3 This is an internal detail. But it's very handy, we should make it public.
4 It already has leaked in the documentation by accident.
7 There is no test with "Parse on stack %ld rejected by rule %d" in it.
10 Tests with typed %merge: 716 717 718 740 741 742 746 747 748
12 716: Duplicate representation of merged trees: glr.c FAILED (glr-regression.at:517)
13 740: Leaked semantic values if user action cuts parse: glr.c FAILED (glr-regression.at:1230)
14 746: Incorrect lookahead during nondeterministic GLR: glr.c FAILED (glr-regression.at:1610)
16 Document typed merges.
19 Clarify that rule numbers in the skeletons are 1-based.
22 There are many macros that should obey api.prefix: YY_CPLUSPLUS, YY_MOVE,
25 ** YYDEBUG etc. in C++
26 Discourage the use of YYDEBUG in C++ (see thread with Jot). Stop supporting
27 #define YYSTYPE by the user.
29 Add value_type as a synonym for semantic_type.
32 Why are yylval and yylloc treated differently?
34 yystack.yyglrShift (create_state_set_index(0), 0, 0, yylval, &yylloc);
37 And add tests in calc.at, to prepare work for D.
39 ** YYERROR and yynerrs
40 We are missing some cases. Write a test case, and check all the skeletons.
44 Don't do this (counterexample.c):
46 // This is the fastest way to get the tail node from the gl_list API.
48 list_get_end (gl_list_t list)
50 gl_list_node_t sentinel = gl_list_add_last (list, NULL);
51 gl_list_node_t res = gl_list_previous_node (list, sentinel);
52 gl_list_remove_node (list, sentinel);
56 *** Ambiguous rewriting
57 If the user is stupid enough to have equal rules, then the derivations are
60 Reduce/reduce conflict on tokens $end, "+", "⊕":
64 First derivation exp ::=[ exp "+" exp • ]
66 Second derivation exp ::=[ exp "+" exp • ]
68 Do we care about this? In color, we use twice the same color here, but we
69 could try to use the same color for the same rule.
72 Show the counterexamples. This is going to be really hard and/or painful.
73 Unless we play it dumb (little structure).
76 - How about not evaluating incomplete lines when the text is not finished
81 - Should i18n be part of the Lexer? Currently it's a static method of
84 - is there a migration path that would allow to use TokenKinds in
87 - define the tokens as an enum too.
89 - promote YYEOF rather than EOF.
92 https://git.savannah.gnu.org/gitweb/?p=gettext.git;a=blob;f=gettext-runtime/intl/plural.y;h=a712255af4f2f739c93336d4ff6556d932a426a5;hb=HEAD
94 should be updated to not use YYERRCODE. Returning an undef token is good
99 Stop hard-coding "Calc". Adjust local.at (look for FIXME).
101 ** A dev warning for b4_
102 Maybe we should check for m4_ and b4_ leaking out of the m4 processing, as
103 Autoconf does. It would have caught over-quotation issues.
106 I feel it's ugly to use the GNU style to declare functions in the doc. It
107 generates tons of white space in the page, and may contribute to bad page
114 The YYUNDEFTOK could be assigned a semantic value so that yyerror could be
115 used to report invalid lexemes.
118 Consider deprecating impure push parsers. They add a lot of complexity, for
119 a bad feature. On the other hand, that would make it much harder to sit
120 push parsers on top of pull parser. Which is currently not relevant, since
121 push parsers are measurably slower.
123 ** %define parse.error formatted
124 How about pushing Bistromathic's yyreport_syntax_error as another standard
125 way to generate the error message, and leave to the user the task of
126 providing the message formats? Currently in bistro, it reads:
129 error_format_string (int argc)
133 default: /* Avoid compiler warnings. */
134 case 0: return _("%@: syntax error");
135 case 1: return _("%@: syntax error: unexpected %u");
136 // TRANSLATORS: '%@' is a location in a file, '%u' is an
137 // "unexpected token", and '%0e', '%1e'... are expected tokens
140 // For instance on the expression "1 + * 2", you'd get
142 // 1.5: syntax error: expected - or ( or number or function or variable before *
143 case 2: return _("%@: syntax error: expected %0e before %u");
144 case 3: return _("%@: syntax error: expected %0e or %1e before %u");
145 case 4: return _("%@: syntax error: expected %0e or %1e or %2e before %u");
146 case 5: return _("%@: syntax error: expected %0e or %1e or %2e or %3e before %u");
147 case 6: return _("%@: syntax error: expected %0e or %1e or %2e or %3e or %4e before %u");
148 case 7: return _("%@: syntax error: expected %0e or %1e or %2e or %3e or %4e or %5e before %u");
149 case 8: return _("%@: syntax error: expected %0e or %1e or %2e or %3e or %4e or %5e or %6e before %u");
153 The message would have to be generated in a string, and pushed to yyerror.
154 Which will be a pain in the neck in yacc.c.
156 If we want to do that, we should think very carefully about the syntax of
159 ** yyclearin does not invoke the lookahead token's %destructor
160 https://lists.gnu.org/r/bug-bison/2018-02/msg00000.html
163 > Modifying yyclearin so that it calls yydestruct seems like the simplest
164 > solution to this issue, but it is conceivable that such a change would
165 > break programs which already perform some kind of workaround in order to
166 > destruct the lookahead symbol. So it might be necessary to use some kind of
167 > compatibility %define, or to create a new replacement macro with a
168 > different name such as yydiscardin.
170 > At a minimum, the fact that yyclearin does not invoke the %destructor
171 > should be highlighted in the documentation, since it is not at all obvious.
175 Les catégories d'avertissements incluent :
176 conflicts-sr conflits S/R (activé par défaut)
177 conflicts-rr conflits R/R (activé par défaut)
178 dangling-alias l'alias chaîne n'est pas attaché à un symbole
179 deprecated construction obsolète
180 empty-rule règle vide sans %empty
181 midrule-values valeurs de règle intermédiaire non définies ou inutilisées
182 precedence priorité et associativité inutiles
183 yacc incompatibilités avec POSIX Yacc
184 other tous les autres avertissements (activé par défaut)
185 all tous les avertissements sauf « dangling-alias » et « yacc »
186 no-CATEGORY désactiver les avertissements dans CATEGORIE
187 none désactiver tous les avertissements
188 error[=CATEGORY] traiter les avertissements comme des erreurs
190 Line -1 and -3 should mention CATEGORIE, not CATEGORY.
193 ** Rewrite glr.cc (currently glr2.cc)
195 We can probably replace setjmp/longjmp with exceptions. That would help
196 tremendously other languages such as D and Java that probably have no
197 similar feature. If we remove jumps, we probably no longer need _Noreturn,
198 so simplify `b4_attribute_define([noreturn])` into `b4_attribute_define`.
200 After discussing with Valentin, it was decided that it's better to stay with
201 jumps, since in some places exceptions are ruled out from C++.
204 Move to our coding conventions. In particular names such as yy_glr_stack,
208 It should be a member of the parser object, see lalr1.cc. Let the parser
209 object decide what the debug stream is, rather than open coding std::cerr.
212 There are many places where pointers should be replaced with references.
213 Some occurrences were fixed, but now some have improper names:
215 -yygetToken (int *yycharp, ]b4_namespace_ref[::]b4_parser_class[& yyparser][]b4_pure_if([, glr_stack* yystackp])[]b4_user_formals[)
216 +yygetToken (int& yycharp, ]b4_namespace_ref[::]b4_parser_class[& yyparser][]b4_pure_if([, glr_stack* yystackp])[]b4_user_formals[)
218 yycharp is no longer a Pointer. And yystackp should probably also be a reference.
221 Currently all the assertions are enabled. Once we are confident in glr2.cc,
222 let parse.assert use the same approach as in lalr1.cc.
225 Stop using std::cerr everywhere.
228 When glr2.cc fully replaces glr.cc, get rid of the glr.cc scaffolding in
232 ** Unit rules / Injection rules (Akim Demaille)
233 Maybe we could expand unit rules (or "injections", see
234 https://homepages.cwi.nl/~daybuild/daily-books/syntax/2-sdf/sdf.html), i.e.,
243 exp: exp '+' exp | exp '&' exp;
245 when there are no actions. This can significantly speed up some grammars.
246 I can't find the papers. In particular the book 'LR parsing: Theory and
247 Practice' is impossible to find, but according to 'Parsing Techniques: a
248 Practical Guide', it includes information about this issue. Does anybody
251 ** clean up (Akim Demaille)
252 Do not work on these items now, as I (Akim) have branches with a lot of
253 changes in this area (hitting several files), and no desire to have to fix
254 conflicts. Addressing these items will happen after my branches have been
258 Introduce a goto struct, and use it in place of from_state/to_state.
259 Rename states1 as path, length as pathlen.
260 Introduce inline functions for things such as nullable[*rp - ntokens]
261 where we need to map from symbol number to nterm number.
263 There are probably a significant part of the relations management that
264 should be migrated on top of a bitsetv.
267 It should probably take a "state*" instead of two arguments.
270 The "automaton" and "set" categories are not so useful. We should probably
271 introduce lr(0) and lalr, just the way we have ielr categories. The
272 "closure" function is too verbose, it should probably have its own category.
274 "set" can still be used for summarizing the important sets. That would make
275 tests easy to maintain.
278 Rename these guys as "diagnostics.*" (or "diagnose.*"), since that's the
279 name they have in GCC, clang, etc. Likewise for the complain_* series of
283 states/nstates, rules/nrules, ..., ritem/nritems
286 *** m4: slot, type, type_tag
287 The meaning of type_tag varies depending on api.value.type. We should avoid
288 that and using clear definitions with stable semantics.
290 * D programming language
291 There's a number of features that are missing, here sorted in _suggested_
292 order of implementation.
294 When copying code from other skeletons, keep the comments exactly as they
295 are. Keep the same variable names. If you change the wording in one place,
296 do it in the others too. In other words: make sure to keep the
297 maintenance *simple* by avoiding any gratuitous difference.
300 Check when gdc and ldc.
302 ** Token Constructors
303 It is possible to mix incorrectly kinds and values, and for instance:
305 return parser.Symbol (TokenKind.NUM, "Hello, World!\n");
307 attaches a string value to NUM kind (wrong, of course). When
308 api.token.constructor is set, in C++, Bison generated "token constructors":
309 parser.make_NUM. parser.make_PLUS, parser.make_STRING, etc. The previous
312 return parser.make_NUM ("Hello, World!\n");
314 which would easily be caught by the type checker.
317 Add support for push parser. Do not start a nice skeleton, just enhance the
318 current one to support push parsers. This is going to be a tougher nut to
321 First, you need to understand well how the push parser is expected to work.
324 - look at examples/c/pushcalc
325 - create an example of a Java push parser.
326 - have a look at the generated parser in Java, which has the advantage of
327 being already based on a parser object, instead of just a function.
329 The C case is harder to read, but it may help too. Keep in mind that
330 because there's no object to maintain state, the C push parser uses some
331 struct (yypstate) to preserve this state. We don't need this in D, the
332 parser object will suffice.
334 I think working directly on the skeleton to add push-parser support is not
335 the simplest path. I suggest that you (1) transform a generated parser into
336 a push parser by hand, and then (2) transform lalr1.d to generate such a
339 Use `git commit` frequently to make sure you keep track of your progress.
341 *** (1.a) Prepare pull parser by hand
342 Copy again one of the D examples into say examples/d/pushcalc. Also
343 check-in the generated parser to facilitate experimentation.
345 - find local variables of yyparse should become members of the parser object
346 (so that we preserve state from one call to the next).
348 - do it in your generated D parser. We don't need an equivalent for
349 yypstate, because we already have it: that the parser object itself.
351 - have your *pull*-parser (i.e., the good old yy::parser::parse()) work
352 properly this way. Write and run tests. That's one of the reasons I
353 suggest using examples/d/calc as a starting point: it already has tests,
354 you can/should add more.
356 At this point you have a pull-parser which you prepared to turn into a
359 *** (1.b) Turn pull parser into push parser by hand
361 - look again at how push parsers are implemented in Java/C to see what needs
362 to change in yyparse so that the control is inverted: parse() will
363 be *given* the tokens, instead of having to call yylex itself. When I say
364 "look at C", I think your best option are (i) yacc.c (look for b4_push_if)
365 and (ii) examples/c/pushcalc.
367 - rename parse() as push_parse(Symbol yyla) (or push_parse(TokenKind, Value,
368 Location)) that takes the symbol as argument. That's the push parser we
371 - define a new parse() function which has the same signature as the usual
372 pull-parser, that repeatedly calls the push_parse function. Something
379 status = this->push_parse (yylex());
380 } while (status == YYPUSH_MORE);
384 - show me that parser, so that we can validate the approach.
386 *** (2) Port that into the skeleton
387 - once we agree on the API of the push parser, implement it into lalr1.d.
388 You will probaby need help on this regard, but imitation, again, should
391 - have example/d/pushcalc work properly and pass tests
393 - add tests in the "real" test suite. Do that in tests/calc.at. I can
399 This is very ambitious. That's the final boss. There are currently no
400 "clean" implementation to get inspiration from.
402 glr.c is very clean but:
404 - is a different skeleton from yacc.c
406 glr.cc is (currently) an ugly hack: a C++ shell around glr.c. Valentin
407 Tolmer is currently rewriting glr.cc to be clean C++, but he is not
408 finished. There will be a lot a common code between lalr1.cc and glr.cc, so
409 eventually I would like them to be fused into a single skeleton, supporting
410 both deterministic and generalized parsing.
412 It would be great for D to also support this.
414 The basic ideas of GLR are explained here:
416 https://www.codeproject.com/Articles/5259825/GLR-Parsing-in-Csharp-How-to-Use-The-Most-Powerful
418 * Better error messages
419 The users are not provided with enough tools to forge their error messages.
420 See for instance "Is there an option to change the message produced by
421 YYERROR_VERBOSE?" by Simon Sobisch, on bison-help.
424 https://www.cs.tufts.edu/~nr/cs257/archive/clinton-jefferey/lr-error-messages.pdf
425 https://research.swtch.com/yyerror
426 http://gallium.inria.fr/~fpottier/publis/fpottier-reachability-cc2016.pdf
429 Fix data/skeletons/yacc.c so that it defines YYPTRDIFF_T properly for modern
430 and older C++ compilers. Currently the code defaults to defining it to
431 'long' for non-GCC compilers, but it should use the proper C++ magic to
432 define it to the same type as the C ptrdiff_t type.
435 Several features are not available in all the back-ends.
437 - push parsers: glr.c, glr.cc, lalr1.cc (not very difficult)
438 - token constructors: Java, C, D (a bit difficult)
439 - glr: D, Java (super difficult)
442 ** Autotest has quotation issues
443 tests/input.at:1730:AT_SETUP([%define errors])
447 $ ./tests/testsuite -l | grep errors | sed q
448 38: input.at:1730 errors
451 ** Get rid of YYPRINT and b4_toknum
452 Besides yytoknum is wrong when api.token.raw is defined.
454 ** Better design for diagnostics
455 The current implementation of diagnostics is ad hoc, it grew organically.
456 It works as a series of calls to several functions, with dependency of the
457 latter calls on the former. For instance:
459 complain (&sym->location,
460 sym->content->status == needed ? complaint : Wother,
461 _("symbol %s is used, but is not defined as a token"
462 " and has no rules; did you mean %s?"),
463 quote_n (0, sym->tag),
464 quote_n (1, best->tag));
465 if (feature_flag & feature_caret)
466 location_caret_suggestion (sym->location, best->tag, stderr);
468 We should rewrite this in a more FP way:
470 1. build a rich structure that denotes the (complete) diagnostic.
471 "Complete" in the sense that it also contains the suggestions, the list
472 of possible matches, etc.
474 2. send this to the pretty-printing routine. The diagnostic structure
475 should be sufficient so that we can generate all the 'format' of
476 diagnostics, including the fixits.
478 If properly done, this diagnostic module can be detached from Bison and be
479 put in gnulib. It could be used, for instance, for errors caught by
482 There's certainly already something alike in GCC. At least that's the
483 impression I get from reading the "-fdiagnostics-format=FORMAT" part of this
486 https://gcc.gnu.org/onlinedocs/gcc/Diagnostic-Message-Formatting-Options.html
488 ** Graphviz display code thoughts
489 The code for the --graph option is over two files: print_graph, and
490 graphviz. This is because Bison used to also produce VCG graphs, but since
491 this is no longer true, maybe we could consider these files for fusion.
493 An other consideration worth noting is that print_graph.c (correct me if I
494 am wrong) should contain generic functions, whereas graphviz.c and other
495 potential files should contain just the specific code for that output
496 format. It will probably prove difficult to tell if the implementation is
497 actually generic whilst only having support for a single format, but it
498 would be nice to keep stuff a bit tidier: right now, the construction of the
499 bitset used to show reductions is in the graphviz-specific code, and on the
500 opposite side we have some use of \l, which is graphviz-specific, in what
501 should be generic code.
503 Little effort seems to have been given to factoring these files and their
504 print{,-xml} counterpart. We would very much like to re-use the pretty format
505 of states from .output for the graphs, etc.
507 Since graphviz dies on medium-to-big grammars, maybe consider an other tool?
510 Check it too when checking the different kinds of parsers. And be
511 sure to check that the initial-action is performed once per parsing.
514 b4_shared_declarations is no longer what it is. Make it
515 b4_parser_declaration for instance.
517 ** yychar in lalr1.cc
518 There is a large difference bw maint and master on the handling of
519 yychar (which was removed in lalr1.cc). See what needs to be
523 /* User semantic actions sometimes alter yychar, and that requires
524 that yytoken be updated with the new translation. We take the
525 approach of translating immediately before every use of yytoken.
526 One alternative is translating here after every semantic action,
527 but that translation would be missed if the semantic action
528 invokes YYABORT, YYACCEPT, or YYERROR immediately after altering
529 yychar. In the case of YYABORT or YYACCEPT, an incorrect
530 destructor might then be invoked immediately. In the case of
531 YYERROR, subsequent parser actions might lead to an incorrect
532 destructor call or verbose syntax error message before the
533 lookahead is translated. */
535 /* Make sure we have latest lookahead translation. See comments at
536 user semantic actions for why this is necessary. */
537 yytoken = yytranslate_ (yychar);
540 ** Get rid of fake #lines [Bison: ...]
541 Possibly as simple as checking whether the column number is nonnegative.
543 I have seen messages like the following from GCC.
545 <built-in>:0: fatal error: opening dependency file .deps/libltdl/argz.Tpo: No such file or directory
548 ** Discuss about %printer/%destroy in the case of C++.
549 It would be very nice to provide the symbol classes with an operator<<
550 and a destructor. Unfortunately the syntax we have chosen for
551 %destroy and %printer make them hard to reuse. For instance, the user
552 is invited to write something like
554 %printer { debug_stream() << $$; } <my_type>;
556 which is hard to reuse elsewhere since it wants to use
557 "debug_stream()" to find the stream to use. The same applies to
558 %destroy: we told the user she could use the members of the Parser
559 class in the printers/destructors, which is not good for an operator<<
560 since it is no longer bound to a particular parser, it's just a
564 ** Rewrite glr.cc in C++ (Valentin Tolmer)
565 As a matter of fact, it would be very interesting to see how much we can
566 share between lalr1.cc and glr.cc. Most of the skeletons should be common.
567 It would be a very nice source of inspiration for the other languages.
569 Valentin Tolmer is working on this.
571 * From lalr1.cc to yacc.c
573 Merging the three stacks in lalr1.cc simplified the code, prompted for
574 other improvements and also made it faster (probably because memory
575 management is performed once instead of three times). I suggest that
576 we do the same in yacc.c.
578 (Some time later): it's also very nice to have three stacks: it's more dense
579 as we don't lose bits to padding. For instance the typical stack for states
580 will use 8 bits, while it is likely to consume 32 bits in a struct.
582 We need trustworthy benchmarks for Bison, for all our backends. Akim has a
583 few things scattered around; we need to put them in the repo, and make them
589 Some statistics about the grammar and the parser would be useful,
590 especially when asking the user to send some information about the
591 grammars she is working on. We should probably also include some
592 information about the variables (I'm not sure for instance we even
593 specify what LR variant was used).
596 How would Paul like to display the conflicted actions? In particular,
597 what when two reductions are possible on a given lookahead token, but one is
598 part of $default. Should we make the two reductions explicit, or just
599 keep $default? See the following point.
601 ** Disabled Reductions
602 See 'tests/conflicts.at (Defaulted Conflicted Reduction)', and decide
606 Extend with error productions. The hard part will probably be finding
607 the right rule so that a single state does not exhibit too many yet
608 undocumented ''features''. Maybe an empty action ought to be
609 presented too. Shall we try to make a single grammar with all these
610 features, or should we have several very small grammars?
614 Well, only if there is really some demand for it.
617 https://github.com/scfc/bison-php/blob/master/data/lalr1.php
620 https://lists.gnu.org/r/bison-patches/2013-09/msg00000.html and following
622 ** Multiple start symbols
623 Would be very useful when parsing closely related languages. The idea is to
624 declare several start symbols, for instance
631 and to generate parse(), parse_stmt() and parse_expr(). Technically, the
632 above grammar would be transformed into
635 %token YY_START_STMT YY_START_EXPR
637 yy_start: YY_START_STMT stmt | YY_START_EXPR expr
639 so that there are no new conflicts in the grammar (as would undoubtedly
640 happen with yy_start: stmt | expr). Then adjust the skeletons so that this
641 initial token (YY_START_STMT, YY_START_EXPR) be shifted first in the
642 corresponding parse function.
644 *** Number of useless symbols
648 [[input.y: warning: 2 nonterminals useless in grammar [-Wother]
649 input.y: warning: 2 rules useless in grammar [-Wother]
650 input.y:2.8-10: error: start symbol exp does not derive any sentence]])
652 We should say "1 nonterminal": the other one is $accept, which should not
653 participate in the count.
656 Do we want to disallow terminal start symbols? The limitation is not
657 technical. Can it be useful to someone to "parse" a token?
660 This is a popular demand. We already made many changes in the parser that
661 should make this reasonably easy to implement.
663 Bruce Mardle <marblypup@yahoo.co.uk>
664 https://lists.gnu.org/archive/html/bison-patches/2015-09/msg00000.html
666 However, there are many other things to do before having such a feature,
667 because I don't want a % equivalent to #include (which we all learned to
668 hate). I want something that builds "modules" of grammars, and assembles
669 them together, paying attention to keep separate bits separated, in pseudo
673 There is demand for push parsers in C++.
675 ** Generate code instead of tables
676 This is certainly quite a lot of work. See
677 http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.50.4539.
680 We should find a means to provide an access to values deep in the
681 stack. For instance, instead of
683 baz: qux { $$ = $<foo>-1 + $<bar>0 + $1; }
685 we should be able to have:
687 foo($foo) bar($bar) baz($bar): qux($qux) { $baz = $foo + $bar + $qux; }
689 Or something like this.
692 It should be possible to have %if/%else/%endif. The implementation is
693 not clear: should it be lexical or syntactic. Vadim Maslow thinks it
694 must be in the scanner: we must not parse what is in a switched off
695 part of %if. Akim Demaille thinks it should be in the parser, so as
696 to avoid falling into another CPP mistake.
698 (Later): I'm sure there's actually good case for this. People who need that
699 feature can use m4/cpp on top of Bison. I don't think it is worth the
700 trouble in Bison itself.
703 There are couple of available extensions of Bison targeting some XML
704 output. Some day we should consider including them. One issue is
705 that they seem to be quite orthogonal to the parsing technique, and
706 seem to depend mostly on the possibility to have some code triggered
707 for each reduction. As a matter of fact, such hooks could also be
708 used to generate the yydebug traces. Some generic scheme probably
711 XML output for GNU Bison and gcc
712 http://www.cs.may.ie/~jpower/Research/bisonXML/
714 XML output for GNU Bison
715 http://yaxx.sourceforge.net/
717 * Coding system independence
720 Currently Bison assumes 8-bit bytes (i.e. that UCHAR_MAX is
721 255). It also assumes that the 8-bit character encoding is
722 the same for the invocation of 'bison' as it is for the
723 invocation of 'cc', but this is not necessarily true when
724 people run bison on an ASCII host and then use cc on an EBCDIC
725 host. I don't think these topics are worth our time
726 addressing (unless we find a gung-ho volunteer for EBCDIC or
727 PDP-10 ports :-) but they should probably be documented
730 More importantly, Bison does not currently allow NUL bytes in
731 tokens, either via escapes (e.g., "x\0y") or via a NUL byte in
732 the source code. This should get fixed.
737 Must we keep %token-table?
742 It is unfortunate that there is a total order for precedence. It
743 makes it impossible to have modular precedence information. We should
744 move to partial orders (sounds like series/parallel orders to me).
746 This is a prerequisite for modules.
748 * Pre and post actions.
749 From: Florian Krohm <florian@edamail.fishkill.ibm.com>
750 Subject: YYACT_EPILOGUE
751 To: bug-bison@gnu.org
752 X-Sent: 1 week, 4 days, 14 hours, 38 minutes, 11 seconds ago
754 The other day I had the need for explicitly building the parse tree. I
755 used %locations for that and defined YYLLOC_DEFAULT to call a function
756 that returns the tree node for the production. Easy. But I also needed
757 to assign the S-attribute to the tree node. That cannot be done in
758 YYLLOC_DEFAULT, because it is invoked before the action is executed.
759 The way I solved this was to define a macro YYACT_EPILOGUE that would
760 be invoked after the action. For reasons of symmetry I also added
761 YYACT_PROLOGUE. Although I had no use for that I can envision how it
762 might come in handy for debugging purposes.
763 All is needed is to add
766 YYACT_EPILOGUE (yyval, (yyvsp - yylen), yylen, yyloc, (yylsp - yylen));
768 YYACT_EPILOGUE (yyval, (yyvsp - yylen), yylen);
771 at the proper place to bison.simple. Ditto for YYACT_PROLOGUE.
773 I was wondering what you think about adding YYACT_PROLOGUE/EPILOGUE
774 to bison. If you're interested, I'll work on a patch.
777 Equip the parser with a means to create the (visual) parse tree.
782 # LocalWords: Cex gnulib gl Bistromathic TokenKinds yylex enum YYEOF EOF
783 # LocalWords: YYerror gettext af hb YYERRCODE undef calc FIXME dev yyerror
784 # LocalWords: Autoconf YYUNDEFTOK lexemes parsers Bistromathic's yyreport
785 # LocalWords: const argc yacc yyclearin lookahead destructor Rici incluent
786 # LocalWords: yydestruct yydiscardin catégories d'avertissements sr activé
787 # LocalWords: conflits défaut rr l'alias chaîne n'est attaché un symbole
788 # LocalWords: obsolète règle vide midrule valeurs de intermédiaire ou avec
789 # LocalWords: définies inutilisées priorité associativité inutiles POSIX
790 # LocalWords: incompatibilités tous les autres avertissements sauf dans rp
791 # LocalWords: désactiver CATEGORIE traiter comme des erreurs glr Akim bool
792 # LocalWords: Demaille arith lalr goto struct pathlen nullable ntokens lr
793 # LocalWords: nterm bitsetv ielr ritem nstates nrules nritems yysymbol EQ
794 # LocalWords: SymbolKind YYEMPTY YYUNDEF YYTNAME NUM yyntokens yytname sed
795 # LocalWords: nonterminals yykind yycode YYNAMES yynames init getName conv
796 # LocalWords: TokenKind ival yychar yylval yylexer Tolmer hoc
797 # LocalWords: Sobisch YYPTRDIFF ptrdiff Autotest YYPRINT toknum yytoknum
798 # LocalWords: sym Wother stderr FP fixits xgettext fdiagnostics Graphviz
799 # LocalWords: graphviz VCG bitset xml bw maint yytoken YYABORT deps
800 # LocalWords: YYACCEPT yytranslate nonnegative destructors yyerrlab repo
801 # LocalWords: backends stmt expr yy Mardle baz qux Vadim Maslow CPP cpp
802 # LocalWords: yydebug gcc UCHAR EBCDIC gung PDP NUL Pre Florian Krohm utf
803 # LocalWords: YYACT YYLLOC YYLSP yyval yyvsp yylen yyloc yylsp endif
804 # LocalWords: ispell american
810 ispell-dictionary: "american"
813 Copyright (C) 2001-2004, 2006, 2008-2015, 2018-2020 Free Software
816 This file is part of Bison, the GNU Compiler Compiler.
818 Permission is granted to copy, distribute and/or modify this document
819 under the terms of the GNU Free Documentation License, Version 1.3 or
820 any later version published by the Free Software Foundation; with no
821 Invariant Sections, with no Front-Cover Texts, and with no Back-Cover
822 Texts. A copy of the license is included in the "GNU Free
823 Documentation License" file as part of this distribution.