]> git.saurik.com Git - bison.git/blobdiff - ChangeLog
Let --trace have arguments.
[bison.git] / ChangeLog
index 980b219377830dd6eb3cb6bf2adbbe88762ba45e..0586bf00923c2d05b44042df322dd249af0f6623 100644 (file)
--- a/ChangeLog
+++ b/ChangeLog
+2002-07-31  Akim Demaille  <akim@epita.fr>
+
+       Let --trace have arguments.
+
+       * src/getargs.h (enum trace_e): New.
+       * src/getargs.c (trace_args, trace_types, trace_argmatch): New.
+       (long_options, short_options): --trace/-T takes an optional
+       argument.
+       Change all the uses of trace_flag to reflect the new flags.
+       * tests/sets.at (Firsts, Nullable, Broken Closure): Use --trace=sets.
+
+       Strengthen `stage' portability.
+
+       * m4/stage.m4 (BISON_PREREQ_STAGE): New.
+       * configure.in: Use it.
+       Don't check for malloc.h and sys/times.h.
+       * src/system.h: Include them when appropriate.
+       * src/main.c (stage): Compile only when mallinfo, struct mallinfo,
+       times and struct tms are available.
+
+2002-07-30  Akim Demaille  <akim@epita.fr>
+
+       In verbose parse error message, don't report `error' as an
+       expected token.
+       * tests/actions.at (Printers and Destructors): Adjust.
+       * tests/calc.at (Calculator $1): Adjust.
+       * data/yacc.c, data/glr.c, data/lalr1.c: When making the verbose
+       error message, do not report the parser accepts the error token in
+       that state.
+
+2002-07-30  Akim Demaille  <akim@epita.fr>
+
+       Normalize conflict related messages.
+
+       * src/complain.h, src/complain.c (warn, complain): New.
+       * src/conflicts.c (conflicts_print): Use them.
+       (conflict_report_yacc): New, extracted from...
+       (conflicts_print): here.
+       * tests/conflicts.at, tests/existing.at: Adjust.
+
+2002-07-30  Akim Demaille  <akim@epita.fr>
+
+       Report rules which are never reduced by the parser: those hidden
+       by conflicts.
+
+       * src/LR0.c (save_reductions): Don't make the final state too
+       different: save its reduction (accept) instead of having a state
+       without any action (no shift or goto, no reduce).
+       Note: the final state is now a ``regular'' state, i.e., the
+       parsers now contain `reduce 0' as default reduction.
+       Nevertheless, since they decide to `accept' when yystate =
+       final_state, they still will not reduce rule 0.
+       * src/print.c (print_actions, print_reduction): Adjust.
+       * src/output.c (action_row): Track reduced rules.
+       (token_actions): Report rules never reduced.
+       * tests/conflicts.at, tests/regression.at: Adjust.
+
+2002-07-30  Akim Demaille  <akim@epita.fr>
+
+       `stage' was accidently included in a previous patch.
+       Initiate its autoconfiscation.
+
+       * configure.in: Look for malloc.h and sys/times.h.
+       * src/main.c (stage): Adjust.
+       Report only when trace_flag.
+
+2002-07-29  Akim Demaille  <akim@epita.fr>
+
+       * src/state.h, src/state.c (transitions_t): Holds state_t*'s, not
+       state_number_t.
+       (errs_t): symbol_t*, not symbol_number_t.
+       (reductions_t): rule_t*, not rule_number_t.
+       (FOR_EACH_SHIFT): New.
+       * src/LR0.c, src/conflicts.c, src/lalr.c, src/output.c
+       * src/print.c, src/print_graph.c: Adjust.
+
+2002-07-29  Akim Demaille  <akim@epita.fr>
+
+       Use $accept and $end, as BYacc and BTYacc do, instead of $axiom and $.
+
+       * src/symtab.h, src/symtab.c (eoftoken, axiom): Rename as...
+       (endtoken, accept): these.
+       * src/reader.c (reader): Set endtoken's default tag to "$end".
+       Set undeftoken's tag to "$undefined" instead of "$undefined.".
+       * doc/bison.texinfo (Table of Symbols): Mention $accept and $end.
+       Adjust.
+
+2002-07-29  Akim Demaille  <akim@epita.fr>
+
+       * src/reduce.c (reduce_grammar): When the language is empty,
+       complain about the start symbol, not the axiom.
+       Use its location.
+       * tests/reduce.at (Empty Language): New.
+
+2002-07-26  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.h, src/reader.c (gram_error): ... can't get
+       yycontrol without making too strong assumptions on the parser
+       itself.
+       * src/output.c (prepare_tokens): Use the real 0th value of
+       token_translations instead of `0'.
+       * src/parse-gram.y (yyerror): Don't rely on yycontrol being
+       visible here.
+       * data/yacc.c (yyreport_parse_error): Rename yylocation as yylloc
+       for the time being: %locations ought to provide it to yyerror.
+
+2002-07-25  Akim Demaille  <akim@epita.fr>
+
+       * src/output.c (prepare_tokens): Go up to ntokens, not ntokens + 1.
+       * doc/bison.texinfo (Decl Summary): s/$illegal/$undefined./.
+       * tests/regression.at (Web2c Actions): Adjust.
+
+2002-07-25  Akim Demaille  <akim@epita.fr>
+
+       Stop storing rules from 1 to nrules + 1.
+
+       * src/LR0.c, src/closure.c, src/derives.c, src/gram.c, src/lalr.c
+       * src/nullable.c, src/output.c, src/print.c, src/reader.c
+       * src/reduce.c: Allocate and free from &rules[0], not &rules[1].
+       Iterate from 0 to nrules.
+       Use rule_number_as_item_number and item_number_as_rule_number.
+       Adjust to `derive' now containing possibly 0.
+       * src/gram.h (rule_number_as_item_number, item_number_as_rule_number):
+       Handle the `- 1' part in rule numbers from/to item numbers.
+       * src/conflicts.c (log_resolution): Fix the message which reversed
+       shift and reduce.
+       * src/output.c (action_row): Initialize default_rule to -1.
+       (token_actions): Adjust.
+       * tests/sets.at (Nullable, Firsts): Fix the previously bogus
+       expected output.
+       * tests/conflicts.at (Resolved SR Conflicts): Likewise.
+
+2002-07-25  Akim Demaille  <akim@epita.fr>
+
+       * data/c.m4 (b4_c_function, b4_c_ansi_args, b4_c_ansi_arg)
+       (b4_c_knr_arg_names, b4_c_knr_arg_name, b4_c_knr_arg_decls)
+       (b4_c_knr_arg_decl): New.
+       * data/yacc.c: Use it to define yysymprint, yydestruct, and
+       yyreport_parse_error.
+
+2002-07-25  Akim Demaille  <akim@epita.fr>
+
+       * data/yacc.c (yyreport_parse_error): New, extracted from...
+       (yyparse): here.
+       (yydestruct, yysymprint): Move above yyparse.
+       Be K&R compliant.
+
+2002-07-25  Akim Demaille  <akim@epita.fr>
+
+       * data/c.m4 (b4_ints_in, b4_int_type, b4_int_type_for): New,
+       replace...
+       (b4_sint_type, b4_uint_type): these.
+       * data/yacc.c, data/glr.c, data/lalr1.cc: Use b4_int_type_for.
+       * tests/regression.at (Web2c Actions): Adjust.
+
+2002-07-25  Akim Demaille  <akim@epita.fr>
+
+       * src/gram.h (TIEM_NUMBER_MAX): New.
+       (item_number_of_rule_number, rule_number_of_item_number): Rename
+       as...
+       (rule_number_as_item_number, item_number_as_rule_number): these.
+       Adjust dependencies.
+       * src/output.c (vector_number_t, VECTOR_NUMBER_MAX)
+       (VECTOR_NUMBER_MIN, state_number_to_vector_number)
+       (symbol_number_to_vector_number): New.
+       (order): Of vector_number_t* type.
+       (base_t, BASE_MAX, BASE_MIN): New.
+       (froms, tos, width, pos, check): Of base_t type.
+       (action_number_t, ACTION_MIN, ACTION_MAX): New.
+       (actrow): Of action_number_t type.
+       (conflrow): Of unsigned int type.
+       (table_ninf, base_ninf): New.
+       (GENERATE_MUSCLE_INSERT_TABLE): Also output the `*_min' value.
+       (muscle_insert_int_table, muscle_insert_base_table)
+       (muscle_insert_rule_number_table): New.
+       (prepare_tokens): Output `toknum' as int_table.
+       (action_row): Returns a rule_number_t.
+       Use ACTION_MIN, not SHRT_MIN.
+       (token_actions): yydefact is rule_number_t*.
+       (table_ninf_remap): New.
+       (pack_table): Use it for `base' and `table'.
+       * data/yacc.c, data/glr.c, data/lalr1.cc (YYFLAG): Remove,
+       replaced with...
+       (YYPACT_NINF, YYTABLE_NINF): these.
+       (yypact, yytable): Compute their types instead of hard-coded
+       `short'.
+       * tests/regression.at (Web2c Actions): Adjust.
+
+2002-07-19  Akim Demaille  <akim@epita.fr>
+
+       * src/scan-gram.l (id): Can start with an underscore.
+
+2002-07-16  Akim Demaille  <akim@epita.fr>
+
+       * src/assoc.c, src/asssoc.h (assoc_t, assoc_to_string): New.
+       Adjust all former `associativity' dependencies.
+       * src/symtab.c (symbol_new): Default associativity is `undef', not
+       `right'.
+       (symbol_check_alias_consistence): Adjust.
+
+2002-07-09  Akim Demaille  <akim@epita.fr>
+
+       * doc/bison.texinfo: Properly set the ``header'' part.
+       Use @dircategory ``GNU programming tools'' as per Texinfo's
+       documentation.
+       Use @copying.
+
+2002-07-09  Akim Demaille  <akim@epita.fr>
+
+       * lib/quotearg.h: Protect against multiple inclusions.
+       * src/location.h (location_t): Add a `file' member.
+       (LOCATION_RESET, LOCATION_PRINT): Adjust.
+       * src/complain.c (warn_at, complain_at, fatal_at): Drop
+       `error_one_per_line' support.
+
+2002-07-09  Akim Demaille  <akim@epita.fr>
+
+       * src/complain.h, src/complain.c (warn, complain): Remove, unused.
+       * src/reader.c (lineno): Remove.
+       Adjust all dependencies.
+       (get_merge_function): Take a location and use complain_at.
+       * src/symtab.h, src/symtab.c (symbol_make_alias): Likewise.
+       * tests/regression.at (Invalid inputs, Mixing %token styles):
+       Adjust.
+
+2002-07-09  Akim Demaille  <akim@epita.fr>
+
+       * src/parse-gram.y (rules_or_grammar_declaration): Add an error
+       recovery rule, and forbid extensions when --yacc.
+       (gram_error): Use complain_at.
+       * src/reader.c (reader): Exit if there were parse errors.
+
+2002-07-09  Akim Demaille  <akim@epita.fr>
+
+       * tests/synclines.at (AT_SYNCLINES_COMPILE): New.
+       (AT_TEST_SYNCLINE): Adjust to unusual GCC outputs.
+       Reported by R Blake <blakers@mac.com>.
+
+2002-07-09  Akim Demaille  <akim@epita.fr>
+
+       * data/yacc.c: Output the copyright notive in the header.
+
+2002-07-03  Akim Demaille  <akim@epita.fr>
+
+       * src/output.c (froms, tos): Are state_number_t.
+       (save_column): sp, sp1, and sp2 are state_number_t.
+       (prepare): Rename `final' as `final_state_number', `nnts' as
+       `nterms_number', `nrules' as `rules_number', `nstates' as
+       `states_number', and `ntokens' as `tokens_number'.  Remove `nsym',
+       unused.
+       * data/yacc.c, data/glr.c, data/lalr1.cc: Adjust.
+       * data/lalr1.cc (nsym_): Remove, unused.
+
+2002-07-03  Akim Demaille  <akim@epita.fr>
+
+       * src/lalr.h, src/lalr.c (goto_number_t): New.
+       * src/lalr.c (goto_list_t): New.
+       Propagate them.
+       * src/nullable.c (rule_list_t): New.
+       Propagate.
+       * src/types.h: Remove.
+
+2002-07-03  Akim Demaille  <akim@epita.fr>
+
+       * src/closure.c (print_fderives): Use rule_rhs_print.
+       * src/derives.c (print_derives): Use rule_rhs_print.
+       (rule_list_t): New, replaces `shorts'.
+       (set_derives): Add comments.
+       * tests/sets.at (Nullable, Firsts): Adjust.
+
+2002-07-03  Akim Demaille  <akim@epita.fr>
+
+       * src/output.c (prepare_actions): Free `tally' and `width'.
+       (prepare_actions): Allocate and free `order'.
+       * src/symtab.c (symbols_free): Free `symbols'.
+       * src/scan-gram.l (scanner_free): Clear Flex's scanners memory.
+       * src/output.c (m4_invoke): Move to...
+       * src/scan-skel.l: here.
+       (<<EOF>>): Close yyout, and free its name.
+
+2002-07-03  Akim Demaille  <akim@epita.fr>
+
+       Fix some memory leaks, and fix a bug: state 0 was examined twice.
+
+       * src/LR0.c (new_state): Merge into...
+       (state_list_append): this.
+       (new_states): Merge into...
+       (generate_states): here.
+       (set_states): Don't ensure a proper `errs' state member here, do it...
+       * src/conflicts.c (conflicts_solve): here.
+       * src/state.h, src/state.c: Comment changes.
+       (state_t): Rename member `shifts' as `transitions'.
+       Adjust all dependencies.
+       (errs_new): For consistency, also take the values as argument.
+       (errs_dup): Remove.
+       (state_errs_set): New.
+       (state_reductions_set, state_transitions_set): Assert that no
+       previous value was assigned.
+       (state_free): New.
+       (states_free): Use it.
+       * src/conflicts.c (resolve_sr_conflict): Don't use an `errs_t' as
+       temporary storage: use `errs' and `nerrs' as elsewhere.
+       (set_conflicts): Allocate and free this `errs'.
+
+2002-07-02  Akim Demaille  <akim@epita.fr>
+
+       * lib/libiberty.h: New.
+       * lib: Update the bitset implementation from upstream.
+       * src/closure.c, src/lalr.c, src/output.c, src/print_graph.c,
+       * src/state.c: Use BITSET_FOR_EACH, not BITSET_EXECUTE.
+       * src/main.c: Adjust bitset stats calls.
+
+2002-07-01  Paul Eggert  <eggert@twinsun.com>
+
+       * src/scan-gram.l (<SC_ESCAPED_CHARACTER>): Convert to unsigned
+       char, so that negative chars don't collide with $.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       Have the GLR tests be `warning' checked, and fix the warnings.
+
+       * data/glr.c (YYFPRINTF): Always define it, not only when YYDEBUG
+       (yyuserAction, yyreportAmbiguity): `Use' all the arguments.
+       (yyremoveDeletes): `yyi' and `yyj' are size_t.
+       Use YYFPRINTF when under if (YYDEBUG) to avoid empty `if' bodies.
+       (yyaddDeferredAction): static.
+       (yyglrReduce): yyi, yyk, amd yyposn are size_t.
+       (yyreportParseError): yyprefix is const.
+       yytokenp is used only when verbose.
+       (yy__GNUC__): Replace with __GNUC__.
+       (yypdumpstack): yyi is size_t.
+       (yypreference): Un-yy local variables and arguments, to avoid
+       clashes with `yyr1'.  Anyway, we are not in the user name space.
+       (yytname_size): be an int, as is compared with ints.
+       * tests/testsuite.at (AT_COMPILE, AT_PARSER_CHECK): New.
+       Use them.
+       * tests/cxx-gram.at: Use quotation to protect $1.
+       Use AT_COMPILE to enable warnings hunts.
+       Prototype yylex and yyerror.
+       `Use' argc.
+       Include `string.h', not `strings.h'.
+       Produce and prototype stmtMerge only when used.
+       yylex takes a location.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       We spend a lot of time in quotearg, in particular when --verbose.
+
+       * src/symtab.c (symbol_get): Store a quoted version of the key.
+       (symbol_tag_get, symbol_tag_get_n, symbol_tag_print): Remove.
+       Adjust all callers.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/state.h (reductions_t): Rename member `nreds' as num.
+       (errs_t): Rename members `nerrs' and `errs' as `num' and `symbols'.
+       * src/state.c (ERRS_ALLOC, REDUCTIONS_ALLOC): Use the correct types.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/state.h, src/state.c (shift_t, SHIFT_SYMBOL, SHIFT_IS_SHIFT)
+       (SHIFT_IS_GOTO, SHIFT_IS_ERROR, SHIFT_DISABLE, SHIFT_IS_DISABLED)
+       (shifts_to): Rename as...
+       (transition_t, TRANSITION_SYMBOL, TRANSITION_IS_TRANSITION)
+       (TRANSITION_IS_GOTO, TRANSITION_IS_ERROR, TRANSITION_DISABLE)
+       (TRANSITION_IS_DISABLED, transitions_to): these.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/print.c (print_shifts, print_gotos): Merge into...
+       (print_transitions): this.
+       (print_transitions, print_errs, print_reductions): Align the
+       lookaheads columns.
+       (print_core, print_transitions, print_errs, print_state,
+       print_grammar): Output empty lines separator before, not after.
+       (state_default_rule_compute): Rename as...
+       (state_default_rule): this.
+       * tests/conflicts.at (Defaulted Conflicted Reduction),
+       (Unresolved SR Conflicts, Resolved SR Conflicts): Adjust.
+       * tests/regression.at (Rule Line Numbers, Web2c Report): Adjust.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       Display items as we display rules.
+
+       * src/gram.h, src/gram.c (rule_lhs_print): New.
+       * src/gram.c (grammar_rules_partial_print): Use it.
+       * src/print.c (print_core): Likewise.
+       * tests/conflicts.at (Defaulted Conflicted Reduction),
+       (Unresolved SR Conflicts): Adjust.
+       (Unresolved SR Conflicts): Adjust and rename as...
+       (Resolved SR Conflicts): this, as was meant.
+       * tests/regression.at (Web2c Report): Adjust.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/print.c (state_default_rule_compute): New, extracted from...
+       (print_reductions): here.
+       Pessimize, but clarify the code.
+       * tests/conflicts.at (Defaulted Conflicted Reduction): New.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/output.c (action_row): Let default_rule be always a rule
+       number.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/closure.c (print_firsts, print_fderives, closure):
+       Use BITSET_EXECUTE.
+       * src/lalr.c (lookaheads_print): Likewise.
+       * src/state.c (state_rule_lookaheads_print): Likewise.
+       * src/print_graph.c (print_core): Likewise.
+       * src/print.c (print_reductions): Likewise.
+       * src/output.c (action_row): Likewise.
+       Use SHIFT_IS_DISABLED, SHIFT_IS_SHIFT and SHIFT_SYMBOL.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/print_graph.c: Use report_flag.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/lalr.c (traverse, digraph, matrix_print, transpose): Move
+       to...
+       * src/relation.h, src/relation.c (traverse, relation_digraph)
+       (relation_print, relation_transpose): New.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/state.h, src/state.c (shifts_to): New.
+       * src/lalr.c (build_relations): Use it.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/gram.h (rule_number_t, RULE_NUMBER_MAX, int_of_rule_number)
+       (item_number_of_rule_number, rule_number_of_item_number): New.
+       * src/LR0.c, src/closure.c, src/derives.c, src/derives.h,
+       * src/gram.c, src/lalr.c, src/nullable.c, src/output.c, src/print.c,
+       * src/print_graph.c, src/reader.c, src/reduce.c, src/reduce.h:
+       Propagate their use.
+       Much remains to be done, in particular wrt `shorts' from types.h.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/symtab.c (symbol_new): Initialize the `printer' member.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/LR0.c (save_reductions): Remove, replaced by...
+       * src/state.h, src/state.c (state_reductions_set): New.
+       (reductions, errs): Rename as...
+       (reductions_t, errs_t): these.
+       Adjust all dependencies.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/LR0.c (state_list_t, state_list_append): New.
+       (first_state, last_state): Now symbol_list_t.
+       (this_state): Remove.
+       (new_itemsets, append_states, save_reductions): Take a state_t as
+       argument.
+       (set_states, generate_states): Adjust.
+       (save_shifts): Remove, replaced by...
+       * src/state.h, src/state.c (state_shifts_set): New.
+       (shifts): Rename as...
+       (shifts_t): this.
+       Adjust all dependencies.
+       * src/state.h (state_t): Remove the `next' member.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/vcg.c (quote): Use slot 2, since we often pass symbol tag
+       escaped in slot 0.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       Use hash.h for the state hash table.
+
+       * src/LR0.c (STATE_HASH_SIZE, state_hash): Remove.
+       (allocate_storage): Use state_hash_new.
+       (free_storage): Use state_hash_free.
+       (new_state, get_state): Adjust.
+       * src/lalr.h, src/lalr.c (states): Move to...
+       * src/states.h (state_t): Remove the `link' member, no longer
+       used.
+       * src/states.h, src/states.c: here.
+       (state_hash_new, state_hash_free, state_hash_lookup)
+       (state_hash_insert, states_free): New.
+       * src/states.c (state_table, state_compare, state_hash): New.
+       * src/output.c (output_actions): Do not free states now, since we
+       still need to know the final_state number in `prepare', called
+       afterwards.  Do it...
+       * src/main.c (main): here: call states_free after `output'.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/state.h, src/state.c (state_new): New, extracted from...
+       * src/LR0.c (new_state): here.
+       * src/state.h (STATE_ALLOC): Move to...
+       * src/state.c: here.
+       * src/LR0.h, src/LR0.c (nstates, final_state): Move to...
+       * src/state.h, src/state.c: here.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.c (gensym): Rename as...
+       * src/symtab.h, src/symtab.c (dummy_symbol_get): this.
+       (getsym): Rename as...
+       (symbol_get): this.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       * src/state.h (state_number_t, STATE_NUMBER_MAX): New.
+       * src/LR0.c, src/LR0.h, src/conflicts.c, src/lalr.c, src/lalr.h,
+       * src/output.c, src/print.c, src/print_graph.c: Propagate.
+       * src/LR0.h, src/LR0.h (final_state): Is a state_t*.
+
+2002-06-30  Akim Demaille  <akim@epita.fr>
+
+       Make the test suite pass with warnings checked.
+
+       * tests/actions.at (Printers and Destructors): Improve.
+       Avoid unsigned vs. signed issues.
+       * tests/calc.at: Don't exercise the scanner here, do it...
+       * tests/input.at (Torturing the Scanner): here.
+
+2002-06-28  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+
+       * data/glr.c: Correct typo in Emacs-mode directive.  Slightly
+       reorganize first lines parallel to yacc.c.
+
+2002-06-28  Akim Demaille  <akim@epita.fr>
+
+       * data/c.m4 (b4_copyright, b4_sint_type, b4_uint_type, b4_token_define)
+       (b4_token_enum, b4_token_defines): New, factored from...
+       * data/lalr1.cc, data/yacc.c, glr.c: here.
+
+2002-06-28  Akim Demaille  <akim@epita.fr>
+
+       * data/yacc.c (yydestruct, yysymprint): Pacify GCC warnings for
+       unused variables.
+       * src/output.c (merger_output): static.
+
+2002-06-28  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.h: s/grammer_current_rule_merge_set/grammar_.../.
+       * src/conflicts.c (conflicts_total_count): `i' is unsigned, to
+       pacify GCC.
+       * src/output.c (save_row): Initialize all the variables to pacify GCC.
+
+2002-06-27  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+
+       Accumulated changelog for new GLR parsing features.
+
+        * src/conflicts.c (count_total_conflicts): Change name to
+       conflicts_total_count.
+       * src/conflicts.h: Ditto.
+       * src/output.c (token_actions): Use the new name.
+       (output_conflicts): Change conflp => conflict_list_heads, and
+       confl => conflict_list for better readability.
+       * data/glr.c: Use the new names.
+       * NEWS: Add self to GLR announcement.
+
+       * src/reader.c (free_merger_functions): Cleanup: XFREE->free.
+
+       * doc/bison.texinfo (GLR Parsers): Make corrections suggested by
+       Akim Demaille.
+
+       * data/bison.glr: Change name to glr.c
+       * data/glr.c: Renamed from bison.glr.
+       * data/Makefile.am: Add glr.c
+
+       * src/getargs.c:
+
+       * src/symlist.h:  Add dprec and merger fields to symbol_list_s.
+       * src/symlist.c (symbol_list_new): Initialize dprec and merger fields.
+
+       Originally 2002-06-16  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+
+       * data/bison.glr: Be sure to restore the
+       current #line when returning to the skeleton contents after having
+       exposed the input file's #line.
+
+       Originally 2002-06-13  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+
+       * data/bison.glr: Bring up to date with changes to bison.simple.
+
+       Originally 2002-06-03  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+
+       * data/bison.glr: Correct definitions that use b4_prefix.
+       Various reformatting.
+       (GLRStack): Make yychar (in YYPURE case) and yytokenp as part of stack.
+       (yyreportParseError, yyrecoverParseError, yyprocessOneStack): remove
+       yytokenp argument; now part of stack.
+       (yychar): Define to behave as documented.
+       (yyclearin): Ditto.
+
+       Originally 2002-05-14  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+
+       * src/reader.h: Add declaration for free_merger_functions.
+
+       * src/reader.c (merge_functions): New variable.
+       (get_merge_function): New function.
+       (free_merger_functions): New function.
+       (readgram): Check for %prec that is not followed by a symbol.
+       Handle %dprec and %merge declarations.
+       (packgram): Initialize dprec and merger fields in rules array.
+
+       * src/output.c (conflict_tos, conflrow, conflict_table, conflict_list,
+       conflict_list_cnt, conflict_list_free): New variables.
+       (table_grow): Also grow conflict_table.
+       (prepare_rules): Output dprec and merger tables.
+       (conflict_row): New function.
+       (action_row): Output conflict lists for GLR parser.  Don't use
+       default reduction in conflicted states for GLR parser so that there
+       are spaces for the conflict lists.
+       (save_row): Also save conflict information.
+       (token_actions): Allocate conflict list.
+       (merger_output): New function.
+       (pack_vector): Pack conflict table, too.
+       (output_conflicts): New function to output yyconflp and yyconfl.
+       (output_check): Allocate conflict_tos.
+       (output_actions): Output conflict tables, also.
+       (output_skeleton): Output b4_mergers definition.
+       (prepare): Output b4_max_rhs_length definition.
+       Use 'bison.glr' as default skeleton for GLR parsers.
+
+       * src/gram.c (glr_parser): New flag.
+       (grammar_free): Call free_merger_functions.
+
+       * src/conflicts.c (count_rr_conflicts): Augment to optionally count
+       all pairs of conflicting reductions, rather than just all tokens
+       causing conflicts.  Needed to size conflict tables.
+       (conflicts_output): Modify call to count_rr_conflicts for new
+       interface.
+       (conflicts_print): Ditto.
+       (count_total_conflicts): New function.
+
+       * src/reader.h (merger_list): New type.
+       (merge_functions): New variable.
+
+       * src/lex.h (tok_dprec, tok_merge): New token types.
+
+       * src/gram.h (rule_s): Add dprec and merger fields.
+       (glr_parser): New flag.
+
+       * src/conflicts.h (count_total_conflicts): New function.
+
+       * src/options.c (option_table): Add %dprec, %merge, and %glr-parser.
+
+       * doc/bison.texinfo (Generalized LR Parsing): New section.
+       (GLR Parsers): New section.
+       (Language and Grammar): Mention GLR parsing.
+       (Table of Symbols): Add %dprec, %glr-parser, %merge, GLR
+       Correct typo ("tge" -> "the").
+
+       * data/bison.glr: New skeleton for GLR parsing.
+
+       * tests/cxx-gram.at: New tests for GLR parsing.
+
+       * tests/testsuite.at: Include cxx-gram.at.
+
+       * tests/Makefile.am: Add cxx-gram.at.
+
+       * src/parse-gram.y:
+
+       * src/scan-gram.l: Add %dprec, %glr-parser, %merge.
+
+       * src/parse-gram.y: Grammar for %dprec, %merge, %glr-parser.
+
+2002-06-27  Akim Demaille  <akim@epita.fr>
+
+       * src/options.h, src/options.c: Remove.
+       * src/getargs.c (short_options, long_options): New.
+
+2002-06-27  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple, data/bison.c++: Rename as...
+       * data/yacc.c, data/lalr1.cc: these.
+       * doc/bison.texinfo (Environment Variables): Remove.
+
+2002-06-25  Raja R Harinath  <harinath@cs.umn.edu>
+
+       * src/getargs.c (report_argmatch): Initialize strtok().
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple (b4_symbol_actions): New, replaces...
+       (b4_symbol_destructor, b4_symbol_printer): these.
+       (yysymprint): Be sure to call YYPRINT only for tokens, and using
+       user token numbers.
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple (yydestructor): Rename as...
+       (yydestruct): this.
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * src/symtab.h, src/symtab.c (symbol_type_set)
+       (symbol_destructor_set, symbol_precedence_set): The location is
+       the last argument.
+       Adjust all callers.
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * src/parse-gram.y (YYPRINT, yyprint): Don't mess with the parser
+       internals.
+       * src/reader.h, src/reader.c (grammar_current_rule_prec_set):
+       Takes a location.
+       * src/symtab.h, src/symtab.c (symbol_class_set)
+       (symbol_user_token_number_set): Likewise.
+       Adjust all callers.
+       Promote complain_at.
+       * tests/input.at (Type Clashes): Adjust.
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple (YYLEX): Fix the declaration when
+       %pure-parser.
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple (yysymprint): Don't print the token number,
+       just its name.
+       * tests/actions.at (Destructors): Rename as...
+       (Printers and Destructors): this.
+       Also exercise %printer.
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple (YYDSYMPRINT): New.
+       Use it to remove many of the #if YYDEBUG/if (yydebug).
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * src/symtab.h, src/symtab.c (symbol_t): printer and
+       printer_location are new members.
+       (symbol_printer_set): New.
+       * src/parse-gram.y (PERCENT_PRINTER): New token.
+       Handle its associated rule.
+       * src/scan-gram.l: Adjust.
+       (handle_destructor_at, handle_destructor_dollar): Rename as...
+       (handle_symbol_code_at, handle_symbol_code_dollar): these.
+       * src/output.c (symbol_printers_output): New.
+       (output_skeleton): Call it.
+       * data/bison.simple (yysymprint): New.  Cannot be named yyprint
+       since there are already many grammar files with a user `yyprint'.
+       Replace the calls to YYPRINT to calls to yysymprint.
+       * tests/calc.at: Adjust.
+       * tests/torture.at (AT_DATA_STACK_TORTURE): Remove YYPRINT: it was
+       taking advantage of parser very internal details (stack size!).
+
+2002-06-20  Akim Demaille  <akim@epita.fr>
+
+       * src/scan-gram.l: Complete the scanner with the missing patterns
+       to pacify Flex.
+       Use `quote' and `symbol_tag_get' where appropriate.
+
+2002-06-19  Akim Demaille  <akim@epita.fr>
+
+       * tests/actions.at (Destructors): Augment to test locations.
+       * data/bison.simple (yydestructor): Pass it the current location
+       if locations are enabled.
+       Prototype only when __STDC__ or C++.
+       Change the argument names to move into the yy name space: there is
+       user code here.
+
+2002-06-19  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple (b4_pure_if): New.
+       Use it instead of #ifdef YYPURE.
+
+2002-06-19  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple (b4_location_if): New.
+       Use it instead of #ifdef YYLSP_NEEDED.
+
+2002-06-19  Akim Demaille  <akim@epita.fr>
+
+       Prepare @$ in %destructor, but currently don't bind it in the
+       skeleton, as %location use is not cleaned up yet.
+
+       * src/scan-gram.l (handle_dollar, handle_destructor_at)
+       (handle_action_at): New.
+       (handle_at, handle_action_dollar, handle_destructor_dollar): Take
+       a braced_code_t and a location as additional arguments.
+       (handle_destructor_dollar): Instead of requiring `b4_eval', just
+       unquote one when outputting `b4_dollar_dollar'.
+       Adjust callers.
+       * data/bison.simple (b4_eval): Remove.
+       (b4_symbol_destructor): Adjust.
+       * tests/input.at (Invalid @n): Adjust.
+
+2002-06-19  Zack Weinberg  <zack@codesourcery.com>
+
+       * doc/bison.texinfo: Document ability to have multiple
+       prologue sections.
+
+2002-06-18  Akim Demaille  <akim@epita.fr>
+
+       * src/files.c (compute_base_names): When computing the output file
+       names from the input file name, strip the directory part.
+
+2002-06-18  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple.new: Comment changes.
+       Reported by Andreas Schwab.
+
+2002-06-18  Matt Kraai  <kraai@alumni.cmu.edu>
+
+       * data/bison.simple (yyoverflowlab): #ifndef yyoverflow, so that
+       there are no `label `yyoverflowlab' defined but not used' warnings
+       when yyoverflow is defined.
+
+2002-06-18  Akim Demaille  <akim@epita.fr>
+
+       * src/symtab.h, src/symtab.c (symbol_t): destructor_location is a
+       new member.
+       (symbol_destructor_set): Adjust.
+       * src/output.c (symbol_destructors_output): Output the destructor
+       locations.
+       Output the symbol name.
+       * data/bison.simple (b4_symbol_destructor): Adjust.
+
+2002-06-18  Cris Bailiff  <c.bailiff@awayweb.com>
+       and Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple.new (yyerrlab1): Be sure to pop and destroy
+       what's left on the stack when the error recovery hits EOF.
+       * tests/actions.at (Destructors): Complete to exercise this case.
+
+2002-06-17  Akim Demaille  <akim@epita.fr>
+
+       * data/m4sugar/m4sugar.m4 (m4_map): Recognize when the list of
+       arguments is really empty, not only equal to `[]'.
+       * src/symtab.h, src/symtab.c (symbol_t): `destructor' is a new
+       member.
+       (symbol_destructor_set): New.
+       * src/output.c (symbol_destructors_output): New.
+       * src/reader.h (brace_code_t, current_braced_code): New.
+       * src/scan-gram.l (BRACED_CODE): Use it to branch on...
+       (handle_dollar): Rename as...
+       (handle_action_dollar): this.
+       (handle_destructor_dollar): New.
+       * src/parse-gram.y (PERCENT_DESTRUCTOR): New.
+       (grammar_declaration): Use it.
+       * data/bison.simple (yystos): Is always defined.
+       (yydestructor): New.
+       * tests/actions.at (Destructors): New.
+       * tests/calc.at (_AT_CHECK_CALC_ERROR): Don't rely on egrep.
+
+2002-06-17  Akim Demaille  <akim@epita.fr>
+
+       * src/symlist.h, src/symlist.c (symbol_list_length): New.
+       * src/scan-gram.l (handle_dollar, handle_at): Compute the
+       rule_length only when needed.
+       * src/output.c (actions_output, token_definitions_output): Output
+       the full M4 block.
+       * src/symtab.c: Don't access directly to the symbol tag, use
+       symbol_tag_get.
+       * src/parse-gram.y: Use symbol_list_free.
+
+2002-06-17  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.h, src/reader.c (symbol_list, symbol_list_new)
+       (symbol_list_prepend, get_type_name): Move to...
+       * src/symlist.h, src/symlist.c (symbol_list_t, symbol_list_new)
+       (symbol_list_prepend, symbol_list_n_type_name_get): here.
+       Adjust all callers.
+       (symbol_list_free): New.
+       * src/scan-gram.l (handle_dollar): Takes a location.
+       * tests/input.at (Invalid $n): Adjust.
+
+2002-06-17  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.h, src/reader.c (symbol_list_new): Export it.
+       (symbol_list_prepend): New.
+       * src/parse-gram.y (%union): `list' is a new member.
+       (symbols.1): New, replaces...
+       (terms_to_prec.1, nterms_to_type.1): these.
+       * src/symtab.h, src/symtab.c (symbol_type_set, symbol_precedence_set)
+       Take a location as additional argument.
+       Adjust all callers.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       * src/parse-gram.y: Move %token in the declaration section so that
+       we don't depend upon CVS Bison.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       * src/state.h, src/state.c (state_rule_lookaheads_print): New.
+       * src/print.c (print_core): Use it.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       * src/conflicts.c (log_resolution): Accept the rule involved in
+       the sr conflicts instead of the lookahead number that points to
+       that rule.
+       (flush_reduce): Accept the current lookahead vector as argument,
+       instead of the index in LA.
+       (resolve_sr_conflict): Accept the current number of lookahead
+       bitset to consider for the STATE, instead of the index in LA.
+       (set_conflicts): Adjust.
+       * src/lalr.c, src/lalr.h, src/state.h: Comment changes.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       * src/state.h (state_t): Replace the `lookaheadsp' member, a
+       short, with `lookaheads' (bitsetv), `lookaheads_rule' (rule_t**).
+       Adjust all dependencies.
+       * src/lalr.c (initialize_lookaheads): Split into...
+       (states_lookaheads_count, states_lookaheads_initialize): these.
+       (lalr): Adjust.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       * src/gram.h, src/gram.c (grammar_rules_partial_print): New, eved
+       out of...
+       (grammar_rules_print): here.
+       * src/reduce.c (reduce_output): Use it.
+       * tests/reduce.at (Useless Rules, Reduced Automaton)
+       (Underivable Rules): Adjust.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       Copy BYacc's nice way to report the grammar.
+
+       * src/gram.h, src/gram.c (grammar_rhs_print, grammar_rules_print):
+       New.
+       Don't print the rules' location, it is confusing and useless.
+       (rule_print): Use grammar_rhs_print.
+       * src/print.c (print_grammar): Use grammar_rules_print.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       Complete and rationalize `useless thing' warnings.
+
+       * src/symtab.h, src/symtab.c (symbol_tag_get, symbol_tag_get_n)
+       (symbol_tag_print): New.
+       Use them everywhere in place of accessing directly the tag member.
+       * src/gram.h, src/gram.c (rule_print): New.
+       Use it where a rule used to be printed `by hand'.
+       * src/reduce.c (nonterminals_reduce): Report the use nonterminals.
+       (reduce_grammar_tables): Report the useless rules.
+       (reduce_print): Useless things are a warning, not an error.
+       Report it as such.
+       * tests/reduce.at (Useless Nonterminals, Useless Rules):
+       (Reduced Automaton, Underivable Rules): Adjust.
+       * tests/regression.at (Web2c Report, Web2c Report): Adjust.
+       * tests/conflicts.at (Unresolved SR Conflicts)
+       (Solved SR Conflicts): Adjust.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       Let symbols have a location.
+
+       * src/symtab.h, src/symtab.c (symbol_t): Location is a new member.
+       (getsym): Adjust.
+       Adjust all callers.
+       * src/complain.h, src/complain.c (complain_at, fatal_at, warn_at):
+       Use location_t, not int.
+       * src/symtab.c (symbol_check_defined): Take advantage of the
+       location.
+       * tests/regression.at (Invalid inputs): Adjust.
+
+2002-06-15  Akim Demaille  <akim@epita.fr>
+
+       * src/parse-gram.y (YYLLOC_DEFAULT, current_lhs_location): New.
+       (input): Don't try to initialize yylloc here, do it in the
+       scanner.
+       * src/scan-gram.l (YY_USER_INIT): Initialize yylloc.
+       * src/gram.h (rule_t): Change line and action_line into location
+       and action_location, of location_t type.
+       Adjust all dependencies.
+       * src/location.h, src/location.c (empty_location): New.
+       * src/reader.h, src/reader.c (grammar_start_symbol_set)
+       (grammar_symbol_append, grammar_rule_begin, grammar_rule_end)
+       (grammar_current_rule_symbol_append)
+       (grammar_current_rule_action_append): Expect a location as argument.
+       * src/reader.c (grammar_midrule_action): Adjust to attach an
+       action's location as dummy symbol location.
+       * src/symtab.h, src/symtab.c (startsymbol_location): New.
+       * tests/regression.at (Web2c Report, Rule Line Numbers): Adjust
+       the line numbers.
+
+2002-06-14  Akim Demaille  <akim@epita.fr>
+
+       Grammar declarations may be found in the grammar section.
+
+       * src/parse-gram.y (rules_or_grammar_declaration): New.
+       (declarations): Each declaration may end with a semicolon, not
+       just...
+       (grammar_declaration): `"%union"'.
+       (grammar): Branch to rules_or_grammar_declaration.
+
+2002-06-14  Akim Demaille  <akim@epita.fr>
+
+       * src/main.c (main): Invoke scanner_free.
+
+2002-06-14  Akim Demaille  <akim@epita.fr>
+
+       * src/output.c (m4_invoke): Extracted from...
+       (output_skeleton): here.
+       Free tempfile.
+
+2002-06-14  Akim Demaille  <akim@epita.fr>
+
+       * src/parse-gram.y (directives, directive, gram)
+       (grammar_directives, precedence_directives, precedence_directive):
+       Rename as...
+       (declarations, declaration, grammar, grammar_declaration)
+       (precedence_declaration, precedence_declarator): these.
+       (symbol_declaration): New.
+
+2002-06-14  Akim Demaille  <akim@epita.fr>
+
+       * src/files.c (action_obstack): Remove, unused.
+       (output_obstack): Remove it, and all its dependencies, as it is no
+       longer needed.
+       * src/reader.c (epilogue_set): Build the epilogue in the
+       muscle_obstack.
+       * src/output.h, src/output.c (muscle_obstack): Move to...
+       * src/muscle_tab.h, src/muscle_tab.h: here.
+       (muscle_init): Initialize muscle_obstack.
+       (muscle_free): New.
+       * src/main.c (main): Call it.
+
+2002-06-14  Akim Demaille  <akim@epita.fr>
+
+       * src/location.h: New, extracted from...
+       * src/reader.h: here.
+       * src/Makefile.am (noinst_HEADERS): Merge into
+       (bison_SOURCES): this.
+       Add location.h.
+       * src/parse-gram.y: Use location_t instead of Bison's.
+       * src/reader.h, src/reader.c (prologue_augment, epilogue_set):
+       Use location_t instead of ints.
+
+2002-06-14  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.simple, data/bison.c++: Be sure to restore the
+       current #line when returning to the skeleton contents after having
+       exposed the input file's #line.
+
+2002-06-12  Akim Demaille  <akim@epita.fr>
+
+       * src/scan-gram.l (SC_BRACED_CODE): Don't use `<.*>', it is too
+       eager.
+       * tests/actions.at (Exotic Dollars): New.
+
+2002-06-12  Akim Demaille  <akim@epita.fr>
+
+       * src/scan-gram.l (SC_PROLOGUE): Don't eat characters amongst
+       ['"/] too eagerly.
+       * tests/input.at (Torturing the Scanner): New.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       * src/scan-gram.l (YY_OBS_INIT): Remove, replace with...
+       [SC_COMMENT,SC_STRING,SC_CHARACTER,SC_BRACED_CODE,SC_PROLOGUE]
+       [SC_EPILOGUE]: Output the quadrigraphs only when not in a comment.
+       * src/reader.h, src/scan-gram.l (scanner_initialize): this.
+       * src/reader.c (reader): Use it.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       * src/scan-gram.l (YY_OBS_FINISH): Don't set yylval.
+       Adjust all callers.
+       (scanner_last_string_free): New.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       * src/scan-gram.l (YY_INIT, YY_GROW, YY_FINISH): Rename as...
+       (YY_OBS_INIT, YY_OBS_GROW, YY_OBS_FINISH): these.
+       (last_string, YY_OBS_FREE): New.
+       Use them when returning an ID.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       Have Bison grammars parsed by a Bison grammar.
+
+       * src/reader.c, src/reader.h (prologue_augment): New.
+       * src/reader.c (copy_definition): Remove.
+
+       * src/reader.h, src/reader.c (gram_start_symbol_set, prologue_augment)
+       (grammar_symbol_append, grammar_rule_begin, grammar_midrule_action)
+       (grammar_current_rule_prec_set, grammar_current_rule_check)
+       (grammar_current_rule_symbol_append)
+       (grammar_current_rule_action_append): Export.
+       * src/parse-gram.y (symbol_list_new, symbol_list_symbol_append_
+       (symbol_list_action_append): Remove.
+       Hook the routines from reader.
+       * src/scan-gram.l: In INITIAL, characters and strings are tokens.
+       * src/system.h (ATTRIBUTE_NORETURN, ATTRIBUTE_UNUSED): Now.
+
+       * src/reader.c (read_declarations): Remove, unused.
+
+       * src/parse-gram.y: Handle the epilogue.
+       * src/reader.h, src/reader.c (gram_start_symbol_set): Rename as...
+       (grammar_start_symbol_set): this.
+       * src/scan-gram.l: Be sure to ``use'' yycontrol to keep GCC quiet.
+       * src/reader.c (readgram): Remove, unused.
+       (reader): Adjust to insert eoftoken and axiom where appropriate.
+
+       * src/reader.c (copy_dollar): Replace with...
+       * src/scan-gram.h (handle_dollar): this.
+       * src/parse-gram.y: Remove `%thong'.
+
+       * src/reader.c (copy_at): Replace with...
+       * src/scan-gram.h (handle_at): this.
+
+       * src/complain.h, src/complain.c (warn_at, complain_at, fatal_at):
+       New.
+
+       * src/scan-gram.l (YY_LINES): Keep lineno synchronized for the
+       time being.
+
+       * src/reader.h, src/reader.c (grammar_rule_end): New.
+
+       * src/parse.y (current_type, current_class): New.
+       Implement `%nterm', `%token' support.
+       Merge `%term' into `%token'.
+       (string_as_id): New.
+       * src/symtab.h, src/symtab.c (symbol_make_alias): Don't pass the
+       type name.
+
+       * src/parse-gram.y: Be sure to handle properly the beginning of
+       rules.
+
+       * src/parse-gram.y: Handle %type.
+       * src/reader.c (grammar_rule_end): Call grammar_current_rule_check.
+
+       * src/parse-gram.y: More directives support.
+       * src/options.c: No longer handle source directives.
+
+       * src/parse-gram.y: Fix %output.
+
+       * src/parse-gram.y: Handle %union.
+       Use the prologue locations.
+       * src/reader.c (parse_union_decl): Remove.
+
+       * src/reader.h, src/reader.c (epilogue_set): New.
+       * src/parse-gram.y: Use it.
+
+       * data/bison.simple, data/bison.c++: b4_stype is now either not
+       defined, then default to int, or to the contents of %union,
+       without `union' itself.
+       Adjust.
+       * src/muscle_tab.c (muscle_init): Don't predefine `stype'.
+
+       * src/output.c (actions_output): Don't output braces, as they are
+       already handled by the scanner.
+
+       * src/scan-gram.l (SC_CHARACTER): Set the user_token_number of
+       characters to themselves.
+
+       * tests/reduce.at (Reduced Automaton): End the grammars with %% so
+       that the epilogue has a proper #line.
+
+       * src/parse-gram.y: Handle precedence/associativity.
+
+       * src/symtab.c (symbol_precedence_set): Requires the symbol to be
+       a terminal.
+       * src/scan-gram.l (SC_BRACED_CODE): Catch strings and characters.
+       * tests/calc.at: Do not use `%token "foo"' as it makes not sense
+       at all to define terminals that cannot be emitted.
+
+       * src/scan-gram.l: Escape M4 characters.
+
+       * src/scan-gram.l: Working properly with escapes in user
+       strings/characters.
+
+       * tests/torture.at (AT_DATA_TRIANGULAR_GRAMMAR)
+       (AT_DATA_HORIZONTAL_GRAMMAR): Respect the `%token ID NUM STRING'
+       grammar.
+       Use more modest sizes, as for the time being the parser does not
+       release memory, and therefore the process swallows a huge amount
+       of memory.
+
+       * tests/torture.at (AT_DATA_LOOKAHEADS_GRAMMAR): Adjust to the
+       stricter %token grammar.
+
+       * src/symtab.h (associativity): Add `undef_assoc'.
+       (symbol_precedence_set): Do nothing when passed an undef_assoc.
+       * src/symtab.c (symbol_check_alias_consistence): Adjust.
+
+       * tests/regression.at (Invalid %directive): Remove, as it is now
+       meaningless.
+       (Invalid inputs): Adjust to the new error messages.
+       (Token definitions): The new grammar doesn't allow too many
+       eccentricities.
+
+       * src/lex.h, src/lex.c: Remove.
+       * src/reader.c (lastprec, skip_to_char, read_signed_integer)
+       (copy_character, copy_string2, copy_string, copy_identifier)
+       (copy_comment, parse_token_decl, parse_type_decl, parse_assoc_decl)
+       (parse_muscle_decl, parse_dquoted_param, parse_skel_decl)
+       (parse_action): Remove.
+       * po/POTFILES.in: Adjust.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.c  (parse_action): Don't store directly into the
+       rule's action member: return the action as a string.
+       Don't require `rule_length' as an argument: compute it.
+       (grammar_current_rule_symbol_append)
+       (grammar_current_rule_action_append): New, eved out from
+       (readgram): here.
+       Remove `action_flag', `rulelength', unused now.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.c (grammar_current_rule_prec_set).
+       (grammar_current_rule_check): New, eved out from...
+       (readgram): here.
+       Remove `xaction', `first_rhs': useless.
+       * tests/input.at (Type clashes): New.
+       * tests/existing.at (GNU Cim Grammar): Adjust.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.c (grammar_midrule_action): New, Eved out from
+       (readgram): here.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.c (grammar_rule_begin, previous_rule, current_rule):
+       New.
+       (readgram): Use them as replacement of inlined code, crule and
+       crule1.
+
+2002-06-11  Akim Demaille  <akim@epita.fr>
+
+       * src/reader.c (grammar_end, grammar_symbol_append): New.
+       (readgram): Use them.
+       Make the use of `p' as local as possible.
+
+2002-06-10  Akim Demaille  <akim@epita.fr>
+
+       GCJ's parser requires the tokens to be defined before the prologue.
+
+       * data/bison.simple: Output the token definition before the user's
+       prologue.
+       * tests/regression.at (Braces parsing, Duplicate string)
+       (Mixing %token styles): Check the output from bison.
+       (Early token definitions): New.
+
+2002-06-10  Akim Demaille  <akim@epita.fr>
+
+       * src/symtab.c (symbol_user_token_number_set): Don't complain when
+       assigning twice the same user number to a token, so that we can
+       use it in...
+       * src/lex.c (lex): here.
+       Also use `symbol_class_set' instead of hand written code.
+       * src/reader.c (parse_assoc_decl): Likewise.
+
+2002-06-10  Akim Demaille  <akim@epita.fr>
+
+       * src/symtab.c, src/symtab.c (symbol_class_set)
+       (symbol_user_token_number_set): New.
+       * src/reader.c (parse_token_decl): Use them.
+       Use a switch instead of ifs.
+       Use a single argument.
+
+2002-06-10  Akim Demaille  <akim@epita.fr>
+
+       Remove `%thong' support as it is undocumented, unused, duplicates
+       `%token's job, and creates useless e-mail traffic with people who
+       want to know what it is, why it is undocumented, unused, and
+       duplicates `%token's job.
+
+       * src/reader.c (parse_thong_decl): Remove.
+       * src/options.c (option_table): Remove "thong".
+       * src/lex.h (tok_thong): Remove.
+
+2002-06-10  Akim Demaille  <akim@epita.fr>
+
+       * src/symtab.c, src/symtab.c (symbol_type_set)
+       (symbol_precedence_set): New.
+       * src/reader.c (parse_type_decl, parse_assoc_decl): Use them.
+       (value_components_used): Remove, unused.
+
+2002-06-09  Akim Demaille  <akim@epita.fr>
+
+       Move symbols handling code out of the reader.
+
+       * src/reader.h, src/reader.c (errtoken, undeftoken, eoftoken)
+       (axiom): Move to...
+       * src/symtab.h, src/symtab.c: here.
+
+       * src/gram.c (start_symbol): Remove: use startsymbol->number.
+       * src/reader.c (startval): Rename as...
+       * src/symtab.h, src/symtab.c (startsymbol): this.
+       * src/reader.c: Adjust.
+
+       * src/reader.c (symbol_check_defined, symbol_make_alias)
+       (symbol_check_alias_consistence, symbol_pack, symbol_translation)
+       (token_translations_init)
+       Move to...
+       * src/symtab.c: here.
+       * src/reader.c (packsymbols): Move to...
+       * src/symtab.h, src/symtab.c (symbols_pack): here.
+       * src/symtab.h, src/symtab.c (symbol_make_alias): Takes SYMVAL as
+       argument.
+
+2002-06-03  Akim Demaille  <akim@epita.fr>
+
+       * src/muscle_tab.c (muscle_insert, muscle_find): Declarations,
+       then statements.
+
+2002-06-03  Akim Demaille  <akim@epita.fr>
+
+       * src/muscle_tab.c (muscle_find, muscle_insert): Don't initialize
+       structs with non literals.
+       * src/scan-skel.l: never-interactive.
+       * src/conflicts.c (enum conflict_resolution_e): No trailing
+       comma.
+       * src/getargs.c (usage): Split long literal strings.
+       Reported by Hans Aberg.
+
+2002-05-28  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.c++: Use C++ ostreams.
+       (cdebug_): New member.
+
+2002-05-28  Akim Demaille  <akim@epita.fr>
+
+       * src/output.c (output_skeleton): Be sure to allocate enough room
+       for `/' _and_ for `\0' in full_skeleton.
+
+2002-05-28  Akim Demaille  <akim@epita.fr>
+
+       * data/bison.c++: Catch up with bison.simple:
+       2002-05-24  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+       and Paul Eggert  <eggert@twinsun.com>: `error' handing.
+       2002-05-26  Akim Demaille  <akim@epita.fr>: stos_, token_number_,
+       and popping traces.
+
+2002-05-27  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+
+       * src/output.c (output_skeleton): Put an explicit path in front of
+       the skeleton file name, rather than relying on the -I directory,
+       to partially alleviate effects of having a skeleton file lying around
+       in the current directory.
+
+2002-05-27  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
+
+       * src/conflicts.c (log_resolution): Correct typo:
+       obstack_printf should be obstack_fgrow1.
+
 2002-05-26  Akim Demaille  <akim@epita.fr>
 
        * src/state.h (state_t): `solved_conflicts' is a new member.
 
 2002-05-24  Paul Hilfinger  <Hilfinger@CS.Berkeley.EDU>
        and Paul Eggert  <eggert@twinsun.com>
-       
+
        * data/bison.simple (yyparse): Correct error handling to conform to
        POSIX and yacc.  Specifically, after syntax error is discovered,
        do not reduce further before shifting the error token.
        * data/bison.simple (yyr1): Don't use yy_token_number_type as element
        type; it isn't always big enough, since it doesn't necessarily
        include non-terminals.
-       (yytranslate): Expand definition of yy_token_number_type, so that 
+       (yytranslate): Expand definition of yy_token_number_type, so that
        the latter can be removed.
        (yy_token_number_type): Remove, only one use.
        * data/bison.c++ (r1_): Parallel change to yyr1 in bison.simple---
        don't use TokenNumberType as element type.
-       
+
        * tests/regression.at: Modify expected output to agree with change
        to yyr1 and yytranslate.
-       
+
 2002-05-13  Florian Krohm  <florian@edamail.fishkill.ibm.com>
 
        * src/reader.c (parse_action): Use copy_character instead of
 
 Copyright (C) 1997, 1998, 1999, 2000, 2001, 2002 Free Software Foundation, Inc.
 
-This file is part of GNU Autoconf.
+This file is part of GNU Bison.
 
-GNU Autoconf is free software; you can redistribute it and/or modify
+GNU Bison is free software; you can redistribute it and/or modify
 it under the terms of the GNU General Public License as published by
 the Free Software Foundation; either version 2, or (at your option)
 any later version.
 
-GNU Autoconf is distributed in the hope that it will be useful,
+GNU Bison is distributed in the hope that it will be useful,
 but WITHOUT ANY WARRANTY; without even the implied warranty of
 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 GNU General Public License for more details.
 
 You should have received a copy of the GNU General Public License
-along with autoconf; see the file COPYING.  If not, write to
+along with GNU Bison; see the file COPYING.  If not, write to
 the Free Software Foundation, Inc., 59 Temple Place - Suite 330,
 Boston, MA 02111-1307, USA.