| 1 | Bison News |
| 2 | ---------- |
| 3 | |
| 4 | * Changes in version 2.5 (????-??-??): |
| 5 | |
| 6 | ** Grammar symbol names can now contain non-initial dashes: |
| 7 | |
| 8 | Consistently with directives (such as %error-verbose) and with |
| 9 | %define variables (e.g. push-pull), grammar symbol names may contain |
| 10 | dashes in any position except the beginning. This is a GNU |
| 11 | extension over POSIX Yacc. Thus, use of this extension is reported |
| 12 | by -Wyacc and rejected in Yacc mode (--yacc). |
| 13 | |
| 14 | ** Named references: |
| 15 | |
| 16 | Historically, Yacc and Bison have supported positional references |
| 17 | ($n, $$) to allow access to symbol values from inside of semantic |
| 18 | actions code. |
| 19 | |
| 20 | Starting from this version, Bison can also accept named references. |
| 21 | When no ambiguity is possible, original symbol names may be used |
| 22 | as named references: |
| 23 | |
| 24 | if_stmt : "if" cond_expr "then" then_stmt ';' |
| 25 | { $if_stmt = mk_if_stmt($cond_expr, $then_stmt); } |
| 26 | |
| 27 | In the more common case, explicit names may be declared: |
| 28 | |
| 29 | stmt[res] : "if" expr[cond] "then" stmt[then] "else" stmt[else] ';' |
| 30 | { $res = mk_if_stmt($cond, $then, $else); } |
| 31 | |
| 32 | Location information is also accessible using @name syntax. When |
| 33 | accessing symbol names containing dots or dashes, explicit bracketing |
| 34 | ($[sym.1]) must be used. |
| 35 | |
| 36 | These features are experimental in this version. More user feedback |
| 37 | will help to stabilize them. |
| 38 | |
| 39 | ** IELR(1) and canonical LR(1): |
| 40 | |
| 41 | IELR(1) is a minimal LR(1) parser table generation algorithm. That |
| 42 | is, given any context-free grammar, IELR(1) generates parser tables |
| 43 | with the full language recognition power of canonical LR(1) but with |
| 44 | nearly the same number of parser states as LALR(1). This reduction |
| 45 | in parser states is often an order of magnitude. More importantly, |
| 46 | because canonical LR(1)'s extra parser states may contain duplicate |
| 47 | conflicts in the case of non-LR(1) grammars, the number of conflicts |
| 48 | for IELR(1) is often an order of magnitude less as well. This can |
| 49 | significantly reduce the complexity of developing of a grammar. |
| 50 | |
| 51 | Bison can now generate IELR(1) and canonical LR(1) parser tables in |
| 52 | place of its traditional LALR(1) parser tables, which remain the |
| 53 | default. You can specify the type of parser tables in the grammar |
| 54 | file with these directives: |
| 55 | |
| 56 | %define lr.type lalr |
| 57 | %define lr.type ielr |
| 58 | %define lr.type canonical-lr |
| 59 | |
| 60 | The default-reduction optimization in the parser tables can also be |
| 61 | adjusted using `%define lr.default-reductions'. For details on both |
| 62 | of these features, see the new section `Tuning LR' in the Bison |
| 63 | manual. |
| 64 | |
| 65 | These features are experimental. More user feedback will help to |
| 66 | stabilize them. |
| 67 | |
| 68 | ** LAC (Lookahead Correction) for syntax error handling: |
| 69 | |
| 70 | Canonical LR, IELR, and LALR can suffer from a couple of problems |
| 71 | upon encountering a syntax error. First, the parser might perform |
| 72 | additional parser stack reductions before discovering the syntax |
| 73 | error. Such reductions can perform user semantic actions that are |
| 74 | unexpected because they are based on an invalid token, and they |
| 75 | cause error recovery to begin in a different syntactic context than |
| 76 | the one in which the invalid token was encountered. Second, when |
| 77 | verbose error messages are enabled (with %error-verbose or the |
| 78 | obsolete `#define YYERROR_VERBOSE'), the expected token list in the |
| 79 | syntax error message can both contain invalid tokens and omit valid |
| 80 | tokens. |
| 81 | |
| 82 | The culprits for the above problems are %nonassoc, default |
| 83 | reductions in inconsistent states, and parser state merging. Thus, |
| 84 | IELR and LALR suffer the most. Canonical LR can suffer only if |
| 85 | %nonassoc is used or if default reductions are enabled for |
| 86 | inconsistent states. |
| 87 | |
| 88 | LAC is a new mechanism within the parsing algorithm that solves |
| 89 | these problems for canonical LR, IELR, and LALR without sacrificing |
| 90 | %nonassoc, default reductions, or state merging. When LAC is in |
| 91 | use, canonical LR and IELR behave almost exactly the same for both |
| 92 | syntactically acceptable and syntactically unacceptable input. |
| 93 | While LALR still does not support the full language-recognition |
| 94 | power of canonical LR and IELR, LAC at least enables LALR's syntax |
| 95 | error handling to correctly reflect LALR's language-recognition |
| 96 | power. |
| 97 | |
| 98 | Currently, LAC is only supported for deterministic parsers in C. |
| 99 | You can enable LAC with the following directive: |
| 100 | |
| 101 | %define parse.lac full |
| 102 | |
| 103 | See the new section `LAC' in the Bison manual for additional |
| 104 | details including a few caveats. |
| 105 | |
| 106 | LAC is an experimental feature. More user feedback will help to |
| 107 | stabilize it. |
| 108 | |
| 109 | ** %define improvements: |
| 110 | |
| 111 | *** Can now be invoked via the command line: |
| 112 | |
| 113 | Each of these command-line options |
| 114 | |
| 115 | -D NAME[=VALUE] |
| 116 | --define=NAME[=VALUE] |
| 117 | |
| 118 | -F NAME[=VALUE] |
| 119 | --force-define=NAME[=VALUE] |
| 120 | |
| 121 | is equivalent to this grammar file declaration |
| 122 | |
| 123 | %define NAME ["VALUE"] |
| 124 | |
| 125 | except that the manner in which Bison processes multiple definitions |
| 126 | for the same NAME differs. Most importantly, -F and --force-define |
| 127 | quietly override %define, but -D and --define do not. For further |
| 128 | details, see the section `Bison Options' in the Bison manual. |
| 129 | |
| 130 | *** Variables renamed: |
| 131 | |
| 132 | The following %define variables |
| 133 | |
| 134 | api.push_pull |
| 135 | lr.keep_unreachable_states |
| 136 | |
| 137 | have been renamed to |
| 138 | |
| 139 | api.push-pull |
| 140 | lr.keep-unreachable-states |
| 141 | |
| 142 | The old names are now deprecated but will be maintained indefinitely |
| 143 | for backward compatibility. |
| 144 | |
| 145 | *** Values no longer need to be quoted in grammar file: |
| 146 | |
| 147 | If a %define value is an identifier, it no longer needs to be placed |
| 148 | within quotations marks. For example, |
| 149 | |
| 150 | %define api.push-pull "push" |
| 151 | |
| 152 | can be rewritten as |
| 153 | |
| 154 | %define api.push-pull push |
| 155 | |
| 156 | *** Unrecognized variables are now errors not warnings. |
| 157 | |
| 158 | *** Multiple invocations for any variable is now an error not a warning. |
| 159 | |
| 160 | ** Unrecognized %code qualifiers are now errors not warnings. |
| 161 | |
| 162 | ** Character literals not of length one: |
| 163 | |
| 164 | Previously, Bison quietly converted all character literals to length |
| 165 | one. For example, without warning, Bison interpreted the operators in |
| 166 | the following grammar to be the same token: |
| 167 | |
| 168 | exp: exp '++' |
| 169 | | exp '+' exp |
| 170 | ; |
| 171 | |
| 172 | Bison now warns when a character literal is not of length one. In |
| 173 | some future release, Bison will start reporting an error instead. |
| 174 | |
| 175 | ** Destructor calls fixed for lookaheads altered in semantic actions: |
| 176 | |
| 177 | Previously for deterministic parsers in C, if a user semantic action |
| 178 | altered yychar, the parser in some cases used the old yychar value to |
| 179 | determine which destructor to call for the lookahead upon a syntax |
| 180 | error or upon parser return. This bug has been fixed. |
| 181 | |
| 182 | ** C++ parsers use YYRHSLOC: |
| 183 | |
| 184 | Similarly to the C parsers, the C++ parsers now define the YYRHSLOC |
| 185 | macro and use it in the default YYLLOC_DEFAULT. You are encouraged |
| 186 | to use it. If, for instance, your location structure has `first' |
| 187 | and `last' members, instead of |
| 188 | |
| 189 | # define YYLLOC_DEFAULT(Current, Rhs, N) \ |
| 190 | do \ |
| 191 | if (N) \ |
| 192 | { \ |
| 193 | (Current).first = (Rhs)[1].location.first; \ |
| 194 | (Current).last = (Rhs)[N].location.last; \ |
| 195 | } \ |
| 196 | else \ |
| 197 | { \ |
| 198 | (Current).first = (Current).last = (Rhs)[0].location.last; \ |
| 199 | } \ |
| 200 | while (false) |
| 201 | |
| 202 | use: |
| 203 | |
| 204 | # define YYLLOC_DEFAULT(Current, Rhs, N) \ |
| 205 | do \ |
| 206 | if (N) \ |
| 207 | { \ |
| 208 | (Current).first = YYRHSLOC (Rhs, 1).first; \ |
| 209 | (Current).last = YYRHSLOC (Rhs, N).last; \ |
| 210 | } \ |
| 211 | else \ |
| 212 | { \ |
| 213 | (Current).first = (Current).last = YYRHSLOC (Rhs, 0).last; \ |
| 214 | } \ |
| 215 | while (false) |
| 216 | |
| 217 | ** YYLLOC_DEFAULT in C++: |
| 218 | |
| 219 | The default implementation of YYLLOC_DEFAULT used to be issued in |
| 220 | the header file. It is now output in the implementation file, after |
| 221 | the user %code sections so that its #ifndef guard does not try to |
| 222 | override the user's YYLLOC_DEFAULT if provided. |
| 223 | |
| 224 | ** YYFAIL now produces warnings and Java parsers no longer implement it: |
| 225 | |
| 226 | YYFAIL has existed for many years as an undocumented feature of |
| 227 | deterministic parsers in C generated by Bison. More recently, it was |
| 228 | a documented feature of Bison's experimental Java parsers. As |
| 229 | promised in Bison 2.4.2's NEWS entry, any appearance of YYFAIL in a |
| 230 | semantic action now produces a deprecation warning, and Java parsers |
| 231 | no longer implement YYFAIL at all. For further details, including a |
| 232 | discussion of how to suppress C preprocessor warnings about YYFAIL |
| 233 | being unused, see the Bison 2.4.2 NEWS entry. |
| 234 | |
| 235 | ** Temporary hack for adding a semicolon to the user action: |
| 236 | |
| 237 | Previously, Bison appended a semicolon to every user action for |
| 238 | reductions when the output language defaulted to C (specifically, when |
| 239 | neither %yacc, %language, %skeleton, or equivalent command-line |
| 240 | options were specified). This allowed actions such as |
| 241 | |
| 242 | exp: exp "+" exp { $$ = $1 + $3 }; |
| 243 | |
| 244 | instead of |
| 245 | |
| 246 | exp: exp "+" exp { $$ = $1 + $3; }; |
| 247 | |
| 248 | As a first step in removing this misfeature, Bison now issues a |
| 249 | warning when it appends a semicolon. Moreover, in cases where Bison |
| 250 | cannot easily determine whether a semicolon is needed (for example, an |
| 251 | action ending with a cpp directive or a braced compound initializer), |
| 252 | it no longer appends one. Thus, the C compiler might now complain |
| 253 | about a missing semicolon where it did not before. Future releases of |
| 254 | Bison will cease to append semicolons entirely. |
| 255 | |
| 256 | ** Verbose syntax error message fixes: |
| 257 | |
| 258 | When %error-verbose or the obsolete `#define YYERROR_VERBOSE' is |
| 259 | specified, syntax error messages produced by the generated parser |
| 260 | include the unexpected token as well as a list of expected tokens. |
| 261 | The effect of %nonassoc on these verbose messages has been corrected |
| 262 | in two ways, but a more complete fix requires LAC, described above: |
| 263 | |
| 264 | *** When %nonassoc is used, there can exist parser states that accept no |
| 265 | tokens, and so the parser does not always require a lookahead token |
| 266 | in order to detect a syntax error. Because no unexpected token or |
| 267 | expected tokens can then be reported, the verbose syntax error |
| 268 | message described above is suppressed, and the parser instead |
| 269 | reports the simpler message, `syntax error'. Previously, this |
| 270 | suppression was sometimes erroneously triggered by %nonassoc when a |
| 271 | lookahead was actually required. Now verbose messages are |
| 272 | suppressed only when all previous lookaheads have already been |
| 273 | shifted or discarded. |
| 274 | |
| 275 | *** Previously, the list of expected tokens erroneously included tokens |
| 276 | that would actually induce a syntax error because conflicts for them |
| 277 | were resolved with %nonassoc in the current parser state. Such |
| 278 | tokens are now properly omitted from the list. |
| 279 | |
| 280 | *** Expected token lists are still often wrong due to state merging |
| 281 | (from LALR or IELR) and default reductions, which can both add |
| 282 | invalid tokens and subtract valid tokens. Canonical LR almost |
| 283 | completely fixes this problem by eliminating state merging and |
| 284 | default reductions. However, there is one minor problem left even |
| 285 | when using canonical LR and even after the fixes above. That is, |
| 286 | if the resolution of a conflict with %nonassoc appears in a later |
| 287 | parser state than the one at which some syntax error is |
| 288 | discovered, the conflicted token is still erroneously included in |
| 289 | the expected token list. Bison's new LAC implementation, |
| 290 | described above, eliminates this problem and the need for |
| 291 | canonical LR. However, LAC is still experimental and is disabled |
| 292 | by default. |
| 293 | |
| 294 | ** Java skeleton fixes: |
| 295 | |
| 296 | *** A location handling bug has been fixed. |
| 297 | |
| 298 | *** The top element of each of the value stack and location stack is now |
| 299 | cleared when popped so that it can be garbage collected. |
| 300 | |
| 301 | *** Parser traces now print the top element of the stack. |
| 302 | |
| 303 | ** -W/--warnings fixes: |
| 304 | |
| 305 | *** Bison now properly recognizes the `no-' versions of categories: |
| 306 | |
| 307 | For example, given the following command line, Bison now enables all |
| 308 | warnings except warnings for incompatibilities with POSIX Yacc: |
| 309 | |
| 310 | bison -Wall,no-yacc gram.y |
| 311 | |
| 312 | *** Bison now treats S/R and R/R conflicts like other warnings: |
| 313 | |
| 314 | Previously, conflict reports were independent of Bison's normal |
| 315 | warning system. Now, Bison recognizes the warning categories |
| 316 | `conflicts-sr' and `conflicts-rr'. This change has important |
| 317 | consequences for the -W and --warnings command-line options. For |
| 318 | example: |
| 319 | |
| 320 | bison -Wno-conflicts-sr gram.y # S/R conflicts not reported |
| 321 | bison -Wno-conflicts-rr gram.y # R/R conflicts not reported |
| 322 | bison -Wnone gram.y # no conflicts are reported |
| 323 | bison -Werror gram.y # any conflict is an error |
| 324 | |
| 325 | However, as before, if the %expect or %expect-rr directive is |
| 326 | specified, an unexpected number of conflicts is an error, and an |
| 327 | expected number of conflicts is not reported, so -W and --warning |
| 328 | then have no effect on the conflict report. |
| 329 | |
| 330 | *** The `none' category no longer disables a preceding `error': |
| 331 | |
| 332 | For example, for the following command line, Bison now reports |
| 333 | errors instead of warnings for incompatibilities with POSIX Yacc: |
| 334 | |
| 335 | bison -Werror,none,yacc gram.y |
| 336 | |
| 337 | *** The `none' category now disables all Bison warnings. |
| 338 | |
| 339 | Previously, the `none' category disabled only Bison warnings for |
| 340 | which there existed a specific -W/--warning category. However, |
| 341 | given the following command line, Bison is now guaranteed to |
| 342 | suppress all warnings: |
| 343 | |
| 344 | bison -Wnone gram.y |
| 345 | |
| 346 | * Changes in version 2.4.3 (2010-08-05): |
| 347 | |
| 348 | ** Bison now obeys -Werror and --warnings=error for warnings about |
| 349 | grammar rules that are useless in the parser due to conflicts. |
| 350 | |
| 351 | ** Problems with spawning M4 on at least FreeBSD 8 and FreeBSD 9 have |
| 352 | been fixed. |
| 353 | |
| 354 | ** Failures in the test suite for GCC 4.5 have been fixed. |
| 355 | |
| 356 | ** Failures in the test suite for some versions of Sun Studio C++ have |
| 357 | been fixed. |
| 358 | |
| 359 | ** Contrary to Bison 2.4.2's NEWS entry, it has been decided that |
| 360 | warnings about undefined %prec identifiers will not be converted to |
| 361 | errors in Bison 2.5. They will remain warnings, which should be |
| 362 | sufficient for POSIX while avoiding backward compatibility issues. |
| 363 | |
| 364 | ** Minor documentation fixes. |
| 365 | |
| 366 | * Changes in version 2.4.2 (2010-03-20): |
| 367 | |
| 368 | ** Some portability problems that resulted in failures and livelocks |
| 369 | in the test suite on some versions of at least Solaris, AIX, HP-UX, |
| 370 | RHEL4, and Tru64 have been addressed. As a result, fatal Bison |
| 371 | errors should no longer cause M4 to report a broken pipe on the |
| 372 | affected platforms. |
| 373 | |
| 374 | ** `%prec IDENTIFIER' requires IDENTIFIER to be defined separately. |
| 375 | |
| 376 | POSIX specifies that an error be reported for any identifier that does |
| 377 | not appear on the LHS of a grammar rule and that is not defined by |
| 378 | %token, %left, %right, or %nonassoc. Bison 2.3b and later lost this |
| 379 | error report for the case when an identifier appears only after a |
| 380 | %prec directive. It is now restored. However, for backward |
| 381 | compatibility with recent Bison releases, it is only a warning for |
| 382 | now. In Bison 2.5 and later, it will return to being an error. |
| 383 | [Between the 2.4.2 and 2.4.3 releases, it was decided that this |
| 384 | warning will not be converted to an error in Bison 2.5.] |
| 385 | |
| 386 | ** Detection of GNU M4 1.4.6 or newer during configure is improved. |
| 387 | |
| 388 | ** Warnings from gcc's -Wundef option about undefined YYENABLE_NLS, |
| 389 | YYLTYPE_IS_TRIVIAL, and __STRICT_ANSI__ in C/C++ parsers are now |
| 390 | avoided. |
| 391 | |
| 392 | ** %code is now a permanent feature. |
| 393 | |
| 394 | A traditional Yacc prologue directive is written in the form: |
| 395 | |
| 396 | %{CODE%} |
| 397 | |
| 398 | To provide a more flexible alternative, Bison 2.3b introduced the |
| 399 | %code directive with the following forms for C/C++: |
| 400 | |
| 401 | %code {CODE} |
| 402 | %code requires {CODE} |
| 403 | %code provides {CODE} |
| 404 | %code top {CODE} |
| 405 | |
| 406 | These forms are now considered permanent features of Bison. See the |
| 407 | %code entries in the section "Bison Declaration Summary" in the Bison |
| 408 | manual for a summary of their functionality. See the section |
| 409 | "Prologue Alternatives" for a detailed discussion including the |
| 410 | advantages of %code over the traditional Yacc prologue directive. |
| 411 | |
| 412 | Bison's Java feature as a whole including its current usage of %code |
| 413 | is still considered experimental. |
| 414 | |
| 415 | ** YYFAIL is deprecated and will eventually be removed. |
| 416 | |
| 417 | YYFAIL has existed for many years as an undocumented feature of |
| 418 | deterministic parsers in C generated by Bison. Previously, it was |
| 419 | documented for Bison's experimental Java parsers. YYFAIL is no longer |
| 420 | documented for Java parsers and is formally deprecated in both cases. |
| 421 | Users are strongly encouraged to migrate to YYERROR, which is |
| 422 | specified by POSIX. |
| 423 | |
| 424 | Like YYERROR, you can invoke YYFAIL from a semantic action in order to |
| 425 | induce a syntax error. The most obvious difference from YYERROR is |
| 426 | that YYFAIL will automatically invoke yyerror to report the syntax |
| 427 | error so that you don't have to. However, there are several other |
| 428 | subtle differences between YYERROR and YYFAIL, and YYFAIL suffers from |
| 429 | inherent flaws when %error-verbose or `#define YYERROR_VERBOSE' is |
| 430 | used. For a more detailed discussion, see: |
| 431 | |
| 432 | http://lists.gnu.org/archive/html/bison-patches/2009-12/msg00024.html |
| 433 | |
| 434 | The upcoming Bison 2.5 will remove YYFAIL from Java parsers, but |
| 435 | deterministic parsers in C will continue to implement it. However, |
| 436 | because YYFAIL is already flawed, it seems futile to try to make new |
| 437 | Bison features compatible with it. Thus, during parser generation, |
| 438 | Bison 2.5 will produce a warning whenever it discovers YYFAIL in a |
| 439 | rule action. In a later release, YYFAIL will be disabled for |
| 440 | %error-verbose and `#define YYERROR_VERBOSE'. Eventually, YYFAIL will |
| 441 | be removed altogether. |
| 442 | |
| 443 | There exists at least one case where Bison 2.5's YYFAIL warning will |
| 444 | be a false positive. Some projects add phony uses of YYFAIL and other |
| 445 | Bison-defined macros for the sole purpose of suppressing C |
| 446 | preprocessor warnings (from GCC cpp's -Wunused-macros, for example). |
| 447 | To avoid Bison's future warning, such YYFAIL uses can be moved to the |
| 448 | epilogue (that is, after the second `%%') in the Bison input file. In |
| 449 | this release (2.4.2), Bison already generates its own code to suppress |
| 450 | C preprocessor warnings for YYFAIL, so projects can remove their own |
| 451 | phony uses of YYFAIL if compatibility with Bison releases prior to |
| 452 | 2.4.2 is not necessary. |
| 453 | |
| 454 | ** Internationalization. |
| 455 | |
| 456 | Fix a regression introduced in Bison 2.4: Under some circumstances, |
| 457 | message translations were not installed although supported by the |
| 458 | host system. |
| 459 | |
| 460 | * Changes in version 2.4.1 (2008-12-11): |
| 461 | |
| 462 | ** In the GLR defines file, unexpanded M4 macros in the yylval and yylloc |
| 463 | declarations have been fixed. |
| 464 | |
| 465 | ** Temporary hack for adding a semicolon to the user action. |
| 466 | |
| 467 | Bison used to prepend a trailing semicolon at the end of the user |
| 468 | action for reductions. This allowed actions such as |
| 469 | |
| 470 | exp: exp "+" exp { $$ = $1 + $3 }; |
| 471 | |
| 472 | instead of |
| 473 | |
| 474 | exp: exp "+" exp { $$ = $1 + $3; }; |
| 475 | |
| 476 | Some grammars still depend on this `feature'. Bison 2.4.1 restores |
| 477 | the previous behavior in the case of C output (specifically, when |
| 478 | neither %language or %skeleton or equivalent command-line options |
| 479 | are used) to leave more time for grammars depending on the old |
| 480 | behavior to be adjusted. Future releases of Bison will disable this |
| 481 | feature. |
| 482 | |
| 483 | ** A few minor improvements to the Bison manual. |
| 484 | |
| 485 | * Changes in version 2.4 (2008-11-02): |
| 486 | |
| 487 | ** %language is an experimental feature. |
| 488 | |
| 489 | We first introduced this feature in test release 2.3b as a cleaner |
| 490 | alternative to %skeleton. Since then, we have discussed the possibility of |
| 491 | modifying its effect on Bison's output file names. Thus, in this release, |
| 492 | we consider %language to be an experimental feature that will likely evolve |
| 493 | in future releases. |
| 494 | |
| 495 | ** Forward compatibility with GNU M4 has been improved. |
| 496 | |
| 497 | ** Several bugs in the C++ skeleton and the experimental Java skeleton have been |
| 498 | fixed. |
| 499 | |
| 500 | * Changes in version 2.3b (2008-05-27): |
| 501 | |
| 502 | ** The quotes around NAME that used to be required in the following directive |
| 503 | are now deprecated: |
| 504 | |
| 505 | %define NAME "VALUE" |
| 506 | |
| 507 | ** The directive `%pure-parser' is now deprecated in favor of: |
| 508 | |
| 509 | %define api.pure |
| 510 | |
| 511 | which has the same effect except that Bison is more careful to warn about |
| 512 | unreasonable usage in the latter case. |
| 513 | |
| 514 | ** Push Parsing |
| 515 | |
| 516 | Bison can now generate an LALR(1) parser in C with a push interface. That |
| 517 | is, instead of invoking `yyparse', which pulls tokens from `yylex', you can |
| 518 | push one token at a time to the parser using `yypush_parse', which will |
| 519 | return to the caller after processing each token. By default, the push |
| 520 | interface is disabled. Either of the following directives will enable it: |
| 521 | |
| 522 | %define api.push_pull "push" // Just push; does not require yylex. |
| 523 | %define api.push_pull "both" // Push and pull; requires yylex. |
| 524 | |
| 525 | See the new section `A Push Parser' in the Bison manual for details. |
| 526 | |
| 527 | The current push parsing interface is experimental and may evolve. More user |
| 528 | feedback will help to stabilize it. |
| 529 | |
| 530 | ** The -g and --graph options now output graphs in Graphviz DOT format, |
| 531 | not VCG format. Like --graph, -g now also takes an optional FILE argument |
| 532 | and thus cannot be bundled with other short options. |
| 533 | |
| 534 | ** Java |
| 535 | |
| 536 | Bison can now generate an LALR(1) parser in Java. The skeleton is |
| 537 | `data/lalr1.java'. Consider using the new %language directive instead of |
| 538 | %skeleton to select it. |
| 539 | |
| 540 | See the new section `Java Parsers' in the Bison manual for details. |
| 541 | |
| 542 | The current Java interface is experimental and may evolve. More user |
| 543 | feedback will help to stabilize it. |
| 544 | |
| 545 | ** %language |
| 546 | |
| 547 | This new directive specifies the programming language of the generated |
| 548 | parser, which can be C (the default), C++, or Java. Besides the skeleton |
| 549 | that Bison uses, the directive affects the names of the generated files if |
| 550 | the grammar file's name ends in ".y". |
| 551 | |
| 552 | ** XML Automaton Report |
| 553 | |
| 554 | Bison can now generate an XML report of the LALR(1) automaton using the new |
| 555 | `--xml' option. The current XML schema is experimental and may evolve. More |
| 556 | user feedback will help to stabilize it. |
| 557 | |
| 558 | ** The grammar file may now specify the name of the parser header file using |
| 559 | %defines. For example: |
| 560 | |
| 561 | %defines "parser.h" |
| 562 | |
| 563 | ** When reporting useless rules, useless nonterminals, and unused terminals, |
| 564 | Bison now employs the terms "useless in grammar" instead of "useless", |
| 565 | "useless in parser" instead of "never reduced", and "unused in grammar" |
| 566 | instead of "unused". |
| 567 | |
| 568 | ** Unreachable State Removal |
| 569 | |
| 570 | Previously, Bison sometimes generated parser tables containing unreachable |
| 571 | states. A state can become unreachable during conflict resolution if Bison |
| 572 | disables a shift action leading to it from a predecessor state. Bison now: |
| 573 | |
| 574 | 1. Removes unreachable states. |
| 575 | |
| 576 | 2. Does not report any conflicts that appeared in unreachable states. |
| 577 | WARNING: As a result, you may need to update %expect and %expect-rr |
| 578 | directives in existing grammar files. |
| 579 | |
| 580 | 3. For any rule used only in such states, Bison now reports the rule as |
| 581 | "useless in parser due to conflicts". |
| 582 | |
| 583 | This feature can be disabled with the following directive: |
| 584 | |
| 585 | %define lr.keep_unreachable_states |
| 586 | |
| 587 | See the %define entry in the `Bison Declaration Summary' in the Bison manual |
| 588 | for further discussion. |
| 589 | |
| 590 | ** Lookahead Set Correction in the `.output' Report |
| 591 | |
| 592 | When instructed to generate a `.output' file including lookahead sets |
| 593 | (using `--report=lookahead', for example), Bison now prints each reduction's |
| 594 | lookahead set only next to the associated state's one item that (1) is |
| 595 | associated with the same rule as the reduction and (2) has its dot at the end |
| 596 | of its RHS. Previously, Bison also erroneously printed the lookahead set |
| 597 | next to all of the state's other items associated with the same rule. This |
| 598 | bug affected only the `.output' file and not the generated parser source |
| 599 | code. |
| 600 | |
| 601 | ** --report-file=FILE is a new option to override the default `.output' file |
| 602 | name. |
| 603 | |
| 604 | ** The `=' that used to be required in the following directives is now |
| 605 | deprecated: |
| 606 | |
| 607 | %file-prefix "parser" |
| 608 | %name-prefix "c_" |
| 609 | %output "parser.c" |
| 610 | |
| 611 | ** An Alternative to `%{...%}' -- `%code QUALIFIER {CODE}' |
| 612 | |
| 613 | Bison 2.3a provided a new set of directives as a more flexible alternative to |
| 614 | the traditional Yacc prologue blocks. Those have now been consolidated into |
| 615 | a single %code directive with an optional qualifier field, which identifies |
| 616 | the purpose of the code and thus the location(s) where Bison should generate |
| 617 | it: |
| 618 | |
| 619 | 1. `%code {CODE}' replaces `%after-header {CODE}' |
| 620 | 2. `%code requires {CODE}' replaces `%start-header {CODE}' |
| 621 | 3. `%code provides {CODE}' replaces `%end-header {CODE}' |
| 622 | 4. `%code top {CODE}' replaces `%before-header {CODE}' |
| 623 | |
| 624 | See the %code entries in section `Bison Declaration Summary' in the Bison |
| 625 | manual for a summary of the new functionality. See the new section `Prologue |
| 626 | Alternatives' for a detailed discussion including the advantages of %code |
| 627 | over the traditional Yacc prologues. |
| 628 | |
| 629 | The prologue alternatives are experimental. More user feedback will help to |
| 630 | determine whether they should become permanent features. |
| 631 | |
| 632 | ** Revised warning: unset or unused mid-rule values |
| 633 | |
| 634 | Since Bison 2.2, Bison has warned about mid-rule values that are set but not |
| 635 | used within any of the actions of the parent rule. For example, Bison warns |
| 636 | about unused $2 in: |
| 637 | |
| 638 | exp: '1' { $$ = 1; } '+' exp { $$ = $1 + $4; }; |
| 639 | |
| 640 | Now, Bison also warns about mid-rule values that are used but not set. For |
| 641 | example, Bison warns about unset $$ in the mid-rule action in: |
| 642 | |
| 643 | exp: '1' { $1 = 1; } '+' exp { $$ = $2 + $4; }; |
| 644 | |
| 645 | However, Bison now disables both of these warnings by default since they |
| 646 | sometimes prove to be false alarms in existing grammars employing the Yacc |
| 647 | constructs $0 or $-N (where N is some positive integer). |
| 648 | |
| 649 | To enable these warnings, specify the option `--warnings=midrule-values' or |
| 650 | `-W', which is a synonym for `--warnings=all'. |
| 651 | |
| 652 | ** Default %destructor or %printer with `<*>' or `<>' |
| 653 | |
| 654 | Bison now recognizes two separate kinds of default %destructor's and |
| 655 | %printer's: |
| 656 | |
| 657 | 1. Place `<*>' in a %destructor/%printer symbol list to define a default |
| 658 | %destructor/%printer for all grammar symbols for which you have formally |
| 659 | declared semantic type tags. |
| 660 | |
| 661 | 2. Place `<>' in a %destructor/%printer symbol list to define a default |
| 662 | %destructor/%printer for all grammar symbols without declared semantic |
| 663 | type tags. |
| 664 | |
| 665 | Bison no longer supports the `%symbol-default' notation from Bison 2.3a. |
| 666 | `<*>' and `<>' combined achieve the same effect with one exception: Bison no |
| 667 | longer applies any %destructor to a mid-rule value if that mid-rule value is |
| 668 | not actually ever referenced using either $$ or $n in a semantic action. |
| 669 | |
| 670 | The default %destructor's and %printer's are experimental. More user |
| 671 | feedback will help to determine whether they should become permanent |
| 672 | features. |
| 673 | |
| 674 | See the section `Freeing Discarded Symbols' in the Bison manual for further |
| 675 | details. |
| 676 | |
| 677 | ** %left, %right, and %nonassoc can now declare token numbers. This is required |
| 678 | by POSIX. However, see the end of section `Operator Precedence' in the Bison |
| 679 | manual for a caveat concerning the treatment of literal strings. |
| 680 | |
| 681 | ** The nonfunctional --no-parser, -n, and %no-parser options have been |
| 682 | completely removed from Bison. |
| 683 | |
| 684 | * Changes in version 2.3a, 2006-09-13: |
| 685 | |
| 686 | ** Instead of %union, you can define and use your own union type |
| 687 | YYSTYPE if your grammar contains at least one <type> tag. |
| 688 | Your YYSTYPE need not be a macro; it can be a typedef. |
| 689 | This change is for compatibility with other Yacc implementations, |
| 690 | and is required by POSIX. |
| 691 | |
| 692 | ** Locations columns and lines start at 1. |
| 693 | In accordance with the GNU Coding Standards and Emacs. |
| 694 | |
| 695 | ** You may now declare per-type and default %destructor's and %printer's: |
| 696 | |
| 697 | For example: |
| 698 | |
| 699 | %union { char *string; } |
| 700 | %token <string> STRING1 |
| 701 | %token <string> STRING2 |
| 702 | %type <string> string1 |
| 703 | %type <string> string2 |
| 704 | %union { char character; } |
| 705 | %token <character> CHR |
| 706 | %type <character> chr |
| 707 | %destructor { free ($$); } %symbol-default |
| 708 | %destructor { free ($$); printf ("%d", @$.first_line); } STRING1 string1 |
| 709 | %destructor { } <character> |
| 710 | |
| 711 | guarantees that, when the parser discards any user-defined symbol that has a |
| 712 | semantic type tag other than `<character>', it passes its semantic value to |
| 713 | `free'. However, when the parser discards a `STRING1' or a `string1', it |
| 714 | also prints its line number to `stdout'. It performs only the second |
| 715 | `%destructor' in this case, so it invokes `free' only once. |
| 716 | |
| 717 | [Although we failed to mention this here in the 2.3a release, the default |
| 718 | %destructor's and %printer's were experimental, and they were rewritten in |
| 719 | future versions.] |
| 720 | |
| 721 | ** Except for LALR(1) parsers in C with POSIX Yacc emulation enabled (with `-y', |
| 722 | `--yacc', or `%yacc'), Bison no longer generates #define statements for |
| 723 | associating token numbers with token names. Removing the #define statements |
| 724 | helps to sanitize the global namespace during preprocessing, but POSIX Yacc |
| 725 | requires them. Bison still generates an enum for token names in all cases. |
| 726 | |
| 727 | ** Handling of traditional Yacc prologue blocks is now more consistent but |
| 728 | potentially incompatible with previous releases of Bison. |
| 729 | |
| 730 | As before, you declare prologue blocks in your grammar file with the |
| 731 | `%{ ... %}' syntax. To generate the pre-prologue, Bison concatenates all |
| 732 | prologue blocks that you've declared before the first %union. To generate |
| 733 | the post-prologue, Bison concatenates all prologue blocks that you've |
| 734 | declared after the first %union. |
| 735 | |
| 736 | Previous releases of Bison inserted the pre-prologue into both the header |
| 737 | file and the code file in all cases except for LALR(1) parsers in C. In the |
| 738 | latter case, Bison inserted it only into the code file. For parsers in C++, |
| 739 | the point of insertion was before any token definitions (which associate |
| 740 | token numbers with names). For parsers in C, the point of insertion was |
| 741 | after the token definitions. |
| 742 | |
| 743 | Now, Bison never inserts the pre-prologue into the header file. In the code |
| 744 | file, it always inserts it before the token definitions. |
| 745 | |
| 746 | ** Bison now provides a more flexible alternative to the traditional Yacc |
| 747 | prologue blocks: %before-header, %start-header, %end-header, and |
| 748 | %after-header. |
| 749 | |
| 750 | For example, the following declaration order in the grammar file reflects the |
| 751 | order in which Bison will output these code blocks. However, you are free to |
| 752 | declare these code blocks in your grammar file in whatever order is most |
| 753 | convenient for you: |
| 754 | |
| 755 | %before-header { |
| 756 | /* Bison treats this block like a pre-prologue block: it inserts it into |
| 757 | * the code file before the contents of the header file. It does *not* |
| 758 | * insert it into the header file. This is a good place to put |
| 759 | * #include's that you want at the top of your code file. A common |
| 760 | * example is `#include "system.h"'. */ |
| 761 | } |
| 762 | %start-header { |
| 763 | /* Bison inserts this block into both the header file and the code file. |
| 764 | * In both files, the point of insertion is before any Bison-generated |
| 765 | * token, semantic type, location type, and class definitions. This is a |
| 766 | * good place to define %union dependencies, for example. */ |
| 767 | } |
| 768 | %union { |
| 769 | /* Unlike the traditional Yacc prologue blocks, the output order for the |
| 770 | * new %*-header blocks is not affected by their declaration position |
| 771 | * relative to any %union in the grammar file. */ |
| 772 | } |
| 773 | %end-header { |
| 774 | /* Bison inserts this block into both the header file and the code file. |
| 775 | * In both files, the point of insertion is after the Bison-generated |
| 776 | * definitions. This is a good place to declare or define public |
| 777 | * functions or data structures that depend on the Bison-generated |
| 778 | * definitions. */ |
| 779 | } |
| 780 | %after-header { |
| 781 | /* Bison treats this block like a post-prologue block: it inserts it into |
| 782 | * the code file after the contents of the header file. It does *not* |
| 783 | * insert it into the header file. This is a good place to declare or |
| 784 | * define internal functions or data structures that depend on the |
| 785 | * Bison-generated definitions. */ |
| 786 | } |
| 787 | |
| 788 | If you have multiple occurrences of any one of the above declarations, Bison |
| 789 | will concatenate the contents in declaration order. |
| 790 | |
| 791 | [Although we failed to mention this here in the 2.3a release, the prologue |
| 792 | alternatives were experimental, and they were rewritten in future versions.] |
| 793 | |
| 794 | ** The option `--report=look-ahead' has been changed to `--report=lookahead'. |
| 795 | The old spelling still works, but is not documented and may be removed |
| 796 | in a future release. |
| 797 | |
| 798 | * Changes in version 2.3, 2006-06-05: |
| 799 | |
| 800 | ** GLR grammars should now use `YYRECOVERING ()' instead of `YYRECOVERING', |
| 801 | for compatibility with LALR(1) grammars. |
| 802 | |
| 803 | ** It is now documented that any definition of YYSTYPE or YYLTYPE should |
| 804 | be to a type name that does not contain parentheses or brackets. |
| 805 | |
| 806 | * Changes in version 2.2, 2006-05-19: |
| 807 | |
| 808 | ** The distribution terms for all Bison-generated parsers now permit |
| 809 | using the parsers in nonfree programs. Previously, this permission |
| 810 | was granted only for Bison-generated LALR(1) parsers in C. |
| 811 | |
| 812 | ** %name-prefix changes the namespace name in C++ outputs. |
| 813 | |
| 814 | ** The C++ parsers export their token_type. |
| 815 | |
| 816 | ** Bison now allows multiple %union declarations, and concatenates |
| 817 | their contents together. |
| 818 | |
| 819 | ** New warning: unused values |
| 820 | Right-hand side symbols whose values are not used are reported, |
| 821 | if the symbols have destructors. For instance: |
| 822 | |
| 823 | exp: exp "?" exp ":" exp { $1 ? $1 : $3; } |
| 824 | | exp "+" exp |
| 825 | ; |
| 826 | |
| 827 | will trigger a warning about $$ and $5 in the first rule, and $3 in |
| 828 | the second ($1 is copied to $$ by the default rule). This example |
| 829 | most likely contains three errors, and could be rewritten as: |
| 830 | |
| 831 | exp: exp "?" exp ":" exp |
| 832 | { $$ = $1 ? $3 : $5; free ($1 ? $5 : $3); free ($1); } |
| 833 | | exp "+" exp |
| 834 | { $$ = $1 ? $1 : $3; if ($1) free ($3); } |
| 835 | ; |
| 836 | |
| 837 | However, if the original actions were really intended, memory leaks |
| 838 | and all, the warnings can be suppressed by letting Bison believe the |
| 839 | values are used, e.g.: |
| 840 | |
| 841 | exp: exp "?" exp ":" exp { $1 ? $1 : $3; (void) ($$, $5); } |
| 842 | | exp "+" exp { $$ = $1; (void) $3; } |
| 843 | ; |
| 844 | |
| 845 | If there are mid-rule actions, the warning is issued if no action |
| 846 | uses it. The following triggers no warning: $1 and $3 are used. |
| 847 | |
| 848 | exp: exp { push ($1); } '+' exp { push ($3); sum (); }; |
| 849 | |
| 850 | The warning is intended to help catching lost values and memory leaks. |
| 851 | If a value is ignored, its associated memory typically is not reclaimed. |
| 852 | |
| 853 | ** %destructor vs. YYABORT, YYACCEPT, and YYERROR. |
| 854 | Destructors are now called when user code invokes YYABORT, YYACCEPT, |
| 855 | and YYERROR, for all objects on the stack, other than objects |
| 856 | corresponding to the right-hand side of the current rule. |
| 857 | |
| 858 | ** %expect, %expect-rr |
| 859 | Incorrect numbers of expected conflicts are now actual errors, |
| 860 | instead of warnings. |
| 861 | |
| 862 | ** GLR, YACC parsers. |
| 863 | The %parse-params are available in the destructors (and the |
| 864 | experimental printers) as per the documentation. |
| 865 | |
| 866 | ** Bison now warns if it finds a stray `$' or `@' in an action. |
| 867 | |
| 868 | ** %require "VERSION" |
| 869 | This specifies that the grammar file depends on features implemented |
| 870 | in Bison version VERSION or higher. |
| 871 | |
| 872 | ** lalr1.cc: The token and value types are now class members. |
| 873 | The tokens were defined as free form enums and cpp macros. YYSTYPE |
| 874 | was defined as a free form union. They are now class members: |
| 875 | tokens are enumerations of the `yy::parser::token' struct, and the |
| 876 | semantic values have the `yy::parser::semantic_type' type. |
| 877 | |
| 878 | If you do not want or can update to this scheme, the directive |
| 879 | `%define "global_tokens_and_yystype" "1"' triggers the global |
| 880 | definition of tokens and YYSTYPE. This change is suitable both |
| 881 | for previous releases of Bison, and this one. |
| 882 | |
| 883 | If you wish to update, then make sure older version of Bison will |
| 884 | fail using `%require "2.2"'. |
| 885 | |
| 886 | ** DJGPP support added. |
| 887 | \f |
| 888 | * Changes in version 2.1, 2005-09-16: |
| 889 | |
| 890 | ** The C++ lalr1.cc skeleton supports %lex-param. |
| 891 | |
| 892 | ** Bison-generated parsers now support the translation of diagnostics like |
| 893 | "syntax error" into languages other than English. The default |
| 894 | language is still English. For details, please see the new |
| 895 | Internationalization section of the Bison manual. Software |
| 896 | distributors should also see the new PACKAGING file. Thanks to |
| 897 | Bruno Haible for this new feature. |
| 898 | |
| 899 | ** Wording in the Bison-generated parsers has been changed slightly to |
| 900 | simplify translation. In particular, the message "memory exhausted" |
| 901 | has replaced "parser stack overflow", as the old message was not |
| 902 | always accurate for modern Bison-generated parsers. |
| 903 | |
| 904 | ** Destructors are now called when the parser aborts, for all symbols left |
| 905 | behind on the stack. Also, the start symbol is now destroyed after a |
| 906 | successful parse. In both cases, the behavior was formerly inconsistent. |
| 907 | |
| 908 | ** When generating verbose diagnostics, Bison-generated parsers no longer |
| 909 | quote the literal strings associated with tokens. For example, for |
| 910 | a syntax error associated with '%token NUM "number"' they might |
| 911 | print 'syntax error, unexpected number' instead of 'syntax error, |
| 912 | unexpected "number"'. |
| 913 | \f |
| 914 | * Changes in version 2.0, 2004-12-25: |
| 915 | |
| 916 | ** Possibly-incompatible changes |
| 917 | |
| 918 | - Bison-generated parsers no longer default to using the alloca function |
| 919 | (when available) to extend the parser stack, due to widespread |
| 920 | problems in unchecked stack-overflow detection. You can "#define |
| 921 | YYSTACK_USE_ALLOCA 1" to require the use of alloca, but please read |
| 922 | the manual to determine safe values for YYMAXDEPTH in that case. |
| 923 | |
| 924 | - Error token location. |
| 925 | During error recovery, the location of the syntax error is updated |
| 926 | to cover the whole sequence covered by the error token: it includes |
| 927 | the shifted symbols thrown away during the first part of the error |
| 928 | recovery, and the lookahead rejected during the second part. |
| 929 | |
| 930 | - Semicolon changes: |
| 931 | . Stray semicolons are no longer allowed at the start of a grammar. |
| 932 | . Semicolons are now required after in-grammar declarations. |
| 933 | |
| 934 | - Unescaped newlines are no longer allowed in character constants or |
| 935 | string literals. They were never portable, and GCC 3.4.0 has |
| 936 | dropped support for them. Better diagnostics are now generated if |
| 937 | forget a closing quote. |
| 938 | |
| 939 | - NUL bytes are no longer allowed in Bison string literals, unfortunately. |
| 940 | |
| 941 | ** New features |
| 942 | |
| 943 | - GLR grammars now support locations. |
| 944 | |
| 945 | - New directive: %initial-action. |
| 946 | This directive allows the user to run arbitrary code (including |
| 947 | initializing @$) from yyparse before parsing starts. |
| 948 | |
| 949 | - A new directive "%expect-rr N" specifies the expected number of |
| 950 | reduce/reduce conflicts in GLR parsers. |
| 951 | |
| 952 | - %token numbers can now be hexadecimal integers, e.g., `%token FOO 0x12d'. |
| 953 | This is a GNU extension. |
| 954 | |
| 955 | - The option `--report=lookahead' was changed to `--report=look-ahead'. |
| 956 | [However, this was changed back after 2.3.] |
| 957 | |
| 958 | - Experimental %destructor support has been added to lalr1.cc. |
| 959 | |
| 960 | - New configure option --disable-yacc, to disable installation of the |
| 961 | yacc command and -ly library introduced in 1.875 for POSIX conformance. |
| 962 | |
| 963 | ** Bug fixes |
| 964 | |
| 965 | - For now, %expect-count violations are now just warnings, not errors. |
| 966 | This is for compatibility with Bison 1.75 and earlier (when there are |
| 967 | reduce/reduce conflicts) and with Bison 1.30 and earlier (when there |
| 968 | are too many or too few shift/reduce conflicts). However, in future |
| 969 | versions of Bison we plan to improve the %expect machinery so that |
| 970 | these violations will become errors again. |
| 971 | |
| 972 | - Within Bison itself, numbers (e.g., goto numbers) are no longer |
| 973 | arbitrarily limited to 16-bit counts. |
| 974 | |
| 975 | - Semicolons are now allowed before "|" in grammar rules, as POSIX requires. |
| 976 | \f |
| 977 | * Changes in version 1.875, 2003-01-01: |
| 978 | |
| 979 | ** The documentation license has been upgraded to version 1.2 |
| 980 | of the GNU Free Documentation License. |
| 981 | |
| 982 | ** syntax error processing |
| 983 | |
| 984 | - In Yacc-style parsers YYLLOC_DEFAULT is now used to compute error |
| 985 | locations too. This fixes bugs in error-location computation. |
| 986 | |
| 987 | - %destructor |
| 988 | It is now possible to reclaim the memory associated to symbols |
| 989 | discarded during error recovery. This feature is still experimental. |
| 990 | |
| 991 | - %error-verbose |
| 992 | This new directive is preferred over YYERROR_VERBOSE. |
| 993 | |
| 994 | - #defining yyerror to steal internal variables is discouraged. |
| 995 | It is not guaranteed to work forever. |
| 996 | |
| 997 | ** POSIX conformance |
| 998 | |
| 999 | - Semicolons are once again optional at the end of grammar rules. |
| 1000 | This reverts to the behavior of Bison 1.33 and earlier, and improves |
| 1001 | compatibility with Yacc. |
| 1002 | |
| 1003 | - `parse error' -> `syntax error' |
| 1004 | Bison now uniformly uses the term `syntax error'; formerly, the code |
| 1005 | and manual sometimes used the term `parse error' instead. POSIX |
| 1006 | requires `syntax error' in diagnostics, and it was thought better to |
| 1007 | be consistent. |
| 1008 | |
| 1009 | - The documentation now emphasizes that yylex and yyerror must be |
| 1010 | declared before use. C99 requires this. |
| 1011 | |
| 1012 | - Bison now parses C99 lexical constructs like UCNs and |
| 1013 | backslash-newline within C escape sequences, as POSIX 1003.1-2001 requires. |
| 1014 | |
| 1015 | - File names are properly escaped in C output. E.g., foo\bar.y is |
| 1016 | output as "foo\\bar.y". |
| 1017 | |
| 1018 | - Yacc command and library now available |
| 1019 | The Bison distribution now installs a `yacc' command, as POSIX requires. |
| 1020 | Also, Bison now installs a small library liby.a containing |
| 1021 | implementations of Yacc-compatible yyerror and main functions. |
| 1022 | This library is normally not useful, but POSIX requires it. |
| 1023 | |
| 1024 | - Type clashes now generate warnings, not errors. |
| 1025 | |
| 1026 | - If the user does not define YYSTYPE as a macro, Bison now declares it |
| 1027 | using typedef instead of defining it as a macro. |
| 1028 | For consistency, YYLTYPE is also declared instead of defined. |
| 1029 | |
| 1030 | ** Other compatibility issues |
| 1031 | |
| 1032 | - %union directives can now have a tag before the `{', e.g., the |
| 1033 | directive `%union foo {...}' now generates the C code |
| 1034 | `typedef union foo { ... } YYSTYPE;'; this is for Yacc compatibility. |
| 1035 | The default union tag is `YYSTYPE', for compatibility with Solaris 9 Yacc. |
| 1036 | For consistency, YYLTYPE's struct tag is now `YYLTYPE' not `yyltype'. |
| 1037 | This is for compatibility with both Yacc and Bison 1.35. |
| 1038 | |
| 1039 | - `;' is output before the terminating `}' of an action, for |
| 1040 | compatibility with Bison 1.35. |
| 1041 | |
| 1042 | - Bison now uses a Yacc-style format for conflict reports, e.g., |
| 1043 | `conflicts: 2 shift/reduce, 1 reduce/reduce'. |
| 1044 | |
| 1045 | - `yystype' and `yyltype' are now obsolescent macros instead of being |
| 1046 | typedefs or tags; they are no longer documented and are planned to be |
| 1047 | withdrawn in a future release. |
| 1048 | |
| 1049 | ** GLR parser notes |
| 1050 | |
| 1051 | - GLR and inline |
| 1052 | Users of Bison have to decide how they handle the portability of the |
| 1053 | C keyword `inline'. |
| 1054 | |
| 1055 | - `parsing stack overflow...' -> `parser stack overflow' |
| 1056 | GLR parsers now report `parser stack overflow' as per the Bison manual. |
| 1057 | |
| 1058 | ** Bison now warns if it detects conflicting outputs to the same file, |
| 1059 | e.g., it generates a warning for `bison -d -o foo.h foo.y' since |
| 1060 | that command outputs both code and header to foo.h. |
| 1061 | |
| 1062 | ** #line in output files |
| 1063 | - --no-line works properly. |
| 1064 | |
| 1065 | ** Bison can no longer be built by a K&R C compiler; it requires C89 or |
| 1066 | later to be built. This change originally took place a few versions |
| 1067 | ago, but nobody noticed until we recently asked someone to try |
| 1068 | building Bison with a K&R C compiler. |
| 1069 | \f |
| 1070 | * Changes in version 1.75, 2002-10-14: |
| 1071 | |
| 1072 | ** Bison should now work on 64-bit hosts. |
| 1073 | |
| 1074 | ** Indonesian translation thanks to Tedi Heriyanto. |
| 1075 | |
| 1076 | ** GLR parsers |
| 1077 | Fix spurious parse errors. |
| 1078 | |
| 1079 | ** Pure parsers |
| 1080 | Some people redefine yyerror to steal yyparse' private variables. |
| 1081 | Reenable this trick until an official feature replaces it. |
| 1082 | |
| 1083 | ** Type Clashes |
| 1084 | In agreement with POSIX and with other Yaccs, leaving a default |
| 1085 | action is valid when $$ is untyped, and $1 typed: |
| 1086 | |
| 1087 | untyped: ... typed; |
| 1088 | |
| 1089 | but the converse remains an error: |
| 1090 | |
| 1091 | typed: ... untyped; |
| 1092 | |
| 1093 | ** Values of mid-rule actions |
| 1094 | The following code: |
| 1095 | |
| 1096 | foo: { ... } { $$ = $1; } ... |
| 1097 | |
| 1098 | was incorrectly rejected: $1 is defined in the second mid-rule |
| 1099 | action, and is equal to the $$ of the first mid-rule action. |
| 1100 | \f |
| 1101 | * Changes in version 1.50, 2002-10-04: |
| 1102 | |
| 1103 | ** GLR parsing |
| 1104 | The declaration |
| 1105 | %glr-parser |
| 1106 | causes Bison to produce a Generalized LR (GLR) parser, capable of handling |
| 1107 | almost any context-free grammar, ambiguous or not. The new declarations |
| 1108 | %dprec and %merge on grammar rules allow parse-time resolution of |
| 1109 | ambiguities. Contributed by Paul Hilfinger. |
| 1110 | |
| 1111 | Unfortunately Bison 1.50 does not work properly on 64-bit hosts |
| 1112 | like the Alpha, so please stick to 32-bit hosts for now. |
| 1113 | |
| 1114 | ** Output Directory |
| 1115 | When not in Yacc compatibility mode, when the output file was not |
| 1116 | specified, running `bison foo/bar.y' created `foo/bar.c'. It |
| 1117 | now creates `bar.c'. |
| 1118 | |
| 1119 | ** Undefined token |
| 1120 | The undefined token was systematically mapped to 2 which prevented |
| 1121 | the use of 2 by the user. This is no longer the case. |
| 1122 | |
| 1123 | ** Unknown token numbers |
| 1124 | If yylex returned an out of range value, yyparse could die. This is |
| 1125 | no longer the case. |
| 1126 | |
| 1127 | ** Error token |
| 1128 | According to POSIX, the error token must be 256. |
| 1129 | Bison extends this requirement by making it a preference: *if* the |
| 1130 | user specified that one of her tokens is numbered 256, then error |
| 1131 | will be mapped onto another number. |
| 1132 | |
| 1133 | ** Verbose error messages |
| 1134 | They no longer report `..., expecting error or...' for states where |
| 1135 | error recovery is possible. |
| 1136 | |
| 1137 | ** End token |
| 1138 | Defaults to `$end' instead of `$'. |
| 1139 | |
| 1140 | ** Error recovery now conforms to documentation and to POSIX |
| 1141 | When a Bison-generated parser encounters a syntax error, it now pops |
| 1142 | the stack until it finds a state that allows shifting the error |
| 1143 | token. Formerly, it popped the stack until it found a state that |
| 1144 | allowed some non-error action other than a default reduction on the |
| 1145 | error token. The new behavior has long been the documented behavior, |
| 1146 | and has long been required by POSIX. For more details, please see |
| 1147 | Paul Eggert, "Reductions during Bison error handling" (2002-05-20) |
| 1148 | <http://lists.gnu.org/archive/html/bug-bison/2002-05/msg00038.html>. |
| 1149 | |
| 1150 | ** Traces |
| 1151 | Popped tokens and nonterminals are now reported. |
| 1152 | |
| 1153 | ** Larger grammars |
| 1154 | Larger grammars are now supported (larger token numbers, larger grammar |
| 1155 | size (= sum of the LHS and RHS lengths), larger LALR tables). |
| 1156 | Formerly, many of these numbers ran afoul of 16-bit limits; |
| 1157 | now these limits are 32 bits on most hosts. |
| 1158 | |
| 1159 | ** Explicit initial rule |
| 1160 | Bison used to play hacks with the initial rule, which the user does |
| 1161 | not write. It is now explicit, and visible in the reports and |
| 1162 | graphs as rule 0. |
| 1163 | |
| 1164 | ** Useless rules |
| 1165 | Before, Bison reported the useless rules, but, although not used, |
| 1166 | included them in the parsers. They are now actually removed. |
| 1167 | |
| 1168 | ** Useless rules, useless nonterminals |
| 1169 | They are now reported, as a warning, with their locations. |
| 1170 | |
| 1171 | ** Rules never reduced |
| 1172 | Rules that can never be reduced because of conflicts are now |
| 1173 | reported. |
| 1174 | |
| 1175 | ** Incorrect `Token not used' |
| 1176 | On a grammar such as |
| 1177 | |
| 1178 | %token useless useful |
| 1179 | %% |
| 1180 | exp: '0' %prec useful; |
| 1181 | |
| 1182 | where a token was used to set the precedence of the last rule, |
| 1183 | bison reported both `useful' and `useless' as useless tokens. |
| 1184 | |
| 1185 | ** Revert the C++ namespace changes introduced in 1.31 |
| 1186 | as they caused too many portability hassles. |
| 1187 | |
| 1188 | ** Default locations |
| 1189 | By an accident of design, the default computation of @$ was |
| 1190 | performed after another default computation was performed: @$ = @1. |
| 1191 | The latter is now removed: YYLLOC_DEFAULT is fully responsible of |
| 1192 | the computation of @$. |
| 1193 | |
| 1194 | ** Token end-of-file |
| 1195 | The token end of file may be specified by the user, in which case, |
| 1196 | the user symbol is used in the reports, the graphs, and the verbose |
| 1197 | error messages instead of `$end', which remains being the default. |
| 1198 | For instance |
| 1199 | %token MYEOF 0 |
| 1200 | or |
| 1201 | %token MYEOF 0 "end of file" |
| 1202 | |
| 1203 | ** Semantic parser |
| 1204 | This old option, which has been broken for ages, is removed. |
| 1205 | |
| 1206 | ** New translations |
| 1207 | Brazilian Portuguese, thanks to Alexandre Folle de Menezes. |
| 1208 | Croatian, thanks to Denis Lackovic. |
| 1209 | |
| 1210 | ** Incorrect token definitions |
| 1211 | When given `%token 'a' "A"', Bison used to output `#define 'a' 65'. |
| 1212 | |
| 1213 | ** Token definitions as enums |
| 1214 | Tokens are output both as the traditional #define's, and, provided |
| 1215 | the compiler supports ANSI C or is a C++ compiler, as enums. |
| 1216 | This lets debuggers display names instead of integers. |
| 1217 | |
| 1218 | ** Reports |
| 1219 | In addition to --verbose, bison supports --report=THINGS, which |
| 1220 | produces additional information: |
| 1221 | - itemset |
| 1222 | complete the core item sets with their closure |
| 1223 | - lookahead [changed to `look-ahead' in 1.875e through 2.3, but changed back] |
| 1224 | explicitly associate lookahead tokens to items |
| 1225 | - solved |
| 1226 | describe shift/reduce conflicts solving. |
| 1227 | Bison used to systematically output this information on top of |
| 1228 | the report. Solved conflicts are now attached to their states. |
| 1229 | |
| 1230 | ** Type clashes |
| 1231 | Previous versions don't complain when there is a type clash on |
| 1232 | the default action if the rule has a mid-rule action, such as in: |
| 1233 | |
| 1234 | %type <foo> bar |
| 1235 | %% |
| 1236 | bar: '0' {} '0'; |
| 1237 | |
| 1238 | This is fixed. |
| 1239 | |
| 1240 | ** GNU M4 is now required when using Bison. |
| 1241 | \f |
| 1242 | * Changes in version 1.35, 2002-03-25: |
| 1243 | |
| 1244 | ** C Skeleton |
| 1245 | Some projects use Bison's C parser with C++ compilers, and define |
| 1246 | YYSTYPE as a class. The recent adjustment of C parsers for data |
| 1247 | alignment and 64 bit architectures made this impossible. |
| 1248 | |
| 1249 | Because for the time being no real solution for C++ parser |
| 1250 | generation exists, kludges were implemented in the parser to |
| 1251 | maintain this use. In the future, when Bison has C++ parsers, this |
| 1252 | kludge will be disabled. |
| 1253 | |
| 1254 | This kludge also addresses some C++ problems when the stack was |
| 1255 | extended. |
| 1256 | \f |
| 1257 | * Changes in version 1.34, 2002-03-12: |
| 1258 | |
| 1259 | ** File name clashes are detected |
| 1260 | $ bison foo.y -d -o foo.x |
| 1261 | fatal error: header and parser would both be named `foo.x' |
| 1262 | |
| 1263 | ** A missing `;' at the end of a rule triggers a warning |
| 1264 | In accordance with POSIX, and in agreement with other |
| 1265 | Yacc implementations, Bison will mandate this semicolon in the near |
| 1266 | future. This eases the implementation of a Bison parser of Bison |
| 1267 | grammars by making this grammar LALR(1) instead of LR(2). To |
| 1268 | facilitate the transition, this release introduces a warning. |
| 1269 | |
| 1270 | ** Revert the C++ namespace changes introduced in 1.31, as they caused too |
| 1271 | many portability hassles. |
| 1272 | |
| 1273 | ** DJGPP support added. |
| 1274 | |
| 1275 | ** Fix test suite portability problems. |
| 1276 | \f |
| 1277 | * Changes in version 1.33, 2002-02-07: |
| 1278 | |
| 1279 | ** Fix C++ issues |
| 1280 | Groff could not be compiled for the definition of size_t was lacking |
| 1281 | under some conditions. |
| 1282 | |
| 1283 | ** Catch invalid @n |
| 1284 | As is done with $n. |
| 1285 | \f |
| 1286 | * Changes in version 1.32, 2002-01-23: |
| 1287 | |
| 1288 | ** Fix Yacc output file names |
| 1289 | |
| 1290 | ** Portability fixes |
| 1291 | |
| 1292 | ** Italian, Dutch translations |
| 1293 | \f |
| 1294 | * Changes in version 1.31, 2002-01-14: |
| 1295 | |
| 1296 | ** Many Bug Fixes |
| 1297 | |
| 1298 | ** GNU Gettext and %expect |
| 1299 | GNU Gettext asserts 10 s/r conflicts, but there are 7. Now that |
| 1300 | Bison dies on incorrect %expectations, we fear there will be |
| 1301 | too many bug reports for Gettext, so _for the time being_, %expect |
| 1302 | does not trigger an error when the input file is named `plural.y'. |
| 1303 | |
| 1304 | ** Use of alloca in parsers |
| 1305 | If YYSTACK_USE_ALLOCA is defined to 0, then the parsers will use |
| 1306 | malloc exclusively. Since 1.29, but was not NEWS'ed. |
| 1307 | |
| 1308 | alloca is used only when compiled with GCC, to avoid portability |
| 1309 | problems as on AIX. |
| 1310 | |
| 1311 | ** yyparse now returns 2 if memory is exhausted; formerly it dumped core. |
| 1312 | |
| 1313 | ** When the generated parser lacks debugging code, YYDEBUG is now 0 |
| 1314 | (as POSIX requires) instead of being undefined. |
| 1315 | |
| 1316 | ** User Actions |
| 1317 | Bison has always permitted actions such as { $$ = $1 }: it adds the |
| 1318 | ending semicolon. Now if in Yacc compatibility mode, the semicolon |
| 1319 | is no longer output: one has to write { $$ = $1; }. |
| 1320 | |
| 1321 | ** Better C++ compliance |
| 1322 | The output parsers try to respect C++ namespaces. |
| 1323 | [This turned out to be a failed experiment, and it was reverted later.] |
| 1324 | |
| 1325 | ** Reduced Grammars |
| 1326 | Fixed bugs when reporting useless nonterminals. |
| 1327 | |
| 1328 | ** 64 bit hosts |
| 1329 | The parsers work properly on 64 bit hosts. |
| 1330 | |
| 1331 | ** Error messages |
| 1332 | Some calls to strerror resulted in scrambled or missing error messages. |
| 1333 | |
| 1334 | ** %expect |
| 1335 | When the number of shift/reduce conflicts is correct, don't issue |
| 1336 | any warning. |
| 1337 | |
| 1338 | ** The verbose report includes the rule line numbers. |
| 1339 | |
| 1340 | ** Rule line numbers are fixed in traces. |
| 1341 | |
| 1342 | ** Swedish translation |
| 1343 | |
| 1344 | ** Parse errors |
| 1345 | Verbose parse error messages from the parsers are better looking. |
| 1346 | Before: parse error: unexpected `'/'', expecting `"number"' or `'-'' or `'('' |
| 1347 | Now: parse error: unexpected '/', expecting "number" or '-' or '(' |
| 1348 | |
| 1349 | ** Fixed parser memory leaks. |
| 1350 | When the generated parser was using malloc to extend its stacks, the |
| 1351 | previous allocations were not freed. |
| 1352 | |
| 1353 | ** Fixed verbose output file. |
| 1354 | Some newlines were missing. |
| 1355 | Some conflicts in state descriptions were missing. |
| 1356 | |
| 1357 | ** Fixed conflict report. |
| 1358 | Option -v was needed to get the result. |
| 1359 | |
| 1360 | ** %expect |
| 1361 | Was not used. |
| 1362 | Mismatches are errors, not warnings. |
| 1363 | |
| 1364 | ** Fixed incorrect processing of some invalid input. |
| 1365 | |
| 1366 | ** Fixed CPP guards: 9foo.h uses BISON_9FOO_H instead of 9FOO_H. |
| 1367 | |
| 1368 | ** Fixed some typos in the documentation. |
| 1369 | |
| 1370 | ** %token MY_EOF 0 is supported. |
| 1371 | Before, MY_EOF was silently renumbered as 257. |
| 1372 | |
| 1373 | ** doc/refcard.tex is updated. |
| 1374 | |
| 1375 | ** %output, %file-prefix, %name-prefix. |
| 1376 | New. |
| 1377 | |
| 1378 | ** --output |
| 1379 | New, aliasing `--output-file'. |
| 1380 | \f |
| 1381 | * Changes in version 1.30, 2001-10-26: |
| 1382 | |
| 1383 | ** `--defines' and `--graph' have now an optional argument which is the |
| 1384 | output file name. `-d' and `-g' do not change; they do not take any |
| 1385 | argument. |
| 1386 | |
| 1387 | ** `%source_extension' and `%header_extension' are removed, failed |
| 1388 | experiment. |
| 1389 | |
| 1390 | ** Portability fixes. |
| 1391 | \f |
| 1392 | * Changes in version 1.29, 2001-09-07: |
| 1393 | |
| 1394 | ** The output file does not define const, as this caused problems when used |
| 1395 | with common autoconfiguration schemes. If you still use ancient compilers |
| 1396 | that lack const, compile with the equivalent of the C compiler option |
| 1397 | `-Dconst='. autoconf's AC_C_CONST macro provides one way to do this. |
| 1398 | |
| 1399 | ** Added `-g' and `--graph'. |
| 1400 | |
| 1401 | ** The Bison manual is now distributed under the terms of the GNU FDL. |
| 1402 | |
| 1403 | ** The input and the output files has automatically a similar extension. |
| 1404 | |
| 1405 | ** Russian translation added. |
| 1406 | |
| 1407 | ** NLS support updated; should hopefully be less troublesome. |
| 1408 | |
| 1409 | ** Added the old Bison reference card. |
| 1410 | |
| 1411 | ** Added `--locations' and `%locations'. |
| 1412 | |
| 1413 | ** Added `-S' and `--skeleton'. |
| 1414 | |
| 1415 | ** `%raw', `-r', `--raw' is disabled. |
| 1416 | |
| 1417 | ** Special characters are escaped when output. This solves the problems |
| 1418 | of the #line lines with path names including backslashes. |
| 1419 | |
| 1420 | ** New directives. |
| 1421 | `%yacc', `%fixed_output_files', `%defines', `%no_parser', `%verbose', |
| 1422 | `%debug', `%source_extension' and `%header_extension'. |
| 1423 | |
| 1424 | ** @$ |
| 1425 | Automatic location tracking. |
| 1426 | \f |
| 1427 | * Changes in version 1.28, 1999-07-06: |
| 1428 | |
| 1429 | ** Should compile better now with K&R compilers. |
| 1430 | |
| 1431 | ** Added NLS. |
| 1432 | |
| 1433 | ** Fixed a problem with escaping the double quote character. |
| 1434 | |
| 1435 | ** There is now a FAQ. |
| 1436 | \f |
| 1437 | * Changes in version 1.27: |
| 1438 | |
| 1439 | ** The make rule which prevented bison.simple from being created on |
| 1440 | some systems has been fixed. |
| 1441 | \f |
| 1442 | * Changes in version 1.26: |
| 1443 | |
| 1444 | ** Bison now uses automake. |
| 1445 | |
| 1446 | ** New mailing lists: <bug-bison@gnu.org> and <help-bison@gnu.org>. |
| 1447 | |
| 1448 | ** Token numbers now start at 257 as previously documented, not 258. |
| 1449 | |
| 1450 | ** Bison honors the TMPDIR environment variable. |
| 1451 | |
| 1452 | ** A couple of buffer overruns have been fixed. |
| 1453 | |
| 1454 | ** Problems when closing files should now be reported. |
| 1455 | |
| 1456 | ** Generated parsers should now work even on operating systems which do |
| 1457 | not provide alloca(). |
| 1458 | \f |
| 1459 | * Changes in version 1.25, 1995-10-16: |
| 1460 | |
| 1461 | ** Errors in the input grammar are not fatal; Bison keeps reading |
| 1462 | the grammar file, and reports all the errors found in it. |
| 1463 | |
| 1464 | ** Tokens can now be specified as multiple-character strings: for |
| 1465 | example, you could use "<=" for a token which looks like <=, instead |
| 1466 | of chosing a name like LESSEQ. |
| 1467 | |
| 1468 | ** The %token_table declaration says to write a table of tokens (names |
| 1469 | and numbers) into the parser file. The yylex function can use this |
| 1470 | table to recognize multiple-character string tokens, or for other |
| 1471 | purposes. |
| 1472 | |
| 1473 | ** The %no_lines declaration says not to generate any #line preprocessor |
| 1474 | directives in the parser file. |
| 1475 | |
| 1476 | ** The %raw declaration says to use internal Bison token numbers, not |
| 1477 | Yacc-compatible token numbers, when token names are defined as macros. |
| 1478 | |
| 1479 | ** The --no-parser option produces the parser tables without including |
| 1480 | the parser engine; a project can now use its own parser engine. |
| 1481 | The actions go into a separate file called NAME.act, in the form of |
| 1482 | a switch statement body. |
| 1483 | \f |
| 1484 | * Changes in version 1.23: |
| 1485 | |
| 1486 | The user can define YYPARSE_PARAM as the name of an argument to be |
| 1487 | passed into yyparse. The argument should have type void *. It should |
| 1488 | actually point to an object. Grammar actions can access the variable |
| 1489 | by casting it to the proper pointer type. |
| 1490 | |
| 1491 | Line numbers in output file corrected. |
| 1492 | \f |
| 1493 | * Changes in version 1.22: |
| 1494 | |
| 1495 | --help option added. |
| 1496 | \f |
| 1497 | * Changes in version 1.20: |
| 1498 | |
| 1499 | Output file does not redefine const for C++. |
| 1500 | |
| 1501 | Local Variables: |
| 1502 | mode: outline |
| 1503 | End: |
| 1504 | |
| 1505 | ----- |
| 1506 | |
| 1507 | Copyright (C) 1995-2011 Free Software Foundation, Inc. |
| 1508 | |
| 1509 | This file is part of Bison, the GNU Parser Generator. |
| 1510 | |
| 1511 | This program is free software: you can redistribute it and/or modify |
| 1512 | it under the terms of the GNU General Public License as published by |
| 1513 | the Free Software Foundation, either version 3 of the License, or |
| 1514 | (at your option) any later version. |
| 1515 | |
| 1516 | This program is distributed in the hope that it will be useful, |
| 1517 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 1518 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 1519 | GNU General Public License for more details. |
| 1520 | |
| 1521 | You should have received a copy of the GNU General Public License |
| 1522 | along with this program. If not, see <http://www.gnu.org/licenses/>. |