Paul Eggert [Mon, 14 Oct 2002 08:44:02 +0000 (08:44 +0000)]
* tests/action.at, tests/calc.at, tests/conflicts.at,
tests/cxx-type.at, tests/headers.at, tests/input.at,
tests/regression.at, tests/synclines.at, tests/torture.at:
Say "bison -o foo.c foo.y", not "bison foo.y -o foo.c",
so that the tests still work even if POSIXLY_CORRECT is set.
* doc/bison.texinfo (Rpcalc Compile, Invocation): Likewise.
Paul Eggert [Mon, 14 Oct 2002 08:07:01 +0000 (08:07 +0000)]
* data/c.m4 (b4_int_type): Use yysigned_char instead of signed char,
for portability to K&R hosts. Fix typo: signed char is guaranteed
only to 127, not to 128.
* data/glr.c (yysigned_char): New type.
* data/yacc.c (yysigned_char): Likewise.
* tests/regression.at (Web2c Actions): signed char -> yysigned_char.
Paul Eggert [Sun, 13 Oct 2002 20:36:45 +0000 (20:36 +0000)]
* data/c.m4 (b4_token_defines): Protect against double-inclusion
by wrapping enum yytokentype's definition inside #ifndef
YYTOKENTYPE. This undoes a bug I introduced on 2002-10-12.
Paul Eggert [Sun, 13 Oct 2002 20:35:06 +0000 (20:35 +0000)]
(b4_token_defines): Protect against double-inclusion
by wrapping enum yytokentype's definition inside #ifndef
YYTOKENTYPE. This undoes a bug I introduced on 2002-10-12.
Akim Demaille [Sun, 13 Oct 2002 18:50:40 +0000 (18:50 +0000)]
Let nondeterministic skeletons be usable with deterministic
tables.
With the patch, GAWK compiled by GCC without -O2 passes its test
suite using a GLR parser driven by LALR tables. It fails with -O2
because `struct stat' gives two different answers on my machine:
88 (definition of an auto var) and later 96 (memset on this var).
Hence the stack is badly corrumpted. The headers inclusion is to
blame: if I move the awk.h inclusion before GLR's system header
inclusion, the two struct stat have the same size.
* src/tables.c (pack_table): Always create conflict_table.
(token_actions): Always create conflict_list.
* data/glr.c (YYFLAG): Remove, unused.
Akim Demaille [Sun, 13 Oct 2002 18:49:44 +0000 (18:49 +0000)]
* configure.ac (AC_GNU_SOURCE): Use it instead of hand written code.
(O0FLAGS): New.
(VALGRIND, GXX): New.
* tests/atlocal.in (CFLAGS): Use O0FLAGS.
* tests/bison.in: Run $PREBISON a pre-command.
* tests/Makefile.am (maintainer-check, maintainer-check-valgrind)
(maintainer-check-g++): New.
* Makefile.am (maintainer-check): New.
Akim Demaille [Sun, 13 Oct 2002 14:25:14 +0000 (14:25 +0000)]
GLR parsers sometimes raise parse errors instead of performing the
default reduction.
Reported by Charles-Henry de Boysson.
* tests/calc.at (_AT_CHECK_CALC, _AT_CHECK_CALC_ERROR): Don't
check the length of the traces when %glr.
(_AT_CHECK_CALC_ERROR): Also skip `^Stack' lines, coming from
GLR's traces.
(AT_CHECK_CALC_LALR, AT_CHECK_CALC_GLR): New.
Test GLR parsers.
* data/glr.c (YYLEFTMOST_STATE): Fix its value.
(yyltype): Remove the yy prefix from the member names.
(yytable): Complete its comment.
(yygetLRActions): Map error action number from YYTABLE from
YYTABLE_NINF to 0.
(yyisErrorAction): No longer compare YYACTION to YYPACT_NINF
(which was a bug: it should have been YYTABEL_NINF, and yet it was
not satisfying as we could compare an YYACTION computed from
YYDEFACT to YYTABLE_NINF although they are unrelated): 0 is the
only value for error actions.
(yyreportParseError): In verbose parse error messages, don't issue
`error' in the list of expected tokens.
* data/yacc.c (yyparse) <yybackup>: Rewrite the decoding of the
next action to perform to match glr.c's decoding.
(yytable): Complete its comment.
Paul Eggert [Sun, 13 Oct 2002 08:40:10 +0000 (08:40 +0000)]
Fix problem reported by Henrik Grubbstroem in
<http://mail.gnu.org/pipermail/bug-bison/2002-October/001670.html>:
"nonterm: { $$ = 123; } { $$ = $1; };" was wrongly rejected,
because the Bison parser reads the second action before reducing
the first one.
* src/scan-gram.l (rule_length): New static var.
Use it to keep track of the rule length in the scanner, since
we can't expect the parser to be in lock-step sync with the scanner.
(handle_action_dollar, handle_action_at): Use this var.
* tests/actions.at (Exotic Dollars): Test for the problem.
Paul Eggert [Sun, 13 Oct 2002 08:38:39 +0000 (08:38 +0000)]
(rule_length): New static var.
Use it to keep track of the rule length in the scanner, since
we can't expect the parser to be in lock-step sync with the scanner.
(handle_action_dollar, handle_action_at): Use this var.
Paul Eggert [Sun, 13 Oct 2002 05:56:20 +0000 (05:56 +0000)]
[! IN_GCC && HAVE_SYS_TIME_H]: Include <sys/time.h>.
Update copyright date and clarify comments.
(get_time) [IN_GCC]: Keep the GCC version for reference.
Merge GCC changes. Change "GCC" to "Bison" in the copyright notice.
Paul Eggert [Sun, 13 Oct 2002 03:53:18 +0000 (03:53 +0000)]
* src/reader.c (grammar_current_rule_check):
Don't worry about the default action if $$ is untyped.
Prevents bogus warnings reported by Jim Gifford in
<http://mail.gnu.org/pipermail/bug-bison/2002-October/001673.html>.
Paul Eggert [Sun, 13 Oct 2002 03:52:53 +0000 (03:52 +0000)]
(grammar_current_rule_check):
Don't worry about the default action if $$ is untyped.
Prevents bogus warnings reported by Jim Gifford in
<http://mail.gnu.org/pipermail/bug-bison/2002-October/001673.html>.
Paul Eggert [Sat, 12 Oct 2002 23:21:00 +0000 (23:21 +0000)]
* data/c.m4 (b4_token_enum): Do not define YYTOKENTYPE.
* data/glr.c, data/lalr1.cc, data/yacc.c:
Output token definitions before the first part of user declarations.
Fixes compatibility problem reported by Jim Gifford for kbd in
<http://mail.gnu.org/pipermail/bug-bison/2002-October/001672.html>.
Paul Eggert [Fri, 11 Oct 2002 20:54:17 +0000 (20:54 +0000)]
* data/yacc.c (yyreport_parse_error): Remove, putting its body into...
(yyparse): here. This undoes some of the 2002-07-25 change.
Compatibility problem reported by Ralf S. Engelschall with
OSSP cfg <http://www.ossp.org/pkg/lib/cfg/>.
Paul Eggert [Fri, 11 Oct 2002 20:53:39 +0000 (20:53 +0000)]
(yyreport_parse_error): Remove, putting its body into...
(yyparse): here. This undoes some of the 2002-07-25 change.
Compatibility problem reported by Ralf S. Engelschall with
OSSP cfg <http://www.ossp.org/pkg/lib/cfg/>.
Akim Demaille [Fri, 11 Oct 2002 11:23:19 +0000 (11:23 +0000)]
* tests/regression.at Characters Escapes): New.
* src/scan-gram.l (SC_ESCAPED_CHARACTER): Accept ' in strings and
characters.
Reported by Jan Nieuwenhuizen.
Paul Eggert [Thu, 10 Oct 2002 07:25:21 +0000 (07:25 +0000)]
(bitsetv_alloc): Return a size that is aligned properly for vectors of
objects. Do not assume that adding a header size to a multiple of a
word size yields a value that is properly aligned for the whole union.
(bitsetv_alloc): Add a cast to (void *) to avoid a GCC warning.
Paul Eggert [Thu, 10 Oct 2002 07:15:56 +0000 (07:15 +0000)]
(enum_bitset_ops, enum_bitset_type): New types.
All instances of parameters of type enum bitset_opts are now of
type enum_bitset_opts, to conform to the C Standard, and similarly
for enum_bitset_type.
(bitset): Now points to a union, not to a struct.
All instances of "struct bitset_struct *" replaced with "bitset".
(bitset_and_or_, bitset_andn_or_, bitset_or_and_): New decls.
Paul Eggert [Thu, 10 Oct 2002 07:13:58 +0000 (07:13 +0000)]
Include <stddef.h>, for offsetof.
(abitset_bytes): Return a size that is aligned
properly for vectors of objects. Do not assume that adding a
header size to a multiple of a word size yields a value that is
properly aligned for the whole union.
(abitset_ones, abitset_zero, abitset_empty_p, abitset_copy1,
abitset_not, abitset_equal_p, abitset_subset_p, abitset_disjoint_p,
abitset_and, abitset_and_cmp, abitset_andn, abitset_andn_cmp,
abitset_or, abitset_or_cmp, abitset_xor, abitset_xor_cmp,
abitset_and_or, abitset_and_or_cmp, abitset_andn_or,
abitset_andn_or_cmp, abitset_or_and, abitset_or_and_cmp,
abitset_copy): Supply prototype decls, to improve the type-checking
that GCC can do.
Paul Eggert [Mon, 7 Oct 2002 19:58:00 +0000 (19:58 +0000)]
* lib/bitset.h (bitset_reset): Do not assume that bitset_word is
the same width as int. This reapplies a hunk of the 2002-08-12 patch
<http://mail.gnu.org/pipermail/bison-patches/2002-August/001111.html>,
which was inadvertently undone by the 2002-09-30 patch.
* lib/lbitset.c (debug_lbitset): Do not assume that bitset_word is
the same width as int.
Paul Eggert [Mon, 7 Oct 2002 05:43:35 +0000 (05:43 +0000)]
(bitset_reset): Do not assume that bitset_word is
the same width as int. This reapplies a hunk of the 2002-08-12 patch
<http://mail.gnu.org/pipermail/bison-patches/2002-August/001111.html>,
which was inadvertently undone by the 2002-09-30 patch.
Paul Eggert [Sat, 5 Oct 2002 04:45:45 +0000 (04:45 +0000)]
Minor spelling, grammar, and white space fixes.
(Symbols): Mention that any negative value returned from yylex
signifies end-of-input. Warn about negative chars. Mention
the portable Standard C character set.
Paul Eggert [Wed, 2 Oct 2002 07:54:52 +0000 (07:54 +0000)]
(bitsetv_alloc, bitsetv_create, bitsetv_free,
bitsetv_zero, bitsetv_ones, bitsetv_transitive_closure,
bitsetv_reflexive_transitive_closure):
Use bitset_bindex, not int or unsigned int or size_t, to count bits.
(bitsetv_alloc): Use size_t, not unsigned int, to count bytes.
(bitsetv_dump, debug_bitsetv):
Use bitset_windex instead of unsigned int.
Use proper printf formats for widths of integer types.
(bitsetv_alloc): Check for arithmetic overflow when computing sizes.
Paul Eggert [Wed, 2 Oct 2002 07:49:35 +0000 (07:49 +0000)]
(bitset_stats_size, bitset_stats_list,
bitset_stats_list_reverse, bitset_stats_size,
bitset_stats_count, bitset_stats_list, bitset_stat_list_reverse):
Use bitset_bindex, not int or unsigned int or size_t, to count bits.
(bitset_stats_bytes, bitset_stats_init):
Use size_t, not unsigned int, to count bytes.
(bitset_percent_histogram_print,
bitset_log_histogram_print, bitset_stats_print_1):
Use proper printf formats for width of integer types.
Paul Eggert [Wed, 2 Oct 2002 07:45:37 +0000 (07:45 +0000)]
(bitset_bytes, bitset_alloc, bitset_obstack_alloc):
Use size_t, not unsigned int, to count bytes.
(bitset_print): Use proper printf format for bitset types.
(bitset_next, bitset_prev, bitset_first, bitset_last):
Return BITSET_BINDEX_MAX (not -1) for no value,
since we now return the bitset_bindex type (not int).
Paul Eggert [Wed, 2 Oct 2002 07:44:07 +0000 (07:44 +0000)]
(struct bitset_vtable.size, struct bitset_vtable.count,
struct bitset_vtable.list, struct bitset_vtable.list_reverse,
bitset_count_, bitset_next, bitset_prev, bitset_first, bitset_last,
bitset_count_):
Use bitset_bindex, not int or unsigned int or size_t, to count bits.
(BITSET_BINDEX_MAX, BITSET_WINDEX_MAX, BITSET_SIZE_MAX): New macros.
(BITSET_INDEX_MAX): Remove. It wasn't right, since it assumed that
sizeof (bitset_word) == sizeof (bitset_windex). All uses changed
to BITSET_WINDEX_MAX.
Paul Eggert [Wed, 2 Oct 2002 07:41:04 +0000 (07:41 +0000)]
(struct abitset_struct.n_bits, abitset_small_list, abitset_size,
abitset_list, abitset_list_reverse, abitset_list):
Use bitset_bindex, not int or unsigned int or size_t, to count bits.
(abitset_ones, abitset_zero, abitset_bytes):
Use size_t, not unsigned int, to count bytes.