]> git.saurik.com Git - bison.git/blob - NEWS
Add syncline support and location accessor to internal %define
[bison.git] / NEWS
1 Bison News
2 ----------
3
4 Changes in version 2.3a+ (????-??-??):
5
6 * Previously, Bison sometimes generated parser tables with states that were
7 unreachable due to conflicts in predecessor states. Bison now:
8
9 1. Removes unreachable states.
10
11 2. Does not report any conflicts that appeared in unreachable states.
12 WARNING: As a result, you may need to update %expect and %expect-rr
13 directives in existing grammar files.
14
15 3. For any rule used only in such states, Bison now reports the rule as
16 "never reduced because of conflicts".
17
18 * The -g and --graph options now output graphs in Graphviz DOT format,
19 not VCG format.
20
21 * An experimental directive %language specifies the language of the
22 generated parser, which can be C (the default) or C++. This
23 directive affects the skeleton used, and the names of the generated
24 files if the grammar file's name ends in ".y".
25
26 * The grammar file may now specify the name of the parser header file using
27 %defines. For example:
28
29 %defines "parser.h"
30
31 * The `=' that used to be required in the following directives is now
32 deprecated:
33
34 %file-prefix "parser"
35 %name-prefix "c_"
36 %output "parser.c"
37
38 * The quotes around NAME that used to be required in the following directive
39 are now deprecated:
40
41 %define NAME "VALUE"
42
43 * Bison 2.3a provided a new set of directives as a more flexible alternative to
44 the traditional Yacc prologue blocks. Those have now been consolidated into
45 a single %code directive with an optional qualifier field, which identifies
46 the purpose of the code and thus the location(s) where Bison should generate
47 it:
48
49 1. `%code {CODE}' replaces `%after-header {CODE}'
50 2. `%code requires {CODE}' replaces `%start-header {CODE}'
51 3. `%code provides {CODE}' replaces `%end-header {CODE}'
52 4. `%code top {CODE}' replaces `%before-header {CODE}'
53
54 See the %code entries in section `Bison Declaration Summary' in the Bison
55 manual for a summary of the new functionality. See the new section `Prologue
56 Alternatives' for a detailed discussion including the advantages of %code
57 over the traditional Yacc prologues.
58
59 The prologue alternatives are experimental. More user feedback will help to
60 determine whether they should become permanent features.
61
62 * Revised warning: unset or unused mid-rule values
63
64 Since Bison 2.2, Bison has warned about mid-rule values that are set but not
65 used within any of the actions of the parent rule. For example, Bison warns
66 about unused $2 in:
67
68 exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; };
69
70 Now, Bison also warns about mid-rule values that are used but not set. For
71 example, Bison warns about unset $$ in the mid-rule action in:
72
73 exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; };
74
75 However, Bison now disables both of these warnings by default since they
76 sometimes prove to be false alarms in existing grammars employing the Yacc
77 constructs $0 or $-N (where N is some positive integer).
78
79 To enable these warnings, specify the flag `--warnings=midrule-values' or
80 `-W', which is a synonym for `--warnings=all'.
81
82 * Bison now recognizes two separate kinds of default %destructor's and
83 %printer's:
84
85 1. Place `<*>' in a %destructor/%printer symbol list to define a default
86 %destructor/%printer for all grammar symbols for which you have formally
87 declared semantic type tags.
88
89 2. Place `<>' in a %destructor/%printer symbol list to define a default
90 %destructor/%printer for all grammar symbols without declared semantic
91 type tags.
92
93 Bison no longer supports the `%symbol-default' notation from Bison 2.3a.
94 `<*>' and `<>' combined achieve the same effect with one exception: Bison no
95 longer applies any %destructor to a mid-rule value if that mid-rule value is
96 not actually ever referenced using either $$ or $n in a semantic action.
97
98 The default %destructor's and %printer's are experimental. More user
99 feedback will help to determine whether they should become permanent
100 features.
101
102 See the section `Freeing Discarded Symbols' in the Bison manual for further
103 details.
104
105 Changes in version 2.3a, 2006-09-13:
106
107 * Instead of %union, you can define and use your own union type
108 YYSTYPE if your grammar contains at least one <type> tag.
109 Your YYSTYPE need not be a macro; it can be a typedef.
110 This change is for compatibility with other Yacc implementations,
111 and is required by POSIX.
112
113 * Locations columns and lines start at 1.
114 In accordance with the GNU Coding Standards and Emacs.
115
116 * You may now declare per-type and default %destructor's and %printer's:
117
118 For example:
119
120 %union { char *string; }
121 %token <string> STRING1
122 %token <string> STRING2
123 %type <string> string1
124 %type <string> string2
125 %union { char character; }
126 %token <character> CHR
127 %type <character> chr
128 %destructor { free ($$); } %symbol-default
129 %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1
130 %destructor { } <character>
131
132 guarantees that, when the parser discards any user-defined symbol that has a
133 semantic type tag other than `<character>', it passes its semantic value to
134 `free'. However, when the parser discards a `STRING1' or a `string1', it
135 also prints its line number to `stdout'. It performs only the second
136 `%destructor' in this case, so it invokes `free' only once.
137
138 [Although we failed to mention this here in the 2.3a release, the default
139 %destructor's and %printer's were experimental, and they were rewritten in
140 future versions.]
141
142 * Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y',
143 `--yacc', or `%yacc'), Bison no longer generates #define statements for
144 associating token numbers with token names. Removing the #define statements
145 helps to sanitize the global namespace during preprocessing, but POSIX Yacc
146 requires them. Bison still generates an enum for token names in all cases.
147
148 * Handling of traditional Yacc prologue blocks is now more consistent but
149 potentially incompatible with previous releases of Bison.
150
151 As before, you declare prologue blocks in your grammar file with the
152 `%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all
153 prologue blocks that you've declared before the first %union. To generate
154 the post-prologue, Bison concatenates all prologue blocks that you've
155 declared after the first %union.
156
157 Previous releases of Bison inserted the pre-prologue into both the header
158 file and the code file in all cases except for LALR(1) parsers in C. In the
159 latter case, Bison inserted it only into the code file. For parsers in C++,
160 the point of insertion was before any token definitions (which associate
161 token numbers with names). For parsers in C, the point of insertion was
162 after the token definitions.
163
164 Now, Bison never inserts the pre-prologue into the header file. In the code
165 file, it always inserts it before the token definitions.
166
167 * Bison now provides a more flexible alternative to the traditional Yacc
168 prologue blocks: %before-header, %start-header, %end-header, and
169 %after-header.
170
171 For example, the following declaration order in the grammar file reflects the
172 order in which Bison will output these code blocks. However, you are free to
173 declare these code blocks in your grammar file in whatever order is most
174 convenient for you:
175
176 %before-header {
177 /* Bison treats this block like a pre-prologue block: it inserts it into
178 * the code file before the contents of the header file. It does *not*
179 * insert it into the header file. This is a good place to put
180 * #include's that you want at the top of your code file. A common
181 * example is `#include "system.h"'. */
182 }
183 %start-header {
184 /* Bison inserts this block into both the header file and the code file.
185 * In both files, the point of insertion is before any Bison-generated
186 * token, semantic type, location type, and class definitions. This is a
187 * good place to define %union dependencies, for example. */
188 }
189 %union {
190 /* Unlike the traditional Yacc prologue blocks, the output order for the
191 * new %*-header blocks is not affected by their declaration position
192 * relative to any %union in the grammar file. */
193 }
194 %end-header {
195 /* Bison inserts this block into both the header file and the code file.
196 * In both files, the point of insertion is after the Bison-generated
197 * definitions. This is a good place to declare or define public
198 * functions or data structures that depend on the Bison-generated
199 * definitions. */
200 }
201 %after-header {
202 /* Bison treats this block like a post-prologue block: it inserts it into
203 * the code file after the contents of the header file. It does *not*
204 * insert it into the header file. This is a good place to declare or
205 * define internal functions or data structures that depend on the
206 * Bison-generated definitions. */
207 }
208
209 If you have multiple occurrences of any one of the above declarations, Bison
210 will concatenate the contents in declaration order.
211
212 [Although we failed to mention this here in the 2.3a release, the prologue
213 alternatives were experimental, and they were rewritten in future versions.]
214
215 * The option `--report=look-ahead' has been changed to `--report=lookahead'.
216 The old spelling still works, but is not documented and may be removed
217 in a future release.
218
219 Changes in version 2.3, 2006-06-05:
220
221 * GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING',
222 for compatibility with LALR(1) grammars.
223
224 * It is now documented that any definition of YYSTYPE or YYLTYPE should
225 be to a type name that does not contain parentheses or brackets.
226
227 Changes in version 2.2, 2006-05-19:
228
229 * The distribution terms for all Bison-generated parsers now permit
230 using the parsers in nonfree programs. Previously, this permission
231 was granted only for Bison-generated LALR(1) parsers in C.
232
233 * %name-prefix changes the namespace name in C++ outputs.
234
235 * The C++ parsers export their token_type.
236
237 * Bison now allows multiple %union declarations, and concatenates
238 their contents together.
239
240 * New warning: unused values
241 Right-hand side symbols whose values are not used are reported,
242 if the symbols have destructors. For instance:
243
244 exp: exp "?" exp ":" exp { $1 ? $1 : $3; }
245 | exp "+" exp
246 ;
247
248 will trigger a warning about $$ and $5 in the first rule, and $3 in
249 the second ($1 is copied to $$ by the default rule). This example
250 most likely contains three errors, and could be rewritten as:
251
252 exp: exp "?" exp ":" exp
253 { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); }
254 | exp "+" exp
255 { $$ = $1 ? $1 : $3; if ($1) free ($3); }
256 ;
257
258 However, if the original actions were really intended, memory leaks
259 and all, the warnings can be suppressed by letting Bison believe the
260 values are used, e.g.:
261
262 exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); }
263 | exp "+" exp { $$ = $1; (void) $3; }
264 ;
265
266 If there are mid-rule actions, the warning is issued if no action
267 uses it. The following triggers no warning: $1 and $3 are used.
268
269 exp: exp { push ($1); } '+' exp { push ($3); sum (); };
270
271 The warning is intended to help catching lost values and memory leaks.
272 If a value is ignored, its associated memory typically is not reclaimed.
273
274 * %destructor vs. YYABORT, YYACCEPT, and YYERROR.
275 Destructors are now called when user code invokes YYABORT, YYACCEPT,
276 and YYERROR, for all objects on the stack, other than objects
277 corresponding to the right-hand side of the current rule.
278
279 * %expect, %expect-rr
280 Incorrect numbers of expected conflicts are now actual errors,
281 instead of warnings.
282
283 * GLR, YACC parsers.
284 The %parse-params are available in the destructors (and the
285 experimental printers) as per the documentation.
286
287 * Bison now warns if it finds a stray `$' or `@' in an action.
288
289 * %require "VERSION"
290 This specifies that the grammar file depends on features implemented
291 in Bison version VERSION or higher.
292
293 * lalr1.cc: The token and value types are now class members.
294 The tokens were defined as free form enums and cpp macros. YYSTYPE
295 was defined as a free form union. They are now class members:
296 tokens are enumerations of the `yy::parser::token' struct, and the
297 semantic values have the `yy::parser::semantic_type' type.
298
299 If you do not want or can update to this scheme, the directive
300 `%define "global_tokens_and_yystype" "1"' triggers the global
301 definition of tokens and YYSTYPE. This change is suitable both
302 for previous releases of Bison, and this one.
303
304 If you wish to update, then make sure older version of Bison will
305 fail using `%require "2.2"'.
306
307 * DJGPP support added.
308 \f
309 Changes in version 2.1, 2005-09-16:
310
311 * The C++ lalr1.cc skeleton supports %lex-param.
312
313 * Bison-generated parsers now support the translation of diagnostics like
314 "syntax error" into languages other than English. The default
315 language is still English. For details, please see the new
316 Internationalization section of the Bison manual. Software
317 distributors should also see the new PACKAGING file. Thanks to
318 Bruno Haible for this new feature.
319
320 * Wording in the Bison-generated parsers has been changed slightly to
321 simplify translation. In particular, the message "memory exhausted"
322 has replaced "parser stack overflow", as the old message was not
323 always accurate for modern Bison-generated parsers.
324
325 * Destructors are now called when the parser aborts, for all symbols left
326 behind on the stack. Also, the start symbol is now destroyed after a
327 successful parse. In both cases, the behavior was formerly inconsistent.
328
329 * When generating verbose diagnostics, Bison-generated parsers no longer
330 quote the literal strings associated with tokens. For example, for
331 a syntax error associated with '%token NUM "number"' they might
332 print 'syntax error, unexpected number' instead of 'syntax error,
333 unexpected "number"'.
334 \f
335 Changes in version 2.0, 2004-12-25:
336
337 * Possibly-incompatible changes
338
339 - Bison-generated parsers no longer default to using the alloca function
340 (when available) to extend the parser stack, due to widespread
341 problems in unchecked stack-overflow detection. You can "#define
342 YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read
343 the manual to determine safe values for YYMAXDEPTH in that case.
344
345 - Error token location.
346 During error recovery, the location of the syntax error is updated
347 to cover the whole sequence covered by the error token: it includes
348 the shifted symbols thrown away during the first part of the error
349 recovery, and the lookahead rejected during the second part.
350
351 - Semicolon changes:
352 . Stray semicolons are no longer allowed at the start of a grammar.
353 . Semicolons are now required after in-grammar declarations.
354
355 - Unescaped newlines are no longer allowed in character constants or
356 string literals. They were never portable, and GCC 3.4.0 has
357 dropped support for them. Better diagnostics are now generated if
358 forget a closing quote.
359
360 - NUL bytes are no longer allowed in Bison string literals, unfortunately.
361
362 * New features
363
364 - GLR grammars now support locations.
365
366 - New directive: %initial-action.
367 This directive allows the user to run arbitrary code (including
368 initializing @$) from yyparse before parsing starts.
369
370 - A new directive "%expect-rr N" specifies the expected number of
371 reduce/reduce conflicts in GLR parsers.
372
373 - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'.
374 This is a GNU extension.
375
376 - The option `--report=lookahead' was changed to `--report=look-ahead'.
377 [However, this was changed back after 2.3.]
378
379 - Experimental %destructor support has been added to lalr1.cc.
380
381 - New configure option --disable-yacc, to disable installation of the
382 yacc command and -ly library introduced in 1.875 for POSIX conformance.
383
384 * Bug fixes
385
386 - For now, %expect-count violations are now just warnings, not errors.
387 This is for compatibility with Bison 1.75 and earlier (when there are
388 reduce/reduce conflicts) and with Bison 1.30 and earlier (when there
389 are too many or too few shift/reduce conflicts). However, in future
390 versions of Bison we plan to improve the %expect machinery so that
391 these violations will become errors again.
392
393 - Within Bison itself, numbers (e.g., goto numbers) are no longer
394 arbitrarily limited to 16-bit counts.
395
396 - Semicolons are now allowed before "|" in grammar rules, as POSIX requires.
397 \f
398 Changes in version 1.875, 2003-01-01:
399
400 * The documentation license has been upgraded to version 1.2
401 of the GNU Free Documentation License.
402
403 * syntax error processing
404
405 - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error
406 locations too. This fixes bugs in error-location computation.
407
408 - %destructor
409 It is now possible to reclaim the memory associated to symbols
410 discarded during error recovery. This feature is still experimental.
411
412 - %error-verbose
413 This new directive is preferred over YYERROR_VERBOSE.
414
415 - #defining yyerror to steal internal variables is discouraged.
416 It is not guaranteed to work forever.
417
418 * POSIX conformance
419
420 - Semicolons are once again optional at the end of grammar rules.
421 This reverts to the behavior of Bison 1.33 and earlier, and improves
422 compatibility with Yacc.
423
424 - `parse error' -> `syntax error'
425 Bison now uniformly uses the term `syntax error'; formerly, the code
426 and manual sometimes used the term `parse error' instead. POSIX
427 requires `syntax error' in diagnostics, and it was thought better to
428 be consistent.
429
430 - The documentation now emphasizes that yylex and yyerror must be
431 declared before use. C99 requires this.
432
433 - Bison now parses C99 lexical constructs like UCNs and
434 backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires.
435
436 - File names are properly escaped in C output. E.g., foo\bar.y is
437 output as "foo\\bar.y".
438
439 - Yacc command and library now available
440 The Bison distribution now installs a `yacc' command, as POSIX requires.
441 Also, Bison now installs a small library liby.a containing
442 implementations of Yacc-compatible yyerror and main functions.
443 This library is normally not useful, but POSIX requires it.
444
445 - Type clashes now generate warnings, not errors.
446
447 - If the user does not define YYSTYPE as a macro, Bison now declares it
448 using typedef instead of defining it as a macro.
449 For consistency, YYLTYPE is also declared instead of defined.
450
451 * Other compatibility issues
452
453 - %union directives can now have a tag before the `{', e.g., the
454 directive `%union foo {...}' now generates the C code
455 `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility.
456 The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc.
457 For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'.
458 This is for compatibility with both Yacc and Bison 1.35.
459
460 - `;' is output before the terminating `}' of an action, for
461 compatibility with Bison 1.35.
462
463 - Bison now uses a Yacc-style format for conflict reports, e.g.,
464 `conflicts: 2 shift/reduce, 1 reduce/reduce'.
465
466 - `yystype' and `yyltype' are now obsolescent macros instead of being
467 typedefs or tags; they are no longer documented and are planned to be
468 withdrawn in a future release.
469
470 * GLR parser notes
471
472 - GLR and inline
473 Users of Bison have to decide how they handle the portability of the
474 C keyword `inline'.
475
476 - `parsing stack overflow...' -> `parser stack overflow'
477 GLR parsers now report `parser stack overflow' as per the Bison manual.
478
479 * Bison now warns if it detects conflicting outputs to the same file,
480 e.g., it generates a warning for `bison -d -o foo.h foo.y' since
481 that command outputs both code and header to foo.h.
482
483 * #line in output files
484 - --no-line works properly.
485
486 * Bison can no longer be built by a K&R C compiler; it requires C89 or
487 later to be built. This change originally took place a few versions
488 ago, but nobody noticed until we recently asked someone to try
489 building Bison with a K&R C compiler.
490 \f
491 Changes in version 1.75, 2002-10-14:
492
493 * Bison should now work on 64-bit hosts.
494
495 * Indonesian translation thanks to Tedi Heriyanto.
496
497 * GLR parsers
498 Fix spurious parse errors.
499
500 * Pure parsers
501 Some people redefine yyerror to steal yyparse' private variables.
502 Reenable this trick until an official feature replaces it.
503
504 * Type Clashes
505 In agreement with POSIX and with other Yaccs, leaving a default
506 action is valid when $$ is untyped, and $1 typed:
507
508 untyped: ... typed;
509
510 but the converse remains an error:
511
512 typed: ... untyped;
513
514 * Values of mid-rule actions
515 The following code:
516
517 foo: { ... } { $$ = $1; } ...
518
519 was incorrectly rejected: $1 is defined in the second mid-rule
520 action, and is equal to the $$ of the first mid-rule action.
521 \f
522 Changes in version 1.50, 2002-10-04:
523
524 * GLR parsing
525 The declaration
526 %glr-parser
527 causes Bison to produce a Generalized LR (GLR) parser, capable of handling
528 almost any context-free grammar, ambiguous or not. The new declarations
529 %dprec and %merge on grammar rules allow parse-time resolution of
530 ambiguities. Contributed by Paul Hilfinger.
531
532 Unfortunately Bison 1.50 does not work properly on 64-bit hosts
533 like the Alpha, so please stick to 32-bit hosts for now.
534
535 * Output Directory
536 When not in Yacc compatibility mode, when the output file was not
537 specified, running `bison foo/bar.y' created `foo/bar.c'. It
538 now creates `bar.c'.
539
540 * Undefined token
541 The undefined token was systematically mapped to 2 which prevented
542 the use of 2 by the user. This is no longer the case.
543
544 * Unknown token numbers
545 If yylex returned an out of range value, yyparse could die. This is
546 no longer the case.
547
548 * Error token
549 According to POSIX, the error token must be 256.
550 Bison extends this requirement by making it a preference: *if* the
551 user specified that one of her tokens is numbered 256, then error
552 will be mapped onto another number.
553
554 * Verbose error messages
555 They no longer report `..., expecting error or...' for states where
556 error recovery is possible.
557
558 * End token
559 Defaults to `$end' instead of `$'.
560
561 * Error recovery now conforms to documentation and to POSIX
562 When a Bison-generated parser encounters a syntax error, it now pops
563 the stack until it finds a state that allows shifting the error
564 token. Formerly, it popped the stack until it found a state that
565 allowed some non-error action other than a default reduction on the
566 error token. The new behavior has long been the documented behavior,
567 and has long been required by POSIX. For more details, please see
568 Paul Eggert, "Reductions during Bison error handling" (2002-05-20)
569 <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>.
570
571 * Traces
572 Popped tokens and nonterminals are now reported.
573
574 * Larger grammars
575 Larger grammars are now supported (larger token numbers, larger grammar
576 size (= sum of the LHS and RHS lengths), larger LALR tables).
577 Formerly, many of these numbers ran afoul of 16-bit limits;
578 now these limits are 32 bits on most hosts.
579
580 * Explicit initial rule
581 Bison used to play hacks with the initial rule, which the user does
582 not write. It is now explicit, and visible in the reports and
583 graphs as rule 0.
584
585 * Useless rules
586 Before, Bison reported the useless rules, but, although not used,
587 included them in the parsers. They are now actually removed.
588
589 * Useless rules, useless nonterminals
590 They are now reported, as a warning, with their locations.
591
592 * Rules never reduced
593 Rules that can never be reduced because of conflicts are now
594 reported.
595
596 * Incorrect `Token not used'
597 On a grammar such as
598
599 %token useless useful
600 %%
601 exp: '0' %prec useful;
602
603 where a token was used to set the precedence of the last rule,
604 bison reported both `useful' and `useless' as useless tokens.
605
606 * Revert the C++ namespace changes introduced in 1.31
607 as they caused too many portability hassles.
608
609 * Default locations
610 By an accident of design, the default computation of @$ was
611 performed after another default computation was performed: @$ = @1.
612 The latter is now removed: YYLLOC_DEFAULT is fully responsible of
613 the computation of @$.
614
615 * Token end-of-file
616 The token end of file may be specified by the user, in which case,
617 the user symbol is used in the reports, the graphs, and the verbose
618 error messages instead of `$end', which remains being the default.
619 For instance
620 %token MYEOF 0
621 or
622 %token MYEOF 0 "end of file"
623
624 * Semantic parser
625 This old option, which has been broken for ages, is removed.
626
627 * New translations
628 Brazilian Portuguese, thanks to Alexandre Folle de Menezes.
629 Croatian, thanks to Denis Lackovic.
630
631 * Incorrect token definitions
632 When given `%token 'a' "A"', Bison used to output `#define 'a' 65'.
633
634 * Token definitions as enums
635 Tokens are output both as the traditional #define's, and, provided
636 the compiler supports ANSI C or is a C++ compiler, as enums.
637 This lets debuggers display names instead of integers.
638
639 * Reports
640 In addition to --verbose, bison supports --report=THINGS, which
641 produces additional information:
642 - itemset
643 complete the core item sets with their closure
644 - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back]
645 explicitly associate lookahead tokens to items
646 - solved
647 describe shift/reduce conflicts solving.
648 Bison used to systematically output this information on top of
649 the report. Solved conflicts are now attached to their states.
650
651 * Type clashes
652 Previous versions don't complain when there is a type clash on
653 the default action if the rule has a mid-rule action, such as in:
654
655 %type <foo> bar
656 %%
657 bar: '0' {} '0';
658
659 This is fixed.
660
661 * GNU M4 is now required when using Bison.
662 \f
663 Changes in version 1.35, 2002-03-25:
664
665 * C Skeleton
666 Some projects use Bison's C parser with C++ compilers, and define
667 YYSTYPE as a class. The recent adjustment of C parsers for data
668 alignment and 64 bit architectures made this impossible.
669
670 Because for the time being no real solution for C++ parser
671 generation exists, kludges were implemented in the parser to
672 maintain this use. In the future, when Bison has C++ parsers, this
673 kludge will be disabled.
674
675 This kludge also addresses some C++ problems when the stack was
676 extended.
677 \f
678 Changes in version 1.34, 2002-03-12:
679
680 * File name clashes are detected
681 $ bison foo.y -d -o foo.x
682 fatal error: header and parser would both be named `foo.x'
683
684 * A missing `;' at the end of a rule triggers a warning
685 In accordance with POSIX, and in agreement with other
686 Yacc implementations, Bison will mandate this semicolon in the near
687 future. This eases the implementation of a Bison parser of Bison
688 grammars by making this grammar LALR(1) instead of LR(2). To
689 facilitate the transition, this release introduces a warning.
690
691 * Revert the C++ namespace changes introduced in 1.31, as they caused too
692 many portability hassles.
693
694 * DJGPP support added.
695
696 * Fix test suite portability problems.
697 \f
698 Changes in version 1.33, 2002-02-07:
699
700 * Fix C++ issues
701 Groff could not be compiled for the definition of size_t was lacking
702 under some conditions.
703
704 * Catch invalid @n
705 As is done with $n.
706 \f
707 Changes in version 1.32, 2002-01-23:
708
709 * Fix Yacc output file names
710
711 * Portability fixes
712
713 * Italian, Dutch translations
714 \f
715 Changes in version 1.31, 2002-01-14:
716
717 * Many Bug Fixes
718
719 * GNU Gettext and %expect
720 GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that
721 Bison dies on incorrect %expectations, we fear there will be
722 too many bug reports for Gettext, so _for the time being_, %expect
723 does not trigger an error when the input file is named `plural.y'.
724
725 * Use of alloca in parsers
726 If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use
727 malloc exclusively. Since 1.29, but was not NEWS'ed.
728
729 alloca is used only when compiled with GCC, to avoid portability
730 problems as on AIX.
731
732 * yyparse now returns 2 if memory is exhausted; formerly it dumped core.
733
734 * When the generated parser lacks debugging code, YYDEBUG is now 0
735 (as POSIX requires) instead of being undefined.
736
737 * User Actions
738 Bison has always permitted actions such as { $$ = $1 }: it adds the
739 ending semicolon. Now if in Yacc compatibility mode, the semicolon
740 is no longer output: one has to write { $$ = $1; }.
741
742 * Better C++ compliance
743 The output parsers try to respect C++ namespaces.
744 [This turned out to be a failed experiment, and it was reverted later.]
745
746 * Reduced Grammars
747 Fixed bugs when reporting useless nonterminals.
748
749 * 64 bit hosts
750 The parsers work properly on 64 bit hosts.
751
752 * Error messages
753 Some calls to strerror resulted in scrambled or missing error messages.
754
755 * %expect
756 When the number of shift/reduce conflicts is correct, don't issue
757 any warning.
758
759 * The verbose report includes the rule line numbers.
760
761 * Rule line numbers are fixed in traces.
762
763 * Swedish translation
764
765 * Parse errors
766 Verbose parse error messages from the parsers are better looking.
767 Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'(''
768 Now: parse error: unexpected '/', expecting "number" or '-' or '('
769
770 * Fixed parser memory leaks.
771 When the generated parser was using malloc to extend its stacks, the
772 previous allocations were not freed.
773
774 * Fixed verbose output file.
775 Some newlines were missing.
776 Some conflicts in state descriptions were missing.
777
778 * Fixed conflict report.
779 Option -v was needed to get the result.
780
781 * %expect
782 Was not used.
783 Mismatches are errors, not warnings.
784
785 * Fixed incorrect processing of some invalid input.
786
787 * Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H.
788
789 * Fixed some typos in the documentation.
790
791 * %token MY_EOF 0 is supported.
792 Before, MY_EOF was silently renumbered as 257.
793
794 * doc/refcard.tex is updated.
795
796 * %output, %file-prefix, %name-prefix.
797 New.
798
799 * --output
800 New, aliasing `--output-file'.
801 \f
802 Changes in version 1.30, 2001-10-26:
803
804 * `--defines' and `--graph' have now an optional argument which is the
805 output file name. `-d' and `-g' do not change; they do not take any
806 argument.
807
808 * `%source_extension' and `%header_extension' are removed, failed
809 experiment.
810
811 * Portability fixes.
812 \f
813 Changes in version 1.29, 2001-09-07:
814
815 * The output file does not define const, as this caused problems when used
816 with common autoconfiguration schemes. If you still use ancient compilers
817 that lack const, compile with the equivalent of the C compiler option
818 `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this.
819
820 * Added `-g' and `--graph'.
821
822 * The Bison manual is now distributed under the terms of the GNU FDL.
823
824 * The input and the output files has automatically a similar extension.
825
826 * Russian translation added.
827
828 * NLS support updated; should hopefully be less troublesome.
829
830 * Added the old Bison reference card.
831
832 * Added `--locations' and `%locations'.
833
834 * Added `-S' and `--skeleton'.
835
836 * `%raw', `-r', `--raw' is disabled.
837
838 * Special characters are escaped when output. This solves the problems
839 of the #line lines with path names including backslashes.
840
841 * New directives.
842 `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose',
843 `%debug', `%source_extension' and `%header_extension'.
844
845 * @$
846 Automatic location tracking.
847 \f
848 Changes in version 1.28, 1999-07-06:
849
850 * Should compile better now with K&R compilers.
851
852 * Added NLS.
853
854 * Fixed a problem with escaping the double quote character.
855
856 * There is now a FAQ.
857 \f
858 Changes in version 1.27:
859
860 * The make rule which prevented bison.simple from being created on
861 some systems has been fixed.
862 \f
863 Changes in version 1.26:
864
865 * Bison now uses automake.
866
867 * New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>.
868
869 * Token numbers now start at 257 as previously documented, not 258.
870
871 * Bison honors the TMPDIR environment variable.
872
873 * A couple of buffer overruns have been fixed.
874
875 * Problems when closing files should now be reported.
876
877 * Generated parsers should now work even on operating systems which do
878 not provide alloca().
879 \f
880 Changes in version 1.25, 1995-10-16:
881
882 * Errors in the input grammar are not fatal; Bison keeps reading
883 the grammar file, and reports all the errors found in it.
884
885 * Tokens can now be specified as multiple-character strings: for
886 example, you could use "<=" for a token which looks like <=, instead
887 of chosing a name like LESSEQ.
888
889 * The %token_table declaration says to write a table of tokens (names
890 and numbers) into the parser file. The yylex function can use this
891 table to recognize multiple-character string tokens, or for other
892 purposes.
893
894 * The %no_lines declaration says not to generate any #line preprocessor
895 directives in the parser file.
896
897 * The %raw declaration says to use internal Bison token numbers, not
898 Yacc-compatible token numbers, when token names are defined as macros.
899
900 * The --no-parser option produces the parser tables without including
901 the parser engine; a project can now use its own parser engine.
902 The actions go into a separate file called NAME.act, in the form of
903 a switch statement body.
904 \f
905 Changes in version 1.23:
906
907 The user can define YYPARSE_PARAM as the name of an argument to be
908 passed into yyparse. The argument should have type void *. It should
909 actually point to an object. Grammar actions can access the variable
910 by casting it to the proper pointer type.
911
912 Line numbers in output file corrected.
913 \f
914 Changes in version 1.22:
915
916 --help option added.
917 \f
918 Changes in version 1.20:
919
920 Output file does not redefine const for C++.
921
922 Local Variables:
923 mode: outline
924 End:
925
926 -----
927
928 Copyright (C) 1995, 1996, 1997, 1998, 1999, 2000, 2001, 2002, 2003,
929 2004, 2005, 2006, 2007 Free Software Foundation, Inc.
930
931 This file is part of Bison, the GNU Compiler Compiler.
932
933 This program is free software: you can redistribute it and/or modify
934 it under the terms of the GNU General Public License as published by
935 the Free Software Foundation, either version 3 of the License, or
936 (at your option) any later version.
937
938 This program is distributed in the hope that it will be useful,
939 but WITHOUT ANY WARRANTY; without even the implied warranty of
940 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
941 GNU General Public License for more details.
942
943 You should have received a copy of the GNU General Public License
944 along with this program. If not, see <http://www.gnu.org/licenses/>.