...). As a consequence, some runs of "make"
14 could fail the first time, but not the second (as the files were generated
17 This is fixed: bison no longer generates this source files, but, of
18 course, still produces the various reports (*.output, *.xml, etc.).
20 *** %empty is used in reports
22 Empty right-hand sides are denoted by '%empty' in all the reports (text,
23 dot, XML and formats derived from it).
25 *** YYERROR and variants
27 When C++ variant support is enabled, an error triggered via YYERROR, but
28 not caught via error recovery, resulted in a double deletion.
30 * Noteworthy changes in release 3.0.1 (2013-11-12) [stable]
34 *** Errors in caret diagnostics
36 On some platforms, some errors could result in endless diagnostics.
38 *** Fixes of the -Werror option
40 Options such as "-Werror -Wno-error=foo" were still turning "foo"
41 diagnostics into errors instead of warnings. This is fixed.
43 Actually, for consistency with GCC, "-Wno-error=foo -Werror" now also
44 leaves "foo" diagnostics as warnings. Similarly, with "-Werror=foo
45 -Wno-error", "foo" diagnostics are now errors.
49 As demonstrated in the documentation, one can now leave spaces between
54 The yacc.1 man page is no longer installed if --disable-yacc was
57 *** Fixes in the test suite
59 Bugs and portability issues.
61 * Noteworthy changes in release 3.0 (2013-07-25) [stable]
63 ** WARNING: Future backward-incompatibilities!
65 Like other GNU packages, Bison will start using some of the C99 features
66 for its own code, especially the definition of variables after statements.
67 The generated C parsers still aim at C90.
69 ** Backward incompatible changes
73 Support for YYFAIL is removed (deprecated in Bison 2.4.2): use YYERROR.
75 Support for yystype and yyltype is removed (deprecated in Bison 1.875):
76 use YYSTYPE and YYLTYPE.
78 Support for YYLEX_PARAM and YYPARSE_PARAM is removed (deprecated in Bison
79 1.875): use %lex-param, %parse-param, or %param.
81 Missing semicolons at the end of actions are no longer added (as announced
84 *** Use of YACC='bison -y'
86 TL;DR: With Autoconf <= 2.69, pass -Wno-yacc to (AM_)YFLAGS if you use
89 Traditional Yacc generates 'y.tab.c' whatever the name of the input file.
90 Therefore Makefiles written for Yacc expect 'y.tab.c' (and possibly
91 'y.tab.h' and 'y.outout') to be generated from 'foo.y'.
93 To this end, for ages, AC_PROG_YACC, Autoconf's macro to look for an
94 implementation of Yacc, was using Bison as 'bison -y'. While it does
95 ensure compatible output file names, it also enables warnings for
96 incompatibilities with POSIX Yacc. In other words, 'bison -y' triggers
97 warnings for Bison extensions.
99 Autoconf 2.70+ fixes this incompatibility by using YACC='bison -o y.tab.c'
100 (which also generates 'y.tab.h' and 'y.output' when needed).
101 Alternatively, disable Yacc warnings by passing '-Wno-yacc' to your Yacc
102 flags (YFLAGS, or AM_YFLAGS with Automake).
106 *** The epilogue is no longer affected by internal #defines (glr.c)
108 The glr.c skeleton uses defines such as #define yylval (yystackp->yyval) in
109 generated code. These weren't properly undefined before the inclusion of
110 the user epilogue, so functions such as the following were butchered by the
111 preprocessor expansion:
113 int yylex (YYSTYPE *yylval);
115 This is fixed: yylval, yynerrs, yychar, and yylloc are now valid
116 identifiers for user-provided variables.
118 *** stdio.h is no longer needed when locations are enabled (yacc.c)
120 Changes in Bison 2.7 introduced a dependency on FILE and fprintf when
121 locations are enabled. This is fixed.
123 *** Warnings about useless %pure-parser/%define api.pure are restored
125 ** Diagnostics reported by Bison
127 Most of these features were contributed by Théophile Ranquet and Victor
132 Version 2.7 introduced caret errors, for a prettier output. These are now
133 activated by default. The old format can still be used by invoking Bison
134 with -fno-caret (or -fnone).
136 Some error messages that reproduced excerpts of the grammar are now using
137 the caret information only. For instance on:
144 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
145 in.y:2.12-14: warning: rule useless in parser due to conflicts: exp: 'a' [-Wother]
149 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
150 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
154 and "bison -fno-caret" reports:
156 in.y: warning: 1 reduce/reduce conflict [-Wconflicts-rr]
157 in.y:2.12-14: warning: rule useless in parser due to conflicts [-Wother]
159 *** Enhancements of the -Werror option
161 The -Werror=CATEGORY option is now recognized, and will treat specified
162 warnings as errors. The warnings need not have been explicitly activated
163 using the -W option, this is similar to what GCC 4.7 does.
165 For example, given the following command line, Bison will treat both
166 warnings related to POSIX Yacc incompatibilities and S/R conflicts as
167 errors (and only those):
169 $ bison -Werror=yacc,error=conflicts-sr input.y
171 If no categories are specified, -Werror will make all active warnings into
172 errors. For example, the following line does the same the previous example:
174 $ bison -Werror -Wnone -Wyacc -Wconflicts-sr input.y
176 (By default -Wconflicts-sr,conflicts-rr,deprecated,other is enabled.)
178 Note that the categories in this -Werror option may not be prefixed with
179 "no-". However, -Wno-error[=CATEGORY] is valid.
181 Note that -y enables -Werror=yacc. Therefore it is now possible to require
182 Yacc-like behavior (e.g., always generate y.tab.c), but to report
183 incompatibilities as warnings: "-y -Wno-error=yacc".
185 *** The display of warnings is now richer
187 The option that controls a given warning is now displayed:
189 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
191 In the case of warnings treated as errors, the prefix is changed from
192 "warning: " to "error: ", and the suffix is displayed, in a manner similar
193 to GCC, as [-Werror=CATEGORY].
195 For instance, where the previous version of Bison would report (and exit
198 bison: warnings being treated as errors
199 input.y:1.1: warning: stray ',' treated as white space
203 input.y:1.1: error: stray ',' treated as white space [-Werror=other]
205 *** Deprecated constructs
207 The new 'deprecated' warning category flags obsolete constructs whose
208 support will be discontinued. It is enabled by default. These warnings
209 used to be reported as 'other' warnings.
211 *** Useless semantic types
213 Bison now warns about useless (uninhabited) semantic types. Since
214 semantic types are not declared to Bison (they are defined in the opaque
215 %union structure), it is %printer/%destructor directives about useless
216 types that trigger the warning:
220 %printer {} <type1> <type3>
221 %destructor {} <type2> <type4>
223 nterm: term { $$ = $1; };
225 3.28-34: warning: type <type3> is used, but is not associated to any symbol
226 4.28-34: warning: type <type4> is used, but is not associated to any symbol
228 *** Undefined but unused symbols
230 Bison used to raise an error for undefined symbols that are not used in
231 the grammar. This is now only a warning.
234 %destructor {} symbol2
239 *** Useless destructors or printers
241 Bison now warns about useless destructors or printers. In the following
242 example, the printer for <type1>, and the destructor for <type2> are
243 useless: all symbols of <type1> (token1) already have a printer, and all
244 symbols of type <type2> (token2) already have a destructor.
246 %token <type1> token1
250 %printer {} token1 <type1> <type3>
251 %destructor {} token2 <type2> <type4>
255 The warnings and error messages about shift/reduce and reduce/reduce
256 conflicts have been normalized. For instance on the following foo.y file:
260 exp: exp '+' exp | '0' | '0';
262 compare the previous version of bison:
265 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
266 $ bison -Werror foo.y
267 bison: warnings being treated as errors
268 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
270 with the new behavior:
273 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
274 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
275 $ bison -Werror foo.y
276 foo.y: error: 1 shift/reduce conflict [-Werror=conflicts-sr]
277 foo.y: error: 2 reduce/reduce conflicts [-Werror=conflicts-rr]
279 When %expect or %expect-rr is used, such as with bar.y:
284 exp: exp '+' exp | '0' | '0';
289 bar.y: conflicts: 1 shift/reduce, 2 reduce/reduce
290 bar.y: expected 0 shift/reduce conflicts
291 bar.y: expected 0 reduce/reduce conflicts
296 bar.y: error: shift/reduce conflicts: 1 found, 0 expected
297 bar.y: error: reduce/reduce conflicts: 2 found, 0 expected
299 ** Incompatibilities with POSIX Yacc
301 The 'yacc' category is no longer part of '-Wall', enable it explicitly
304 ** Additional yylex/yyparse arguments
306 The new directive %param declares additional arguments to both yylex and
307 yyparse. The %lex-param, %parse-param, and %param directives support one
308 or more arguments. Instead of
310 %lex-param {arg1_type *arg1}
311 %lex-param {arg2_type *arg2}
312 %parse-param {arg1_type *arg1}
313 %parse-param {arg2_type *arg2}
317 %param {arg1_type *arg1} {arg2_type *arg2}
319 ** Types of values for %define variables
321 Bison used to make no difference between '%define foo bar' and '%define
322 foo "bar"'. The former is now called a 'keyword value', and the latter a
323 'string value'. A third kind was added: 'code values', such as '%define
326 Keyword variables are used for fixed value sets, e.g.,
330 Code variables are used for value in the target language, e.g.,
332 %define api.value.type {struct semantic_type}
334 String variables are used remaining cases, e.g. file names.
336 ** Variable api.token.prefix
338 The variable api.token.prefix changes the way tokens are identified in
339 the generated files. This is especially useful to avoid collisions
340 with identifiers in the target language. For instance
342 %token FILE for ERROR
343 %define api.token.prefix {TOK_}
345 start: FILE for ERROR;
347 will generate the definition of the symbols TOK_FILE, TOK_for, and
348 TOK_ERROR in the generated sources. In particular, the scanner must
349 use these prefixed token names, although the grammar itself still
350 uses the short names (as in the sample rule given above).
352 ** Variable api.value.type
354 This new %define variable supersedes the #define macro YYSTYPE. The use
355 of YYSTYPE is discouraged. In particular, #defining YYSTYPE *and* either
356 using %union or %defining api.value.type results in undefined behavior.
358 Either define api.value.type, or use "%union":
365 %token <ival> INT "integer"
366 %token <sval> STRING "string"
367 %printer { fprintf (yyo, "%d", $$); } <ival>
368 %destructor { free ($$); } <sval>
371 yylval.ival = 42; return INT;
372 yylval.sval = "42"; return STRING;
374 The %define variable api.value.type supports both keyword and code values.
376 The keyword value 'union' means that the user provides genuine types, not
377 union member names such as "ival" and "sval" above (WARNING: will fail if
378 -y/--yacc/%yacc is enabled).
380 %define api.value.type union
381 %token <int> INT "integer"
382 %token <char *> STRING "string"
383 %printer { fprintf (yyo, "%d", $$); } <int>
384 %destructor { free ($$); } <char *>
387 yylval.INT = 42; return INT;
388 yylval.STRING = "42"; return STRING;
390 The keyword value variant is somewhat equivalent, but for C++ special
391 provision is made to allow classes to be used (more about this below).
393 %define api.value.type variant
394 %token <int> INT "integer"
395 %token <std::string> STRING "string"
397 Code values (in braces) denote user defined types. This is where YYSTYPE
415 %define api.value.type {struct my_value}
416 %token <u.ival> INT "integer"
417 %token <u.sval> STRING "string"
418 %printer { fprintf (yyo, "%d", $$); } <u.ival>
419 %destructor { free ($$); } <u.sval>
422 yylval.u.ival = 42; return INT;
423 yylval.u.sval = "42"; return STRING;
425 ** Variable parse.error
427 This variable controls the verbosity of error messages. The use of the
428 %error-verbose directive is deprecated in favor of "%define parse.error
431 ** Renamed %define variables
433 The following variables have been renamed for consistency. Backward
434 compatibility is ensured, but upgrading is recommended.
436 lr.default-reductions -> lr.default-reduction
437 lr.keep-unreachable-states -> lr.keep-unreachable-state
438 namespace -> api.namespace
439 stype -> api.value.type
441 ** Semantic predicates
443 Contributed by Paul Hilfinger.
445 The new, experimental, semantic-predicate feature allows actions of the
446 form "%?{ BOOLEAN-EXPRESSION }", which cause syntax errors (as for
447 YYERROR) if the expression evaluates to 0, and are evaluated immediately
448 in GLR parsers, rather than being deferred. The result is that they allow
449 the programmer to prune possible parses based on the values of run-time
452 ** The directive %expect-rr is now an error in non GLR mode
454 It used to be an error only if used in non GLR mode, _and_ if there are
455 reduce/reduce conflicts.
457 ** Tokens are numbered in their order of appearance
459 Contributed by Valentin Tolmer.
461 With '%token A B', A had a number less than the one of B. However,
462 precedence declarations used to generate a reversed order. This is now
463 fixed, and introducing tokens with any of %token, %left, %right,
464 %precedence, or %nonassoc yields the same result.
466 When mixing declarations of tokens with a litteral character (e.g., 'a')
467 or with an identifier (e.g., B) in a precedence declaration, Bison
468 numbered the litteral characters first. For example
472 would lead to the tokens declared in this order: 'c' 'd' A B. Again, the
473 input order is now preserved.
475 These changes were made so that one can remove useless precedence and
476 associativity declarations (i.e., map %nonassoc, %left or %right to
477 %precedence, or to %token) and get exactly the same output.
479 ** Useless precedence and associativity
481 Contributed by Valentin Tolmer.
483 When developing and maintaining a grammar, useless associativity and
484 precedence directives are common. They can be a nuisance: new ambiguities
485 arising are sometimes masked because their conflicts are resolved due to
486 the extra precedence or associativity information. Furthermore, it can
487 hinder the comprehension of a new grammar: one will wonder about the role
488 of a precedence, where in fact it is useless. The following changes aim
489 at detecting and reporting these extra directives.
491 *** Precedence warning category
493 A new category of warning, -Wprecedence, was introduced. It flags the
494 useless precedence and associativity directives.
496 *** Useless associativity
498 Bison now warns about symbols with a declared associativity that is never
499 used to resolve conflicts. In that case, using %precedence is sufficient;
500 the parsing tables will remain unchanged. Solving these warnings may raise
501 useless precedence warnings, as the symbols no longer have associativity.
515 warning: useless associativity for '+', use %precedence [-Wprecedence]
519 *** Useless precedence
521 Bison now warns about symbols with a declared precedence and no declared
522 associativity (i.e., declared with %precedence), and whose precedence is
523 never used. In that case, the symbol can be safely declared with %token
524 instead, without modifying the parsing tables. For example:
528 exp: "var" '=' "number";
532 warning: useless precedence for '=' [-Wprecedence]
536 *** Useless precedence and associativity
538 In case of both useless precedence and associativity, the issue is flagged
543 exp: "var" '=' "number";
547 warning: useless precedence and associativity for '=' [-Wprecedence]
553 With help from Joel E. Denny and Gabriel Rassoul.
555 Empty rules (i.e., with an empty right-hand side) can now be explicitly
556 marked by the new %empty directive. Using %empty on a non-empty rule is
557 an error. The new -Wempty-rule warning reports empty rules without
558 %empty. On the following grammar:
568 3.4-5: warning: empty rule without %empty [-Wempty-rule]
571 5.8-13: error: %empty on non-empty rule
575 ** Java skeleton improvements
577 The constants for token names were moved to the Lexer interface. Also, it
578 is possible to add code to the parser's constructors using "%code init"
579 and "%define init_throws".
580 Contributed by Paolo Bonzini.
582 The Java skeleton now supports push parsing.
583 Contributed by Dennis Heimbigner.
585 ** C++ skeletons improvements
587 *** The parser header is no longer mandatory (lalr1.cc, glr.cc)
589 Using %defines is now optional. Without it, the needed support classes
590 are defined in the generated parser, instead of additional files (such as
591 location.hh, position.hh and stack.hh).
593 *** Locations are no longer mandatory (lalr1.cc, glr.cc)
595 Both lalr1.cc and glr.cc no longer require %location.
597 *** syntax_error exception (lalr1.cc)
599 The C++ parser features a syntax_error exception, which can be
600 thrown from the scanner or from user rules to raise syntax errors.
601 This facilitates reporting errors caught in sub-functions (e.g.,
602 rejecting too large integral literals from a conversion function
603 used by the scanner, or rejecting invalid combinations from a
604 factory invoked by the user actions).
606 *** %define api.value.type variant
608 This is based on a submission from Michiel De Wilde. With help
609 from Théophile Ranquet.
611 In this mode, complex C++ objects can be used as semantic values. For
614 %token <::std::string> TEXT;
617 %type <::std::string> item;
618 %type <::std::list<std::string>> list;
621 list { std::cout << $1 << std::endl; }
625 %empty { /* Generates an empty string list. */ }
626 | list item ";" { std::swap ($$, $1); $$.push_back ($2); }
630 TEXT { std::swap ($$, $1); }
631 | NUMBER { $$ = string_cast ($1); }
634 *** %define api.token.constructor
636 When variants are enabled, Bison can generate functions to build the
637 tokens. This guarantees that the token type (e.g., NUMBER) is consistent
638 with the semantic value (e.g., int):
640 parser::symbol_type yylex ()
642 parser::location_type loc = ...;
644 return parser::make_TEXT ("Hello, world!", loc);
646 return parser::make_NUMBER (42, loc);
648 return parser::make_SEMICOLON (loc);
654 There are operator- and operator-= for 'location'. Negative line/column
655 increments can no longer underflow the resulting value.
657 * Noteworthy changes in release 2.7.1 (2013-04-15) [stable]
661 *** Fix compiler attribute portability (yacc.c)
663 With locations enabled, __attribute__ was used unprotected.
665 *** Fix some compiler warnings (lalr1.cc)
667 * Noteworthy changes in release 2.7 (2012-12-12) [stable]
671 Warnings about uninitialized yylloc in yyparse have been fixed.
673 Restored C90 compliance (yet no report was ever made).
675 ** Diagnostics are improved
677 Contributed by Théophile Ranquet.
679 *** Changes in the format of error messages
681 This used to be the format of many error reports:
683 input.y:2.7-12: %type redeclaration for exp
684 input.y:1.7-12: previous declaration
688 input.y:2.7-12: error: %type redeclaration for exp
689 input.y:1.7-12: previous declaration
691 *** New format for error reports: carets
693 Caret errors have been added to Bison:
695 input.y:2.7-12: error: %type redeclaration for exp
698 input.y:1.7-12: previous declaration
704 input.y:3.20-23: error: ambiguous reference: '$exp'
705 exp: exp '+' exp { $exp = $1 + $3; };
707 input.y:3.1-3: refers to: $exp at $$
708 exp: exp '+' exp { $exp = $1 + $3; };
710 input.y:3.6-8: refers to: $exp at $1
711 exp: exp '+' exp { $exp = $1 + $3; };
713 input.y:3.14-16: refers to: $exp at $3
714 exp: exp '+' exp { $exp = $1 + $3; };
717 The default behavior for now is still not to display these unless
718 explicitly asked with -fcaret (or -fall). However, in a later release, it
719 will be made the default behavior (but may still be deactivated with
722 ** New value for %define variable: api.pure full
724 The %define variable api.pure requests a pure (reentrant) parser. However,
725 for historical reasons, using it in a location-tracking Yacc parser
726 resulted in a yyerror function that did not take a location as a
727 parameter. With this new value, the user may request a better pure parser,
728 where yyerror does take a location as a parameter (in location-tracking
731 The use of "%define api.pure true" is deprecated in favor of this new
732 "%define api.pure full".
734 ** New %define variable: api.location.type (glr.cc, lalr1.cc, lalr1.java)
736 The %define variable api.location.type defines the name of the type to use
737 for locations. When defined, Bison no longer generates the position.hh
738 and location.hh files, nor does the parser will include them: the user is
739 then responsible to define her type.
741 This can be used in programs with several parsers to factor their location
742 and position files: let one of them generate them, and the others just use
745 This feature was actually introduced, but not documented, in Bison 2.5,
746 under the name "location_type" (which is maintained for backward
749 For consistency, lalr1.java's %define variables location_type and
750 position_type are deprecated in favor of api.location.type and
753 ** Exception safety (lalr1.cc)
755 The parse function now catches exceptions, uses the %destructors to
756 release memory (the lookahead symbol and the symbols pushed on the stack)
757 before re-throwing the exception.
759 This feature is somewhat experimental. User feedback would be
762 ** Graph improvements in DOT and XSLT
764 Contributed by Théophile Ranquet.
766 The graphical presentation of the states is more readable: their shape is
767 now rectangular, the state number is clearly displayed, and the items are
768 numbered and left-justified.
770 The reductions are now explicitly represented as transitions to other
771 diamond shaped nodes.
773 These changes are present in both --graph output and xml2dot.xsl XSLT
774 processing, with minor (documented) differences.
776 ** %language is no longer an experimental feature.
778 The introduction of this feature, in 2.4, was four years ago. The
779 --language option and the %language directive are no longer experimental.
783 The sections about shift/reduce and reduce/reduce conflicts resolution
784 have been fixed and extended.
786 Although introduced more than four years ago, XML and Graphviz reports
787 were not properly documented.
789 The translation of mid-rule actions is now described.
791 * Noteworthy changes in release 2.6.5 (2012-11-07) [stable]
793 We consider compiler warnings about Bison generated parsers to be bugs.
794 Rather than working around them in your own project, please consider
795 reporting them to us.
799 Warnings about uninitialized yylval and/or yylloc for push parsers with a
800 pure interface have been fixed for GCC 4.0 up to 4.8, and Clang 2.9 to
803 Other issues in the test suite have been addressed.
805 Null characters are correctly displayed in error messages.
807 When possible, yylloc is correctly initialized before calling yylex. It
808 is no longer necessary to initialize it in the %initial-action.
810 * Noteworthy changes in release 2.6.4 (2012-10-23) [stable]
812 Bison 2.6.3's --version was incorrect. This release fixes this issue.
814 * Noteworthy changes in release 2.6.3 (2012-10-22) [stable]
818 Bugs and portability issues in the test suite have been fixed.
820 Some errors in translations have been addressed, and --help now directs
821 users to the appropriate place to report them.
823 Stray Info files shipped by accident are removed.
825 Incorrect definitions of YY_, issued by yacc.c when no parser header is
826 generated, are removed.
828 All the generated headers are self-contained.
830 ** Header guards (yacc.c, glr.c, glr.cc)
832 In order to avoid collisions, the header guards are now
833 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
834 For instance the header generated from
836 %define api.prefix "calc"
837 %defines "lib/parse.h"
839 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
841 ** Fix compiler warnings in the generated parser (yacc.c, glr.c)
843 The compilation of pure parsers (%define api.pure) can trigger GCC
846 input.c: In function 'yyparse':
847 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
848 function [-Wmaybe-uninitialized]
852 This is now fixed; pragmas to avoid these warnings are no longer needed.
854 Warnings from clang ("equality comparison with extraneous parentheses" and
855 "function declared 'noreturn' should not return") have also been
858 * Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
862 Buffer overruns, complaints from Flex, and portability issues in the test
863 suite have been fixed.
865 ** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
867 Trailing end-of-lines in %parse-param or %lex-param would result in
868 invalid C++. This is fixed.
870 ** Spurious spaces and end-of-lines
872 The generated files no longer end (nor start) with empty lines.
874 * Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
876 Bison no longer executes user-specified M4 code when processing a grammar.
880 In addition to the removal of the features announced in Bison 2.6, the
881 next major release will remove the "Temporary hack for adding a semicolon
882 to the user action", as announced in the release 2.5. Instead of:
884 exp: exp "+" exp { $$ = $1 + $3 };
888 exp: exp "+" exp { $$ = $1 + $3; };
892 *** Type names are now properly escaped.
894 *** glr.cc: set_debug_level and debug_level work as expected.
896 *** Stray @ or $ in actions
898 While Bison used to warn about stray $ or @ in action rules, it did not
899 for other actions such as printers, destructors, or initial actions. It
902 ** Type names in actions
904 For consistency with rule actions, it is now possible to qualify $$ by a
905 type-name in destructors, printers, and initial actions. For instance:
907 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
909 will display two values for each typed and untyped symbol (provided
910 that YYSTYPE has both "ival" and "fval" fields).
912 * Noteworthy changes in release 2.6 (2012-07-19) [stable]
916 The next major release of Bison will drop support for the following
917 deprecated features. Please report disagreements to bug-bison@gnu.org.
921 Support for generating parsers in K&R C will be removed. Parsers
922 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
925 *** Features deprecated since Bison 1.875
927 The definitions of yystype and yyltype will be removed; use YYSTYPE and
930 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
931 %lex-param, will no longer be supported.
933 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
936 *** The generated header will be included (yacc.c)
938 Instead of duplicating the content of the generated header (definition of
939 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
940 as is already the case for GLR or C++ parsers. This change is deferred
941 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
944 ** Generated Parser Headers
946 *** Guards (yacc.c, glr.c, glr.cc)
948 The generated headers are now guarded, as is already the case for C++
949 parsers (lalr1.cc). For instance, with --defines=foo.h:
954 #endif /* !YY_FOO_H */
956 *** New declarations (yacc.c, glr.c)
958 The generated header now declares yydebug and yyparse. Both honor
959 --name-prefix=bar_, and yield
961 int bar_parse (void);
965 #define yyparse bar_parse
968 in order to facilitate the inclusion of several parser headers inside a
969 single compilation unit.
971 *** Exported symbols in C++
973 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
974 header, are removed, as they prevent the possibility of including several
975 generated headers from a single compilation unit.
979 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
982 ** New %define variable: api.prefix
984 Now that the generated headers are more complete and properly protected
985 against multiple inclusions, constant names, such as YYSTYPE are a
986 problem. While yyparse and others are properly renamed by %name-prefix,
987 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
988 would introduce backward compatibility issues in projects not expecting
989 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
990 it is deprecated in favor of a new %define variable: api.prefix.
992 The following examples compares both:
994 %name-prefix "bar_" | %define api.prefix "bar_"
995 %token <ival> FOO %token <ival> FOO
996 %union { int ival; } %union { int ival; }
1002 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
1003 # define BAR_FOO_H # define BAR_FOO_H
1005 /* Enabling traces. */ /* Enabling traces. */
1006 # ifndef YYDEBUG | # ifndef BAR_DEBUG
1007 > # if defined YYDEBUG
1009 > # define BAR_DEBUG 1
1011 > # define BAR_DEBUG 0
1014 # define YYDEBUG 0 | # define BAR_DEBUG 0
1018 # if YYDEBUG | # if BAR_DEBUG
1019 extern int bar_debug; extern int bar_debug;
1022 /* Tokens. */ /* Tokens. */
1023 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
1024 # define YYTOKENTYPE | # define BAR_TOKENTYPE
1025 enum yytokentype { | enum bar_tokentype {
1030 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
1031 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
1032 typedef union YYSTYPE | typedef union BAR_STYPE
1035 } YYSTYPE; | } BAR_STYPE;
1036 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
1039 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
1041 int bar_parse (void); int bar_parse (void);
1043 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
1045 * Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
1049 The next major release will drop support for generating parsers in K&R C.
1051 ** yacc.c: YYBACKUP works as expected.
1053 ** glr.c improvements:
1055 *** Location support is eliminated when not requested:
1057 GLR parsers used to include location-related code even when locations were
1058 not requested, and therefore not even usable.
1060 *** __attribute__ is preserved:
1062 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
1063 when -std is passed to GCC).
1065 ** lalr1.java: several fixes:
1067 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
1068 first token leads to a syntax error. Some minor clean ups.
1072 *** C++11 compatibility:
1074 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
1079 The header files such as "parser.hh", "location.hh", etc. used a constant
1080 name for preprocessor guards, for instance:
1082 #ifndef BISON_LOCATION_HH
1083 # define BISON_LOCATION_HH
1085 #endif // !BISON_LOCATION_HH
1087 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
1088 case characters are converted to upper case, and series of
1089 non-alphanumerical characters are converted to an underscore.
1091 With "bison -o lang++/parser.cc", "location.hh" would now include:
1093 #ifndef YY_LANG_LOCATION_HH
1094 # define YY_LANG_LOCATION_HH
1096 #endif // !YY_LANG_LOCATION_HH
1100 The position and location constructors (and their initialize methods)
1101 accept new arguments for line and column. Several issues in the
1102 documentation were fixed.
1104 ** liby is no longer asking for "rpl_fprintf" on some platforms.
1106 ** Changes in the manual:
1108 *** %printer is documented
1110 The "%printer" directive, supported since at least Bison 1.50, is finally
1111 documented. The "mfcalc" example is extended to demonstrate it.
1113 For consistency with the C skeletons, the C++ parsers now also support
1114 "yyoutput" (as an alias to "debug_stream ()").
1116 *** Several improvements have been made:
1118 The layout for grammar excerpts was changed to a more compact scheme.
1119 Named references are motivated. The description of the automaton
1120 description file (*.output) is updated to the current format. Incorrect
1121 index entries were fixed. Some other errors were fixed.
1125 *** Conflicting prototypes with recent/modified Flex.
1127 Fixed build problems with the current, unreleased, version of Flex, and
1128 some modified versions of 2.5.35, which have modified function prototypes.
1130 *** Warnings during the build procedure have been eliminated.
1132 *** Several portability problems in the test suite have been fixed:
1134 This includes warnings with some compilers, unexpected behavior of tools
1135 such as diff, warning messages from the test suite itself, etc.
1137 *** The install-pdf target works properly:
1139 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
1140 halts in the middle of its course.
1142 * Changes in version 2.5 (2011-05-14):
1144 ** Grammar symbol names can now contain non-initial dashes:
1146 Consistently with directives (such as %error-verbose) and with
1147 %define variables (e.g. push-pull), grammar symbol names may contain
1148 dashes in any position except the beginning. This is a GNU
1149 extension over POSIX Yacc. Thus, use of this extension is reported
1150 by -Wyacc and rejected in Yacc mode (--yacc).
1152 ** Named references:
1154 Historically, Yacc and Bison have supported positional references
1155 ($n, $$) to allow access to symbol values from inside of semantic
1158 Starting from this version, Bison can also accept named references.
1159 When no ambiguity is possible, original symbol names may be used
1160 as named references:
1162 if_stmt : "if" cond_expr "then" then_stmt ';'
1163 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
1165 In the more common case, explicit names may be declared:
1167 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
1168 { $res = mk_if_stmt($cond, $then, $else); }
1170 Location information is also accessible using @name syntax. When
1171 accessing symbol names containing dots or dashes, explicit bracketing
1172 ($[sym.1]) must be used.
1174 These features are experimental in this version. More user feedback
1175 will help to stabilize them.
1176 Contributed by Alex Rozenman.
1178 ** IELR(1) and canonical LR(1):
1180 IELR(1) is a minimal LR(1) parser table generation algorithm. That
1181 is, given any context-free grammar, IELR(1) generates parser tables
1182 with the full language-recognition power of canonical LR(1) but with
1183 nearly the same number of parser states as LALR(1). This reduction
1184 in parser states is often an order of magnitude. More importantly,
1185 because canonical LR(1)'s extra parser states may contain duplicate
1186 conflicts in the case of non-LR(1) grammars, the number of conflicts
1187 for IELR(1) is often an order of magnitude less as well. This can
1188 significantly reduce the complexity of developing of a grammar.
1190 Bison can now generate IELR(1) and canonical LR(1) parser tables in
1191 place of its traditional LALR(1) parser tables, which remain the
1192 default. You can specify the type of parser tables in the grammar
1193 file with these directives:
1195 %define lr.type lalr
1196 %define lr.type ielr
1197 %define lr.type canonical-lr
1199 The default-reduction optimization in the parser tables can also be
1200 adjusted using "%define lr.default-reductions". For details on both
1201 of these features, see the new section "Tuning LR" in the Bison
1204 These features are experimental. More user feedback will help to
1207 ** LAC (Lookahead Correction) for syntax error handling
1209 Contributed by Joel E. Denny.
1211 Canonical LR, IELR, and LALR can suffer from a couple of problems
1212 upon encountering a syntax error. First, the parser might perform
1213 additional parser stack reductions before discovering the syntax
1214 error. Such reductions can perform user semantic actions that are
1215 unexpected because they are based on an invalid token, and they
1216 cause error recovery to begin in a different syntactic context than
1217 the one in which the invalid token was encountered. Second, when
1218 verbose error messages are enabled (with %error-verbose or the
1219 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
1220 syntax error message can both contain invalid tokens and omit valid
1223 The culprits for the above problems are %nonassoc, default
1224 reductions in inconsistent states, and parser state merging. Thus,
1225 IELR and LALR suffer the most. Canonical LR can suffer only if
1226 %nonassoc is used or if default reductions are enabled for
1227 inconsistent states.
1229 LAC is a new mechanism within the parsing algorithm that solves
1230 these problems for canonical LR, IELR, and LALR without sacrificing
1231 %nonassoc, default reductions, or state merging. When LAC is in
1232 use, canonical LR and IELR behave almost exactly the same for both
1233 syntactically acceptable and syntactically unacceptable input.
1234 While LALR still does not support the full language-recognition
1235 power of canonical LR and IELR, LAC at least enables LALR's syntax
1236 error handling to correctly reflect LALR's language-recognition
1239 Currently, LAC is only supported for deterministic parsers in C.
1240 You can enable LAC with the following directive:
1242 %define parse.lac full
1244 See the new section "LAC" in the Bison manual for additional
1245 details including a few caveats.
1247 LAC is an experimental feature. More user feedback will help to
1250 ** %define improvements:
1252 *** Can now be invoked via the command line:
1254 Each of these command-line options
1257 --define=NAME[=VALUE]
1260 --force-define=NAME[=VALUE]
1262 is equivalent to this grammar file declaration
1264 %define NAME ["VALUE"]
1266 except that the manner in which Bison processes multiple definitions
1267 for the same NAME differs. Most importantly, -F and --force-define
1268 quietly override %define, but -D and --define do not. For further
1269 details, see the section "Bison Options" in the Bison manual.
1271 *** Variables renamed:
1273 The following %define variables
1276 lr.keep_unreachable_states
1278 have been renamed to
1281 lr.keep-unreachable-states
1283 The old names are now deprecated but will be maintained indefinitely
1284 for backward compatibility.
1286 *** Values no longer need to be quoted in the grammar file:
1288 If a %define value is an identifier, it no longer needs to be placed
1289 within quotations marks. For example,
1291 %define api.push-pull "push"
1295 %define api.push-pull push
1297 *** Unrecognized variables are now errors not warnings.
1299 *** Multiple invocations for any variable is now an error not a warning.
1301 ** Unrecognized %code qualifiers are now errors not warnings.
1303 ** Character literals not of length one:
1305 Previously, Bison quietly converted all character literals to length
1306 one. For example, without warning, Bison interpreted the operators in
1307 the following grammar to be the same token:
1313 Bison now warns when a character literal is not of length one. In
1314 some future release, Bison will start reporting an error instead.
1316 ** Destructor calls fixed for lookaheads altered in semantic actions:
1318 Previously for deterministic parsers in C, if a user semantic action
1319 altered yychar, the parser in some cases used the old yychar value to
1320 determine which destructor to call for the lookahead upon a syntax
1321 error or upon parser return. This bug has been fixed.
1323 ** C++ parsers use YYRHSLOC:
1325 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
1326 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
1327 to use it. If, for instance, your location structure has "first"
1328 and "last" members, instead of
1330 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1334 (Current).first = (Rhs)[1].location.first; \
1335 (Current).last = (Rhs)[N].location.last; \
1339 (Current).first = (Current).last = (Rhs)[0].location.last; \
1345 # define YYLLOC_DEFAULT(Current, Rhs, N) \
1349 (Current).first = YYRHSLOC (Rhs, 1).first; \
1350 (Current).last = YYRHSLOC (Rhs, N).last; \
1354 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
1358 ** YYLLOC_DEFAULT in C++:
1360 The default implementation of YYLLOC_DEFAULT used to be issued in
1361 the header file. It is now output in the implementation file, after
1362 the user %code sections so that its #ifndef guard does not try to
1363 override the user's YYLLOC_DEFAULT if provided.
1365 ** YYFAIL now produces warnings and Java parsers no longer implement it:
1367 YYFAIL has existed for many years as an undocumented feature of
1368 deterministic parsers in C generated by Bison. More recently, it was
1369 a documented feature of Bison's experimental Java parsers. As
1370 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
1371 semantic action now produces a deprecation warning, and Java parsers
1372 no longer implement YYFAIL at all. For further details, including a
1373 discussion of how to suppress C preprocessor warnings about YYFAIL
1374 being unused, see the Bison 2.4.2 NEWS entry.
1376 ** Temporary hack for adding a semicolon to the user action:
1378 Previously, Bison appended a semicolon to every user action for
1379 reductions when the output language defaulted to C (specifically, when
1380 neither %yacc, %language, %skeleton, or equivalent command-line
1381 options were specified). This allowed actions such as
1383 exp: exp "+" exp { $$ = $1 + $3 };
1387 exp: exp "+" exp { $$ = $1 + $3; };
1389 As a first step in removing this misfeature, Bison now issues a
1390 warning when it appends a semicolon. Moreover, in cases where Bison
1391 cannot easily determine whether a semicolon is needed (for example, an
1392 action ending with a cpp directive or a braced compound initializer),
1393 it no longer appends one. Thus, the C compiler might now complain
1394 about a missing semicolon where it did not before. Future releases of
1395 Bison will cease to append semicolons entirely.
1397 ** Verbose syntax error message fixes:
1399 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
1400 specified, syntax error messages produced by the generated parser
1401 include the unexpected token as well as a list of expected tokens.
1402 The effect of %nonassoc on these verbose messages has been corrected
1403 in two ways, but a more complete fix requires LAC, described above:
1405 *** When %nonassoc is used, there can exist parser states that accept no
1406 tokens, and so the parser does not always require a lookahead token
1407 in order to detect a syntax error. Because no unexpected token or
1408 expected tokens can then be reported, the verbose syntax error
1409 message described above is suppressed, and the parser instead
1410 reports the simpler message, "syntax error". Previously, this
1411 suppression was sometimes erroneously triggered by %nonassoc when a
1412 lookahead was actually required. Now verbose messages are
1413 suppressed only when all previous lookaheads have already been
1414 shifted or discarded.
1416 *** Previously, the list of expected tokens erroneously included tokens
1417 that would actually induce a syntax error because conflicts for them
1418 were resolved with %nonassoc in the current parser state. Such
1419 tokens are now properly omitted from the list.
1421 *** Expected token lists are still often wrong due to state merging
1422 (from LALR or IELR) and default reductions, which can both add
1423 invalid tokens and subtract valid tokens. Canonical LR almost
1424 completely fixes this problem by eliminating state merging and
1425 default reductions. However, there is one minor problem left even
1426 when using canonical LR and even after the fixes above. That is,
1427 if the resolution of a conflict with %nonassoc appears in a later
1428 parser state than the one at which some syntax error is
1429 discovered, the conflicted token is still erroneously included in
1430 the expected token list. Bison's new LAC implementation,
1431 described above, eliminates this problem and the need for
1432 canonical LR. However, LAC is still experimental and is disabled
1435 ** Java skeleton fixes:
1437 *** A location handling bug has been fixed.
1439 *** The top element of each of the value stack and location stack is now
1440 cleared when popped so that it can be garbage collected.
1442 *** Parser traces now print the top element of the stack.
1444 ** -W/--warnings fixes:
1446 *** Bison now properly recognizes the "no-" versions of categories:
1448 For example, given the following command line, Bison now enables all
1449 warnings except warnings for incompatibilities with POSIX Yacc:
1451 bison -Wall,no-yacc gram.y
1453 *** Bison now treats S/R and R/R conflicts like other warnings:
1455 Previously, conflict reports were independent of Bison's normal
1456 warning system. Now, Bison recognizes the warning categories
1457 "conflicts-sr" and "conflicts-rr". This change has important
1458 consequences for the -W and --warnings command-line options. For
1461 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
1462 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
1463 bison -Wnone gram.y # no conflicts are reported
1464 bison -Werror gram.y # any conflict is an error
1466 However, as before, if the %expect or %expect-rr directive is
1467 specified, an unexpected number of conflicts is an error, and an
1468 expected number of conflicts is not reported, so -W and --warning
1469 then have no effect on the conflict report.
1471 *** The "none" category no longer disables a preceding "error":
1473 For example, for the following command line, Bison now reports
1474 errors instead of warnings for incompatibilities with POSIX Yacc:
1476 bison -Werror,none,yacc gram.y
1478 *** The "none" category now disables all Bison warnings:
1480 Previously, the "none" category disabled only Bison warnings for
1481 which there existed a specific -W/--warning category. However,
1482 given the following command line, Bison is now guaranteed to
1483 suppress all warnings:
1487 ** Precedence directives can now assign token number 0:
1489 Since Bison 2.3b, which restored the ability of precedence
1490 directives to assign token numbers, doing so for token number 0 has
1491 produced an assertion failure. For example:
1495 This bug has been fixed.
1497 * Changes in version 2.4.3 (2010-08-05):
1499 ** Bison now obeys -Werror and --warnings=error for warnings about
1500 grammar rules that are useless in the parser due to conflicts.
1502 ** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
1505 ** Failures in the test suite for GCC 4.5 have been fixed.
1507 ** Failures in the test suite for some versions of Sun Studio C++ have
1510 ** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
1511 warnings about undefined %prec identifiers will not be converted to
1512 errors in Bison 2.5. They will remain warnings, which should be
1513 sufficient for POSIX while avoiding backward compatibility issues.
1515 ** Minor documentation fixes.
1517 * Changes in version 2.4.2 (2010-03-20):
1519 ** Some portability problems that resulted in failures and livelocks
1520 in the test suite on some versions of at least Solaris, AIX, HP-UX,
1521 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
1522 errors should no longer cause M4 to report a broken pipe on the
1525 ** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
1527 POSIX specifies that an error be reported for any identifier that does
1528 not appear on the LHS of a grammar rule and that is not defined by
1529 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
1530 error report for the case when an identifier appears only after a
1531 %prec directive. It is now restored. However, for backward
1532 compatibility with recent Bison releases, it is only a warning for
1533 now. In Bison 2.5 and later, it will return to being an error.
1534 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
1535 warning will not be converted to an error in Bison 2.5.]
1537 ** Detection of GNU M4 1.4.6 or newer during configure is improved.
1539 ** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
1540 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
1543 ** %code is now a permanent feature.
1545 A traditional Yacc prologue directive is written in the form:
1549 To provide a more flexible alternative, Bison 2.3b introduced the
1550 %code directive with the following forms for C/C++:
1553 %code requires {CODE}
1554 %code provides {CODE}
1557 These forms are now considered permanent features of Bison. See the
1558 %code entries in the section "Bison Declaration Summary" in the Bison
1559 manual for a summary of their functionality. See the section
1560 "Prologue Alternatives" for a detailed discussion including the
1561 advantages of %code over the traditional Yacc prologue directive.
1563 Bison's Java feature as a whole including its current usage of %code
1564 is still considered experimental.
1566 ** YYFAIL is deprecated and will eventually be removed.
1568 YYFAIL has existed for many years as an undocumented feature of
1569 deterministic parsers in C generated by Bison. Previously, it was
1570 documented for Bison's experimental Java parsers. YYFAIL is no longer
1571 documented for Java parsers and is formally deprecated in both cases.
1572 Users are strongly encouraged to migrate to YYERROR, which is
1575 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
1576 induce a syntax error. The most obvious difference from YYERROR is
1577 that YYFAIL will automatically invoke yyerror to report the syntax
1578 error so that you don't have to. However, there are several other
1579 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
1580 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1581 used. For a more detailed discussion, see:
1583 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
1585 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
1586 deterministic parsers in C will continue to implement it. However,
1587 because YYFAIL is already flawed, it seems futile to try to make new
1588 Bison features compatible with it. Thus, during parser generation,
1589 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
1590 rule action. In a later release, YYFAIL will be disabled for
1591 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1592 be removed altogether.
1594 There exists at least one case where Bison 2.5's YYFAIL warning will
1595 be a false positive. Some projects add phony uses of YYFAIL and other
1596 Bison-defined macros for the sole purpose of suppressing C
1597 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
1598 To avoid Bison's future warning, such YYFAIL uses can be moved to the
1599 epilogue (that is, after the second "%%") in the Bison input file. In
1600 this release (2.4.2), Bison already generates its own code to suppress
1601 C preprocessor warnings for YYFAIL, so projects can remove their own
1602 phony uses of YYFAIL if compatibility with Bison releases prior to
1603 2.4.2 is not necessary.
1605 ** Internationalization.
1607 Fix a regression introduced in Bison 2.4: Under some circumstances,
1608 message translations were not installed although supported by the
1611 * Changes in version 2.4.1 (2008-12-11):
1613 ** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
1614 declarations have been fixed.
1616 ** Temporary hack for adding a semicolon to the user action.
1618 Bison used to prepend a trailing semicolon at the end of the user
1619 action for reductions. This allowed actions such as
1621 exp: exp "+" exp { $$ = $1 + $3 };
1625 exp: exp "+" exp { $$ = $1 + $3; };
1627 Some grammars still depend on this "feature". Bison 2.4.1 restores
1628 the previous behavior in the case of C output (specifically, when
1629 neither %language or %skeleton or equivalent command-line options
1630 are used) to leave more time for grammars depending on the old
1631 behavior to be adjusted. Future releases of Bison will disable this
1634 ** A few minor improvements to the Bison manual.
1636 * Changes in version 2.4 (2008-11-02):
1638 ** %language is an experimental feature.
1640 We first introduced this feature in test release 2.3b as a cleaner
1641 alternative to %skeleton. Since then, we have discussed the possibility of
1642 modifying its effect on Bison's output file names. Thus, in this release,
1643 we consider %language to be an experimental feature that will likely evolve
1646 ** Forward compatibility with GNU M4 has been improved.
1648 ** Several bugs in the C++ skeleton and the experimental Java skeleton have been
1651 * Changes in version 2.3b (2008-05-27):
1653 ** The quotes around NAME that used to be required in the following directive
1656 %define NAME "VALUE"
1658 ** The directive "%pure-parser" is now deprecated in favor of:
1662 which has the same effect except that Bison is more careful to warn about
1663 unreasonable usage in the latter case.
1667 Bison can now generate an LALR(1) parser in C with a push interface. That
1668 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
1669 push one token at a time to the parser using "yypush_parse", which will
1670 return to the caller after processing each token. By default, the push
1671 interface is disabled. Either of the following directives will enable it:
1673 %define api.push_pull "push" // Just push; does not require yylex.
1674 %define api.push_pull "both" // Push and pull; requires yylex.
1676 See the new section "A Push Parser" in the Bison manual for details.
1678 The current push parsing interface is experimental and may evolve. More user
1679 feedback will help to stabilize it.
1681 ** The -g and --graph options now output graphs in Graphviz DOT format,
1682 not VCG format. Like --graph, -g now also takes an optional FILE argument
1683 and thus cannot be bundled with other short options.
1687 Bison can now generate an LALR(1) parser in Java. The skeleton is
1688 "data/lalr1.java". Consider using the new %language directive instead of
1689 %skeleton to select it.
1691 See the new section "Java Parsers" in the Bison manual for details.
1693 The current Java interface is experimental and may evolve. More user
1694 feedback will help to stabilize it.
1695 Contributed by Paolo Bonzini.
1699 This new directive specifies the programming language of the generated
1700 parser, which can be C (the default), C++, or Java. Besides the skeleton
1701 that Bison uses, the directive affects the names of the generated files if
1702 the grammar file's name ends in ".y".
1704 ** XML Automaton Report
1706 Bison can now generate an XML report of the LALR(1) automaton using the new
1707 "--xml" option. The current XML schema is experimental and may evolve. More
1708 user feedback will help to stabilize it.
1709 Contributed by Wojciech Polak.
1711 ** The grammar file may now specify the name of the parser header file using
1712 %defines. For example:
1716 ** When reporting useless rules, useless nonterminals, and unused terminals,
1717 Bison now employs the terms "useless in grammar" instead of "useless",
1718 "useless in parser" instead of "never reduced", and "unused in grammar"
1719 instead of "unused".
1721 ** Unreachable State Removal
1723 Previously, Bison sometimes generated parser tables containing unreachable
1724 states. A state can become unreachable during conflict resolution if Bison
1725 disables a shift action leading to it from a predecessor state. Bison now:
1727 1. Removes unreachable states.
1729 2. Does not report any conflicts that appeared in unreachable states.
1730 WARNING: As a result, you may need to update %expect and %expect-rr
1731 directives in existing grammar files.
1733 3. For any rule used only in such states, Bison now reports the rule as
1734 "useless in parser due to conflicts".
1736 This feature can be disabled with the following directive:
1738 %define lr.keep_unreachable_states
1740 See the %define entry in the "Bison Declaration Summary" in the Bison manual
1741 for further discussion.
1743 ** Lookahead Set Correction in the ".output" Report
1745 When instructed to generate a ".output" file including lookahead sets
1746 (using "--report=lookahead", for example), Bison now prints each reduction's
1747 lookahead set only next to the associated state's one item that (1) is
1748 associated with the same rule as the reduction and (2) has its dot at the end
1749 of its RHS. Previously, Bison also erroneously printed the lookahead set
1750 next to all of the state's other items associated with the same rule. This
1751 bug affected only the ".output" file and not the generated parser source
1754 ** --report-file=FILE is a new option to override the default ".output" file
1757 ** The "=" that used to be required in the following directives is now
1760 %file-prefix "parser"
1764 ** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
1766 Bison 2.3a provided a new set of directives as a more flexible alternative to
1767 the traditional Yacc prologue blocks. Those have now been consolidated into
1768 a single %code directive with an optional qualifier field, which identifies
1769 the purpose of the code and thus the location(s) where Bison should generate
1772 1. "%code {CODE}" replaces "%after-header {CODE}"
1773 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1774 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1775 4. "%code top {CODE}" replaces "%before-header {CODE}"
1777 See the %code entries in section "Bison Declaration Summary" in the Bison
1778 manual for a summary of the new functionality. See the new section "Prologue
1779 Alternatives" for a detailed discussion including the advantages of %code
1780 over the traditional Yacc prologues.
1782 The prologue alternatives are experimental. More user feedback will help to
1783 determine whether they should become permanent features.
1785 ** Revised warning: unset or unused mid-rule values
1787 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1788 used within any of the actions of the parent rule. For example, Bison warns
1791 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1793 Now, Bison also warns about mid-rule values that are used but not set. For
1794 example, Bison warns about unset $$ in the mid-rule action in:
1796 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1798 However, Bison now disables both of these warnings by default since they
1799 sometimes prove to be false alarms in existing grammars employing the Yacc
1800 constructs $0 or $-N (where N is some positive integer).
1802 To enable these warnings, specify the option "--warnings=midrule-values" or
1803 "-W", which is a synonym for "--warnings=all".
1805 ** Default %destructor or %printer with "<*>" or "<>"
1807 Bison now recognizes two separate kinds of default %destructor's and
1810 1. Place "<*>" in a %destructor/%printer symbol list to define a default
1811 %destructor/%printer for all grammar symbols for which you have formally
1812 declared semantic type tags.
1814 2. Place "<>" in a %destructor/%printer symbol list to define a default
1815 %destructor/%printer for all grammar symbols without declared semantic
1818 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1819 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
1820 longer applies any %destructor to a mid-rule value if that mid-rule value is
1821 not actually ever referenced using either $$ or $n in a semantic action.
1823 The default %destructor's and %printer's are experimental. More user
1824 feedback will help to determine whether they should become permanent
1827 See the section "Freeing Discarded Symbols" in the Bison manual for further
1830 ** %left, %right, and %nonassoc can now declare token numbers. This is required
1831 by POSIX. However, see the end of section "Operator Precedence" in the Bison
1832 manual for a caveat concerning the treatment of literal strings.
1834 ** The nonfunctional --no-parser, -n, and %no-parser options have been
1835 completely removed from Bison.
1837 * Changes in version 2.3a, 2006-09-13:
1839 ** Instead of %union, you can define and use your own union type
1840 YYSTYPE if your grammar contains at least one <type> tag.
1841 Your YYSTYPE need not be a macro; it can be a typedef.
1842 This change is for compatibility with other Yacc implementations,
1843 and is required by POSIX.
1845 ** Locations columns and lines start at 1.
1846 In accordance with the GNU Coding Standards and Emacs.
1848 ** You may now declare per-type and default %destructor's and %printer's:
1852 %union { char *string; }
1853 %token <string> STRING1
1854 %token <string> STRING2
1855 %type <string> string1
1856 %type <string> string2
1857 %union { char character; }
1858 %token <character> CHR
1859 %type <character> chr
1860 %destructor { free ($$); } %symbol-default
1861 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1862 %destructor { } <character>
1864 guarantees that, when the parser discards any user-defined symbol that has a
1865 semantic type tag other than "<character>", it passes its semantic value to
1866 "free". However, when the parser discards a "STRING1" or a "string1", it
1867 also prints its line number to "stdout". It performs only the second
1868 "%destructor" in this case, so it invokes "free" only once.
1870 [Although we failed to mention this here in the 2.3a release, the default
1871 %destructor's and %printer's were experimental, and they were rewritten in
1874 ** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1875 "--yacc", or "%yacc"), Bison no longer generates #define statements for
1876 associating token numbers with token names. Removing the #define statements
1877 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1878 requires them. Bison still generates an enum for token names in all cases.
1880 ** Handling of traditional Yacc prologue blocks is now more consistent but
1881 potentially incompatible with previous releases of Bison.
1883 As before, you declare prologue blocks in your grammar file with the
1884 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
1885 prologue blocks that you've declared before the first %union. To generate
1886 the post-prologue, Bison concatenates all prologue blocks that you've
1887 declared after the first %union.
1889 Previous releases of Bison inserted the pre-prologue into both the header
1890 file and the code file in all cases except for LALR(1) parsers in C. In the
1891 latter case, Bison inserted it only into the code file. For parsers in C++,
1892 the point of insertion was before any token definitions (which associate
1893 token numbers with names). For parsers in C, the point of insertion was
1894 after the token definitions.
1896 Now, Bison never inserts the pre-prologue into the header file. In the code
1897 file, it always inserts it before the token definitions.
1899 ** Bison now provides a more flexible alternative to the traditional Yacc
1900 prologue blocks: %before-header, %start-header, %end-header, and
1903 For example, the following declaration order in the grammar file reflects the
1904 order in which Bison will output these code blocks. However, you are free to
1905 declare these code blocks in your grammar file in whatever order is most
1909 /* Bison treats this block like a pre-prologue block: it inserts it into
1910 * the code file before the contents of the header file. It does *not*
1911 * insert it into the header file. This is a good place to put
1912 * #include's that you want at the top of your code file. A common
1913 * example is '#include "system.h"'. */
1916 /* Bison inserts this block into both the header file and the code file.
1917 * In both files, the point of insertion is before any Bison-generated
1918 * token, semantic type, location type, and class definitions. This is a
1919 * good place to define %union dependencies, for example. */
1922 /* Unlike the traditional Yacc prologue blocks, the output order for the
1923 * new %*-header blocks is not affected by their declaration position
1924 * relative to any %union in the grammar file. */
1927 /* Bison inserts this block into both the header file and the code file.
1928 * In both files, the point of insertion is after the Bison-generated
1929 * definitions. This is a good place to declare or define public
1930 * functions or data structures that depend on the Bison-generated
1934 /* Bison treats this block like a post-prologue block: it inserts it into
1935 * the code file after the contents of the header file. It does *not*
1936 * insert it into the header file. This is a good place to declare or
1937 * define internal functions or data structures that depend on the
1938 * Bison-generated definitions. */
1941 If you have multiple occurrences of any one of the above declarations, Bison
1942 will concatenate the contents in declaration order.
1944 [Although we failed to mention this here in the 2.3a release, the prologue
1945 alternatives were experimental, and they were rewritten in future versions.]
1947 ** The option "--report=look-ahead" has been changed to "--report=lookahead".
1948 The old spelling still works, but is not documented and may be removed
1949 in a future release.
1951 * Changes in version 2.3, 2006-06-05:
1953 ** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
1954 for compatibility with LALR(1) grammars.
1956 ** It is now documented that any definition of YYSTYPE or YYLTYPE should
1957 be to a type name that does not contain parentheses or brackets.
1959 * Changes in version 2.2, 2006-05-19:
1961 ** The distribution terms for all Bison-generated parsers now permit
1962 using the parsers in nonfree programs. Previously, this permission
1963 was granted only for Bison-generated LALR(1) parsers in C.
1965 ** %name-prefix changes the namespace name in C++ outputs.
1967 ** The C++ parsers export their token_type.
1969 ** Bison now allows multiple %union declarations, and concatenates
1970 their contents together.
1972 ** New warning: unused values
1973 Right-hand side symbols whose values are not used are reported,
1974 if the symbols have destructors. For instance:
1976 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
1980 will trigger a warning about $$ and $5 in the first rule, and $3 in
1981 the second ($1 is copied to $$ by the default rule). This example
1982 most likely contains three errors, and could be rewritten as:
1984 exp: exp "?" exp ":" exp
1985 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
1987 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
1990 However, if the original actions were really intended, memory leaks
1991 and all, the warnings can be suppressed by letting Bison believe the
1992 values are used, e.g.:
1994 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
1995 | exp "+" exp { $$ = $1; (void) $3; }
1998 If there are mid-rule actions, the warning is issued if no action
1999 uses it. The following triggers no warning: $1 and $3 are used.
2001 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
2003 The warning is intended to help catching lost values and memory leaks.
2004 If a value is ignored, its associated memory typically is not reclaimed.
2006 ** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
2007 Destructors are now called when user code invokes YYABORT, YYACCEPT,
2008 and YYERROR, for all objects on the stack, other than objects
2009 corresponding to the right-hand side of the current rule.
2011 ** %expect, %expect-rr
2012 Incorrect numbers of expected conflicts are now actual errors,
2013 instead of warnings.
2015 ** GLR, YACC parsers.
2016 The %parse-params are available in the destructors (and the
2017 experimental printers) as per the documentation.
2019 ** Bison now warns if it finds a stray "$" or "@" in an action.
2021 ** %require "VERSION"
2022 This specifies that the grammar file depends on features implemented
2023 in Bison version VERSION or higher.
2025 ** lalr1.cc: The token and value types are now class members.
2026 The tokens were defined as free form enums and cpp macros. YYSTYPE
2027 was defined as a free form union. They are now class members:
2028 tokens are enumerations of the "yy::parser::token" struct, and the
2029 semantic values have the "yy::parser::semantic_type" type.
2031 If you do not want or can update to this scheme, the directive
2032 '%define "global_tokens_and_yystype" "1"' triggers the global
2033 definition of tokens and YYSTYPE. This change is suitable both
2034 for previous releases of Bison, and this one.
2036 If you wish to update, then make sure older version of Bison will
2037 fail using '%require "2.2"'.
2039 ** DJGPP support added.
2041 * Changes in version 2.1, 2005-09-16:
2043 ** The C++ lalr1.cc skeleton supports %lex-param.
2045 ** Bison-generated parsers now support the translation of diagnostics like
2046 "syntax error" into languages other than English. The default
2047 language is still English. For details, please see the new
2048 Internationalization section of the Bison manual. Software
2049 distributors should also see the new PACKAGING file. Thanks to
2050 Bruno Haible for this new feature.
2052 ** Wording in the Bison-generated parsers has been changed slightly to
2053 simplify translation. In particular, the message "memory exhausted"
2054 has replaced "parser stack overflow", as the old message was not
2055 always accurate for modern Bison-generated parsers.
2057 ** Destructors are now called when the parser aborts, for all symbols left
2058 behind on the stack. Also, the start symbol is now destroyed after a
2059 successful parse. In both cases, the behavior was formerly inconsistent.
2061 ** When generating verbose diagnostics, Bison-generated parsers no longer
2062 quote the literal strings associated with tokens. For example, for
2063 a syntax error associated with '%token NUM "number"' they might
2064 print 'syntax error, unexpected number' instead of 'syntax error,
2065 unexpected "number"'.
2067 * Changes in version 2.0, 2004-12-25:
2069 ** Possibly-incompatible changes
2071 - Bison-generated parsers no longer default to using the alloca function
2072 (when available) to extend the parser stack, due to widespread
2073 problems in unchecked stack-overflow detection. You can "#define
2074 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
2075 the manual to determine safe values for YYMAXDEPTH in that case.
2077 - Error token location.
2078 During error recovery, the location of the syntax error is updated
2079 to cover the whole sequence covered by the error token: it includes
2080 the shifted symbols thrown away during the first part of the error
2081 recovery, and the lookahead rejected during the second part.
2083 - Semicolon changes:
2084 . Stray semicolons are no longer allowed at the start of a grammar.
2085 . Semicolons are now required after in-grammar declarations.
2087 - Unescaped newlines are no longer allowed in character constants or
2088 string literals. They were never portable, and GCC 3.4.0 has
2089 dropped support for them. Better diagnostics are now generated if
2090 forget a closing quote.
2092 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
2096 - GLR grammars now support locations.
2098 - New directive: %initial-action.
2099 This directive allows the user to run arbitrary code (including
2100 initializing @$) from yyparse before parsing starts.
2102 - A new directive "%expect-rr N" specifies the expected number of
2103 reduce/reduce conflicts in GLR parsers.
2105 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
2106 This is a GNU extension.
2108 - The option "--report=lookahead" was changed to "--report=look-ahead".
2109 [However, this was changed back after 2.3.]
2111 - Experimental %destructor support has been added to lalr1.cc.
2113 - New configure option --disable-yacc, to disable installation of the
2114 yacc command and -ly library introduced in 1.875 for POSIX conformance.
2118 - For now, %expect-count violations are now just warnings, not errors.
2119 This is for compatibility with Bison 1.75 and earlier (when there are
2120 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
2121 are too many or too few shift/reduce conflicts). However, in future
2122 versions of Bison we plan to improve the %expect machinery so that
2123 these violations will become errors again.
2125 - Within Bison itself, numbers (e.g., goto numbers) are no longer
2126 arbitrarily limited to 16-bit counts.
2128 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
2130 * Changes in version 1.875, 2003-01-01:
2132 ** The documentation license has been upgraded to version 1.2
2133 of the GNU Free Documentation License.
2135 ** syntax error processing
2137 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
2138 locations too. This fixes bugs in error-location computation.
2141 It is now possible to reclaim the memory associated to symbols
2142 discarded during error recovery. This feature is still experimental.
2145 This new directive is preferred over YYERROR_VERBOSE.
2147 - #defining yyerror to steal internal variables is discouraged.
2148 It is not guaranteed to work forever.
2150 ** POSIX conformance
2152 - Semicolons are once again optional at the end of grammar rules.
2153 This reverts to the behavior of Bison 1.33 and earlier, and improves
2154 compatibility with Yacc.
2156 - "parse error" -> "syntax error"
2157 Bison now uniformly uses the term "syntax error"; formerly, the code
2158 and manual sometimes used the term "parse error" instead. POSIX
2159 requires "syntax error" in diagnostics, and it was thought better to
2162 - The documentation now emphasizes that yylex and yyerror must be
2163 declared before use. C99 requires this.
2165 - Bison now parses C99 lexical constructs like UCNs and
2166 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
2168 - File names are properly escaped in C output. E.g., foo\bar.y is
2169 output as "foo\\bar.y".
2171 - Yacc command and library now available
2172 The Bison distribution now installs a "yacc" command, as POSIX requires.
2173 Also, Bison now installs a small library liby.a containing
2174 implementations of Yacc-compatible yyerror and main functions.
2175 This library is normally not useful, but POSIX requires it.
2177 - Type clashes now generate warnings, not errors.
2179 - If the user does not define YYSTYPE as a macro, Bison now declares it
2180 using typedef instead of defining it as a macro.
2181 For consistency, YYLTYPE is also declared instead of defined.
2183 ** Other compatibility issues
2185 - %union directives can now have a tag before the "{", e.g., the
2186 directive "%union foo {...}" now generates the C code
2187 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
2188 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
2189 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
2190 This is for compatibility with both Yacc and Bison 1.35.
2192 - ";" is output before the terminating "}" of an action, for
2193 compatibility with Bison 1.35.
2195 - Bison now uses a Yacc-style format for conflict reports, e.g.,
2196 "conflicts: 2 shift/reduce, 1 reduce/reduce".
2198 - "yystype" and "yyltype" are now obsolescent macros instead of being
2199 typedefs or tags; they are no longer documented and are planned to be
2200 withdrawn in a future release.
2205 Users of Bison have to decide how they handle the portability of the
2208 - "parsing stack overflow..." -> "parser stack overflow"
2209 GLR parsers now report "parser stack overflow" as per the Bison manual.
2211 ** %parse-param and %lex-param
2212 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
2213 additional context to yyparse and yylex. They suffer from several
2216 - a single argument only can be added,
2217 - their types are weak (void *),
2218 - this context is not passed to ancillary functions such as yyerror,
2219 - only yacc.c parsers support them.
2221 The new %parse-param/%lex-param directives provide a more precise control.
2224 %parse-param {int *nastiness}
2225 %lex-param {int *nastiness}
2226 %parse-param {int *randomness}
2228 results in the following signatures:
2230 int yylex (int *nastiness);
2231 int yyparse (int *nastiness, int *randomness);
2233 or, if both %pure-parser and %locations are used:
2235 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
2236 int yyparse (int *nastiness, int *randomness);
2238 ** Bison now warns if it detects conflicting outputs to the same file,
2239 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
2240 that command outputs both code and header to foo.h.
2242 ** #line in output files
2243 - --no-line works properly.
2245 ** Bison can no longer be built by a K&R C compiler; it requires C89 or
2246 later to be built. This change originally took place a few versions
2247 ago, but nobody noticed until we recently asked someone to try
2248 building Bison with a K&R C compiler.
2250 * Changes in version 1.75, 2002-10-14:
2252 ** Bison should now work on 64-bit hosts.
2254 ** Indonesian translation thanks to Tedi Heriyanto.
2257 Fix spurious parse errors.
2260 Some people redefine yyerror to steal yyparse' private variables.
2261 Reenable this trick until an official feature replaces it.
2264 In agreement with POSIX and with other Yaccs, leaving a default
2265 action is valid when $$ is untyped, and $1 typed:
2269 but the converse remains an error:
2273 ** Values of mid-rule actions
2276 foo: { ... } { $$ = $1; } ...
2278 was incorrectly rejected: $1 is defined in the second mid-rule
2279 action, and is equal to the $$ of the first mid-rule action.
2281 * Changes in version 1.50, 2002-10-04:
2286 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
2287 almost any context-free grammar, ambiguous or not. The new declarations
2288 %dprec and %merge on grammar rules allow parse-time resolution of
2289 ambiguities. Contributed by Paul Hilfinger.
2291 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
2292 like the Alpha, so please stick to 32-bit hosts for now.
2295 When not in Yacc compatibility mode, when the output file was not
2296 specified, running "bison foo/bar.y" created "foo/bar.c". It
2297 now creates "bar.c".
2300 The undefined token was systematically mapped to 2 which prevented
2301 the use of 2 by the user. This is no longer the case.
2303 ** Unknown token numbers
2304 If yylex returned an out of range value, yyparse could die. This is
2308 According to POSIX, the error token must be 256.
2309 Bison extends this requirement by making it a preference: *if* the
2310 user specified that one of her tokens is numbered 256, then error
2311 will be mapped onto another number.
2313 ** Verbose error messages
2314 They no longer report "..., expecting error or..." for states where
2315 error recovery is possible.
2318 Defaults to "$end" instead of "$".
2320 ** Error recovery now conforms to documentation and to POSIX
2321 When a Bison-generated parser encounters a syntax error, it now pops
2322 the stack until it finds a state that allows shifting the error
2323 token. Formerly, it popped the stack until it found a state that
2324 allowed some non-error action other than a default reduction on the
2325 error token. The new behavior has long been the documented behavior,
2326 and has long been required by POSIX. For more details, please see
2327 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
2328 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
2331 Popped tokens and nonterminals are now reported.
2334 Larger grammars are now supported (larger token numbers, larger grammar
2335 size (= sum of the LHS and RHS lengths), larger LALR tables).
2336 Formerly, many of these numbers ran afoul of 16-bit limits;
2337 now these limits are 32 bits on most hosts.
2339 ** Explicit initial rule
2340 Bison used to play hacks with the initial rule, which the user does
2341 not write. It is now explicit, and visible in the reports and
2345 Before, Bison reported the useless rules, but, although not used,
2346 included them in the parsers. They are now actually removed.
2348 ** Useless rules, useless nonterminals
2349 They are now reported, as a warning, with their locations.
2351 ** Rules never reduced
2352 Rules that can never be reduced because of conflicts are now
2355 ** Incorrect "Token not used"
2356 On a grammar such as
2358 %token useless useful
2360 exp: '0' %prec useful;
2362 where a token was used to set the precedence of the last rule,
2363 bison reported both "useful" and "useless" as useless tokens.
2365 ** Revert the C++ namespace changes introduced in 1.31
2366 as they caused too many portability hassles.
2368 ** Default locations
2369 By an accident of design, the default computation of @$ was
2370 performed after another default computation was performed: @$ = @1.
2371 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
2372 the computation of @$.
2374 ** Token end-of-file
2375 The token end of file may be specified by the user, in which case,
2376 the user symbol is used in the reports, the graphs, and the verbose
2377 error messages instead of "$end", which remains being the default.
2381 %token MYEOF 0 "end of file"
2384 This old option, which has been broken for ages, is removed.
2387 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
2388 Croatian, thanks to Denis Lackovic.
2390 ** Incorrect token definitions
2393 bison used to output
2396 ** Token definitions as enums
2397 Tokens are output both as the traditional #define's, and, provided
2398 the compiler supports ANSI C or is a C++ compiler, as enums.
2399 This lets debuggers display names instead of integers.
2402 In addition to --verbose, bison supports --report=THINGS, which
2403 produces additional information:
2405 complete the core item sets with their closure
2406 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
2407 explicitly associate lookahead tokens to items
2409 describe shift/reduce conflicts solving.
2410 Bison used to systematically output this information on top of
2411 the report. Solved conflicts are now attached to their states.
2414 Previous versions don't complain when there is a type clash on
2415 the default action if the rule has a mid-rule action, such as in:
2423 ** GNU M4 is now required when using Bison.
2425 * Changes in version 1.35, 2002-03-25:
2428 Some projects use Bison's C parser with C++ compilers, and define
2429 YYSTYPE as a class. The recent adjustment of C parsers for data
2430 alignment and 64 bit architectures made this impossible.
2432 Because for the time being no real solution for C++ parser
2433 generation exists, kludges were implemented in the parser to
2434 maintain this use. In the future, when Bison has C++ parsers, this
2435 kludge will be disabled.
2437 This kludge also addresses some C++ problems when the stack was
2440 * Changes in version 1.34, 2002-03-12:
2442 ** File name clashes are detected
2443 $ bison foo.y -d -o foo.x
2444 fatal error: header and parser would both be named "foo.x"
2446 ** A missing ";" at the end of a rule triggers a warning
2447 In accordance with POSIX, and in agreement with other
2448 Yacc implementations, Bison will mandate this semicolon in the near
2449 future. This eases the implementation of a Bison parser of Bison
2450 grammars by making this grammar LALR(1) instead of LR(2). To
2451 facilitate the transition, this release introduces a warning.
2453 ** Revert the C++ namespace changes introduced in 1.31, as they caused too
2454 many portability hassles.
2456 ** DJGPP support added.
2458 ** Fix test suite portability problems.
2460 * Changes in version 1.33, 2002-02-07:
2463 Groff could not be compiled for the definition of size_t was lacking
2464 under some conditions.
2469 * Changes in version 1.32, 2002-01-23:
2471 ** Fix Yacc output file names
2473 ** Portability fixes
2475 ** Italian, Dutch translations
2477 * Changes in version 1.31, 2002-01-14:
2481 ** GNU Gettext and %expect
2482 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
2483 Bison dies on incorrect %expectations, we fear there will be
2484 too many bug reports for Gettext, so _for the time being_, %expect
2485 does not trigger an error when the input file is named "plural.y".
2487 ** Use of alloca in parsers
2488 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
2489 malloc exclusively. Since 1.29, but was not NEWS'ed.
2491 alloca is used only when compiled with GCC, to avoid portability
2494 ** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
2496 ** When the generated parser lacks debugging code, YYDEBUG is now 0
2497 (as POSIX requires) instead of being undefined.
2500 Bison has always permitted actions such as { $$ = $1 }: it adds the
2501 ending semicolon. Now if in Yacc compatibility mode, the semicolon
2502 is no longer output: one has to write { $$ = $1; }.
2504 ** Better C++ compliance
2505 The output parsers try to respect C++ namespaces.
2506 [This turned out to be a failed experiment, and it was reverted later.]
2509 Fixed bugs when reporting useless nonterminals.
2512 The parsers work properly on 64 bit hosts.
2515 Some calls to strerror resulted in scrambled or missing error messages.
2518 When the number of shift/reduce conflicts is correct, don't issue
2521 ** The verbose report includes the rule line numbers.
2523 ** Rule line numbers are fixed in traces.
2525 ** Swedish translation
2528 Verbose parse error messages from the parsers are better looking.
2529 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
2530 Now: parse error: unexpected '/', expecting "number" or '-' or '('
2532 ** Fixed parser memory leaks.
2533 When the generated parser was using malloc to extend its stacks, the
2534 previous allocations were not freed.
2536 ** Fixed verbose output file.
2537 Some newlines were missing.
2538 Some conflicts in state descriptions were missing.
2540 ** Fixed conflict report.
2541 Option -v was needed to get the result.
2545 Mismatches are errors, not warnings.
2547 ** Fixed incorrect processing of some invalid input.
2549 ** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
2551 ** Fixed some typos in the documentation.
2553 ** %token MY_EOF 0 is supported.
2554 Before, MY_EOF was silently renumbered as 257.
2556 ** doc/refcard.tex is updated.
2558 ** %output, %file-prefix, %name-prefix.
2562 New, aliasing "--output-file".
2564 * Changes in version 1.30, 2001-10-26:
2566 ** "--defines" and "--graph" have now an optional argument which is the
2567 output file name. "-d" and "-g" do not change; they do not take any
2570 ** "%source_extension" and "%header_extension" are removed, failed
2573 ** Portability fixes.
2575 * Changes in version 1.29, 2001-09-07:
2577 ** The output file does not define const, as this caused problems when used
2578 with common autoconfiguration schemes. If you still use ancient compilers
2579 that lack const, compile with the equivalent of the C compiler option
2580 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
2582 ** Added "-g" and "--graph".
2584 ** The Bison manual is now distributed under the terms of the GNU FDL.
2586 ** The input and the output files has automatically a similar extension.
2588 ** Russian translation added.
2590 ** NLS support updated; should hopefully be less troublesome.
2592 ** Added the old Bison reference card.
2594 ** Added "--locations" and "%locations".
2596 ** Added "-S" and "--skeleton".
2598 ** "%raw", "-r", "--raw" is disabled.
2600 ** Special characters are escaped when output. This solves the problems
2601 of the #line lines with path names including backslashes.
2604 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
2605 "%debug", "%source_extension" and "%header_extension".
2608 Automatic location tracking.
2610 * Changes in version 1.28, 1999-07-06:
2612 ** Should compile better now with K&R compilers.
2616 ** Fixed a problem with escaping the double quote character.
2618 ** There is now a FAQ.
2620 * Changes in version 1.27:
2622 ** The make rule which prevented bison.simple from being created on
2623 some systems has been fixed.
2625 * Changes in version 1.26:
2627 ** Bison now uses Automake.
2629 ** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
2631 ** Token numbers now start at 257 as previously documented, not 258.
2633 ** Bison honors the TMPDIR environment variable.
2635 ** A couple of buffer overruns have been fixed.
2637 ** Problems when closing files should now be reported.
2639 ** Generated parsers should now work even on operating systems which do
2640 not provide alloca().
2642 * Changes in version 1.25, 1995-10-16:
2644 ** Errors in the input grammar are not fatal; Bison keeps reading
2645 the grammar file, and reports all the errors found in it.
2647 ** Tokens can now be specified as multiple-character strings: for
2648 example, you could use "<=" for a token which looks like <=, instead
2649 of choosing a name like LESSEQ.
2651 ** The %token_table declaration says to write a table of tokens (names
2652 and numbers) into the parser file. The yylex function can use this
2653 table to recognize multiple-character string tokens, or for other
2656 ** The %no_lines declaration says not to generate any #line preprocessor
2657 directives in the parser file.
2659 ** The %raw declaration says to use internal Bison token numbers, not
2660 Yacc-compatible token numbers, when token names are defined as macros.
2662 ** The --no-parser option produces the parser tables without including
2663 the parser engine; a project can now use its own parser engine.
2664 The actions go into a separate file called NAME.act, in the form of
2665 a switch statement body.
2667 * Changes in version 1.23:
2669 The user can define YYPARSE_PARAM as the name of an argument to be
2670 passed into yyparse. The argument should have type void *. It should
2671 actually point to an object. Grammar actions can access the variable
2672 by casting it to the proper pointer type.
2674 Line numbers in output file corrected.
2676 * Changes in version 1.22:
2678 --help option added.
2680 * Changes in version 1.20:
2682 Output file does not redefine const for C++.
2686 Copyright (C) 1995-2013 Free Software Foundation, Inc.
2688 This file is part of Bison, the GNU Parser Generator.
2690 This program is free software: you can redistribute it and/or modify
2691 it under the terms of the GNU General Public License as published by
2692 the Free Software Foundation, either version 3 of the License, or
2693 (at your option) any later version.
2695 This program is distributed in the hope that it will be useful,
2696 but WITHOUT ANY WARRANTY; without even the implied warranty of
2697 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2698 GNU General Public License for more details.
2700 You should have received a copy of the GNU General Public License
2701 along with this program. If not, see <http://www.gnu.org/licenses/>.
2703 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2704 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2705 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2706 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2707 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2708 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2709 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2710 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2711 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2712 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2713 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2714 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2715 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
2716 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
2717 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
2718 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2719 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
2720 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
2721 LocalWords: TOK calc yyo fval Wconflicts parsers yystackp yyval yynerrs
2722 LocalWords: Théophile Ranquet Santet fno fnone stype associativity Tolmer
2723 LocalWords: Wprecedence Rassoul Wempty Paolo Bonzini parser's Michiel loc
2724 LocalWords: redeclaration sval fcaret reentrant XSLT xsl Wmaybe yyvsp Tedi
2725 LocalWords: pragmas noreturn untyped Rozenman unexpanded Wojciech Polak
2726 LocalWords: Alexandre MERCHANTABILITY yytype