]> git.saurik.com Git - bison.git/blame - NEWS
todo: check push parsers
[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
34904c57 16*** Warning categories are now displayed
73370a9d
VS
17
18 For instance:
19
20 foo.y:4.6: warning: type clash on default action: <foo> != <bar> [-Wother]
21
22*** Useless semantic types
9641b918
VS
23
24 Bison now warns about useless (uninhabited) semantic types. Since
25 semantic types are not declared to Bison (they are defined in the opaque
26 %union structure), it is %printer/%destructor directives about useless
27 types that trigger the warning:
28
29 %token <type1> term
30 %type <type2> nterm
31 %printer {} <type1> <type3>
32 %destructor {} <type2> <type4>
33 %%
34 nterm: term { $$ = $1; };
35
36 3.28-34: warning: type <type3> is used, but is not associated to any symbol
37 4.28-34: warning: type <type4> is used, but is not associated to any symbol
38
73370a9d 39*** Undeclared symbols
b921d92f
VS
40
41 Bison used to raise an error for %printer and %destructor directives for
42 undefined symbols.
43
44 %printer {} symbol1
45 %destructor {} symbol2
46 %%
47 exp: "a";
48
49 This is now only a warning.
50
73370a9d 51*** Useless destructors or printers
ea9a35c6
VS
52
53 Bison now warns about useless destructors or printers. In the following
54 example, the printer for <type1>, and the destructor for <type2> are
55 useless: all symbols of <type1> (token1) already have a printer, and all
56 symbols of type <type2> (token2) already have a destructor.
57
58 %token <type1> token1
59 <type2> token2
60 <type3> token3
61 <type4> token4
62 %printer {} token1 <type1> <type3>
63 %destructor {} token2 <type2> <type4>
64
2055a44e
AD
65** Additional yylex/yyparse arguments
66
6dc4663d
AD
67 The new directive %param declares additional arguments to both yylex and
68 yyparse. The %lex-param, %parse-param, and %param directives support one
69 or more arguments. Instead of
2055a44e 70
e436fa67
AD
71 %lex-param {arg1_type *arg1}
72 %lex-param {arg2_type *arg2}
73 %parse-param {arg1_type *arg1}
74 %parse-param {arg2_type *arg2}
2055a44e
AD
75
76 one may now declare
77
e436fa67 78 %param {arg1_type *arg1} {arg2_type *arg2}
2055a44e
AD
79
80** Java skeleton improvements
0ea583d2
AD
81
82 The constants for token names were moved to the Lexer interface.
83 Also, it is possible to add code to the parser's constructors using
84 "%code init" and "%define init_throws".
85
34904c57
AD
86** C++ skeletons improvements
87
93549bcd
AD
88*** parser header (%defines) is no longer mandatory (lalr1.cc)
89
90 In which case, if needed, the support classes are defined in the generated
5de5b987
AD
91 parser, instead of additional files (location.hh, position.hh and
92 stack.hh).
93549bcd 93
34904c57
AD
94*** locations are no longer mandatory (lalr1.cc, glr.cc)
95
96 Both lalr1.cc and glr.cc no longer require %location.
97
98*** syntax_error exception (lalr1.cc)
a6552c5d
AD
99
100 The C++ parser features a syntax_error exception, which can be
101 thrown from the scanner or from user rules to raise syntax errors.
102 This facilitates reporting errors caught in sub-functions (e.g.,
103 rejecting too large integral literals from a conversion function
104 used by the scanner, or rejecting invalid combinations from a
105 factory invoked by the user actions).
106
4c6622c2 107** Variable api.tokens.prefix
99c08fb6 108
4c6622c2 109 The variable api.tokens.prefix changes the way tokens are identified in
99c08fb6
AD
110 the generated files. This is especially useful to avoid collisions
111 with identifiers in the target language. For instance
112
e436fa67
AD
113 %token FILE for ERROR
114 %define api.tokens.prefix "TOK_"
115 %%
116 start: FILE for ERROR;
99c08fb6
AD
117
118 will generate the definition of the symbols TOK_FILE, TOK_for, and
119 TOK_ERROR in the generated sources. In particular, the scanner must
120 use these prefixed token names, although the grammar itself still
121 uses the short names (as in the sample rule given above).
122
67501061
AD
123** Variable api.namespace
124
125 The "namespace" variable is renamed "api.namespace". Backward
126 compatibility is ensured, but upgrading is recommended.
127
31b850d2
AD
128** Variable parse.error
129
130 The variable error controls the verbosity of error messages. The
131 use of the %error-verbose directive is deprecated in favor of
132 %define parse.error "verbose".
133
ca2a6d15
PH
134** Semantic predicates
135
136 The new, experimental, semantic-predicate feature allows actions of
137 the form %?{ BOOLEAN-EXPRESSION }, which cause syntax errors (as for
138 YYERROR) if the expression evaluates to 0, and are evaluated immediately
139 in GLR parsers, rather than being deferred. The result is that they
140 allow the programmer to prune possible parses based on the values of
dcb366b1 141 run-time expressions.
ca2a6d15 142
e436fa67
AD
143* Noteworthy changes in release ?.? (????-??-??) [?]
144
9c26b8fc 145
e1eeecd3 146* Noteworthy changes in release 2.6.2 (2012-08-03) [stable]
9c26b8fc 147
43ca8040
AD
148** Bug fixes
149
150 Buffer overruns, complaints from Flex, and portability issues in the test
151 suite have been fixed.
152
c9d546b2
AD
153** Spaces in %lex- and %parse-param (lalr1.cc, glr.cc)
154
155 Trailing end-of-lines in %parse-param or %lex-param would result in
156 invalid C++. This is fixed.
9c26b8fc 157
dcd5344d
AD
158** Spurious spaces and end-of-lines
159
160 The generated files no longer end (nor start) with empty lines.
161
77b214ef 162* Noteworthy changes in release 2.6.1 (2012-07-30) [stable]
a4107f24 163
8617d87e
AD
164 Bison no longer executes user-specified M4 code when processing a grammar.
165
e20e6a50
AD
166** Future Changes
167
168 In addition to the removal of the features announced in Bison 2.6, the
169 next major release will remove the "Temporary hack for adding a semicolon
170 to the user action", as announced in the release 2.5. Instead of:
171
172 exp: exp "+" exp { $$ = $1 + $3 };
173
174 write:
175
176 exp: exp "+" exp { $$ = $1 + $3; };
177
8617d87e
AD
178** Bug fixes
179
0e164d43
AD
180*** Type names are now properly escaped.
181
182*** glr.cc: set_debug_level and debug_level work as expected.
a4107f24 183
26313726
AD
184*** Stray @ or $ in actions
185
186 While Bison used to warn about stray $ or @ in action rules, it did not
187 for other actions such as printers, destructors, or initial actions. It
188 now does.
189
cd735a8c 190** Type names in actions
4982f078
AD
191
192 For consistency with rule actions, it is now possible to qualify $$ by a
cd735a8c 193 type-name in destructors, printers, and initial actions. For instance:
4982f078
AD
194
195 %printer { fprintf (yyo, "(%d, %f)", $<ival>$, $<fval>$); } <*> <>;
196
197 will display two values for each typed and untyped symbol (provided
cd735a8c 198 that YYSTYPE has both "ival" and "fval" fields).
60aa04a2 199
1505e8bb 200* Noteworthy changes in release 2.6 (2012-07-19) [stable]
0f11eec2 201
9553083c
AD
202** Future changes:
203
55d1006f
AD
204 The next major release of Bison will drop support for the following
205 deprecated features. Please report disagreements to bug-bison@gnu.org.
0f11eec2 206
aaf61036 207*** K&R C parsers
55d1006f
AD
208
209 Support for generating parsers in K&R C will be removed. Parsers
242cc08e 210 generated for C support ISO C90, and are tested with ISO C99 and ISO C11
55d1006f
AD
211 compilers.
212
258cddbc 213*** Features deprecated since Bison 1.875
0f11eec2 214
258cddbc
AD
215 The definitions of yystype and yyltype will be removed; use YYSTYPE and
216 YYLTYPE.
0f11eec2 217
258cddbc
AD
218 YYPARSE_PARAM and YYLEX_PARAM, deprecated in favor of %parse-param and
219 %lex-param, will no longer be supported.
220
221 Support for the preprocessor symbol YYERROR_VERBOSE will be removed, use
222 %error-verbose.
55d1006f
AD
223
224*** The generated header will be included (yacc.c)
0f11eec2
AD
225
226 Instead of duplicating the content of the generated header (definition of
55d1006f
AD
227 YYSTYPE, yyparse declaration etc.), the generated parser will include it,
228 as is already the case for GLR or C++ parsers. This change is deferred
229 because existing versions of ylwrap (e.g., Automake 1.12.1) do not support
230 it.
0f11eec2 231
c2425191 232** Generated Parser Headers
56ca3d8f 233
258cddbc 234*** Guards (yacc.c, glr.c, glr.cc)
c3e9f08f
AD
235
236 The generated headers are now guarded, as is already the case for C++
242cc08e 237 parsers (lalr1.cc). For instance, with --defines=foo.h:
c3e9f08f 238
e29f0771
AD
239 #ifndef YY_FOO_H
240 # define YY_FOO_H
241 ...
242 #endif /* !YY_FOO_H */
c3e9f08f 243
258cddbc 244*** New declarations (yacc.c, glr.c)
56ca3d8f
AD
245
246 The generated header now declares yydebug and yyparse. Both honor
247 --name-prefix=bar_, and yield
248
e29f0771 249 int bar_parse (void);
56ca3d8f
AD
250
251 rather than
252
e29f0771
AD
253 #define yyparse bar_parse
254 int yyparse (void);
56ca3d8f
AD
255
256 in order to facilitate the inclusion of several parser headers inside a
257 single compilation unit.
c3e9f08f 258
258cddbc
AD
259*** Exported symbols in C++
260
261 The symbols YYTOKEN_TABLE and YYERROR_VERBOSE, which were defined in the
262 header, are removed, as they prevent the possibility of including several
263 generated headers from a single compilation unit.
264
694af10c
AD
265*** YYLSP_NEEDED
266
267 For the same reasons, the undocumented and unused macro YYLSP_NEEDED is no
268 longer defined.
269
4b3847c3
AD
270** New %define variable: api.prefix
271
272 Now that the generated headers are more complete and properly protected
273 against multiple inclusions, constant names, such as YYSTYPE are a
274 problem. While yyparse and others are properly renamed by %name-prefix,
275 YYSTYPE, YYDEBUG and others have never been affected by it. Because it
276 would introduce backward compatibility issues in projects not expecting
277 YYSTYPE to be renamed, instead of changing the behavior of %name-prefix,
278 it is deprecated in favor of a new %define variable: api.prefix.
279
280 The following examples compares both:
281
282 %name-prefix "bar_" | %define api.prefix "bar_"
283 %token <ival> FOO %token <ival> FOO
284 %union { int ival; } %union { int ival; }
285 %% %%
286 exp: 'a'; exp: 'a';
287
288 bison generates:
289
290 #ifndef BAR_FOO_H #ifndef BAR_FOO_H
291 # define BAR_FOO_H # define BAR_FOO_H
292
293 /* Enabling traces. */ /* Enabling traces. */
5f108727
AD
294 # ifndef YYDEBUG | # ifndef BAR_DEBUG
295 > # if defined YYDEBUG
296 > # if YYDEBUG
297 > # define BAR_DEBUG 1
298 > # else
299 > # define BAR_DEBUG 0
300 > # endif
301 > # else
302 # define YYDEBUG 0 | # define BAR_DEBUG 0
303 > # endif
304 # endif | # endif
305
306 # if YYDEBUG | # if BAR_DEBUG
4b3847c3
AD
307 extern int bar_debug; extern int bar_debug;
308 # endif # endif
309
310 /* Tokens. */ /* Tokens. */
311 # ifndef YYTOKENTYPE | # ifndef BAR_TOKENTYPE
312 # define YYTOKENTYPE | # define BAR_TOKENTYPE
313 enum yytokentype { | enum bar_tokentype {
314 FOO = 258 FOO = 258
315 }; };
316 # endif # endif
317
318 #if ! defined YYSTYPE \ | #if ! defined BAR_STYPE \
319 && ! defined YYSTYPE_IS_DECLARED | && ! defined BAR_STYPE_IS_DECLARED
320 typedef union YYSTYPE | typedef union BAR_STYPE
321 { {
322 int ival; int ival;
323 } YYSTYPE; | } BAR_STYPE;
324 # define YYSTYPE_IS_DECLARED 1 | # define BAR_STYPE_IS_DECLARED 1
325 #endif #endif
326
327 extern YYSTYPE bar_lval; | extern BAR_STYPE bar_lval;
328
329 int bar_parse (void); int bar_parse (void);
330
331 #endif /* !BAR_FOO_H */ #endif /* !BAR_FOO_H */
332
dfaac272 333* Noteworthy changes in release 2.5.1 (2012-06-05) [stable]
df6e3db0 334
debe2c03 335** Future changes:
765e1bd4 336
e4ab1254 337 The next major release will drop support for generating parsers in K&R C.
041308d0 338
466b4cf2 339** yacc.c: YYBACKUP works as expected.
ef51bfa7 340
d834eca0 341** glr.c improvements:
041308d0 342
d834eca0 343*** Location support is eliminated when not requested:
041308d0 344
e4ab1254
AD
345 GLR parsers used to include location-related code even when locations were
346 not requested, and therefore not even usable.
378e917c 347
d834eca0 348*** __attribute__ is preserved:
d115aad9 349
e4ab1254
AD
350 __attribute__ is no longer disabled when __STRICT_ANSI__ is defined (i.e.,
351 when -std is passed to GCC).
041308d0 352
466b4cf2 353** lalr1.java: several fixes:
041308d0 354
e4ab1254
AD
355 The Java parser no longer throws ArrayIndexOutOfBoundsException if the
356 first token leads to a syntax error. Some minor clean ups.
041308d0 357
22172d47 358** Changes for C++:
ef51bfa7 359
22172d47 360*** C++11 compatibility:
ef51bfa7 361
e4ab1254
AD
362 C and C++ parsers use "nullptr" instead of "0" when __cplusplus is 201103L
363 or higher.
936c88d1 364
22172d47
AD
365*** Header guards
366
367 The header files such as "parser.hh", "location.hh", etc. used a constant
368 name for preprocessor guards, for instance:
369
e29f0771
AD
370 #ifndef BISON_LOCATION_HH
371 # define BISON_LOCATION_HH
372 ...
373 #endif // !BISON_LOCATION_HH
22172d47
AD
374
375 The inclusion guard is now computed from "PREFIX/FILE-NAME", where lower
376 case characters are converted to upper case, and series of
377 non-alphanumerical characters are converted to an underscore.
378
379 With "bison -o lang++/parser.cc", "location.hh" would now include:
380
e29f0771
AD
381 #ifndef YY_LANG_LOCATION_HH
382 # define YY_LANG_LOCATION_HH
383 ...
384 #endif // !YY_LANG_LOCATION_HH
22172d47
AD
385
386*** C++ locations:
936c88d1 387
e4ab1254
AD
388 The position and location constructors (and their initialize methods)
389 accept new arguments for line and column. Several issues in the
390 documentation were fixed.
936c88d1 391
466b4cf2
AD
392** liby is no longer asking for "rpl_fprintf" on some platforms.
393
7e508a2b
AD
394** Changes in the manual:
395
396*** %printer is documented
397
e4ab1254
AD
398 The "%printer" directive, supported since at least Bison 1.50, is finally
399 documented. The "mfcalc" example is extended to demonstrate it.
7e508a2b 400
e4ab1254
AD
401 For consistency with the C skeletons, the C++ parsers now also support
402 "yyoutput" (as an alias to "debug_stream ()").
7e508a2b
AD
403
404*** Several improvements have been made:
466b4cf2 405
e4ab1254
AD
406 The layout for grammar excerpts was changed to a more compact scheme.
407 Named references are motivated. The description of the automaton
408 description file (*.output) is updated to the current format. Incorrect
409 index entries were fixed. Some other errors were fixed.
466b4cf2 410
86b08b49
AD
411** Building bison:
412
413*** Conflicting prototypes with recent/modified Flex.
466b4cf2 414
e4ab1254
AD
415 Fixed build problems with the current, unreleased, version of Flex, and
416 some modified versions of 2.5.35, which have modified function prototypes.
466b4cf2 417
8ef26c2a
AD
418*** Warnings during the build procedure have been eliminated.
419
420*** Several portability problems in the test suite have been fixed:
466b4cf2 421
e4ab1254
AD
422 This includes warnings with some compilers, unexpected behavior of tools
423 such as diff, warning messages from the test suite itself, etc.
466b4cf2 424
91aadcc7 425*** The install-pdf target works properly:
8ef26c2a 426
e4ab1254
AD
427 Running "make install-pdf" (or -dvi, -html, -info, and -ps) no longer
428 halts in the middle of its course.
8ef26c2a 429
28801043 430* Changes in version 2.5 (2011-05-14):
50cca368 431
82f3355e
JD
432** Grammar symbol names can now contain non-initial dashes:
433
434 Consistently with directives (such as %error-verbose) and with
435 %define variables (e.g. push-pull), grammar symbol names may contain
436 dashes in any position except the beginning. This is a GNU
437 extension over POSIX Yacc. Thus, use of this extension is reported
438 by -Wyacc and rejected in Yacc mode (--yacc).
439
f1b238df 440** Named references:
66381412
AR
441
442 Historically, Yacc and Bison have supported positional references
443 ($n, $$) to allow access to symbol values from inside of semantic
444 actions code.
445
446 Starting from this version, Bison can also accept named references.
447 When no ambiguity is possible, original symbol names may be used
448 as named references:
449
4b568fc0 450 if_stmt : "if" cond_expr "then" then_stmt ';'
66381412
AR
451 { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); }
452
453 In the more common case, explicit names may be declared:
454
4b568fc0 455 stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';'
66381412
AR
456 { $res = mk_if_stmt($cond, $then, $else); }
457
5b1ff423 458 Location information is also accessible using @name syntax. When
66381412
AR
459 accessing symbol names containing dots or dashes, explicit bracketing
460 ($[sym.1]) must be used.
461
5b1ff423 462 These features are experimental in this version. More user feedback
66381412
AR
463 will help to stabilize them.
464
f1b238df 465** IELR(1) and canonical LR(1):
eb45ef3b
JD
466
467 IELR(1) is a minimal LR(1) parser table generation algorithm. That
468 is, given any context-free grammar, IELR(1) generates parser tables
7262f54f 469 with the full language-recognition power of canonical LR(1) but with
f1b238df
JD
470 nearly the same number of parser states as LALR(1). This reduction
471 in parser states is often an order of magnitude. More importantly,
eb45ef3b
JD
472 because canonical LR(1)'s extra parser states may contain duplicate
473 conflicts in the case of non-LR(1) grammars, the number of conflicts
474 for IELR(1) is often an order of magnitude less as well. This can
475 significantly reduce the complexity of developing of a grammar.
476
477 Bison can now generate IELR(1) and canonical LR(1) parser tables in
478 place of its traditional LALR(1) parser tables, which remain the
479 default. You can specify the type of parser tables in the grammar
480 file with these directives:
481
cf499cff
JD
482 %define lr.type lalr
483 %define lr.type ielr
484 %define lr.type canonical-lr
eb45ef3b 485
7fceb615 486 The default-reduction optimization in the parser tables can also be
e4ab1254
AD
487 adjusted using "%define lr.default-reductions". For details on both
488 of these features, see the new section "Tuning LR" in the Bison
7fceb615 489 manual.
eb45ef3b
JD
490
491 These features are experimental. More user feedback will help to
492 stabilize them.
493
7fceb615 494** LAC (Lookahead Correction) for syntax error handling:
fcf834f9
JD
495
496 Canonical LR, IELR, and LALR can suffer from a couple of problems
497 upon encountering a syntax error. First, the parser might perform
498 additional parser stack reductions before discovering the syntax
7fceb615 499 error. Such reductions can perform user semantic actions that are
fcf834f9
JD
500 unexpected because they are based on an invalid token, and they
501 cause error recovery to begin in a different syntactic context than
502 the one in which the invalid token was encountered. Second, when
7fceb615 503 verbose error messages are enabled (with %error-verbose or the
e4ab1254 504 obsolete "#define YYERROR_VERBOSE"), the expected token list in the
7fceb615
JD
505 syntax error message can both contain invalid tokens and omit valid
506 tokens.
fcf834f9
JD
507
508 The culprits for the above problems are %nonassoc, default
509 reductions in inconsistent states, and parser state merging. Thus,
510 IELR and LALR suffer the most. Canonical LR can suffer only if
511 %nonassoc is used or if default reductions are enabled for
512 inconsistent states.
513
7fceb615
JD
514 LAC is a new mechanism within the parsing algorithm that solves
515 these problems for canonical LR, IELR, and LALR without sacrificing
516 %nonassoc, default reductions, or state merging. When LAC is in
517 use, canonical LR and IELR behave almost exactly the same for both
518 syntactically acceptable and syntactically unacceptable input.
fcf834f9
JD
519 While LALR still does not support the full language-recognition
520 power of canonical LR and IELR, LAC at least enables LALR's syntax
521 error handling to correctly reflect LALR's language-recognition
522 power.
523
524 Currently, LAC is only supported for deterministic parsers in C.
525 You can enable LAC with the following directive:
526
527 %define parse.lac full
528
e4ab1254 529 See the new section "LAC" in the Bison manual for additional
7fceb615 530 details including a few caveats.
fcf834f9
JD
531
532 LAC is an experimental feature. More user feedback will help to
533 stabilize it.
534
d397d9f0 535** %define improvements:
cf499cff 536
f1b238df 537*** Can now be invoked via the command line:
50cca368 538
de5ab940 539 Each of these command-line options
50cca368 540
de5ab940
JD
541 -D NAME[=VALUE]
542 --define=NAME[=VALUE]
543
544 -F NAME[=VALUE]
545 --force-define=NAME[=VALUE]
50cca368
JD
546
547 is equivalent to this grammar file declaration
548
de5ab940 549 %define NAME ["VALUE"]
50cca368 550
de5ab940
JD
551 except that the manner in which Bison processes multiple definitions
552 for the same NAME differs. Most importantly, -F and --force-define
553 quietly override %define, but -D and --define do not. For further
e4ab1254 554 details, see the section "Bison Options" in the Bison manual.
50cca368 555
f1b238df 556*** Variables renamed:
67212941
JD
557
558 The following %define variables
559
560 api.push_pull
561 lr.keep_unreachable_states
562
563 have been renamed to
564
565 api.push-pull
566 lr.keep-unreachable-states
567
568 The old names are now deprecated but will be maintained indefinitely
569 for backward compatibility.
570
7262f54f 571*** Values no longer need to be quoted in the grammar file:
cf499cff
JD
572
573 If a %define value is an identifier, it no longer needs to be placed
574 within quotations marks. For example,
575
576 %define api.push-pull "push"
577
578 can be rewritten as
579
580 %define api.push-pull push
581
d397d9f0 582*** Unrecognized variables are now errors not warnings.
cdf3f113 583
d397d9f0
JD
584*** Multiple invocations for any variable is now an error not a warning.
585
586** Unrecognized %code qualifiers are now errors not warnings.
587
588** Character literals not of length one:
589
590 Previously, Bison quietly converted all character literals to length
591 one. For example, without warning, Bison interpreted the operators in
592 the following grammar to be the same token:
593
594 exp: exp '++'
595 | exp '+' exp
596 ;
597
598 Bison now warns when a character literal is not of length one. In
599 some future release, Bison will start reporting an error instead.
600
601** Destructor calls fixed for lookaheads altered in semantic actions:
602
603 Previously for deterministic parsers in C, if a user semantic action
604 altered yychar, the parser in some cases used the old yychar value to
605 determine which destructor to call for the lookahead upon a syntax
606 error or upon parser return. This bug has been fixed.
607
608** C++ parsers use YYRHSLOC:
609
610 Similarly to the C parsers, the C++ parsers now define the YYRHSLOC
611 macro and use it in the default YYLLOC_DEFAULT. You are encouraged
e4ab1254
AD
612 to use it. If, for instance, your location structure has "first"
613 and "last" members, instead of
d397d9f0 614
e29f0771
AD
615 # define YYLLOC_DEFAULT(Current, Rhs, N) \
616 do \
617 if (N) \
618 { \
619 (Current).first = (Rhs)[1].location.first; \
620 (Current).last = (Rhs)[N].location.last; \
621 } \
622 else \
623 { \
624 (Current).first = (Current).last = (Rhs)[0].location.last; \
625 } \
626 while (false)
d397d9f0
JD
627
628 use:
629
e29f0771
AD
630 # define YYLLOC_DEFAULT(Current, Rhs, N) \
631 do \
632 if (N) \
633 { \
634 (Current).first = YYRHSLOC (Rhs, 1).first; \
635 (Current).last = YYRHSLOC (Rhs, N).last; \
636 } \
637 else \
638 { \
639 (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \
640 } \
641 while (false)
d397d9f0
JD
642
643** YYLLOC_DEFAULT in C++:
644
645 The default implementation of YYLLOC_DEFAULT used to be issued in
646 the header file. It is now output in the implementation file, after
647 the user %code sections so that its #ifndef guard does not try to
648 override the user's YYLLOC_DEFAULT if provided.
cdf3f113 649
f1b238df 650** YYFAIL now produces warnings and Java parsers no longer implement it:
4395a9ff
JD
651
652 YYFAIL has existed for many years as an undocumented feature of
653 deterministic parsers in C generated by Bison. More recently, it was
654 a documented feature of Bison's experimental Java parsers. As
655 promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a
656 semantic action now produces a deprecation warning, and Java parsers
657 no longer implement YYFAIL at all. For further details, including a
658 discussion of how to suppress C preprocessor warnings about YYFAIL
659 being unused, see the Bison 2.4.2 NEWS entry.
660
f1b238df 661** Temporary hack for adding a semicolon to the user action:
197b82ba
JD
662
663 Previously, Bison appended a semicolon to every user action for
664 reductions when the output language defaulted to C (specifically, when
665 neither %yacc, %language, %skeleton, or equivalent command-line
666 options were specified). This allowed actions such as
667
668 exp: exp "+" exp { $$ = $1 + $3 };
669
670 instead of
671
672 exp: exp "+" exp { $$ = $1 + $3; };
673
674 As a first step in removing this misfeature, Bison now issues a
675 warning when it appends a semicolon. Moreover, in cases where Bison
676 cannot easily determine whether a semicolon is needed (for example, an
677 action ending with a cpp directive or a braced compound initializer),
678 it no longer appends one. Thus, the C compiler might now complain
679 about a missing semicolon where it did not before. Future releases of
680 Bison will cease to append semicolons entirely.
681
d2060f06
JD
682** Verbose syntax error message fixes:
683
e4ab1254 684 When %error-verbose or the obsolete "#define YYERROR_VERBOSE" is
7fceb615
JD
685 specified, syntax error messages produced by the generated parser
686 include the unexpected token as well as a list of expected tokens.
687 The effect of %nonassoc on these verbose messages has been corrected
688 in two ways, but a more complete fix requires LAC, described above:
d2060f06
JD
689
690*** When %nonassoc is used, there can exist parser states that accept no
691 tokens, and so the parser does not always require a lookahead token
692 in order to detect a syntax error. Because no unexpected token or
693 expected tokens can then be reported, the verbose syntax error
694 message described above is suppressed, and the parser instead
e4ab1254 695 reports the simpler message, "syntax error". Previously, this
d2060f06
JD
696 suppression was sometimes erroneously triggered by %nonassoc when a
697 lookahead was actually required. Now verbose messages are
698 suppressed only when all previous lookaheads have already been
699 shifted or discarded.
700
701*** Previously, the list of expected tokens erroneously included tokens
702 that would actually induce a syntax error because conflicts for them
703 were resolved with %nonassoc in the current parser state. Such
704 tokens are now properly omitted from the list.
705
706*** Expected token lists are still often wrong due to state merging
fcf834f9
JD
707 (from LALR or IELR) and default reductions, which can both add
708 invalid tokens and subtract valid tokens. Canonical LR almost
709 completely fixes this problem by eliminating state merging and
710 default reductions. However, there is one minor problem left even
711 when using canonical LR and even after the fixes above. That is,
712 if the resolution of a conflict with %nonassoc appears in a later
713 parser state than the one at which some syntax error is
714 discovered, the conflicted token is still erroneously included in
715 the expected token list. Bison's new LAC implementation,
716 described above, eliminates this problem and the need for
717 canonical LR. However, LAC is still experimental and is disabled
718 by default.
53f036ce 719
1a33f4f6
JD
720** Java skeleton fixes:
721
722*** A location handling bug has been fixed.
723
724*** The top element of each of the value stack and location stack is now
725 cleared when popped so that it can be garbage collected.
6771a463 726
02803d55
JD
727*** Parser traces now print the top element of the stack.
728
86408959
JD
729** -W/--warnings fixes:
730
e4ab1254 731*** Bison now properly recognizes the "no-" versions of categories:
86408959
JD
732
733 For example, given the following command line, Bison now enables all
734 warnings except warnings for incompatibilities with POSIX Yacc:
735
736 bison -Wall,no-yacc gram.y
737
786743d5
JD
738*** Bison now treats S/R and R/R conflicts like other warnings:
739
740 Previously, conflict reports were independent of Bison's normal
741 warning system. Now, Bison recognizes the warning categories
e4ab1254 742 "conflicts-sr" and "conflicts-rr". This change has important
786743d5
JD
743 consequences for the -W and --warnings command-line options. For
744 example:
745
746 bison -Wno-conflicts-sr gram.y # S/R conflicts not reported
747 bison -Wno-conflicts-rr gram.y # R/R conflicts not reported
748 bison -Wnone gram.y # no conflicts are reported
749 bison -Werror gram.y # any conflict is an error
750
751 However, as before, if the %expect or %expect-rr directive is
752 specified, an unexpected number of conflicts is an error, and an
753 expected number of conflicts is not reported, so -W and --warning
754 then have no effect on the conflict report.
755
e4ab1254 756*** The "none" category no longer disables a preceding "error":
bf0e44e8
JD
757
758 For example, for the following command line, Bison now reports
759 errors instead of warnings for incompatibilities with POSIX Yacc:
760
761 bison -Werror,none,yacc gram.y
762
e4ab1254 763*** The "none" category now disables all Bison warnings:
c39014ae 764
e4ab1254 765 Previously, the "none" category disabled only Bison warnings for
c39014ae
JD
766 which there existed a specific -W/--warning category. However,
767 given the following command line, Bison is now guaranteed to
768 suppress all warnings:
769
770 bison -Wnone gram.y
771
1f36f544
JD
772** Precedence directives can now assign token number 0:
773
774 Since Bison 2.3b, which restored the ability of precedence
775 directives to assign token numbers, doing so for token number 0 has
776 produced an assertion failure. For example:
777
778 %left END 0
779
780 This bug has been fixed.
781
64877e5e 782* Changes in version 2.4.3 (2010-08-05):
8b9e021f 783
2bfcac9a
JD
784** Bison now obeys -Werror and --warnings=error for warnings about
785 grammar rules that are useless in the parser due to conflicts.
786
8b9e021f
JD
787** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have
788 been fixed.
789
4ad3921d
JD
790** Failures in the test suite for GCC 4.5 have been fixed.
791
06cb07d5
JD
792** Failures in the test suite for some versions of Sun Studio C++ have
793 been fixed.
794
9b5049bd
JD
795** Contrary to Bison 2.4.2's NEWS entry, it has been decided that
796 warnings about undefined %prec identifiers will not be converted to
797 errors in Bison 2.5. They will remain warnings, which should be
798 sufficient for POSIX while avoiding backward compatibility issues.
799
93d7dde9
JD
800** Minor documentation fixes.
801
e19a049c 802* Changes in version 2.4.2 (2010-03-20):
74553c98 803
f39ab286
JD
804** Some portability problems that resulted in failures and livelocks
805 in the test suite on some versions of at least Solaris, AIX, HP-UX,
e19a049c
JD
806 RHEL4, and Tru64 have been addressed. As a result, fatal Bison
807 errors should no longer cause M4 to report a broken pipe on the
f39ab286
JD
808 affected platforms.
809
e4ab1254 810** "%prec IDENTIFIER" requires IDENTIFIER to be defined separately.
8bb3a2e7
JD
811
812 POSIX specifies that an error be reported for any identifier that does
813 not appear on the LHS of a grammar rule and that is not defined by
814 %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this
815 error report for the case when an identifier appears only after a
816 %prec directive. It is now restored. However, for backward
817 compatibility with recent Bison releases, it is only a warning for
818 now. In Bison 2.5 and later, it will return to being an error.
9b5049bd
JD
819 [Between the 2.4.2 and 2.4.3 releases, it was decided that this
820 warning will not be converted to an error in Bison 2.5.]
8bb3a2e7 821
d8911864
EB
822** Detection of GNU M4 1.4.6 or newer during configure is improved.
823
a603c6e0
JD
824** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS,
825 YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now
826 avoided.
c938d650 827
98a345a2
JD
828** %code is now a permanent feature.
829
830 A traditional Yacc prologue directive is written in the form:
831
832 %{CODE%}
833
834 To provide a more flexible alternative, Bison 2.3b introduced the
835 %code directive with the following forms for C/C++:
836
837 %code {CODE}
838 %code requires {CODE}
839 %code provides {CODE}
840 %code top {CODE}
841
842 These forms are now considered permanent features of Bison. See the
843 %code entries in the section "Bison Declaration Summary" in the Bison
844 manual for a summary of their functionality. See the section
845 "Prologue Alternatives" for a detailed discussion including the
846 advantages of %code over the traditional Yacc prologue directive.
847
848 Bison's Java feature as a whole including its current usage of %code
849 is still considered experimental.
850
1625df5b
JD
851** YYFAIL is deprecated and will eventually be removed.
852
853 YYFAIL has existed for many years as an undocumented feature of
854 deterministic parsers in C generated by Bison. Previously, it was
855 documented for Bison's experimental Java parsers. YYFAIL is no longer
856 documented for Java parsers and is formally deprecated in both cases.
857 Users are strongly encouraged to migrate to YYERROR, which is
858 specified by POSIX.
859
860 Like YYERROR, you can invoke YYFAIL from a semantic action in order to
861 induce a syntax error. The most obvious difference from YYERROR is
862 that YYFAIL will automatically invoke yyerror to report the syntax
863 error so that you don't have to. However, there are several other
864 subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from
e4ab1254 865 inherent flaws when %error-verbose or "#define YYERROR_VERBOSE" is
1625df5b
JD
866 used. For a more detailed discussion, see:
867
868 http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html
869
870 The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but
871 deterministic parsers in C will continue to implement it. However,
872 because YYFAIL is already flawed, it seems futile to try to make new
873 Bison features compatible with it. Thus, during parser generation,
874 Bison 2.5 will produce a warning whenever it discovers YYFAIL in a
875 rule action. In a later release, YYFAIL will be disabled for
e4ab1254 876 %error-verbose and "#define YYERROR_VERBOSE". Eventually, YYFAIL will
1625df5b
JD
877 be removed altogether.
878
879 There exists at least one case where Bison 2.5's YYFAIL warning will
880 be a false positive. Some projects add phony uses of YYFAIL and other
881 Bison-defined macros for the sole purpose of suppressing C
882 preprocessor warnings (from GCC cpp's -Wunused-macros, for example).
883 To avoid Bison's future warning, such YYFAIL uses can be moved to the
e4ab1254 884 epilogue (that is, after the second "%%") in the Bison input file. In
1625df5b
JD
885 this release (2.4.2), Bison already generates its own code to suppress
886 C preprocessor warnings for YYFAIL, so projects can remove their own
887 phony uses of YYFAIL if compatibility with Bison releases prior to
888 2.4.2 is not necessary.
889
2755de8f
AD
890** Internationalization.
891
892 Fix a regression introduced in Bison 2.4: Under some circumstances,
893 message translations were not installed although supported by the
894 host system.
895
74553c98 896* Changes in version 2.4.1 (2008-12-11):
c9ba9e59 897
0ea583d2
AD
898** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc
899 declarations have been fixed.
1979121c 900
0ea583d2
AD
901** Temporary hack for adding a semicolon to the user action.
902
903 Bison used to prepend a trailing semicolon at the end of the user
904 action for reductions. This allowed actions such as
905
906 exp: exp "+" exp { $$ = $1 + $3 };
907
908 instead of
909
910 exp: exp "+" exp { $$ = $1 + $3; };
911
e4ab1254 912 Some grammars still depend on this "feature". Bison 2.4.1 restores
0ea583d2
AD
913 the previous behavior in the case of C output (specifically, when
914 neither %language or %skeleton or equivalent command-line options
915 are used) to leave more time for grammars depending on the old
916 behavior to be adjusted. Future releases of Bison will disable this
917 feature.
918
919** A few minor improvements to the Bison manual.
c9ba9e59 920
402b123d 921* Changes in version 2.4 (2008-11-02):
7bd1665a 922
402b123d 923** %language is an experimental feature.
ed4d67dc
JD
924
925 We first introduced this feature in test release 2.3b as a cleaner
926 alternative to %skeleton. Since then, we have discussed the possibility of
927 modifying its effect on Bison's output file names. Thus, in this release,
928 we consider %language to be an experimental feature that will likely evolve
929 in future releases.
7bd1665a 930
402b123d 931** Forward compatibility with GNU M4 has been improved.
241fda7a 932
402b123d 933** Several bugs in the C++ skeleton and the experimental Java skeleton have been
241fda7a
JD
934 fixed.
935
402b123d 936* Changes in version 2.3b (2008-05-27):
35fe0834 937
402b123d 938** The quotes around NAME that used to be required in the following directive
d9df47b6
JD
939 are now deprecated:
940
941 %define NAME "VALUE"
942
e4ab1254 943** The directive "%pure-parser" is now deprecated in favor of:
d9df47b6
JD
944
945 %define api.pure
946
947 which has the same effect except that Bison is more careful to warn about
948 unreasonable usage in the latter case.
949
402b123d 950** Push Parsing
c373bf8b
JD
951
952 Bison can now generate an LALR(1) parser in C with a push interface. That
e4ab1254
AD
953 is, instead of invoking "yyparse", which pulls tokens from "yylex", you can
954 push one token at a time to the parser using "yypush_parse", which will
c373bf8b
JD
955 return to the caller after processing each token. By default, the push
956 interface is disabled. Either of the following directives will enable it:
957
958 %define api.push_pull "push" // Just push; does not require yylex.
959 %define api.push_pull "both" // Push and pull; requires yylex.
960
e4ab1254 961 See the new section "A Push Parser" in the Bison manual for details.
c373bf8b 962
59da312b
JD
963 The current push parsing interface is experimental and may evolve. More user
964 feedback will help to stabilize it.
965
402b123d 966** The -g and --graph options now output graphs in Graphviz DOT format,
8e55b3aa
JD
967 not VCG format. Like --graph, -g now also takes an optional FILE argument
968 and thus cannot be bundled with other short options.
c373bf8b 969
402b123d 970** Java
59da312b
JD
971
972 Bison can now generate an LALR(1) parser in Java. The skeleton is
e4ab1254 973 "data/lalr1.java". Consider using the new %language directive instead of
59da312b
JD
974 %skeleton to select it.
975
e4ab1254 976 See the new section "Java Parsers" in the Bison manual for details.
59da312b
JD
977
978 The current Java interface is experimental and may evolve. More user
979 feedback will help to stabilize it.
980
402b123d 981** %language
59da312b
JD
982
983 This new directive specifies the programming language of the generated
d43f77e7
PB
984 parser, which can be C (the default), C++, or Java. Besides the skeleton
985 that Bison uses, the directive affects the names of the generated files if
986 the grammar file's name ends in ".y".
59da312b 987
402b123d 988** XML Automaton Report
59da312b
JD
989
990 Bison can now generate an XML report of the LALR(1) automaton using the new
e4ab1254 991 "--xml" option. The current XML schema is experimental and may evolve. More
59da312b 992 user feedback will help to stabilize it.
c373bf8b 993
402b123d 994** The grammar file may now specify the name of the parser header file using
c373bf8b
JD
995 %defines. For example:
996
997 %defines "parser.h"
998
402b123d 999** When reporting useless rules, useless nonterminals, and unused terminals,
d80fb37a
JD
1000 Bison now employs the terms "useless in grammar" instead of "useless",
1001 "useless in parser" instead of "never reduced", and "unused in grammar"
1002 instead of "unused".
cff03fb2 1003
402b123d 1004** Unreachable State Removal
c373bf8b
JD
1005
1006 Previously, Bison sometimes generated parser tables containing unreachable
31984206
JD
1007 states. A state can become unreachable during conflict resolution if Bison
1008 disables a shift action leading to it from a predecessor state. Bison now:
75ad86ee
JD
1009
1010 1. Removes unreachable states.
1011
1012 2. Does not report any conflicts that appeared in unreachable states.
1013 WARNING: As a result, you may need to update %expect and %expect-rr
1014 directives in existing grammar files.
1015
1016 3. For any rule used only in such states, Bison now reports the rule as
cff03fb2 1017 "useless in parser due to conflicts".
75ad86ee 1018
31984206
JD
1019 This feature can be disabled with the following directive:
1020
1021 %define lr.keep_unreachable_states
1022
e4ab1254 1023 See the %define entry in the "Bison Declaration Summary" in the Bison manual
31984206
JD
1024 for further discussion.
1025
e4ab1254 1026** Lookahead Set Correction in the ".output" Report
b1cc23c4 1027
e4ab1254
AD
1028 When instructed to generate a ".output" file including lookahead sets
1029 (using "--report=lookahead", for example), Bison now prints each reduction's
88c78747
JD
1030 lookahead set only next to the associated state's one item that (1) is
1031 associated with the same rule as the reduction and (2) has its dot at the end
1032 of its RHS. Previously, Bison also erroneously printed the lookahead set
1033 next to all of the state's other items associated with the same rule. This
e4ab1254 1034 bug affected only the ".output" file and not the generated parser source
88c78747
JD
1035 code.
1036
e4ab1254 1037** --report-file=FILE is a new option to override the default ".output" file
59da312b 1038 name.
1bb2bd75 1039
e4ab1254 1040** The "=" that used to be required in the following directives is now
02975b9a
JD
1041 deprecated:
1042
1043 %file-prefix "parser"
1044 %name-prefix "c_"
1045 %output "parser.c"
1046
e4ab1254 1047** An Alternative to "%{...%}" -- "%code QUALIFIER {CODE}"
c373bf8b
JD
1048
1049 Bison 2.3a provided a new set of directives as a more flexible alternative to
8e0a5e9e
JD
1050 the traditional Yacc prologue blocks. Those have now been consolidated into
1051 a single %code directive with an optional qualifier field, which identifies
1052 the purpose of the code and thus the location(s) where Bison should generate
1053 it:
1054
e4ab1254
AD
1055 1. "%code {CODE}" replaces "%after-header {CODE}"
1056 2. "%code requires {CODE}" replaces "%start-header {CODE}"
1057 3. "%code provides {CODE}" replaces "%end-header {CODE}"
1058 4. "%code top {CODE}" replaces "%before-header {CODE}"
8e0a5e9e 1059
e4ab1254
AD
1060 See the %code entries in section "Bison Declaration Summary" in the Bison
1061 manual for a summary of the new functionality. See the new section "Prologue
1062 Alternatives" for a detailed discussion including the advantages of %code
8e0a5e9e
JD
1063 over the traditional Yacc prologues.
1064
1065 The prologue alternatives are experimental. More user feedback will help to
1066 determine whether they should become permanent features.
1067
402b123d 1068** Revised warning: unset or unused mid-rule values
17bd8a73
JD
1069
1070 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
1071 used within any of the actions of the parent rule. For example, Bison warns
1072 about unused $2 in:
1073
1074 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
1075
1076 Now, Bison also warns about mid-rule values that are used but not set. For
1077 example, Bison warns about unset $$ in the mid-rule action in:
1078
1079 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
1080
1081 However, Bison now disables both of these warnings by default since they
1082 sometimes prove to be false alarms in existing grammars employing the Yacc
1083 constructs $0 or $-N (where N is some positive integer).
1084
e4ab1254
AD
1085 To enable these warnings, specify the option "--warnings=midrule-values" or
1086 "-W", which is a synonym for "--warnings=all".
17bd8a73 1087
e4ab1254 1088** Default %destructor or %printer with "<*>" or "<>"
c373bf8b
JD
1089
1090 Bison now recognizes two separate kinds of default %destructor's and
12e35840
JD
1091 %printer's:
1092
e4ab1254 1093 1. Place "<*>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1094 %destructor/%printer for all grammar symbols for which you have formally
1095 declared semantic type tags.
1096
e4ab1254 1097 2. Place "<>" in a %destructor/%printer symbol list to define a default
12e35840
JD
1098 %destructor/%printer for all grammar symbols without declared semantic
1099 type tags.
1100
e4ab1254
AD
1101 Bison no longer supports the "%symbol-default" notation from Bison 2.3a.
1102 "<*>" and "<>" combined achieve the same effect with one exception: Bison no
12e35840
JD
1103 longer applies any %destructor to a mid-rule value if that mid-rule value is
1104 not actually ever referenced using either $$ or $n in a semantic action.
1105
85894313
JD
1106 The default %destructor's and %printer's are experimental. More user
1107 feedback will help to determine whether they should become permanent
1108 features.
1109
e4ab1254 1110 See the section "Freeing Discarded Symbols" in the Bison manual for further
12e35840
JD
1111 details.
1112
402b123d 1113** %left, %right, and %nonassoc can now declare token numbers. This is required
e4ab1254 1114 by POSIX. However, see the end of section "Operator Precedence" in the Bison
ab7f29f8
JD
1115 manual for a caveat concerning the treatment of literal strings.
1116
402b123d 1117** The nonfunctional --no-parser, -n, and %no-parser options have been
b1cc23c4
JD
1118 completely removed from Bison.
1119
402b123d 1120* Changes in version 2.3a, 2006-09-13:
742e4900 1121
402b123d 1122** Instead of %union, you can define and use your own union type
ddc8ede1
PE
1123 YYSTYPE if your grammar contains at least one <type> tag.
1124 Your YYSTYPE need not be a macro; it can be a typedef.
1125 This change is for compatibility with other Yacc implementations,
1126 and is required by POSIX.
1127
402b123d 1128** Locations columns and lines start at 1.
cd48d21d
AD
1129 In accordance with the GNU Coding Standards and Emacs.
1130
402b123d 1131** You may now declare per-type and default %destructor's and %printer's:
ec5479ce
JD
1132
1133 For example:
1134
b2a0b7ca
JD
1135 %union { char *string; }
1136 %token <string> STRING1
1137 %token <string> STRING2
1138 %type <string> string1
1139 %type <string> string2
1140 %union { char character; }
1141 %token <character> CHR
1142 %type <character> chr
1143 %destructor { free ($$); } %symbol-default
1144 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
1145 %destructor { } <character>
1146
1147 guarantees that, when the parser discards any user-defined symbol that has a
e4ab1254
AD
1148 semantic type tag other than "<character>", it passes its semantic value to
1149 "free". However, when the parser discards a "STRING1" or a "string1", it
1150 also prints its line number to "stdout". It performs only the second
1151 "%destructor" in this case, so it invokes "free" only once.
ec5479ce 1152
85894313
JD
1153 [Although we failed to mention this here in the 2.3a release, the default
1154 %destructor's and %printer's were experimental, and they were rewritten in
1155 future versions.]
1156
e4ab1254
AD
1157** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with "-y",
1158 "--yacc", or "%yacc"), Bison no longer generates #define statements for
b931235e
JD
1159 associating token numbers with token names. Removing the #define statements
1160 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
1161 requires them. Bison still generates an enum for token names in all cases.
1162
402b123d 1163** Handling of traditional Yacc prologue blocks is now more consistent but
34f98f46 1164 potentially incompatible with previous releases of Bison.
9bc0dd67
JD
1165
1166 As before, you declare prologue blocks in your grammar file with the
e4ab1254 1167 "%{ ... %}" syntax. To generate the pre-prologue, Bison concatenates all
34f98f46
JD
1168 prologue blocks that you've declared before the first %union. To generate
1169 the post-prologue, Bison concatenates all prologue blocks that you've
ddc8ede1 1170 declared after the first %union.
9bc0dd67 1171
34f98f46 1172 Previous releases of Bison inserted the pre-prologue into both the header
9bc0dd67
JD
1173 file and the code file in all cases except for LALR(1) parsers in C. In the
1174 latter case, Bison inserted it only into the code file. For parsers in C++,
1175 the point of insertion was before any token definitions (which associate
1176 token numbers with names). For parsers in C, the point of insertion was
1177 after the token definitions.
1178
1179 Now, Bison never inserts the pre-prologue into the header file. In the code
1180 file, it always inserts it before the token definitions.
1181
402b123d 1182** Bison now provides a more flexible alternative to the traditional Yacc
34f98f46
JD
1183 prologue blocks: %before-header, %start-header, %end-header, and
1184 %after-header.
1185
1186 For example, the following declaration order in the grammar file reflects the
1187 order in which Bison will output these code blocks. However, you are free to
1188 declare these code blocks in your grammar file in whatever order is most
1189 convenient for you:
1190
1191 %before-header {
1192 /* Bison treats this block like a pre-prologue block: it inserts it into
1193 * the code file before the contents of the header file. It does *not*
1194 * insert it into the header file. This is a good place to put
1195 * #include's that you want at the top of your code file. A common
e4ab1254 1196 * example is '#include "system.h"'. */
34f98f46
JD
1197 }
1198 %start-header {
1199 /* Bison inserts this block into both the header file and the code file.
1200 * In both files, the point of insertion is before any Bison-generated
1201 * token, semantic type, location type, and class definitions. This is a
1202 * good place to define %union dependencies, for example. */
9bc0dd67
JD
1203 }
1204 %union {
34f98f46
JD
1205 /* Unlike the traditional Yacc prologue blocks, the output order for the
1206 * new %*-header blocks is not affected by their declaration position
1207 * relative to any %union in the grammar file. */
9bc0dd67 1208 }
34f98f46
JD
1209 %end-header {
1210 /* Bison inserts this block into both the header file and the code file.
1211 * In both files, the point of insertion is after the Bison-generated
1212 * definitions. This is a good place to declare or define public
1213 * functions or data structures that depend on the Bison-generated
1214 * definitions. */
9bc0dd67 1215 }
34f98f46
JD
1216 %after-header {
1217 /* Bison treats this block like a post-prologue block: it inserts it into
1218 * the code file after the contents of the header file. It does *not*
1219 * insert it into the header file. This is a good place to declare or
1220 * define internal functions or data structures that depend on the
1221 * Bison-generated definitions. */
1222 }
1223
1224 If you have multiple occurrences of any one of the above declarations, Bison
1225 will concatenate the contents in declaration order.
9bc0dd67 1226
85894313
JD
1227 [Although we failed to mention this here in the 2.3a release, the prologue
1228 alternatives were experimental, and they were rewritten in future versions.]
1229
e4ab1254 1230** The option "--report=look-ahead" has been changed to "--report=lookahead".
9e6e7ed2
PE
1231 The old spelling still works, but is not documented and may be removed
1232 in a future release.
742e4900 1233
402b123d 1234* Changes in version 2.3, 2006-06-05:
4ad3ed84 1235
e4ab1254 1236** GLR grammars should now use "YYRECOVERING ()" instead of "YYRECOVERING",
4ad3ed84
PE
1237 for compatibility with LALR(1) grammars.
1238
402b123d 1239** It is now documented that any definition of YYSTYPE or YYLTYPE should
4ad3ed84
PE
1240 be to a type name that does not contain parentheses or brackets.
1241
402b123d 1242* Changes in version 2.2, 2006-05-19:
193d7c70 1243
402b123d 1244** The distribution terms for all Bison-generated parsers now permit
193d7c70
PE
1245 using the parsers in nonfree programs. Previously, this permission
1246 was granted only for Bison-generated LALR(1) parsers in C.
5f4236a0 1247
402b123d 1248** %name-prefix changes the namespace name in C++ outputs.
aa08666d 1249
402b123d 1250** The C++ parsers export their token_type.
5f4236a0 1251
402b123d 1252** Bison now allows multiple %union declarations, and concatenates
d6ca7905
PE
1253 their contents together.
1254
402b123d 1255** New warning: unused values
4d7bc38c
PE
1256 Right-hand side symbols whose values are not used are reported,
1257 if the symbols have destructors. For instance:
affac613 1258
8f3596a6 1259 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
e9690142
JD
1260 | exp "+" exp
1261 ;
affac613 1262
8f3596a6
AD
1263 will trigger a warning about $$ and $5 in the first rule, and $3 in
1264 the second ($1 is copied to $$ by the default rule). This example
4e26c69e 1265 most likely contains three errors, and could be rewritten as:
affac613 1266
4e26c69e 1267 exp: exp "?" exp ":" exp
e9690142
JD
1268 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
1269 | exp "+" exp
1270 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
1271 ;
affac613 1272
4e26c69e
PE
1273 However, if the original actions were really intended, memory leaks
1274 and all, the warnings can be suppressed by letting Bison believe the
1275 values are used, e.g.:
721be13c 1276
8f3596a6 1277 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
e9690142
JD
1278 | exp "+" exp { $$ = $1; (void) $3; }
1279 ;
721be13c 1280
84866159
AD
1281 If there are mid-rule actions, the warning is issued if no action
1282 uses it. The following triggers no warning: $1 and $3 are used.
1283
1284 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
1285
721be13c
PE
1286 The warning is intended to help catching lost values and memory leaks.
1287 If a value is ignored, its associated memory typically is not reclaimed.
affac613 1288
402b123d 1289** %destructor vs. YYABORT, YYACCEPT, and YYERROR.
9d9b8b70
PE
1290 Destructors are now called when user code invokes YYABORT, YYACCEPT,
1291 and YYERROR, for all objects on the stack, other than objects
1292 corresponding to the right-hand side of the current rule.
a85284cf 1293
402b123d 1294** %expect, %expect-rr
035aa4a0
PE
1295 Incorrect numbers of expected conflicts are now actual errors,
1296 instead of warnings.
1297
402b123d 1298** GLR, YACC parsers.
4e26c69e
PE
1299 The %parse-params are available in the destructors (and the
1300 experimental printers) as per the documentation.
4b367315 1301
e4ab1254 1302** Bison now warns if it finds a stray "$" or "@" in an action.
ad6a9b97 1303
402b123d 1304** %require "VERSION"
4e26c69e
PE
1305 This specifies that the grammar file depends on features implemented
1306 in Bison version VERSION or higher.
b50d2359 1307
402b123d 1308** lalr1.cc: The token and value types are now class members.
e14d0ab6
AD
1309 The tokens were defined as free form enums and cpp macros. YYSTYPE
1310 was defined as a free form union. They are now class members:
e4ab1254
AD
1311 tokens are enumerations of the "yy::parser::token" struct, and the
1312 semantic values have the "yy::parser::semantic_type" type.
fb9712a9
AD
1313
1314 If you do not want or can update to this scheme, the directive
e4ab1254 1315 '%define "global_tokens_and_yystype" "1"' triggers the global
b50d2359
AD
1316 definition of tokens and YYSTYPE. This change is suitable both
1317 for previous releases of Bison, and this one.
fb9712a9 1318
b50d2359 1319 If you wish to update, then make sure older version of Bison will
e4ab1254 1320 fail using '%require "2.2"'.
fb9712a9 1321
402b123d 1322** DJGPP support added.
193d7c70 1323\f
402b123d 1324* Changes in version 2.1, 2005-09-16:
1ce59070 1325
402b123d 1326** The C++ lalr1.cc skeleton supports %lex-param.
e14d0ab6 1327
402b123d 1328** Bison-generated parsers now support the translation of diagnostics like
baf785db
PE
1329 "syntax error" into languages other than English. The default
1330 language is still English. For details, please see the new
0410a6e0
PE
1331 Internationalization section of the Bison manual. Software
1332 distributors should also see the new PACKAGING file. Thanks to
1333 Bruno Haible for this new feature.
1ce59070 1334
402b123d 1335** Wording in the Bison-generated parsers has been changed slightly to
1a059451
PE
1336 simplify translation. In particular, the message "memory exhausted"
1337 has replaced "parser stack overflow", as the old message was not
1338 always accurate for modern Bison-generated parsers.
1339
402b123d 1340** Destructors are now called when the parser aborts, for all symbols left
258b75ca
PE
1341 behind on the stack. Also, the start symbol is now destroyed after a
1342 successful parse. In both cases, the behavior was formerly inconsistent.
1343
402b123d 1344** When generating verbose diagnostics, Bison-generated parsers no longer
72f000b0
PE
1345 quote the literal strings associated with tokens. For example, for
1346 a syntax error associated with '%token NUM "number"' they might
1347 print 'syntax error, unexpected number' instead of 'syntax error,
1348 unexpected "number"'.
193d7c70 1349\f
402b123d 1350* Changes in version 2.0, 2004-12-25:
efeed023 1351
402b123d 1352** Possibly-incompatible changes
d7e14fc0 1353
82de6b0d
PE
1354 - Bison-generated parsers no longer default to using the alloca function
1355 (when available) to extend the parser stack, due to widespread
1356 problems in unchecked stack-overflow detection. You can "#define
1357 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
1358 the manual to determine safe values for YYMAXDEPTH in that case.
8dd162d3 1359
82de6b0d
PE
1360 - Error token location.
1361 During error recovery, the location of the syntax error is updated
1362 to cover the whole sequence covered by the error token: it includes
1363 the shifted symbols thrown away during the first part of the error
1364 recovery, and the lookahead rejected during the second part.
18d192f0 1365
82de6b0d
PE
1366 - Semicolon changes:
1367 . Stray semicolons are no longer allowed at the start of a grammar.
1368 . Semicolons are now required after in-grammar declarations.
e342c3be 1369
82de6b0d
PE
1370 - Unescaped newlines are no longer allowed in character constants or
1371 string literals. They were never portable, and GCC 3.4.0 has
1372 dropped support for them. Better diagnostics are now generated if
1373 forget a closing quote.
8dd162d3 1374
82de6b0d 1375 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
f74b6f91 1376
402b123d 1377** New features
1452af69 1378
82de6b0d 1379 - GLR grammars now support locations.
4febdd96 1380
82de6b0d
PE
1381 - New directive: %initial-action.
1382 This directive allows the user to run arbitrary code (including
1383 initializing @$) from yyparse before parsing starts.
1452af69 1384
82de6b0d
PE
1385 - A new directive "%expect-rr N" specifies the expected number of
1386 reduce/reduce conflicts in GLR parsers.
1452af69 1387
e4ab1254 1388 - %token numbers can now be hexadecimal integers, e.g., "%token FOO 0x12d".
82de6b0d 1389 This is a GNU extension.
4febdd96 1390
e4ab1254 1391 - The option "--report=lookahead" was changed to "--report=look-ahead".
9e6e7ed2 1392 [However, this was changed back after 2.3.]
1452af69 1393
82de6b0d 1394 - Experimental %destructor support has been added to lalr1.cc.
1452af69 1395
82de6b0d
PE
1396 - New configure option --disable-yacc, to disable installation of the
1397 yacc command and -ly library introduced in 1.875 for POSIX conformance.
6040d338 1398
402b123d 1399** Bug fixes
d5a3fe37 1400
82de6b0d
PE
1401 - For now, %expect-count violations are now just warnings, not errors.
1402 This is for compatibility with Bison 1.75 and earlier (when there are
1403 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
1404 are too many or too few shift/reduce conflicts). However, in future
1405 versions of Bison we plan to improve the %expect machinery so that
1406 these violations will become errors again.
3473d0f8 1407
82de6b0d
PE
1408 - Within Bison itself, numbers (e.g., goto numbers) are no longer
1409 arbitrarily limited to 16-bit counts.
d600ee67 1410
82de6b0d 1411 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
d600ee67 1412\f
402b123d 1413* Changes in version 1.875, 2003-01-01:
963fcc17 1414
402b123d 1415** The documentation license has been upgraded to version 1.2
dc546b0f 1416 of the GNU Free Documentation License.
75eb3bc4 1417
402b123d 1418** syntax error processing
75eb3bc4 1419
dc546b0f
PE
1420 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
1421 locations too. This fixes bugs in error-location computation.
75eb3bc4 1422
dc546b0f
PE
1423 - %destructor
1424 It is now possible to reclaim the memory associated to symbols
1425 discarded during error recovery. This feature is still experimental.
20daca06 1426
dc546b0f
PE
1427 - %error-verbose
1428 This new directive is preferred over YYERROR_VERBOSE.
74724a70 1429
dc546b0f
PE
1430 - #defining yyerror to steal internal variables is discouraged.
1431 It is not guaranteed to work forever.
d1de5372 1432
402b123d 1433** POSIX conformance
d1de5372 1434
dc546b0f
PE
1435 - Semicolons are once again optional at the end of grammar rules.
1436 This reverts to the behavior of Bison 1.33 and earlier, and improves
1437 compatibility with Yacc.
74724a70 1438
e4ab1254
AD
1439 - "parse error" -> "syntax error"
1440 Bison now uniformly uses the term "syntax error"; formerly, the code
1441 and manual sometimes used the term "parse error" instead. POSIX
1442 requires "syntax error" in diagnostics, and it was thought better to
dc546b0f 1443 be consistent.
74724a70 1444
dc546b0f
PE
1445 - The documentation now emphasizes that yylex and yyerror must be
1446 declared before use. C99 requires this.
d1de5372 1447
dc546b0f
PE
1448 - Bison now parses C99 lexical constructs like UCNs and
1449 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
d1de5372 1450
dc546b0f
PE
1451 - File names are properly escaped in C output. E.g., foo\bar.y is
1452 output as "foo\\bar.y".
6780ca7a 1453
dc546b0f 1454 - Yacc command and library now available
e4ab1254 1455 The Bison distribution now installs a "yacc" command, as POSIX requires.
dc546b0f
PE
1456 Also, Bison now installs a small library liby.a containing
1457 implementations of Yacc-compatible yyerror and main functions.
1458 This library is normally not useful, but POSIX requires it.
6e649e65 1459
dc546b0f 1460 - Type clashes now generate warnings, not errors.
6e649e65 1461
dc546b0f
PE
1462 - If the user does not define YYSTYPE as a macro, Bison now declares it
1463 using typedef instead of defining it as a macro.
1464 For consistency, YYLTYPE is also declared instead of defined.
9501dc6e 1465
402b123d 1466** Other compatibility issues
886a425c 1467
e4ab1254
AD
1468 - %union directives can now have a tag before the "{", e.g., the
1469 directive "%union foo {...}" now generates the C code
1470 "typedef union foo { ... } YYSTYPE;"; this is for Yacc compatibility.
1471 The default union tag is "YYSTYPE", for compatibility with Solaris 9 Yacc.
1472 For consistency, YYLTYPE's struct tag is now "YYLTYPE" not "yyltype".
dc546b0f 1473 This is for compatibility with both Yacc and Bison 1.35.
72f889cc 1474
e4ab1254 1475 - ";" is output before the terminating "}" of an action, for
dc546b0f 1476 compatibility with Bison 1.35.
886a425c 1477
dc546b0f 1478 - Bison now uses a Yacc-style format for conflict reports, e.g.,
e4ab1254 1479 "conflicts: 2 shift/reduce, 1 reduce/reduce".
437c2d80 1480
e4ab1254 1481 - "yystype" and "yyltype" are now obsolescent macros instead of being
dc546b0f
PE
1482 typedefs or tags; they are no longer documented and are planned to be
1483 withdrawn in a future release.
2a8d363a 1484
402b123d 1485** GLR parser notes
2a8d363a 1486
dc546b0f
PE
1487 - GLR and inline
1488 Users of Bison have to decide how they handle the portability of the
e4ab1254 1489 C keyword "inline".
959e5f51 1490
e4ab1254
AD
1491 - "parsing stack overflow..." -> "parser stack overflow"
1492 GLR parsers now report "parser stack overflow" as per the Bison manual.
900c5db5 1493
18ad57b3
AD
1494** %parse-param and %lex-param
1495 The macros YYPARSE_PARAM and YYLEX_PARAM provide a means to pass
1496 additional context to yyparse and yylex. They suffer from several
1497 shortcomings:
1498
1499 - a single argument only can be added,
1500 - their types are weak (void *),
242cc08e 1501 - this context is not passed to ancillary functions such as yyerror,
18ad57b3
AD
1502 - only yacc.c parsers support them.
1503
1504 The new %parse-param/%lex-param directives provide a more precise control.
1505 For instance:
1506
1507 %parse-param {int *nastiness}
1508 %lex-param {int *nastiness}
1509 %parse-param {int *randomness}
1510
1511 results in the following signatures:
1512
1513 int yylex (int *nastiness);
1514 int yyparse (int *nastiness, int *randomness);
1515
1516 or, if both %pure-parser and %locations are used:
1517
1518 int yylex (YYSTYPE *lvalp, YYLTYPE *llocp, int *nastiness);
1519 int yyparse (int *nastiness, int *randomness);
1520
402b123d 1521** Bison now warns if it detects conflicting outputs to the same file,
e4ab1254 1522 e.g., it generates a warning for "bison -d -o foo.h foo.y" since
dc546b0f 1523 that command outputs both code and header to foo.h.
6e40b4eb 1524
402b123d 1525** #line in output files
dc546b0f 1526 - --no-line works properly.
6e40b4eb 1527
402b123d 1528** Bison can no longer be built by a K&R C compiler; it requires C89 or
6e40b4eb
AD
1529 later to be built. This change originally took place a few versions
1530 ago, but nobody noticed until we recently asked someone to try
1531 building Bison with a K&R C compiler.
d600ee67 1532\f
402b123d 1533* Changes in version 1.75, 2002-10-14:
7933f2b5 1534
402b123d 1535** Bison should now work on 64-bit hosts.
7933f2b5 1536
402b123d 1537** Indonesian translation thanks to Tedi Heriyanto.
7933f2b5 1538
402b123d 1539** GLR parsers
f50adbbd
AD
1540 Fix spurious parse errors.
1541
402b123d 1542** Pure parsers
f50adbbd
AD
1543 Some people redefine yyerror to steal yyparse' private variables.
1544 Reenable this trick until an official feature replaces it.
1545
402b123d 1546** Type Clashes
d90c934c
AD
1547 In agreement with POSIX and with other Yaccs, leaving a default
1548 action is valid when $$ is untyped, and $1 typed:
1549
e9690142 1550 untyped: ... typed;
d90c934c
AD
1551
1552 but the converse remains an error:
1553
e9690142 1554 typed: ... untyped;
d90c934c 1555
402b123d 1556** Values of mid-rule actions
d90c934c
AD
1557 The following code:
1558
e9690142 1559 foo: { ... } { $$ = $1; } ...
d90c934c
AD
1560
1561 was incorrectly rejected: $1 is defined in the second mid-rule
1562 action, and is equal to the $$ of the first mid-rule action.
d600ee67 1563\f
402b123d 1564* Changes in version 1.50, 2002-10-04:
adc8c848 1565
402b123d 1566** GLR parsing
676385e2
PH
1567 The declaration
1568 %glr-parser
1569 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
1570 almost any context-free grammar, ambiguous or not. The new declarations
e8832397 1571 %dprec and %merge on grammar rules allow parse-time resolution of
676385e2
PH
1572 ambiguities. Contributed by Paul Hilfinger.
1573
7933f2b5 1574 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
420f93c8
PE
1575 like the Alpha, so please stick to 32-bit hosts for now.
1576
402b123d 1577** Output Directory
8c165d89 1578 When not in Yacc compatibility mode, when the output file was not
e4ab1254
AD
1579 specified, running "bison foo/bar.y" created "foo/bar.c". It
1580 now creates "bar.c".
8c165d89 1581
402b123d 1582** Undefined token
007a50a4 1583 The undefined token was systematically mapped to 2 which prevented
e88dbdbf 1584 the use of 2 by the user. This is no longer the case.
007a50a4 1585
402b123d 1586** Unknown token numbers
e88dbdbf 1587 If yylex returned an out of range value, yyparse could die. This is
007a50a4
AD
1588 no longer the case.
1589
402b123d 1590** Error token
e88dbdbf 1591 According to POSIX, the error token must be 256.
23c5a174
AD
1592 Bison extends this requirement by making it a preference: *if* the
1593 user specified that one of her tokens is numbered 256, then error
1594 will be mapped onto another number.
1595
402b123d 1596** Verbose error messages
e4ab1254 1597 They no longer report "..., expecting error or..." for states where
217598da
AD
1598 error recovery is possible.
1599
402b123d 1600** End token
e4ab1254 1601 Defaults to "$end" instead of "$".
217598da 1602
402b123d 1603** Error recovery now conforms to documentation and to POSIX
68cd8af3
PE
1604 When a Bison-generated parser encounters a syntax error, it now pops
1605 the stack until it finds a state that allows shifting the error
1606 token. Formerly, it popped the stack until it found a state that
1607 allowed some non-error action other than a default reduction on the
1608 error token. The new behavior has long been the documented behavior,
1609 and has long been required by POSIX. For more details, please see
337116ba
PE
1610 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
1611 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
68cd8af3 1612
402b123d 1613** Traces
5504898e
AD
1614 Popped tokens and nonterminals are now reported.
1615
402b123d 1616** Larger grammars
a861a339
PE
1617 Larger grammars are now supported (larger token numbers, larger grammar
1618 size (= sum of the LHS and RHS lengths), larger LALR tables).
1619 Formerly, many of these numbers ran afoul of 16-bit limits;
1620 now these limits are 32 bits on most hosts.
355e7c1c 1621
402b123d 1622** Explicit initial rule
643a5994
AD
1623 Bison used to play hacks with the initial rule, which the user does
1624 not write. It is now explicit, and visible in the reports and
1625 graphs as rule 0.
23c5a174 1626
402b123d 1627** Useless rules
643a5994 1628 Before, Bison reported the useless rules, but, although not used,
77714df2 1629 included them in the parsers. They are now actually removed.
23c5a174 1630
402b123d 1631** Useless rules, useless nonterminals
6b98e4b5
AD
1632 They are now reported, as a warning, with their locations.
1633
402b123d 1634** Rules never reduced
e8832397
AD
1635 Rules that can never be reduced because of conflicts are now
1636 reported.
1637
e4ab1254 1638** Incorrect "Token not used"
11652ab3
AD
1639 On a grammar such as
1640
e29f0771
AD
1641 %token useless useful
1642 %%
1643 exp: '0' %prec useful;
11652ab3
AD
1644
1645 where a token was used to set the precedence of the last rule,
e4ab1254 1646 bison reported both "useful" and "useless" as useless tokens.
11652ab3 1647
402b123d 1648** Revert the C++ namespace changes introduced in 1.31
77714df2 1649 as they caused too many portability hassles.
0179dd65 1650
402b123d 1651** Default locations
b2d52318
AD
1652 By an accident of design, the default computation of @$ was
1653 performed after another default computation was performed: @$ = @1.
1654 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
1655 the computation of @$.
adc8c848 1656
402b123d 1657** Token end-of-file
b7c49edf
AD
1658 The token end of file may be specified by the user, in which case,
1659 the user symbol is used in the reports, the graphs, and the verbose
e4ab1254 1660 error messages instead of "$end", which remains being the default.
b7c49edf 1661 For instance
e29f0771 1662 %token MYEOF 0
b7c49edf 1663 or
e29f0771 1664 %token MYEOF 0 "end of file"
fdbcd8e2 1665
402b123d 1666** Semantic parser
fdbcd8e2
AD
1667 This old option, which has been broken for ages, is removed.
1668
402b123d 1669** New translations
a861a339 1670 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
84614e13
AD
1671 Croatian, thanks to Denis Lackovic.
1672
402b123d 1673** Incorrect token definitions
e4ab1254
AD
1674 When given
1675 %token 'a' "A"
1676 bison used to output
1677 #define 'a' 65
b87f8b21 1678
402b123d 1679** Token definitions as enums
77714df2
AD
1680 Tokens are output both as the traditional #define's, and, provided
1681 the compiler supports ANSI C or is a C++ compiler, as enums.
e88dbdbf 1682 This lets debuggers display names instead of integers.
77714df2 1683
402b123d 1684** Reports
ec3bc396
AD
1685 In addition to --verbose, bison supports --report=THINGS, which
1686 produces additional information:
b408954b
AD
1687 - itemset
1688 complete the core item sets with their closure
e4ab1254 1689 - lookahead [changed to "look-ahead" in 1.875e through 2.3, but changed back]
9e6e7ed2 1690 explicitly associate lookahead tokens to items
b408954b
AD
1691 - solved
1692 describe shift/reduce conflicts solving.
1693 Bison used to systematically output this information on top of
1694 the report. Solved conflicts are now attached to their states.
ec3bc396 1695
402b123d 1696** Type clashes
9af3fbce
AD
1697 Previous versions don't complain when there is a type clash on
1698 the default action if the rule has a mid-rule action, such as in:
1699
e29f0771
AD
1700 %type <foo> bar
1701 %%
1702 bar: '0' {} '0';
9af3fbce
AD
1703
1704 This is fixed.
a861a339 1705
402b123d 1706** GNU M4 is now required when using Bison.
f987e9d2 1707\f
402b123d 1708* Changes in version 1.35, 2002-03-25:
76551463 1709
402b123d 1710** C Skeleton
76551463
AD
1711 Some projects use Bison's C parser with C++ compilers, and define
1712 YYSTYPE as a class. The recent adjustment of C parsers for data
1713 alignment and 64 bit architectures made this impossible.
1714
1715 Because for the time being no real solution for C++ parser
1716 generation exists, kludges were implemented in the parser to
1717 maintain this use. In the future, when Bison has C++ parsers, this
1718 kludge will be disabled.
1719
1720 This kludge also addresses some C++ problems when the stack was
1721 extended.
76551463 1722\f
402b123d 1723* Changes in version 1.34, 2002-03-12:
76551463 1724
402b123d 1725** File name clashes are detected
76551463 1726 $ bison foo.y -d -o foo.x
e4ab1254 1727 fatal error: header and parser would both be named "foo.x"
76551463 1728
e4ab1254 1729** A missing ";" at the end of a rule triggers a warning
76551463
AD
1730 In accordance with POSIX, and in agreement with other
1731 Yacc implementations, Bison will mandate this semicolon in the near
1732 future. This eases the implementation of a Bison parser of Bison
1733 grammars by making this grammar LALR(1) instead of LR(2). To
1734 facilitate the transition, this release introduces a warning.
1735
402b123d 1736** Revert the C++ namespace changes introduced in 1.31, as they caused too
76551463
AD
1737 many portability hassles.
1738
402b123d 1739** DJGPP support added.
76551463 1740
402b123d 1741** Fix test suite portability problems.
76551463 1742\f
402b123d 1743* Changes in version 1.33, 2002-02-07:
76551463 1744
402b123d 1745** Fix C++ issues
76551463
AD
1746 Groff could not be compiled for the definition of size_t was lacking
1747 under some conditions.
1748
402b123d 1749** Catch invalid @n
76551463
AD
1750 As is done with $n.
1751\f
402b123d 1752* Changes in version 1.32, 2002-01-23:
76551463 1753
402b123d 1754** Fix Yacc output file names
76551463 1755
402b123d 1756** Portability fixes
76551463 1757
402b123d 1758** Italian, Dutch translations
76551463 1759\f
402b123d 1760* Changes in version 1.31, 2002-01-14:
52d1aeee 1761
402b123d 1762** Many Bug Fixes
52d1aeee 1763
402b123d 1764** GNU Gettext and %expect
52d1aeee
MA
1765 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
1766 Bison dies on incorrect %expectations, we fear there will be
1767 too many bug reports for Gettext, so _for the time being_, %expect
e4ab1254 1768 does not trigger an error when the input file is named "plural.y".
52d1aeee 1769
402b123d 1770** Use of alloca in parsers
52d1aeee
MA
1771 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
1772 malloc exclusively. Since 1.29, but was not NEWS'ed.
1773
1774 alloca is used only when compiled with GCC, to avoid portability
1775 problems as on AIX.
1776
402b123d 1777** yyparse now returns 2 if memory is exhausted; formerly it dumped core.
b47dbebe 1778
402b123d 1779** When the generated parser lacks debugging code, YYDEBUG is now 0
52d1aeee
MA
1780 (as POSIX requires) instead of being undefined.
1781
402b123d 1782** User Actions
52d1aeee
MA
1783 Bison has always permitted actions such as { $$ = $1 }: it adds the
1784 ending semicolon. Now if in Yacc compatibility mode, the semicolon
1785 is no longer output: one has to write { $$ = $1; }.
1786
402b123d 1787** Better C++ compliance
52d1aeee 1788 The output parsers try to respect C++ namespaces.
76551463 1789 [This turned out to be a failed experiment, and it was reverted later.]
52d1aeee 1790
402b123d 1791** Reduced Grammars
52d1aeee
MA
1792 Fixed bugs when reporting useless nonterminals.
1793
402b123d 1794** 64 bit hosts
52d1aeee
MA
1795 The parsers work properly on 64 bit hosts.
1796
402b123d 1797** Error messages
52d1aeee
MA
1798 Some calls to strerror resulted in scrambled or missing error messages.
1799
402b123d 1800** %expect
52d1aeee
MA
1801 When the number of shift/reduce conflicts is correct, don't issue
1802 any warning.
1803
402b123d 1804** The verbose report includes the rule line numbers.
52d1aeee 1805
402b123d 1806** Rule line numbers are fixed in traces.
52d1aeee 1807
402b123d 1808** Swedish translation
52d1aeee 1809
402b123d 1810** Parse errors
52d1aeee
MA
1811 Verbose parse error messages from the parsers are better looking.
1812 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
1813 Now: parse error: unexpected '/', expecting "number" or '-' or '('
1814
402b123d 1815** Fixed parser memory leaks.
52d1aeee
MA
1816 When the generated parser was using malloc to extend its stacks, the
1817 previous allocations were not freed.
1818
402b123d 1819** Fixed verbose output file.
52d1aeee
MA
1820 Some newlines were missing.
1821 Some conflicts in state descriptions were missing.
1822
402b123d 1823** Fixed conflict report.
52d1aeee
MA
1824 Option -v was needed to get the result.
1825
402b123d 1826** %expect
52d1aeee
MA
1827 Was not used.
1828 Mismatches are errors, not warnings.
1829
402b123d 1830** Fixed incorrect processing of some invalid input.
52d1aeee 1831
402b123d 1832** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
52d1aeee 1833
402b123d 1834** Fixed some typos in the documentation.
52d1aeee 1835
402b123d 1836** %token MY_EOF 0 is supported.
52d1aeee
MA
1837 Before, MY_EOF was silently renumbered as 257.
1838
402b123d 1839** doc/refcard.tex is updated.
52d1aeee 1840
402b123d 1841** %output, %file-prefix, %name-prefix.
52d1aeee
MA
1842 New.
1843
402b123d 1844** --output
e4ab1254 1845 New, aliasing "--output-file".
52d1aeee 1846\f
402b123d 1847* Changes in version 1.30, 2001-10-26:
342b8b6e 1848
e4ab1254
AD
1849** "--defines" and "--graph" have now an optional argument which is the
1850 output file name. "-d" and "-g" do not change; they do not take any
342b8b6e
AD
1851 argument.
1852
e4ab1254 1853** "%source_extension" and "%header_extension" are removed, failed
342b8b6e
AD
1854 experiment.
1855
402b123d 1856** Portability fixes.
f987e9d2 1857\f
402b123d 1858* Changes in version 1.29, 2001-09-07:
342b8b6e 1859
402b123d 1860** The output file does not define const, as this caused problems when used
342b8b6e
AD
1861 with common autoconfiguration schemes. If you still use ancient compilers
1862 that lack const, compile with the equivalent of the C compiler option
e4ab1254 1863 "-Dconst=". Autoconf's AC_C_CONST macro provides one way to do this.
342b8b6e 1864
e4ab1254 1865** Added "-g" and "--graph".
f87a2205 1866
402b123d 1867** The Bison manual is now distributed under the terms of the GNU FDL.
f2b5126e 1868
402b123d 1869** The input and the output files has automatically a similar extension.
234a3be3 1870
402b123d 1871** Russian translation added.
f87a2205 1872
402b123d 1873** NLS support updated; should hopefully be less troublesome.
f87a2205 1874
402b123d 1875** Added the old Bison reference card.
c33638bb 1876
e4ab1254 1877** Added "--locations" and "%locations".
6deb4447 1878
e4ab1254 1879** Added "-S" and "--skeleton".
cd5bd6ac 1880
e4ab1254 1881** "%raw", "-r", "--raw" is disabled.
62ab6972 1882
402b123d 1883** Special characters are escaped when output. This solves the problems
cd5bd6ac
AD
1884 of the #line lines with path names including backslashes.
1885
402b123d 1886** New directives.
e4ab1254
AD
1887 "%yacc", "%fixed_output_files", "%defines", "%no_parser", "%verbose",
1888 "%debug", "%source_extension" and "%header_extension".
f987e9d2 1889
402b123d 1890** @$
f987e9d2 1891 Automatic location tracking.
f87a2205 1892\f
402b123d 1893* Changes in version 1.28, 1999-07-06:
d2e00347 1894
402b123d 1895** Should compile better now with K&R compilers.
d2e00347 1896
402b123d 1897** Added NLS.
d2e00347 1898
402b123d 1899** Fixed a problem with escaping the double quote character.
d2e00347 1900
402b123d 1901** There is now a FAQ.
d2e00347 1902\f
402b123d 1903* Changes in version 1.27:
5c31c3c2 1904
402b123d 1905** The make rule which prevented bison.simple from being created on
5c31c3c2
JT
1906 some systems has been fixed.
1907\f
402b123d 1908* Changes in version 1.26:
4be07551 1909
7e508a2b 1910** Bison now uses Automake.
4be07551 1911
402b123d 1912** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
4be07551 1913
402b123d 1914** Token numbers now start at 257 as previously documented, not 258.
4be07551 1915
402b123d 1916** Bison honors the TMPDIR environment variable.
4be07551 1917
402b123d 1918** A couple of buffer overruns have been fixed.
f51dbca1 1919
402b123d 1920** Problems when closing files should now be reported.
f51dbca1 1921
402b123d 1922** Generated parsers should now work even on operating systems which do
f51dbca1 1923 not provide alloca().
4be07551 1924\f
402b123d 1925* Changes in version 1.25, 1995-10-16:
df8878c5 1926
402b123d 1927** Errors in the input grammar are not fatal; Bison keeps reading
df8878c5 1928the grammar file, and reports all the errors found in it.
8c44d3ec 1929
402b123d 1930** Tokens can now be specified as multiple-character strings: for
df8878c5 1931example, you could use "<=" for a token which looks like <=, instead
7e508a2b 1932of choosing a name like LESSEQ.
df8878c5 1933
402b123d 1934** The %token_table declaration says to write a table of tokens (names
df8878c5
RS
1935and numbers) into the parser file. The yylex function can use this
1936table to recognize multiple-character string tokens, or for other
1937purposes.
1938
402b123d 1939** The %no_lines declaration says not to generate any #line preprocessor
df8878c5
RS
1940directives in the parser file.
1941
402b123d 1942** The %raw declaration says to use internal Bison token numbers, not
df8878c5
RS
1943Yacc-compatible token numbers, when token names are defined as macros.
1944
402b123d 1945** The --no-parser option produces the parser tables without including
df8878c5
RS
1946the parser engine; a project can now use its own parser engine.
1947The actions go into a separate file called NAME.act, in the form of
1948a switch statement body.
1949\f
402b123d 1950* Changes in version 1.23:
6780ca7a 1951
4d019228
DM
1952The user can define YYPARSE_PARAM as the name of an argument to be
1953passed into yyparse. The argument should have type void *. It should
1954actually point to an object. Grammar actions can access the variable
1955by casting it to the proper pointer type.
6780ca7a 1956
6780ca7a 1957Line numbers in output file corrected.
6780ca7a 1958\f
402b123d 1959* Changes in version 1.22:
6780ca7a
DM
1960
1961--help option added.
6780ca7a 1962\f
402b123d 1963* Changes in version 1.20:
6780ca7a
DM
1964
1965Output file does not redefine const for C++.
9f4503d6 1966
76551463
AD
1967-----
1968
34136e65 1969Copyright (C) 1995-2012 Free Software Foundation, Inc.
76551463 1970
74553c98 1971This file is part of Bison, the GNU Parser Generator.
76551463 1972
f16b0819 1973This program is free software: you can redistribute it and/or modify
76551463 1974it under the terms of the GNU General Public License as published by
f16b0819
PE
1975the Free Software Foundation, either version 3 of the License, or
1976(at your option) any later version.
76551463 1977
f16b0819 1978This program is distributed in the hope that it will be useful,
76551463
AD
1979but WITHOUT ANY WARRANTY; without even the implied warranty of
1980MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
1981GNU General Public License for more details.
1982
1983You should have received a copy of the GNU General Public License
f16b0819 1984along with this program. If not, see <http://www.gnu.org/licenses/>.
7e508a2b
AD
1985
1986 LocalWords: yacc YYBACKUP glr GCC lalr ArrayIndexOutOfBoundsException nullptr
1987 LocalWords: cplusplus liby rpl fprintf mfcalc Wyacc stmt cond expr mk sym lr
1988 LocalWords: IELR ielr Lookahead YYERROR nonassoc LALR's api lookaheads yychar
1989 LocalWords: destructor lookahead YYRHSLOC YYLLOC Rhs ifndef YYFAIL cpp sr rr
1990 LocalWords: preprocessor initializer Wno Wnone Werror FreeBSD prec livelocks
1991 LocalWords: Solaris AIX UX RHEL Tru LHS gcc's Wundef YYENABLE NLS YYLTYPE VCG
1992 LocalWords: yyerror cpp's Wunused yylval yylloc prepend yyparse yylex yypush
1993 LocalWords: Graphviz xml nonterminals midrule destructor's YYSTYPE typedef ly
1994 LocalWords: CHR chr printf stdout namespace preprocessing enum pre include's
1995 LocalWords: YYRECOVERING nonfree destructors YYABORT YYACCEPT params enums de
1996 LocalWords: struct yystype DJGPP lex param Haible NUM alloca YYSTACK NUL goto
1997 LocalWords: YYMAXDEPTH Unescaped UCNs YYLTYPE's yyltype typedefs inline Yaccs
1998 LocalWords: Heriyanto Reenable dprec Hilfinger Eggert MYEOF Folle Menezes EOF
242cc08e 1999 LocalWords: Lackovic define's itemset Groff Gettext malloc NEWS'ed YYDEBUG YY
7e508a2b 2000 LocalWords: namespaces strerror const autoconfiguration Dconst Autoconf's FDL
242cc08e
AD
2001 LocalWords: Automake TMPDIR LESSEQ ylwrap endif yydebug YYTOKEN YYLSP ival hh
2002 LocalWords: extern YYTOKENTYPE TOKENTYPE yytokentype tokentype STYPE lval pdf
dcb366b1
AD
2003 LocalWords: lang yyoutput dvi html ps POSIX lvalp llocp Wother nterm arg init
2004 LocalWords: TOK
7e508a2b
AD
2005
2006Local Variables:
2007mode: outline
e4ab1254 2008fill-column: 76
7e508a2b 2009End: