api.tokens.prefix -> api.token.prefix
[bison.git] / NEWS
CommitLineData
ed7658fe 1GNU Bison NEWS
3af4feb2 2
52718713 3* Noteworthy changes in release ?.? (????-??-??) [?]
74553c98 4
47db7ed1
AD
5** Incompatible changes
6
7*** Obsolete features
8
de62ae86 9 Support for YYFAIL is removed (deprecated in Bison 2.4.2).
5f819b2f 10 Support for yystype and yyltype (instead of YYSTYPE and YYLTYPE)
de62ae86
AD
11 is removed (deprecated in Bison 1.875).
12 Support for YYPARSE_PARAM is removed (deprecated in Bison 1.875).
47db7ed1 13
73370a9d
VS
14** Warnings
15
1048a1c9 16*** Enhancements of the -Werror option
518e8830 17
1048a1c9
AD
18 The -Werror=CATEGORY option is now recognized, and will treat specified
19 warnings as errors. The warnings need not have been explictly activated
20 using the -W option, this is similar to what gcc 4.7 does.
21
22 For example, given the following command line, Bison will treat both
23 warnings related to POSIX Yacc incompatiblities and S/R conflicts as
24 errors (and only those):
25
26 $ bison -Werror=yacc,error=conflicts-sr input.y
27
28 If no categories are specified, -Werror will make all active warnings into
29 errors. For example, the following line does the same the previous example:
30
31 $ bison -Werror -Wnone -Wyacc -Wconflicts-sr input.y
32
33 (By default -Wconflicts-sr,conflicts-rr,deprecated,other is enabled.)
34
35 Note that the categories in this -Werror option may not be prefixed with
36 "no-". However, -Wno-error[=CATEGORY] is valid.
37
38 Note that -y enables -Werror=yacc. Therefore it is now possible to require
39 Yacc-like behavior (e.g., always generate y.tab.c), but to report
40 incompatibilities as warnings: "-y -Wno-error=yacc".
518e8830 41
46bdb8ec 42*** The display of warnings is now richer
73370a9d 43
46bdb8ec 44 The option that controls a given warning is now displayed:
73370a9d 45
46bdb8ec 46 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
73370a9d 47
46bdb8ec
TR
48 In the case of warnings treated as errors, the prefix is changed from
49 "warning: " to "error: ", and the suffix is displayed, in a manner similar
50 to gcc, as [-Werror=CATEGORY].
1048a1c9 51
46bdb8ec
TR
52 For instance, where the previous version of Bison would report (and exit
53 with failure):
1048a1c9
AD
54
55 bison: warnings being treated as errors
46bdb8ec 56 input.y:1.1: warning: stray ',' treated as white space
1048a1c9 57
46bdb8ec 58 it now reports:
1048a1c9 59
1048a1c9
AD
60 input.y:1.1: error: stray ',' treated as white space [-Werror=other]
61
62*** Deprecated constructs
63
64 The new 'deprecated' warning category flags obsolete constructs whose
65 support will be discontinued. It is enabled by default. These warnings
66 used to be reported as 'other' warnings.
67
73370a9d 68*** Useless semantic types
9641b918
VS
69
70 Bison now warns about useless (uninhabited) semantic types. Since
71 semantic types are not declared to Bison (they are defined in the opaque
72 %union structure), it is %printer/%destructor directives about useless
73 types that trigger the warning:
74
75 %token <type1> term
76 %type <type2> nterm
77 %printer {} <type1> <type3>
78 %destructor {} <type2> <type4>
79 %%
80 nterm: term { $$ = $1; };
81
82 3.28-34: warning: type <type3> is used, but is not associated to any symbol
83 4.28-34: warning: type <type4> is used, but is not associated to any symbol
84
31557b9e 85*** Undefined but unused symbols
b921d92f 86
31557b9e
AD
87 Bison used to raise an error for undefined symbols that are not used in
88 the grammar. This is now only a warning.
b921d92f
VS
89
90 %printer {} symbol1
91 %destructor {} symbol2
31557b9e 92 %type <type> symbol3
b921d92f
VS
93 %%
94 exp: "a";
95
73370a9d 96*** Useless destructors or printers
ea9a35c6
VS
97
98 Bison now warns about useless destructors or printers. In the following
99 example, the printer for <type1>, and the destructor for <type2> are
100 useless: all symbols of <type1> (token1) already have a printer, and all
101 symbols of type <type2> (token2) already have a destructor.
102
103 %token <type1> token1
104 <type2> token2
105 <type3> token3
106 <type4> token4
107 %printer {} token1 <type1> <type3>
108 %destructor {} token2 <type2> <type4>
109
d87ea54c
AD
110*** Conflicts
111
112 The warnings and error messages about shift/reduce and reduce/reduce
113 conflicts have been normalized. For instance on the following foo.y file:
114
115 %glr-parser
116 %%
117 exp: exp '+' exp | '0' | '0';
118
119 compare the previous version of bison:
120
121 $ bison foo.y
122 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
123 $ bison -Werror foo.y
124 bison: warnings being treated as errors
125 foo.y: conflicts: 1 shift/reduce, 2 reduce/reduce
126
127 with the new behavior:
128
129 $ bison foo.y
130 foo.y: warning: 1 shift/reduce conflict [-Wconflicts-sr]
131 foo.y: warning: 2 reduce/reduce conflicts [-Wconflicts-rr]
132 $ bison -Werror foo.y
9503b0a4
TR
133 foo.y: error: 1 shift/reduce conflict [-Werror=conflicts-sr]
134 foo.y: error: 2 reduce/reduce conflicts [-Werror=conflicts-rr]
d87ea54c
AD
135
136 When %expect or %expect-rr is used, such as with bar.y:
137
138 %expect 0
139 %glr-parser
140 %%
141 exp: exp '+' exp | '0' | '0';
142
143 Former behavior:
144
145 $ bison bar.y
146 bar.y: conflicts: 1 shift/reduce, 2 reduce/reduce
147 bar.y: expected 0 shift/reduce conflicts
148 bar.y: expected 0 reduce/reduce conflicts
149
150 New one:
151
152 $ bison bar.y
9503b0a4
TR
153 bar.y: error: shift/reduce conflicts: 1 found, 0 expected
154 bar.y: error: reduce/reduce conflicts: 2 found, 0 expected
d87ea54c 155
2055a44e
AD
156** Additional yylex/yyparse arguments
157
6dc4663d
AD
158 The new directive %param declares additional arguments to both yylex and
159 yyparse. The %lex-param, %parse-param, and %param directives support one
160 or more arguments. Instead of
2055a44e 161
e436fa67
AD
162 %lex-param {arg1_type *arg1}
163 %lex-param {arg2_type *arg2}
164 %parse-param {arg1_type *arg1}
165 %parse-param {arg2_type *arg2}
2055a44e
AD
166
167 one may now declare
168
e436fa67 169 %param {arg1_type *arg1} {arg2_type *arg2}
2055a44e
AD
170
171** Java skeleton improvements
0ea583d2 172
1f77b2e0
AD
173 The constants for token names were moved to the Lexer interface. Also, it
174 is possible to add code to the parser's constructors using "%code init"
175 and "%define init_throws".
0ea583d2 176
34904c57
AD
177** C++ skeletons improvements
178
e4c0985b 179*** The parser header is no longer mandatory (lalr1.cc, glr.cc)
93549bcd 180
e4c0985b
AD
181 Using %defines is now optional. Without it, the needed support classes
182 are defined in the generated parser, instead of additional files (such as
183 location.hh, position.hh and stack.hh).
93549bcd 184
e4c0985b 185*** Locations are no longer mandatory (lalr1.cc, glr.cc)
34904c57
AD
186
187 Both lalr1.cc and glr.cc no longer require %location.
188
189*** syntax_error exception (lalr1.cc)
a6552c5d
AD
190
191 The C++ parser features a syntax_error exception, which can be
192 thrown from the scanner or from user rules to raise syntax errors.
193 This facilitates reporting errors caught in sub-functions (e.g.,
194 rejecting too large integral literals from a conversion function
195 used by the scanner, or rejecting invalid combinations from a
196 factory invoked by the user actions).
197
2a6b66c5 198** Variable api.token.prefix
99c08fb6 199
2a6b66c5 200 The variable api.token.prefix changes the way tokens are identified in
99c08fb6
AD
201 the generated files. This is especially useful to avoid collisions
202 with identifiers in the target language. For instance
203
e436fa67 204 %token FILE for ERROR
2a6b66c5 205 %define api.token.prefix "TOK_"
e436fa67
AD
206 %%
207 start: FILE for ERROR;
99c08fb6
AD
208
209 will generate the definition of the symbols TOK_FILE, TOK_for, and
210 TOK_ERROR in the generated sources. In particular, the scanner must
211 use these prefixed token names, although the grammar itself still
212 uses the short names (as in the sample rule given above).
213
67501061
AD
214** Variable api.namespace
215
1f77b2e0 216 The 'namespace' variable is renamed 'api.namespace'. Backward
67501061
AD
217 compatibility is ensured, but upgrading is recommended.
218
31b850d2
AD
219** Variable parse.error
220
1f77b2e0
AD
221 This variable controls the verbosity of error messages. The use of the
222 %error-verbose directive is deprecated in favor of "%define parse.error
223 verbose".
31b850d2 224
ca2a6d15
PH
225** Semantic predicates
226
1f77b2e0
AD
227 The new, experimental, semantic-predicate feature allows actions of the
228 form "%?{ BOOLEAN-EXPRESSION }", which cause syntax errors (as for
ca2a6d15 229 YYERROR) if the expression evaluates to 0, and are evaluated immediately
1f77b2e0
AD
230 in GLR parsers, rather than being deferred. The result is that they allow
231 the programmer to prune possible parses based on the values of run-time
232 expressions.
ca2a6d15 233
d1400569
AD
234** The directive %expect-rr is now an error in non GLR mode
235
236 It used to be an error only if used in non GLR mode, _and_ if there are
237 reduce/reduce conflicts.
238
e436fa67
AD
239* Noteworthy changes in release ?.? (????-??-??) [?]
240
6b4cb804
AD
241** Bug fixes
242
a1a77e1f 243 Bugs and portability issues in the test suite have been fixed.
6b4cb804
AD
244
245 Some errors in translations have been addressed, and --help now directs
246 users to the appropriate place to report them.
247
248 Stray Info files shipped by accident are removed.
249
250 Incorrect definitions of YY_, issued by yacc.c when no parser header is
251 generated, are removed.
9c26b8fc 252
a2b3f101
TR
253 All the generated headers are self-contained.
254
5a05f42e 255** Changes in the format of error messages
cbaea010
TR
256
257 This used to be the format of many error reports:
258
259 foo.y:5.10-24: result type clash on merge function 'merge': <t3> != <t2>
260 foo.y:4.13-27: previous declaration
261
262 It is now:
263
264 foo.y:5.10-25: result type clash on merge function 'merge': <t3> != <t2>
265 foo.y:4.13-27: previous declaration
266
c9d5bcc9
AD
267** Header guards (yacc.c, glr.c, glr.cc)
268
269 In order to avoid collisions, the header guards are now
270 YY_<PREFIX>_<FILE>_INCLUDED, instead of merely <PREFIX>_<FILE>.
271 For instance the header generated from
272
273 %define api.prefix "calc"
274 %defines "lib/parse.h"
275
276 will use YY_CALC_LIB_PARSE_H_INCLUDED as guard.
277
d3e4409a
AD
278** Exception safety (lalr1.cc)
279
280 The parse function now catches exceptions, uses the %destructors to
281 release memory (the lookahead symbol and the symbols pushed on the stack)
282 before rethrowing the exception.
283
284 This feature is somewhat experimental. User feedback would be
285 appreciated.
286
c12c4c50 287** Fix compiler warnings in the generated parser (yacc.c, glr.c)
321d3e35
AD
288
289 The compilation of pure parsers (%define api.pure) can trigger GCC
290 warnings such as:
291
292 input.c: In function 'yyparse':
293 input.c:1503:12: warning: 'yylval' may be used uninitialized in this
294 function [-Wmaybe-uninitialized]
295 *++yyvsp = yylval;
296 ^
297
298 This is now fixed; pragmas to avoid these warnings are no longer needed.
299
c12c4c50
AD
300 Warnings from clang ("equality comparison with extraneous parentheses" and
301 "function declared 'noreturn' should not return") have also been
302 addressed.
303
db8ab2be
AD
304** New %define variable: api.location.type (glr.cc, lalr1.cc)
305
306 The %define variable api.location.type defines the name of the type to use
307 for locations. When defined, Bison no longer generates the position.hh
308 and location.hh files, nor does the parser will include them: the user is
309 then responsible to define her type.
310
311 This can be used in programs with several parsers to factor their location
f6b561d9
AD
312 and position files: let one of them generate them, and the others just use
313 them.
db8ab2be
AD
314
315 This feature was actually introduced, but not documented, in Bison 2.5,
316 under the name "location_type" (which is maintained for backward
317 compatibility).
318
e1eeecd3 319* Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
9c26b8fc 320
43ca8040
AD
321** Bug fixes
322
323 Buffer overruns, complaints from Flex, and portability issues in the test
324 suite have been fixed.
325
c9d546b2
AD
326** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
327
328 Trailing end-of-lines in %parse-param or %lex-param would result in
329 invalid C++. This is fixed.
9c26b8fc 330
dcd5344d
AD
331** Spurious spaces and end-of-lines
332
333 The generated files no longer end (nor start) with empty lines.
334
77b214ef 335* Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
a4107f24 336
8617d87e
AD
337 Bison no longer executes user-specified M4 code when processing a grammar.
338
e20e6a50
AD
339** Future Changes
340
341 In addition to the removal of the features announced in Bison 2.6, the
342 next major release will remove the "Temporary hack for adding a semicolon
343 to the user action", as announced in the release 2.5. Instead of:
344
345 exp: exp "+" exp { $$ = $1 + $3 };
346
347 write:
348
349 exp: exp "+" exp { $$ = $1 + $3; };
350
8617d87e
AD
351** Bug fixes
352
0e164d43
AD
353*** Type names are now properly escaped.
354
355*** glr.cc: set_debug_level and debug_level work as expected.
a4107f24 356
26313726
AD
357*** Stray @ or $ in actions
358
359 While Bison used to warn about stray $ or @ in action rules, it did not
360 for other actions such as printers, destructors, or initial actions. It
361 now does.
362
cd735a8c 363** Type names in actions
4982f078
AD
364
365 For consistency with rule actions, it is now possible to qualify $$ by a
cd735a8c 366 type-name in destructors, printers, and initial actions. For instance:
4982f078
AD
367
368 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
369
370 will display two values for each typed and untyped symbol (provided
cd735a8c 371 that YYSTYPE has both "ival" and "fval" fields).
60aa04a2 372
1505e8bb 373* Noteworthy changes in release 2.6 (2012-07-19) [stable]
0f11eec2 374
9553083c
AD
375** Future changes:
376
55d1006f
AD
377 The next major release of Bison will drop support for the following
378 deprecated features. Please report disagreements to bug-bison@gnu.org.
0f11eec2 379
aaf61036 380*** K&R C parsers
55d1006f
AD
381
382 Support for generating parsers in K&R C will be removed. Parsers
242cc08e 383 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
55d1006f
AD
384 compilers.
385
258cddbc 386*** Features deprecated since Bison 1.875
0f11eec2 387
258cddbc
AD
388 The definitions of yystype and yyltype will be removed; use YYSTYPE and
389 YYLTYPE.
0f11eec2 390
258cddbc
AD
391 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
392 %lex-param, will no longer be supported.
393
394 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
395 %error-verbose.
55d1006f
AD
396
397*** The generated header will be included (yacc.c)
0f11eec2
AD
398
399 Instead of duplicating the content of the generated header (definition of
55d1006f
AD
400 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
401 as is already the case for GLR or C++ parsers. This change is deferred
402 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
403 it.
0f11eec2 404
c2425191 405** Generated Parser Headers
56ca3d8f 406
258cddbc 407*** Guards (yacc.c, glr.c, glr.cc)
c3e9f08f
AD
408
409 The generated headers are now guarded, as is already the case for C++
242cc08e 410 parsers (lalr1.cc). For instance, with --defines=foo.h:
c3e9f08f 411
e29f0771
AD
412 #ifndef YY_FOO_H
413 # define YY_FOO_H
414 ...
415 #endif /* !YY_FOO_H */
c3e9f08f 416
258cddbc 417*** New declarations (yacc.c, glr.c)
56ca3d8f
AD
418
419 The generated header now declares yydebug and yyparse. Both honor
420 --name-prefix=bar_, and yield
421
e29f0771 422 int bar_parse (void);
56ca3d8f
AD
423
424 rather than
425
e29f0771
AD
426 #define yyparse bar_parse
427 int yyparse (void);
56ca3d8f
AD
428
429 in order to facilitate the inclusion of several parser headers inside a
430 single compilation unit.
c3e9f08f 431
258cddbc
AD
432*** Exported symbols in C++
433
434 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
435 header, are removed, as they prevent the possibility of including several
436 generated headers from a single compilation unit.
437
694af10c
AD
438*** YYLSP_NEEDED
439
440 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
441 longer defined.
442
4b3847c3
AD
443** New %define variable: api.prefix
444
445 Now that the generated headers are more complete and properly protected
446 against multiple inclusions, constant names, such as YYSTYPE are a
447 problem. While yyparse and others are properly renamed by %name-prefix,
448 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
449 would introduce backward compatibility issues in projects not expecting
450 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
451 it is deprecated in favor of a new %define variable: api.prefix.
452
453 The following examples compares both:
454
455 %name-prefix "bar_" | %define api.prefix "bar_"
456 %token <ival> FOO %token <ival> FOO
457 %union { int ival; } %union { int ival; }
458 %% %%
459 exp: 'a'; exp: 'a';
460
461 bison generates:
462
463 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
464 # define BAR_FOO_H # define BAR_FOO_H
465
466 /* Enabling traces. */ /* Enabling traces. */
5f108727
AD
467 # ifndef YYDEBUG | # ifndef BAR_DEBUG
468 > # if defined YYDEBUG
469 > # if YYDEBUG
470 > # define BAR_DEBUG 1
471 > # else
472 > # define BAR_DEBUG 0
473 > # endif
474 > # else
475 # define YYDEBUG 0 | # define BAR_DEBUG 0
476 > # endif
477 # endif | # endif
478
479 # if YYDEBUG | # if BAR_DEBUG
4b3847c3
AD
480 extern int bar_debug; extern int bar_debug;
481 # endif # endif
482
483 /* Tokens. */ /* Tokens. */
484 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
485 # define YYTOKENTYPE | # define BAR_TOKENTYPE
486 enum yytokentype { | enum bar_tokentype {
487 FOO = 258 FOO = 258
488 }; };
489 # endif # endif
490
491 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
492 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
493 typedef union YYSTYPE | typedef union BAR_STYPE
494 { {
495 int ival; int ival;
496 } YYSTYPE; | } BAR_STYPE;
497 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
498 #endif #endif
499
500 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
501
502 int bar_parse (void); int bar_parse (void);
503
504 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
505
dfaac272 506* Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
df6e3db0 507
debe2c03 508** Future changes:
765e1bd4 509
e4ab1254 510 The next major release will drop support for generating parsers in K&R C.
041308d0 511
466b4cf2 512** yacc.c: YYBACKUP works as expected.
ef51bfa7 513
d834eca0 514** glr.c improvements:
041308d0 515
d834eca0 516*** Location support is eliminated when not requested:
041308d0 517
e4ab1254
AD
518 GLR parsers used to include location-related code even when locations were
519 not requested, and therefore not even usable.
378e917c 520
d834eca0 521*** __attribute__ is preserved:
d115aad9 522
e4ab1254
AD
523 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
524 when -std is passed to GCC).
041308d0 525
466b4cf2 526** lalr1.java: several fixes:
041308d0 527
e4ab1254
AD
528 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
529 first token leads to a syntax error. Some minor clean ups.
041308d0 530
22172d47 531** Changes for C++:
ef51bfa7 532
22172d47 533*** C++11 compatibility:
ef51bfa7 534
e4ab1254
AD
535 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
536 or higher.
936c88d1 537
22172d47
AD
538*** Header guards
539
540 The header files such as "parser.hh", "location.hh", etc. used a constant
541 name for preprocessor guards, for instance:
542
e29f0771
AD
543 #ifndef BISON_LOCATION_HH
544 # define BISON_LOCATION_HH
545 ...
546 #endif // !BISON_LOCATION_HH
22172d47
AD
547
548 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
549 case characters are converted to upper case, and series of
550 non-alphanumerical characters are converted to an underscore.
551
552 With "bison -o lang++/parser.cc", "location.hh" would now include:
553
e29f0771
AD
554 #ifndef YY_LANG_LOCATION_HH
555 # define YY_LANG_LOCATION_HH
556 ...
557 #endif // !YY_LANG_LOCATION_HH
22172d47
AD
558
559*** C++ locations:
936c88d1 560
e4ab1254
AD
561 The position and location constructors (and their initialize methods)
562 accept new arguments for line and column. Several issues in the
563 documentation were fixed.
936c88d1 564
466b4cf2
AD
565** liby is no longer asking for "rpl_fprintf" on some platforms.
566
7e508a2b
AD
567** Changes in the manual:
568
569*** %printer is documented
570
e4ab1254
AD
571 The "%printer" directive, supported since at least Bison 1.50, is finally
572 documented. The "mfcalc" example is extended to demonstrate it.
7e508a2b 573
e4ab1254
AD
574 For consistency with the C skeletons, the C++ parsers now also support
575 "yyoutput" (as an alias to "debug_stream ()").
7e508a2b
AD
576
577*** Several improvements have been made:
466b4cf2 578
e4ab1254
AD
579 The layout for grammar excerpts was changed to a more compact scheme.
580 Named references are motivated. The description of the automaton
581 description file (*.output) is updated to the current format. Incorrect
582 index entries were fixed. Some other errors were fixed.
466b4cf2 583
86b08b49
AD
584** Building bison:
585
586*** Conflicting prototypes with recent/modified Flex.
466b4cf2 587
e4ab1254
AD
588 Fixed build problems with the current, unreleased, version of Flex, and
589 some modified versions of 2.5.35, which have modified function prototypes.
466b4cf2 590
8ef26c2a
AD
591*** Warnings during the build procedure have been eliminated.
592
593*** Several portability problems in the test suite have been fixed:
466b4cf2 594
e4ab1254
AD
595 This includes warnings with some compilers, unexpected behavior of tools
596 such as diff, warning messages from the test suite itself, etc.
466b4cf2 597
91aadcc7 598*** The install-pdf target works properly:
8ef26c2a 599
e4ab1254
AD
600 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
601 halts in the middle of its course.
8ef26c2a 602
28801043 603* Changes in version 2.5 (2011-05-14):
50cca368 604
82f3355e
JD
605** Grammar symbol names can now contain non-initial dashes:
606
607 Consistently with directives (such as %error-verbose) and with
608 %define variables (e.g. push-pull), grammar symbol names may contain
609 dashes in any position except the beginning. This is a GNU
610 extension over POSIX Yacc. Thus, use of this extension is reported
611 by -Wyacc and rejected in Yacc mode (--yacc).
612
f1b238df 613** Named references:
66381412
AR
614
615 Historically, Yacc and Bison have supported positional references
616 ($n, $$) to allow access to symbol values from inside of semantic
617 actions code.
618
619 Starting from this version, Bison can also accept named references.
620 When no ambiguity is possible, original symbol names may be used
621 as named references:
622
4b568fc0 623 if_stmt : "if" cond_expr "then" then_stmt ';'
66381412
AR
624 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
625
626 In the more common case, explicit names may be declared:
627
4b568fc0 628 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
66381412
AR
629 { $res = mk_if_stmt($cond, $then, $else); }
630
5b1ff423 631 Location information is also accessible using @name syntax. When
66381412
AR
632 accessing symbol names containing dots or dashes, explicit bracketing
633 ($[sym.1]) must be used.
634
5b1ff423 635 These features are experimental in this version. More user feedback
66381412
AR
636 will help to stabilize them.
637
f1b238df 638** IELR(1) and canonical LR(1):
eb45ef3b
JD
639
640 IELR(1) is a minimal LR(1) parser table generation algorithm. That
641 is, given any context-free grammar, IELR(1) generates parser tables
7262f54f 642 with the full language-recognition power of canonical LR(1) but with
f1b238df
JD
643 nearly the same number of parser states as LALR(1). This reduction
644 in parser states is often an order of magnitude. More importantly,
eb45ef3b
JD
645 because canonical LR(1)'s extra parser states may contain duplicate
646 conflicts in the case of non-LR(1) grammars, the number of conflicts
647 for IELR(1) is often an order of magnitude less as well. This can
648 significantly reduce the complexity of developing of a grammar.
649
650 Bison can now generate IELR(1) and canonical LR(1) parser tables in
651 place of its traditional LALR(1) parser tables, which remain the
652 default. You can specify the type of parser tables in the grammar
653 file with these directives:
654
cf499cff
JD
655 %define lr.type lalr
656 %define lr.type ielr
657 %define lr.type canonical-lr
eb45ef3b 658
7fceb615 659 The default-reduction optimization in the parser tables can also be
e4ab1254
AD
660 adjusted using "%define lr.default-reductions". For details on both
661 of these features, see the new section "Tuning LR" in the Bison
7fceb615 662 manual.
eb45ef3b
JD
663
664 These features are experimental. More user feedback will help to
665 stabilize them.
666
7fceb615 667** LAC (Lookahead Correction) for syntax error handling:
fcf834f9
JD
668
669 Canonical LR, IELR, and LALR can suffer from a couple of problems
670 upon encountering a syntax error. First, the parser might perform
671 additional parser stack reductions before discovering the syntax
7fceb615 672 error. Such reductions can perform user semantic actions that are
fcf834f9
JD
673 unexpected because they are based on an invalid token, and they
674 cause error recovery to begin in a different syntactic context than
675 the one in which the invalid token was encountered. Second, when
7fceb615 676 verbose error messages are enabled (with %error-verbose or the
e4ab1254 677 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
7fceb615
JD
678 syntax error message can both contain invalid tokens and omit valid
679 tokens.
fcf834f9
JD
680
681 The culprits for the above problems are %nonassoc, default
682 reductions in inconsistent states, and parser state merging. Thus,
683 IELR and LALR suffer the most. Canonical LR can suffer only if
684 %nonassoc is used or if default reductions are enabled for
685 inconsistent states.
686
7fceb615
JD
687 LAC is a new mechanism within the parsing algorithm that solves
688 these problems for canonical LR, IELR, and LALR without sacrificing
689 %nonassoc, default reductions, or state merging. When LAC is in
690 use, canonical LR and IELR behave almost exactly the same for both
691 syntactically acceptable and syntactically unacceptable input.
fcf834f9
JD
692 While LALR still does not support the full language-recognition
693 power of canonical LR and IELR, LAC at least enables LALR's syntax
694 error handling to correctly reflect LALR's language-recognition
695 power.
696
697 Currently, LAC is only supported for deterministic parsers in C.
698 You can enable LAC with the following directive:
699
700 %define parse.lac full
701
e4ab1254 702 See the new section "LAC" in the Bison manual for additional
7fceb615 703 details including a few caveats.
fcf834f9
JD
704
705 LAC is an experimental feature. More user feedback will help to
706 stabilize it.
707
d397d9f0 708** %define improvements:
cf499cff 709
f1b238df 710*** Can now be invoked via the command line:
50cca368 711
de5ab940 712 Each of these command-line options
50cca368 713
de5ab940
JD
714 -D NAME[=VALUE]
715 --define=NAME[=VALUE]
716
717 -F NAME[=VALUE]
718 --force-define=NAME[=VALUE]
50cca368
JD
719
720 is equivalent to this grammar file declaration
721
de5ab940 722 %define NAME ["VALUE"]
50cca368 723
de5ab940
JD
724 except that the manner in which Bison processes multiple definitions
725 for the same NAME differs. Most importantly, -F and --force-define
726 quietly override %define, but -D and --define do not. For further
e4ab1254 727 details, see the section "Bison Options" in the Bison manual.
50cca368 728
f1b238df 729*** Variables renamed:
67212941
JD
730
731 The following %define variables
732
733 api.push_pull
734 lr.keep_unreachable_states
735
736 have been renamed to
737
738 api.push-pull
739 lr.keep-unreachable-states
740
741 The old names are now deprecated but will be maintained indefinitely
742 for backward compatibility.
743
7262f54f 744*** Values no longer need to be quoted in the grammar file:
cf499cff
JD
745
746 If a %define value is an identifier, it no longer needs to be placed
747 within quotations marks. For example,
748
749 %define api.push-pull "push"
750
751 can be rewritten as
752
753 %define api.push-pull push
754
d397d9f0 755*** Unrecognized variables are now errors not warnings.
cdf3f113 756
d397d9f0
JD
757*** Multiple invocations for any variable is now an error not a warning.
758
759** Unrecognized %code qualifiers are now errors not warnings.
760
761** Character literals not of length one:
762
763 Previously, Bison quietly converted all character literals to length
764 one. For example, without warning, Bison interpreted the operators in
765 the following grammar to be the same token:
766
767 exp: exp '++'
768 | exp '+' exp
769 ;
770
771 Bison now warns when a character literal is not of length one. In
772 some future release, Bison will start reporting an error instead.
773
774** Destructor calls fixed for lookaheads altered in semantic actions:
775
776 Previously for deterministic parsers in C, if a user semantic action
777 altered yychar, the parser in some cases used the old yychar value to
778 determine which destructor to call for the lookahead upon a syntax
779 error or upon parser return. This bug has been fixed.
780
781** C++ parsers use YYRHSLOC:
782
783 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
784 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
e4ab1254
AD
785 to use it. If, for instance, your location structure has "first"
786 and "last" members, instead of
d397d9f0 787
e29f0771
AD
788 # define YYLLOC_DEFAULT(Current, Rhs, N) \
789 do \
790 if (N) \
791 { \
792 (Current).first = (Rhs)[1].location.first; \
793 (Current).last = (Rhs)[N].location.last; \
794 } \
795 else \
796 { \
797 (Current).first = (Current).last = (Rhs)[0].location.last; \
798 } \
799 while (false)
d397d9f0
JD
800
801 use:
802
e29f0771
AD
803 # define YYLLOC_DEFAULT(Current, Rhs, N) \
804 do \
805 if (N) \
806 { \
807 (Current).first = YYRHSLOC (Rhs, 1).first; \
808 (Current).last = YYRHSLOC (Rhs, N).last; \
809 } \
810 else \
811 { \
812 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
813 } \
814 while (false)
d397d9f0
JD
815
816** YYLLOC_DEFAULT in C++:
817
818 The default implementation of YYLLOC_DEFAULT used to be issued in
819 the header file. It is now output in the implementation file, after
820 the user %code sections so that its #ifndef guard does not try to
821 override the user's YYLLOC_DEFAULT if provided.
cdf3f113 822
f1b238df 823** YYFAIL now produces warnings and Java parsers no longer implement it:
4395a9ff
JD
824
825 YYFAIL has existed for many years as an undocumented feature of
826 deterministic parsers in C generated by Bison. More recently, it was
827 a documented feature of Bison's experimental Java parsers. As
828 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
829 semantic action now produces a deprecation warning, and Java parsers
830 no longer implement YYFAIL at all. For further details, including a
831 discussion of how to suppress C preprocessor warnings about YYFAIL
832 being unused, see the Bison 2.4.2 NEWS entry.
833
f1b238df 834** Temporary hack for adding a semicolon to the user action:
197b82ba
JD
835
836 Previously, Bison appended a semicolon to every user action for
837 reductions when the output language defaulted to C (specifically, when
838 neither %yacc, %language, %skeleton, or equivalent command-line
839 options were specified). This allowed actions such as
840
841 exp: exp "+" exp { $$ = $1 + $3 };
842
843 instead of
844
845 exp: exp "+" exp { $$ = $1 + $3; };
846
847 As a first step in removing this misfeature, Bison now issues a
848 warning when it appends a semicolon. Moreover, in cases where Bison
849 cannot easily determine whether a semicolon is needed (for example, an
850 action ending with a cpp directive or a braced compound initializer),
851 it no longer appends one. Thus, the C compiler might now complain
852 about a missing semicolon where it did not before. Future releases of
853 Bison will cease to append semicolons entirely.
854
d2060f06
JD
855** Verbose syntax error message fixes:
856
e4ab1254 857 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
7fceb615
JD
858 specified, syntax error messages produced by the generated parser
859 include the unexpected token as well as a list of expected tokens.
860 The effect of %nonassoc on these verbose messages has been corrected
861 in two ways, but a more complete fix requires LAC, described above:
d2060f06
JD
862
863*** When %nonassoc is used, there can exist parser states that accept no
864 tokens, and so the parser does not always require a lookahead token
865 in order to detect a syntax error. Because no unexpected token or
866 expected tokens can then be reported, the verbose syntax error
867 message described above is suppressed, and the parser instead
e4ab1254 868 reports the simpler message, "syntax error". Previously, this
d2060f06
JD
869 suppression was sometimes erroneously triggered by %nonassoc when a
870 lookahead was actually required. Now verbose messages are
871 suppressed only when all previous lookaheads have already been
872 shifted or discarded.
873
874*** Previously, the list of expected tokens erroneously included tokens
875 that would actually induce a syntax error because conflicts for them
876 were resolved with %nonassoc in the current parser state. Such
877 tokens are now properly omitted from the list.
878
879*** Expected token lists are still often wrong due to state merging
fcf834f9
JD
880 (from LALR or IELR) and default reductions, which can both add
881 invalid tokens and subtract valid tokens. Canonical LR almost
882 completely fixes this problem by eliminating state merging and
883 default reductions. However, there is one minor problem left even
884 when using canonical LR and even after the fixes above. That is,
885 if the resolution of a conflict with %nonassoc appears in a later
886 parser state than the one at which some syntax error is
887 discovered, the conflicted token is still erroneously included in
888 the expected token list. Bison's new LAC implementation,
889 described above, eliminates this problem and the need for
890 canonical LR. However, LAC is still experimental and is disabled
891 by default.
53f036ce 892
1a33f4f6
JD
893** Java skeleton fixes:
894
895*** A location handling bug has been fixed.
896
897*** The top element of each of the value stack and location stack is now
898 cleared when popped so that it can be garbage collected.
6771a463 899
02803d55
JD
900*** Parser traces now print the top element of the stack.
901
86408959
JD
902** -W/--warnings fixes:
903
e4ab1254 904*** Bison now properly recognizes the "no-" versions of categories:
86408959
JD
905
906 For example, given the following command line, Bison now enables all
907 warnings except warnings for incompatibilities with POSIX Yacc:
908
909 bison -Wall,no-yacc gram.y
910
786743d5
JD
911*** Bison now treats S/R and R/R conflicts like other warnings:
912
913 Previously, conflict reports were independent of Bison's normal
914 warning system. Now, Bison recognizes the warning categories
e4ab1254 915 "conflicts-sr" and "conflicts-rr". This change has important
786743d5
JD
916 consequences for the -W and --warnings command-line options. For
917 example:
918
919 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
920 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
921 bison -Wnone gram.y # no conflicts are reported
922 bison -Werror gram.y # any conflict is an error
923
924 However, as before, if the %expect or %expect-rr directive is
925 specified, an unexpected number of conflicts is an error, and an
926 expected number of conflicts is not reported, so -W and --warning
927 then have no effect on the conflict report.
928
e4ab1254 929*** The "none" category no longer disables a preceding "error":
bf0e44e8
JD
930
931 For example, for the following command line, Bison now reports
932 errors instead of warnings for incompatibilities with POSIX Yacc:
933
934 bison -Werror,none,yacc gram.y
935
e4ab1254 936*** The "none" category now disables all Bison warnings:
c39014ae 937
e4ab1254 938 Previously, the "none" category disabled only Bison warnings for
c39014ae
JD
939 which there existed a specific -W/--warning category. However,
940 given the following command line, Bison is now guaranteed to
941 suppress all warnings:
942
943 bison -Wnone gram.y
944
1f36f544
JD
945** Precedence directives can now assign token number 0:
946
947 Since Bison 2.3b, which restored the ability of precedence
948 directives to assign token numbers, doing so for token number 0 has
949 produced an assertion failure. For example:
950
951 %left END 0
952
953 This bug has been fixed.
954
64877e5e 955* Changes in version 2.4.3 (2010-08-05):
8b9e021f 956
2bfcac9a
JD
957** Bison now obeys -Werror and --warnings=error for warnings about
958 grammar rules that are useless in the parser due to conflicts.
959
8b9e021f
JD
960** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
961 been fixed.
962
4ad3921d
JD
963** Failures in the test suite for GCC 4.5 have been fixed.
964
06cb07d5
JD
965** Failures in the test suite for some versions of Sun Studio C++ have
966 been fixed.
967
9b5049bd
JD
968** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
969 warnings about undefined %prec identifiers will not be converted to
970 errors in Bison 2.5. They will remain warnings, which should be
971 sufficient for POSIX while avoiding backward compatibility issues.
972
93d7dde9
JD
973** Minor documentation fixes.
974
e19a049c 975* Changes in version 2.4.2 (2010-03-20):
74553c98 976
f39ab286
JD
977** Some portability problems that resulted in failures and livelocks
978 in the test suite on some versions of at least Solaris, AIX, HP-UX,
e19a049c
JD
979 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
980 errors should no longer cause M4 to report a broken pipe on the
f39ab286
JD
981 affected platforms.
982
e4ab1254 983** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
8bb3a2e7
JD
984
985 POSIX specifies that an error be reported for any identifier that does
986 not appear on the LHS of a grammar rule and that is not defined by
987 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
988 error report for the case when an identifier appears only after a
989 %prec directive. It is now restored. However, for backward
990 compatibility with recent Bison releases, it is only a warning for
991 now. In Bison 2.5 and later, it will return to being an error.
9b5049bd
JD
992 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
993 warning will not be converted to an error in Bison 2.5.]
8bb3a2e7 994
d8911864
EB
995** Detection of GNU M4 1.4.6 or newer during configure is improved.
996
a603c6e0
JD
997** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
998 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
999 avoided.
c938d650 1000
98a345a2
JD
1001** %code is now a permanent feature.
1002
1003 A traditional Yacc prologue directive is written in the form:
1004
1005 %{CODE%}
1006
1007 To provide a more flexible alternative, Bison 2.3b introduced the
1008 %code directive with the following forms for C/C++:
1009
1010 %code {CODE}
1011 %code requires {CODE}
1012 %code provides {CODE}
1013 %code top {CODE}
1014
1015 These forms are now considered permanent features of Bison. See the
1016 %code entries in the section "Bison Declaration Summary" in the Bison
1017 manual for a summary of their functionality. See the section
1018 "Prologue Alternatives" for a detailed discussion including the
1019 advantages of %code over the traditional Yacc prologue directive.
1020
1021 Bison's Java feature as a whole including its current usage of %code
1022 is still considered experimental.
1023
1625df5b
JD
1024** YYFAIL is deprecated and will eventually be removed.
1025
1026 YYFAIL has existed for many years as an undocumented feature of
1027 deterministic parsers in C generated by Bison. Previously, it was
1028 documented for Bison's experimental Java parsers. YYFAIL is no longer
1029 documented for Java parsers and is formally deprecated in both cases.
1030 Users are strongly encouraged to migrate to YYERROR, which is
1031 specified by POSIX.
1032
1033 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
1034 induce a syntax error. The most obvious difference from YYERROR is
1035 that YYFAIL will automatically invoke yyerror to report the syntax
1036 error so that you don't have to. However, there are several other
1037 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
e4ab1254 1038 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1625df5b
JD
1039 used. For a more detailed discussion, see:
1040
1041 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
1042
1043 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
1044 deterministic parsers in C will continue to implement it. However,
1045 because YYFAIL is already flawed, it seems futile to try to make new
1046 Bison features compatible with it. Thus, during parser generation,
1047 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
1048 rule action. In a later release, YYFAIL will be disabled for
e4ab1254 1049 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1625df5b
JD
1050 be removed altogether.
1051
1052 There exists at least one case where Bison 2.5's YYFAIL warning will
1053 be a false positive. Some projects add phony uses of YYFAIL and other
1054 Bison-defined macros for the sole purpose of suppressing C
1055 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
1056 To avoid Bison's future warning, such YYFAIL uses can be moved to the
e4ab1254 1057 epilogue (that is, after the second "%%") in the Bison input file. In
1625df5b
JD
1058 this release (2.4.2), Bison already generates its own code to suppress
1059 C preprocessor warnings for YYFAIL, so projects can remove their own
1060 phony uses of YYFAIL if compatibility with Bison releases prior to
1061 2.4.2 is not necessary.
1062
2755de8f
AD
1063** Internationalization.
1064
1065 Fix a regression introduced in Bison 2.4: Under some circumstances,
1066 message translations were not installed although supported by the
1067 host system.
1068
74553c98 1069* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 1070
0ea583d2
AD
1071** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
1072 declarations have been fixed.
1979121c 1073
0ea583d2
AD
1074** Temporary hack for adding a semicolon to the user action.
1075
1076 Bison used to prepend a trailing semicolon at the end of the user
1077 action for reductions. This allowed actions such as
1078
1079 exp: exp "+" exp { $$ = $1 + $3 };
1080
1081 instead of
1082
1083 exp: exp "+" exp { $$ = $1 + $3; };
1084
e4ab1254 1085 Some grammars still depend on this "feature". Bison 2.4.1 restores
0ea583d2
AD
1086 the previous behavior in the case of C output (specifically, when
1087 neither %language or %skeleton or equivalent command-line options
1088 are used) to leave more time for grammars depending on the old
1089 behavior to be adjusted. Future releases of Bison will disable this
1090 feature.
1091
1092** A few minor improvements to the Bison manual.
c9ba9e59 1093
402b123d 1094* Changes in version 2.4 (2008-11-02):
7bd1665a 1095
402b123d 1096** %language is an experimental feature.
ed4d67dc
JD
1097
1098 We first introduced this feature in test release 2.3b as a cleaner
1099 alternative to %skeleton. Since then, we have discussed the possibility of
1100 modifying its effect on Bison's output file names. Thus, in this release,
1101 we consider %language to be an experimental feature that will likely evolve
1102 in future releases.
7bd1665a 1103
402b123d 1104** Forward compatibility with GNU M4 has been improved.
241fda7a 1105
402b123d 1106** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
1107 fixed.
1108
402b123d 1109* Changes in version 2.3b (2008-05-27):
35fe0834 1110
402b123d 1111** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
1112 are now deprecated:
1113
1114 %define NAME "VALUE"
1115
e4ab1254 1116** The directive "%pure-parser" is now deprecated in favor of:
d9df47b6
JD
1117
1118 %define api.pure
1119
1120 which has the same effect except that Bison is more careful to warn about
1121 unreasonable usage in the latter case.
1122
402b123d 1123** Push Parsing
c373bf8b
JD
1124
1125 Bison can now generate an LALR(1) parser in C with a push interface. That
e4ab1254
AD
1126 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
1127 push one token at a time to the parser using "yypush_parse", which will
c373bf8b
JD
1128 return to the caller after processing each token. By default, the push
1129 interface is disabled. Either of the following directives will enable it:
1130
1131 %define api.push_pull "push" // Just push; does not require yylex.
1132 %define api.push_pull "both" // Push and pull; requires yylex.
1133
e4ab1254 1134 See the new section "A Push Parser" in the Bison manual for details.
c373bf8b 1135
59da312b
JD
1136 The current push parsing interface is experimental and may evolve. More user
1137 feedback will help to stabilize it.
1138
402b123d 1139** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
1140 not VCG format. Like --graph, -g now also takes an optional FILE argument
1141 and thus cannot be bundled with other short options.
c373bf8b 1142
402b123d 1143** Java
59da312b
JD
1144
1145 Bison can now generate an LALR(1) parser in Java. The skeleton is
e4ab1254 1146 "data/lalr1.java". Consider using the new %language directive instead of
59da312b
JD
1147 %skeleton to select it.
1148
e4ab1254 1149 See the new section "Java Parsers" in the Bison manual for details.
59da312b
JD
1150
1151 The current Java interface is experimental and may evolve. More user
1152 feedback will help to stabilize it.
1153
402b123d 1154** %language
59da312b
JD
1155
1156 This new directive specifies the programming language of the generated
d43f77e7
PB
1157 parser, which can be C (the default), C++, or Java. Besides the skeleton
1158 that Bison uses, the directive affects the names of the generated files if
1159 the grammar file's name ends in ".y".
59da312b 1160
402b123d 1161** XML Automaton Report
59da312b
JD
1162
1163 Bison can now generate an XML report of the LALR(1) automaton using the new
e4ab1254 1164 "--xml" option. The current XML schema is experimental and may evolve. More
59da312b 1165 user feedback will help to stabilize it.
c373bf8b 1166
402b123d 1167** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
1168 %defines. For example:
1169
1170 %defines "parser.h"
1171
402b123d 1172** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
1173 Bison now employs the terms "useless in grammar" instead of "useless",
1174 "useless in parser" instead of "never reduced", and "unused in grammar"
1175 instead of "unused".
cff03fb2 1176
402b123d 1177** Unreachable State Removal
c373bf8b
JD
1178
1179 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
1180 states. A state can become unreachable during conflict resolution if Bison
1181 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
1182
1183 1. Removes unreachable states.
1184
1185 2. Does not report any conflicts that appeared in unreachable states.
1186 WARNING: As a result, you may need to update %expect and %expect-rr
1187 directives in existing grammar files.
1188
1189 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 1190 "useless in parser due to conflicts".
75ad86ee 1191
31984206
JD
1192 This feature can be disabled with the following directive:
1193
1194 %define lr.keep_unreachable_states
1195
e4ab1254 1196 See the %define entry in the "Bison Declaration Summary" in the Bison manual
31984206
JD
1197 for further discussion.
1198
e4ab1254 1199** Lookahead Set Correction in the ".output" Report
b1cc23c4 1200
e4ab1254
AD
1201 When instructed to generate a ".output" file including lookahead sets
1202 (using "--report=lookahead", for example), Bison now prints each reduction's
88c78747
JD
1203 lookahead set only next to the associated state's one item that (1) is
1204 associated with the same rule as the reduction and (2) has its dot at the end
1205 of its RHS. Previously, Bison also erroneously printed the lookahead set
1206 next to all of the state's other items associated with the same rule. This
e4ab1254 1207 bug affected only the ".output" file and not the generated parser source
88c78747
JD
1208 code.
1209
e4ab1254 1210** --report-file=FILE is a new option to override the default ".output" file
59da312b 1211 name.
1bb2bd75 1212
e4ab1254 1213** The "=" that used to be required in the following directives is now
02975b9a
JD
1214 deprecated:
1215
1216 %file-prefix "parser"
1217 %name-prefix "c_"
1218 %output "parser.c"
1219
e4ab1254 1220** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
c373bf8b
JD
1221
1222 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
1223 the traditional Yacc prologue blocks. Those have now been consolidated into
1224 a single %code directive with an optional qualifier field, which identifies
1225 the purpose of the code and thus the location(s) where Bison should generate
1226 it:
1227
e4ab1254
AD
1228 1. "%code {CODE}" replaces "%after-header {CODE}"
1229 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1230 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1231 4. "%code top {CODE}" replaces "%before-header {CODE}"
8e0a5e9e 1232
e4ab1254
AD
1233 See the %code entries in section "Bison Declaration Summary" in the Bison
1234 manual for a summary of the new functionality. See the new section "Prologue
1235 Alternatives" for a detailed discussion including the advantages of %code
8e0a5e9e
JD
1236 over the traditional Yacc prologues.
1237
1238 The prologue alternatives are experimental. More user feedback will help to
1239 determine whether they should become permanent features.
1240
402b123d 1241** Revised warning: unset or unused mid-rule values
17bd8a73
JD
1242
1243 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1244 used within any of the actions of the parent rule. For example, Bison warns
1245 about unused $2 in:
1246
1247 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1248
1249 Now, Bison also warns about mid-rule values that are used but not set. For
1250 example, Bison warns about unset $$ in the mid-rule action in:
1251
1252 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1253
1254 However, Bison now disables both of these warnings by default since they
1255 sometimes prove to be false alarms in existing grammars employing the Yacc
1256 constructs $0 or $-N (where N is some positive integer).
1257
e4ab1254
AD
1258 To enable these warnings, specify the option "--warnings=midrule-values" or
1259 "-W", which is a synonym for "--warnings=all".
17bd8a73 1260
e4ab1254 1261** Default %destructor or %printer with "<*>" or "<>"
c373bf8b
JD
1262
1263 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
1264 %printer's:
1265
e4ab1254 1266 1. Place "<*>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1267 %destructor/%printer for all grammar symbols for which you have formally
1268 declared semantic type tags.
1269
e4ab1254 1270 2. Place "<>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1271 %destructor/%printer for all grammar symbols without declared semantic
1272 type tags.
1273
e4ab1254
AD
1274 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1275 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
12e35840
JD
1276 longer applies any %destructor to a mid-rule value if that mid-rule value is
1277 not actually ever referenced using either $$ or $n in a semantic action.
1278
85894313
JD
1279 The default %destructor's and %printer's are experimental. More user
1280 feedback will help to determine whether they should become permanent
1281 features.
1282
e4ab1254 1283 See the section "Freeing Discarded Symbols" in the Bison manual for further
12e35840
JD
1284 details.
1285
402b123d 1286** %left, %right, and %nonassoc can now declare token numbers. This is required
e4ab1254 1287 by POSIX. However, see the end of section "Operator Precedence" in the Bison
ab7f29f8
JD
1288 manual for a caveat concerning the treatment of literal strings.
1289
402b123d 1290** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
1291 completely removed from Bison.
1292
402b123d 1293* Changes in version 2.3a, 2006-09-13:
742e4900 1294
402b123d 1295** Instead of %union, you can define and use your own union type
ddc8ede1
PE
1296 YYSTYPE if your grammar contains at least one <type> tag.
1297 Your YYSTYPE need not be a macro; it can be a typedef.
1298 This change is for compatibility with other Yacc implementations,
1299 and is required by POSIX.
1300
402b123d 1301** Locations columns and lines start at 1.
cd48d21d
AD
1302 In accordance with the GNU Coding Standards and Emacs.
1303
402b123d 1304** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
1305
1306 For example:
1307
b2a0b7ca
JD
1308 %union { char *string; }
1309 %token <string> STRING1
1310 %token <string> STRING2
1311 %type <string> string1
1312 %type <string> string2
1313 %union { char character; }
1314 %token <character> CHR
1315 %type <character> chr
1316 %destructor { free ($$); } %symbol-default
1317 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1318 %destructor { } <character>
1319
1320 guarantees that, when the parser discards any user-defined symbol that has a
e4ab1254
AD
1321 semantic type tag other than "<character>", it passes its semantic value to
1322 "free". However, when the parser discards a "STRING1" or a "string1", it
1323 also prints its line number to "stdout". It performs only the second
1324 "%destructor" in this case, so it invokes "free" only once.
ec5479ce 1325
85894313
JD
1326 [Although we failed to mention this here in the 2.3a release, the default
1327 %destructor's and %printer's were experimental, and they were rewritten in
1328 future versions.]
1329
e4ab1254
AD
1330** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1331 "--yacc", or "%yacc"), Bison no longer generates #define statements for
b931235e
JD
1332 associating token numbers with token names. Removing the #define statements
1333 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1334 requires them. Bison still generates an enum for token names in all cases.
1335
402b123d 1336** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 1337 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
1338
1339 As before, you declare prologue blocks in your grammar file with the
e4ab1254 1340 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
1341 prologue blocks that you've declared before the first %union. To generate
1342 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 1343 declared after the first %union.
9bc0dd67 1344
34f98f46 1345 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
1346 file and the code file in all cases except for LALR(1) parsers in C. In the
1347 latter case, Bison inserted it only into the code file. For parsers in C++,
1348 the point of insertion was before any token definitions (which associate
1349 token numbers with names). For parsers in C, the point of insertion was
1350 after the token definitions.
1351
1352 Now, Bison never inserts the pre-prologue into the header file. In the code
1353 file, it always inserts it before the token definitions.
1354
402b123d 1355** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
1356 prologue blocks: %before-header, %start-header, %end-header, and
1357 %after-header.
1358
1359 For example, the following declaration order in the grammar file reflects the
1360 order in which Bison will output these code blocks. However, you are free to
1361 declare these code blocks in your grammar file in whatever order is most
1362 convenient for you:
1363
1364 %before-header {
1365 /* Bison treats this block like a pre-prologue block: it inserts it into
1366 * the code file before the contents of the header file. It does *not*
1367 * insert it into the header file. This is a good place to put
1368 * #include's that you want at the top of your code file. A common
e4ab1254 1369 * example is '#include "system.h"'. */
34f98f46
JD
1370 }
1371 %start-header {
1372 /* Bison inserts this block into both the header file and the code file.
1373 * In both files, the point of insertion is before any Bison-generated
1374 * token, semantic type, location type, and class definitions. This is a
1375 * good place to define %union dependencies, for example. */
9bc0dd67
JD
1376 }
1377 %union {
34f98f46
JD
1378 /* Unlike the traditional Yacc prologue blocks, the output order for the
1379 * new %*-header blocks is not affected by their declaration position
1380 * relative to any %union in the grammar file. */
9bc0dd67 1381 }
34f98f46
JD
1382 %end-header {
1383 /* Bison inserts this block into both the header file and the code file.
1384 * In both files, the point of insertion is after the Bison-generated
1385 * definitions. This is a good place to declare or define public
1386 * functions or data structures that depend on the Bison-generated
1387 * definitions. */
9bc0dd67 1388 }
34f98f46
JD
1389 %after-header {
1390 /* Bison treats this block like a post-prologue block: it inserts it into
1391 * the code file after the contents of the header file. It does *not*
1392 * insert it into the header file. This is a good place to declare or
1393 * define internal functions or data structures that depend on the
1394 * Bison-generated definitions. */
1395 }
1396
1397 If you have multiple occurrences of any one of the above declarations, Bison
1398 will concatenate the contents in declaration order.
9bc0dd67 1399
85894313
JD
1400 [Although we failed to mention this here in the 2.3a release, the prologue
1401 alternatives were experimental, and they were rewritten in future versions.]
1402
e4ab1254 1403** The option "--report=look-ahead" has been changed to "--report=lookahead".
9e6e7ed2
PE
1404 The old spelling still works, but is not documented and may be removed
1405 in a future release.
742e4900 1406
402b123d 1407* Changes in version 2.3, 2006-06-05:
4ad3ed84 1408
e4ab1254 1409** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
4ad3ed84
PE
1410 for compatibility with LALR(1) grammars.
1411
402b123d 1412** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
1413 be to a type name that does not contain parentheses or brackets.
1414
402b123d 1415* Changes in version 2.2, 2006-05-19:
193d7c70 1416
402b123d 1417** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
1418 using the parsers in nonfree programs. Previously, this permission
1419 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 1420
402b123d 1421** %name-prefix changes the namespace name in C++ outputs.
aa08666d 1422
402b123d 1423** The C++ parsers export their token_type.
5f4236a0 1424
402b123d 1425** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
1426 their contents together.
1427
402b123d 1428** New warning: unused values
4d7bc38c
PE
1429 Right-hand side symbols whose values are not used are reported,
1430 if the symbols have destructors. For instance:
affac613 1431
8f3596a6 1432 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
e9690142
JD
1433 | exp "+" exp
1434 ;
affac613 1435
8f3596a6
AD
1436 will trigger a warning about $$ and $5 in the first rule, and $3 in
1437 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 1438 most likely contains three errors, and could be rewritten as:
affac613 1439
4e26c69e 1440 exp: exp "?" exp ":" exp
e9690142
JD
1441 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
1442 | exp "+" exp
1443 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
1444 ;
affac613 1445
4e26c69e
PE
1446 However, if the original actions were really intended, memory leaks
1447 and all, the warnings can be suppressed by letting Bison believe the
1448 values are used, e.g.:
721be13c 1449
8f3596a6 1450 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
e9690142
JD
1451 | exp "+" exp { $$ = $1; (void) $3; }
1452 ;
721be13c 1453
84866159
AD
1454 If there are mid-rule actions, the warning is issued if no action
1455 uses it. The following triggers no warning: $1 and $3 are used.
1456
1457 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
1458
721be13c
PE
1459 The warning is intended to help catching lost values and memory leaks.
1460 If a value is ignored, its associated memory typically is not reclaimed.
affac613 1461
402b123d 1462** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
1463 Destructors are now called when user code invokes YYABORT, YYACCEPT,
1464 and YYERROR, for all objects on the stack, other than objects
1465 corresponding to the right-hand side of the current rule.
a85284cf 1466
402b123d 1467** %expect, %expect-rr
035aa4a0
PE
1468 Incorrect numbers of expected conflicts are now actual errors,
1469 instead of warnings.
1470
402b123d 1471** GLR, YACC parsers.
4e26c69e
PE
1472 The %parse-params are available in the destructors (and the
1473 experimental printers) as per the documentation.
4b367315 1474
e4ab1254 1475** Bison now warns if it finds a stray "$" or "@" in an action.
ad6a9b97 1476
402b123d 1477** %require "VERSION"
4e26c69e
PE
1478 This specifies that the grammar file depends on features implemented
1479 in Bison version VERSION or higher.
b50d2359 1480
402b123d 1481** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
1482 The tokens were defined as free form enums and cpp macros. YYSTYPE
1483 was defined as a free form union. They are now class members:
e4ab1254
AD
1484 tokens are enumerations of the "yy::parser::token" struct, and the
1485 semantic values have the "yy::parser::semantic_type" type.
fb9712a9
AD
1486
1487 If you do not want or can update to this scheme, the directive
e4ab1254 1488 '%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
1489 definition of tokens and YYSTYPE. This change is suitable both
1490 for previous releases of Bison, and this one.
fb9712a9 1491
b50d2359 1492 If you wish to update, then make sure older version of Bison will
e4ab1254 1493 fail using '%require "2.2"'.
fb9712a9 1494
402b123d 1495** DJGPP support added.
193d7c70 1496\f
402b123d 1497* Changes in version 2.1, 2005-09-16:
1ce59070 1498
402b123d 1499** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 1500
402b123d 1501** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
1502 "syntax error" into languages other than English. The default
1503 language is still English. For details, please see the new
0410a6e0
PE
1504 Internationalization section of the Bison manual. Software
1505 distributors should also see the new PACKAGING file. Thanks to
1506 Bruno Haible for this new feature.
1ce59070 1507
402b123d 1508** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
1509 simplify translation. In particular, the message "memory exhausted"
1510 has replaced "parser stack overflow", as the old message was not
1511 always accurate for modern Bison-generated parsers.
1512
402b123d 1513** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
1514 behind on the stack. Also, the start symbol is now destroyed after a
1515 successful parse. In both cases, the behavior was formerly inconsistent.
1516
402b123d 1517** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
1518 quote the literal strings associated with tokens. For example, for
1519 a syntax error associated with '%token NUM "number"' they might
1520 print 'syntax error, unexpected number' instead of 'syntax error,
1521 unexpected "number"'.
193d7c70 1522\f
402b123d 1523* Changes in version 2.0, 2004-12-25:
efeed023 1524
402b123d 1525** Possibly-incompatible changes
d7e14fc0 1526
82de6b0d
PE
1527 - Bison-generated parsers no longer default to using the alloca function
1528 (when available) to extend the parser stack, due to widespread
1529 problems in unchecked stack-overflow detection. You can "#define
1530 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
1531 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 1532
82de6b0d
PE
1533 - Error token location.
1534 During error recovery, the location of the syntax error is updated
1535 to cover the whole sequence covered by the error token: it includes
1536 the shifted symbols thrown away during the first part of the error
1537 recovery, and the lookahead rejected during the second part.
18d192f0 1538
82de6b0d
PE
1539 - Semicolon changes:
1540 . Stray semicolons are no longer allowed at the start of a grammar.
1541 . Semicolons are now required after in-grammar declarations.
e342c3be 1542
82de6b0d
PE
1543 - Unescaped newlines are no longer allowed in character constants or
1544 string literals. They were never portable, and GCC 3.4.0 has
1545 dropped support for them. Better diagnostics are now generated if
1546 forget a closing quote.
8dd162d3 1547
82de6b0d 1548 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 1549
402b123d 1550** New features
1452af69 1551
82de6b0d 1552 - GLR grammars now support locations.
4febdd96 1553
82de6b0d
PE
1554 - New directive: %initial-action.
1555 This directive allows the user to run arbitrary code (including
1556 initializing @$) from yyparse before parsing starts.
1452af69 1557
82de6b0d
PE
1558 - A new directive "%expect-rr N" specifies the expected number of
1559 reduce/reduce conflicts in GLR parsers.
1452af69 1560
e4ab1254 1561 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
82de6b0d 1562 This is a GNU extension.
4febdd96 1563
e4ab1254 1564 - The option "--report=lookahead" was changed to "--report=look-ahead".
9e6e7ed2 1565 [However, this was changed back after 2.3.]
1452af69 1566
82de6b0d 1567 - Experimental %destructor support has been added to lalr1.cc.
1452af69 1568
82de6b0d
PE
1569 - New configure option --disable-yacc, to disable installation of the
1570 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 1571
402b123d 1572** Bug fixes
d5a3fe37 1573
82de6b0d
PE
1574 - For now, %expect-count violations are now just warnings, not errors.
1575 This is for compatibility with Bison 1.75 and earlier (when there are
1576 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
1577 are too many or too few shift/reduce conflicts). However, in future
1578 versions of Bison we plan to improve the %expect machinery so that
1579 these violations will become errors again.
3473d0f8 1580
82de6b0d
PE
1581 - Within Bison itself, numbers (e.g., goto numbers) are no longer
1582 arbitrarily limited to 16-bit counts.
d600ee67 1583
82de6b0d 1584 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 1585\f
402b123d 1586* Changes in version 1.875, 2003-01-01:
963fcc17 1587
402b123d 1588** The documentation license has been upgraded to version 1.2
dc546b0f 1589 of the GNU Free Documentation License.
75eb3bc4 1590
402b123d 1591** syntax error processing
75eb3bc4 1592
dc546b0f
PE
1593 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
1594 locations too. This fixes bugs in error-location computation.
75eb3bc4 1595
dc546b0f
PE
1596 - %destructor
1597 It is now possible to reclaim the memory associated to symbols
1598 discarded during error recovery. This feature is still experimental.
20daca06 1599
dc546b0f
PE
1600 - %error-verbose
1601 This new directive is preferred over YYERROR_VERBOSE.
74724a70 1602
dc546b0f
PE
1603 - #defining yyerror to steal internal variables is discouraged.
1604 It is not guaranteed to work forever.
d1de5372 1605
402b123d 1606** POSIX conformance
d1de5372 1607
dc546b0f
PE
1608 - Semicolons are once again optional at the end of grammar rules.
1609 This reverts to the behavior of Bison 1.33 and earlier, and improves
1610 compatibility with Yacc.
74724a70 1611
e4ab1254
AD
1612 - "parse error" -> "syntax error"
1613 Bison now uniformly uses the term "syntax error"; formerly, the code
1614 and manual sometimes used the term "parse error" instead. POSIX
1615 requires "syntax error" in diagnostics, and it was thought better to
dc546b0f 1616 be consistent.
74724a70 1617
dc546b0f
PE
1618 - The documentation now emphasizes that yylex and yyerror must be
1619 declared before use. C99 requires this.
d1de5372 1620
dc546b0f
PE
1621 - Bison now parses C99 lexical constructs like UCNs and
1622 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 1623
dc546b0f
PE
1624 - File names are properly escaped in C output. E.g., foo\bar.y is
1625 output as "foo\\bar.y".
6780ca7a 1626
dc546b0f 1627 - Yacc command and library now available
e4ab1254 1628 The Bison distribution now installs a "yacc" command, as POSIX requires.
dc546b0f
PE
1629 Also, Bison now installs a small library liby.a containing
1630 implementations of Yacc-compatible yyerror and main functions.
1631 This library is normally not useful, but POSIX requires it.
6e649e65 1632
dc546b0f 1633 - Type clashes now generate warnings, not errors.
6e649e65 1634
dc546b0f
PE
1635 - If the user does not define YYSTYPE as a macro, Bison now declares it
1636 using typedef instead of defining it as a macro.
1637 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 1638
402b123d 1639** Other compatibility issues
886a425c 1640
e4ab1254
AD
1641 - %union directives can now have a tag before the "{", e.g., the
1642 directive "%union foo {...}" now generates the C code
1643 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
1644 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
1645 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
dc546b0f 1646 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 1647
e4ab1254 1648 - ";" is output before the terminating "}" of an action, for
dc546b0f 1649 compatibility with Bison 1.35.
886a425c 1650
dc546b0f 1651 - Bison now uses a Yacc-style format for conflict reports, e.g.,
e4ab1254 1652 "conflicts: 2 shift/reduce, 1 reduce/reduce".
437c2d80 1653
e4ab1254 1654 - "yystype" and "yyltype" are now obsolescent macros instead of being
dc546b0f
PE
1655 typedefs or tags; they are no longer documented and are planned to be
1656 withdrawn in a future release.
2a8d363a 1657
402b123d 1658** GLR parser notes
2a8d363a 1659
dc546b0f
PE
1660 - GLR and inline
1661 Users of Bison have to decide how they handle the portability of the
e4ab1254 1662 C keyword "inline".
959e5f51 1663
e4ab1254
AD
1664 - "parsing stack overflow..." -> "parser stack overflow"
1665 GLR parsers now report "parser stack overflow" as per the Bison manual.
900c5db5 1666
18ad57b3
AD
1667** %parse-param and %lex-param
1668 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
1669 additional context to yyparse and yylex. They suffer from several
1670 shortcomings:
1671
1672 - a single argument only can be added,
1673 - their types are weak (void *),
242cc08e 1674 - this context is not passed to ancillary functions such as yyerror,
18ad57b3
AD
1675 - only yacc.c parsers support them.
1676
1677 The new %parse-param/%lex-param directives provide a more precise control.
1678 For instance:
1679
1680 %parse-param {int *nastiness}
1681 %lex-param {int *nastiness}
1682 %parse-param {int *randomness}
1683
1684 results in the following signatures:
1685
1686 int yylex (int *nastiness);
1687 int yyparse (int *nastiness, int *randomness);
1688
1689 or, if both %pure-parser and %locations are used:
1690
1691 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
1692 int yyparse (int *nastiness, int *randomness);
1693
402b123d 1694** Bison now warns if it detects conflicting outputs to the same file,
e4ab1254 1695 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
dc546b0f 1696 that command outputs both code and header to foo.h.
6e40b4eb 1697
402b123d 1698** #line in output files
dc546b0f 1699 - --no-line works properly.
6e40b4eb 1700
402b123d 1701** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
1702 later to be built. This change originally took place a few versions
1703 ago, but nobody noticed until we recently asked someone to try
1704 building Bison with a K&R C compiler.
d600ee67 1705\f
402b123d 1706* Changes in version 1.75, 2002-10-14:
7933f2b5 1707
402b123d 1708** Bison should now work on 64-bit hosts.
7933f2b5 1709
402b123d 1710** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 1711
402b123d 1712** GLR parsers
f50adbbd
AD
1713 Fix spurious parse errors.
1714
402b123d 1715** Pure parsers
f50adbbd
AD
1716 Some people redefine yyerror to steal yyparse' private variables.
1717 Reenable this trick until an official feature replaces it.
1718
402b123d 1719** Type Clashes
d90c934c
AD
1720 In agreement with POSIX and with other Yaccs, leaving a default
1721 action is valid when $$ is untyped, and $1 typed:
1722
e9690142 1723 untyped: ... typed;
d90c934c
AD
1724
1725 but the converse remains an error:
1726
e9690142 1727 typed: ... untyped;
d90c934c 1728
402b123d 1729** Values of mid-rule actions
d90c934c
AD
1730 The following code:
1731
e9690142 1732 foo: { ... } { $$ = $1; } ...
d90c934c
AD
1733
1734 was incorrectly rejected: $1 is defined in the second mid-rule
1735 action, and is equal to the $$ of the first mid-rule action.
d600ee67 1736\f
402b123d 1737* Changes in version 1.50, 2002-10-04:
adc8c848 1738
402b123d 1739** GLR parsing
676385e2
PH
1740 The declaration
1741 %glr-parser
1742 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
1743 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 1744 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
1745 ambiguities. Contributed by Paul Hilfinger.
1746
7933f2b5 1747 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
1748 like the Alpha, so please stick to 32-bit hosts for now.
1749
402b123d 1750** Output Directory
8c165d89 1751 When not in Yacc compatibility mode, when the output file was not
e4ab1254
AD
1752 specified, running "bison foo/bar.y" created "foo/bar.c". It
1753 now creates "bar.c".
8c165d89 1754
402b123d 1755** Undefined token
007a50a4 1756 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 1757 the use of 2 by the user. This is no longer the case.
007a50a4 1758
402b123d 1759** Unknown token numbers
e88dbdbf 1760 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
1761 no longer the case.
1762
402b123d 1763** Error token
e88dbdbf 1764 According to POSIX, the error token must be 256.
23c5a174
AD
1765 Bison extends this requirement by making it a preference: *if* the
1766 user specified that one of her tokens is numbered 256, then error
1767 will be mapped onto another number.
1768
402b123d 1769** Verbose error messages
e4ab1254 1770 They no longer report "..., expecting error or..." for states where
217598da
AD
1771 error recovery is possible.
1772
402b123d 1773** End token
e4ab1254 1774 Defaults to "$end" instead of "$".
217598da 1775
402b123d 1776** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
1777 When a Bison-generated parser encounters a syntax error, it now pops
1778 the stack until it finds a state that allows shifting the error
1779 token. Formerly, it popped the stack until it found a state that
1780 allowed some non-error action other than a default reduction on the
1781 error token. The new behavior has long been the documented behavior,
1782 and has long been required by POSIX. For more details, please see
337116ba
PE
1783 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
1784 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 1785
402b123d 1786** Traces
5504898e
AD
1787 Popped tokens and nonterminals are now reported.
1788
402b123d 1789** Larger grammars
a861a339
PE
1790 Larger grammars are now supported (larger token numbers, larger grammar
1791 size (= sum of the LHS and RHS lengths), larger LALR tables).
1792 Formerly, many of these numbers ran afoul of 16-bit limits;
1793 now these limits are 32 bits on most hosts.
355e7c1c 1794
402b123d 1795** Explicit initial rule
643a5994
AD
1796 Bison used to play hacks with the initial rule, which the user does
1797 not write. It is now explicit, and visible in the reports and
1798 graphs as rule 0.
23c5a174 1799
402b123d 1800** Useless rules
643a5994 1801 Before, Bison reported the useless rules, but, although not used,
77714df2 1802 included them in the parsers. They are now actually removed.
23c5a174 1803
402b123d 1804** Useless rules, useless nonterminals
6b98e4b5
AD
1805 They are now reported, as a warning, with their locations.
1806
402b123d 1807** Rules never reduced
e8832397
AD
1808 Rules that can never be reduced because of conflicts are now
1809 reported.
1810
e4ab1254 1811** Incorrect "Token not used"
11652ab3
AD
1812 On a grammar such as
1813
e29f0771
AD
1814 %token useless useful
1815 %%
1816 exp: '0' %prec useful;
11652ab3
AD
1817
1818 where a token was used to set the precedence of the last rule,
e4ab1254 1819 bison reported both "useful" and "useless" as useless tokens.
11652ab3 1820
402b123d 1821** Revert the C++ namespace changes introduced in 1.31
77714df2 1822 as they caused too many portability hassles.
0179dd65 1823
402b123d 1824** Default locations
b2d52318
AD
1825 By an accident of design, the default computation of @$ was
1826 performed after another default computation was performed: @$ = @1.
1827 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1828 the computation of @$.
adc8c848 1829
402b123d 1830** Token end-of-file
b7c49edf
AD
1831 The token end of file may be specified by the user, in which case,
1832 the user symbol is used in the reports, the graphs, and the verbose
e4ab1254 1833 error messages instead of "$end", which remains being the default.
b7c49edf 1834 For instance
e29f0771 1835 %token MYEOF 0
b7c49edf 1836 or
e29f0771 1837 %token MYEOF 0 "end of file"
fdbcd8e2 1838
402b123d 1839** Semantic parser
fdbcd8e2
AD
1840 This old option, which has been broken for ages, is removed.
1841
402b123d 1842** New translations
a861a339 1843 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
1844 Croatian, thanks to Denis Lackovic.
1845
402b123d 1846** Incorrect token definitions
e4ab1254
AD
1847 When given
1848 %token 'a' "A"
1849 bison used to output
1850 #define 'a' 65
b87f8b21 1851
402b123d 1852** Token definitions as enums
77714df2
AD
1853 Tokens are output both as the traditional #define's, and, provided
1854 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 1855 This lets debuggers display names instead of integers.
77714df2 1856
402b123d 1857** Reports
ec3bc396
AD
1858 In addition to --verbose, bison supports --report=THINGS, which
1859 produces additional information:
b408954b
AD
1860 - itemset
1861 complete the core item sets with their closure
e4ab1254 1862 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
9e6e7ed2 1863 explicitly associate lookahead tokens to items
b408954b
AD
1864 - solved
1865 describe shift/reduce conflicts solving.
1866 Bison used to systematically output this information on top of
1867 the report. Solved conflicts are now attached to their states.
ec3bc396 1868
402b123d 1869** Type clashes
9af3fbce
AD
1870 Previous versions don't complain when there is a type clash on
1871 the default action if the rule has a mid-rule action, such as in:
1872
e29f0771
AD
1873 %type <foo> bar
1874 %%
1875 bar: '0' {} '0';
9af3fbce
AD
1876
1877 This is fixed.
a861a339 1878
402b123d 1879** GNU M4 is now required when using Bison.
f987e9d2 1880\f
402b123d 1881* Changes in version 1.35, 2002-03-25:
76551463 1882
402b123d 1883** C Skeleton
76551463
AD
1884 Some projects use Bison's C parser with C++ compilers, and define
1885 YYSTYPE as a class. The recent adjustment of C parsers for data
1886 alignment and 64 bit architectures made this impossible.
1887
1888 Because for the time being no real solution for C++ parser
1889 generation exists, kludges were implemented in the parser to
1890 maintain this use. In the future, when Bison has C++ parsers, this
1891 kludge will be disabled.
1892
1893 This kludge also addresses some C++ problems when the stack was
1894 extended.
76551463 1895\f
402b123d 1896* Changes in version 1.34, 2002-03-12:
76551463 1897
402b123d 1898** File name clashes are detected
76551463 1899 $ bison foo.y -d -o foo.x
e4ab1254 1900 fatal error: header and parser would both be named "foo.x"
76551463 1901
e4ab1254 1902** A missing ";" at the end of a rule triggers a warning
76551463
AD
1903 In accordance with POSIX, and in agreement with other
1904 Yacc implementations, Bison will mandate this semicolon in the near
1905 future. This eases the implementation of a Bison parser of Bison
1906 grammars by making this grammar LALR(1) instead of LR(2). To
1907 facilitate the transition, this release introduces a warning.
1908
402b123d 1909** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
1910 many portability hassles.
1911
402b123d 1912** DJGPP support added.
76551463 1913
402b123d 1914** Fix test suite portability problems.
76551463 1915\f
402b123d 1916* Changes in version 1.33, 2002-02-07:
76551463 1917
402b123d 1918** Fix C++ issues
76551463
AD
1919 Groff could not be compiled for the definition of size_t was lacking
1920 under some conditions.
1921
402b123d 1922** Catch invalid @n
76551463
AD
1923 As is done with $n.
1924\f
402b123d 1925* Changes in version 1.32, 2002-01-23:
76551463 1926
402b123d 1927** Fix Yacc output file names
76551463 1928
402b123d 1929** Portability fixes
76551463 1930
402b123d 1931** Italian, Dutch translations
76551463 1932\f
402b123d 1933* Changes in version 1.31, 2002-01-14:
52d1aeee 1934
402b123d 1935** Many Bug Fixes
52d1aeee 1936
402b123d 1937** GNU Gettext and %expect
52d1aeee
MA
1938 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1939 Bison dies on incorrect %expectations, we fear there will be
1940 too many bug reports for Gettext, so _for the time being_, %expect
e4ab1254 1941 does not trigger an error when the input file is named "plural.y".
52d1aeee 1942
402b123d 1943** Use of alloca in parsers
52d1aeee
MA
1944 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1945 malloc exclusively. Since 1.29, but was not NEWS'ed.
1946
1947 alloca is used only when compiled with GCC, to avoid portability
1948 problems as on AIX.
1949
402b123d 1950** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 1951
402b123d 1952** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
1953 (as POSIX requires) instead of being undefined.
1954
402b123d 1955** User Actions
52d1aeee
MA
1956 Bison has always permitted actions such as { $$ = $1 }: it adds the
1957 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1958 is no longer output: one has to write { $$ = $1; }.
1959
402b123d 1960** Better C++ compliance
52d1aeee 1961 The output parsers try to respect C++ namespaces.
76551463 1962 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 1963
402b123d 1964** Reduced Grammars
52d1aeee
MA
1965 Fixed bugs when reporting useless nonterminals.
1966
402b123d 1967** 64 bit hosts
52d1aeee
MA
1968 The parsers work properly on 64 bit hosts.
1969
402b123d 1970** Error messages
52d1aeee
MA
1971 Some calls to strerror resulted in scrambled or missing error messages.
1972
402b123d 1973** %expect
52d1aeee
MA
1974 When the number of shift/reduce conflicts is correct, don't issue
1975 any warning.
1976
402b123d 1977** The verbose report includes the rule line numbers.
52d1aeee 1978
402b123d 1979** Rule line numbers are fixed in traces.
52d1aeee 1980
402b123d 1981** Swedish translation
52d1aeee 1982
402b123d 1983** Parse errors
52d1aeee
MA
1984 Verbose parse error messages from the parsers are better looking.
1985 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1986 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1987
402b123d 1988** Fixed parser memory leaks.
52d1aeee
MA
1989 When the generated parser was using malloc to extend its stacks, the
1990 previous allocations were not freed.
1991
402b123d 1992** Fixed verbose output file.
52d1aeee
MA
1993 Some newlines were missing.
1994 Some conflicts in state descriptions were missing.
1995
402b123d 1996** Fixed conflict report.
52d1aeee
MA
1997 Option -v was needed to get the result.
1998
402b123d 1999** %expect
52d1aeee
MA
2000 Was not used.
2001 Mismatches are errors, not warnings.
2002
402b123d 2003** Fixed incorrect processing of some invalid input.
52d1aeee 2004
402b123d 2005** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 2006
402b123d 2007** Fixed some typos in the documentation.
52d1aeee 2008
402b123d 2009** %token MY_EOF 0 is supported.
52d1aeee
MA
2010 Before, MY_EOF was silently renumbered as 257.
2011
402b123d 2012** doc/refcard.tex is updated.
52d1aeee 2013
402b123d 2014** %output, %file-prefix, %name-prefix.
52d1aeee
MA
2015 New.
2016
402b123d 2017** --output
e4ab1254 2018 New, aliasing "--output-file".
52d1aeee 2019\f
402b123d 2020* Changes in version 1.30, 2001-10-26:
342b8b6e 2021
e4ab1254
AD
2022** "--defines" and "--graph" have now an optional argument which is the
2023 output file name. "-d" and "-g" do not change; they do not take any
342b8b6e
AD
2024 argument.
2025
e4ab1254 2026** "%source_extension" and "%header_extension" are removed, failed
342b8b6e
AD
2027 experiment.
2028
402b123d 2029** Portability fixes.
f987e9d2 2030\f
402b123d 2031* Changes in version 1.29, 2001-09-07:
342b8b6e 2032
402b123d 2033** The output file does not define const, as this caused problems when used
342b8b6e
AD
2034 with common autoconfiguration schemes. If you still use ancient compilers
2035 that lack const, compile with the equivalent of the C compiler option
e4ab1254 2036 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
342b8b6e 2037
e4ab1254 2038** Added "-g" and "--graph".
f87a2205 2039
402b123d 2040** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 2041
402b123d 2042** The input and the output files has automatically a similar extension.
234a3be3 2043
402b123d 2044** Russian translation added.
f87a2205 2045
402b123d 2046** NLS support updated; should hopefully be less troublesome.
f87a2205 2047
402b123d 2048** Added the old Bison reference card.
c33638bb 2049
e4ab1254 2050** Added "--locations" and "%locations".
6deb4447 2051
e4ab1254 2052** Added "-S" and "--skeleton".
cd5bd6ac 2053
e4ab1254 2054** "%raw", "-r", "--raw" is disabled.
62ab6972 2055
402b123d 2056** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
2057 of the #line lines with path names including backslashes.
2058
402b123d 2059** New directives.
e4ab1254
AD
2060 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
2061 "%debug", "%source_extension" and "%header_extension".
f987e9d2 2062
402b123d 2063** @$
f987e9d2 2064 Automatic location tracking.
f87a2205 2065\f
402b123d 2066* Changes in version 1.28, 1999-07-06:
d2e00347 2067
402b123d 2068** Should compile better now with K&R compilers.
d2e00347 2069
402b123d 2070** Added NLS.
d2e00347 2071
402b123d 2072** Fixed a problem with escaping the double quote character.
d2e00347 2073
402b123d 2074** There is now a FAQ.
d2e00347 2075\f
402b123d 2076* Changes in version 1.27:
5c31c3c2 2077
402b123d 2078** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
2079 some systems has been fixed.
2080\f
402b123d 2081* Changes in version 1.26:
4be07551 2082
7e508a2b 2083** Bison now uses Automake.
4be07551 2084
402b123d 2085** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 2086
402b123d 2087** Token numbers now start at 257 as previously documented, not 258.
4be07551 2088
402b123d 2089** Bison honors the TMPDIR environment variable.
4be07551 2090
402b123d 2091** A couple of buffer overruns have been fixed.
f51dbca1 2092
402b123d 2093** Problems when closing files should now be reported.
f51dbca1 2094
402b123d 2095** Generated parsers should now work even on operating systems which do
f51dbca1 2096 not provide alloca().
4be07551 2097\f
402b123d 2098* Changes in version 1.25, 1995-10-16:
df8878c5 2099
402b123d 2100** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 2101the grammar file, and reports all the errors found in it.
8c44d3ec 2102
402b123d 2103** Tokens can now be specified as multiple-character strings: for
df8878c5 2104example, you could use "<=" for a token which looks like <=, instead
7e508a2b 2105of choosing a name like LESSEQ.
df8878c5 2106
402b123d 2107** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
2108and numbers) into the parser file. The yylex function can use this
2109table to recognize multiple-character string tokens, or for other
2110purposes.
2111
402b123d 2112** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
2113directives in the parser file.
2114
402b123d 2115** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
2116Yacc-compatible token numbers, when token names are defined as macros.
2117
402b123d 2118** The --no-parser option produces the parser tables without including
df8878c5
RS
2119the parser engine; a project can now use its own parser engine.
2120The actions go into a separate file called NAME.act, in the form of
2121a switch statement body.
2122\f
402b123d 2123* Changes in version 1.23:
6780ca7a 2124
4d019228
DM
2125The user can define YYPARSE_PARAM as the name of an argument to be
2126passed into yyparse. The argument should have type void *. It should
2127actually point to an object. Grammar actions can access the variable
2128by casting it to the proper pointer type.
6780ca7a 2129
6780ca7a 2130Line numbers in output file corrected.
6780ca7a 2131\f
402b123d 2132* Changes in version 1.22:
6780ca7a
DM
2133
2134--help option added.
6780ca7a 2135\f
402b123d 2136* Changes in version 1.20:
6780ca7a
DM
2137
2138Output file does not redefine const for C++.
9f4503d6 2139
76551463
AD
2140-----
2141
34136e65 2142Copyright (C) 1995-2012 Free Software Foundation, Inc.
76551463 2143
74553c98 2144This file is part of Bison, the GNU Parser Generator.
76551463 2145
f16b0819 2146This program is free software: you can redistribute it and/or modify
76551463 2147it under the terms of the GNU General Public License as published by
f16b0819
PE
2148the Free Software Foundation, either version 3 of the License, or
2149(at your option) any later version.
76551463 2150
f16b0819 2151This program is distributed in the hope that it will be useful,
76551463
AD
2152but WITHOUT ANY WARRANTY; without even the implied warranty of
2153MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
2154GNU General Public License for more details.
2155
2156You should have received a copy of the GNU General Public License
f16b0819 2157along with this program. If not, see <http://www.gnu.org/licenses/>.
7e508a2b
AD
2158
2159 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
2160 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
2161 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
2162 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
2163 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
2164 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
2165 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
2166 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
2167 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
2168 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
2169 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
2170 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
2171 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
242cc08e 2172 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
7e508a2b 2173 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
242cc08e
AD
2174 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2175 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
dcb366b1 2176 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
2b45240d 2177 LocalWords: TOK calc yyo fval
7e508a2b
AD
2178
2179Local Variables:
2180mode: outline
e4ab1254 2181fill-column: 76
7e508a2b 2182End: