Bison supports a union tag, for obscure reasons. But it does a poor
job at it, especially since Bison 3.0.
Reported by Stephen Cameron and Tobias Frost.
It did not ensure that the name was not given several times. An easy
way to do this is to make the %union tag be handled as a %define
variable, as they cannot be defined several times.
Since Bison 3.0, the synclines were wrongly placed, resulting in
invalid code. Addressing this issue, because of the way the union tag
was stored (as a code muscle), would have been tedious. Unless we
rather define the %union tag as a %percent variable, whose synclines
are easier to manipulate.
So replace the b4_union_name muscle by the api.value.union.name
%define variable, document, and check.
* data/bison.m4: Make sure that api.value.union.name has a keyword value.
* data/c++.m4: Make sure that api.value.union.name is not defined.
* data/c.m4 (b4_union_name): No longer use it, use api.value.union.name.
* doc/bison.texi (%define Summary): Document it.
* src/parse-gram.y (union_name): No longer define b4_uion_name, but
api.value.union.name.
* tests/input.at (Redefined %union name): New.
* tests/synclines.at (%union name syncline): New.
* tests/types.at: Check named %unions.
Reported by Rich Wilson.
* data/c.m4 (b4_symbol_type_register): Append to b4_union_members,
not b4_user_union_members.
The latter invokes the former, but it is the former which is reinitialized
to empty by b4_value_type_setup_union.
* tests/types.at: Check it.
This reveals another bug, this time in the case of glr.c parsers.
* data/glr.c: Generate the header file before the implementation file,
to be sure that the setup is run before what depends on it.
clang, with -DNDEBUG and -Werror fails on some functions that might
lack a return. This is because aver is just another assert, discarded
with -DNDEBUG. So use abort.
* src/muscle-tab.c, src/scan-skel.l: here.
Currently, we do not install the various examples extracted from the
documentation. Let's do it, as they are useful starting points.
* configure.ac: When --enable-gcc-warnings is set, enable ENABLE_GCC_WARNINGS.
* examples/extexi: No longer issue synclines by default.
* examples/local.mk: Except if ENABLE_GCC_WARNINGS.
* examples/calc++/local.mk, examples/mfcalc/local.mk,
* examples/rpcalc/local.mk: Install the example files.
Reported by Alexandre Duret-Lutz as "second problem" in:
http://lists.gnu.org/archive/html/bug-bison/2013-09/msg00015.html
* bootstrap.conf: We need the "unlink" module.
* src/files.h, src/files.c (unlink_generated_sources): New.
* src/output.c: Use it.
* tests/output.at: Check the case of late errors.
* src/files.h, src/files.c (output_file_name_check): Take an additional
argument to record whether a file is a source or report file.
* src/files.c (generated_file): New.
(file_names, file_names_count): Replace with...
(generated_files, generated_files_size): these.
* src/scan-skel.l: Adjust.
Reported by Alexandre Duret-Lutz as "second problem" in:
http://lists.gnu.org/archive/html/bug-bison/2013-09/msg00015.html
One problem is that some errors are caught early, before the
generation of output files, while others can only be detected
afterwards (since, for instance, skeletons can raise errors
themselves).
This will be addressed in two steps: early errors do not generate
source files at all, while later errors will remove the files that
have already been generated.
* src/scan-skel.l (yyout): Open to /dev/null when there are errors.
* tests/output.at (AT_CHECK_FILES): Factored out of...
(AT_CHECK_OUTPUT): this.
Fuse the "SHELLIO" argument in the "FLAGS" one.
Use $5 to denote the expected exit status.
Add a test case for early errors.
When variant are enabled, the yylhs variable (the left-hand side of
the rule being reduced, i.e. $$ and @$) is explicitly destroyed when
YYERROR is called. This is because before running the user code, $$
is initialized, so that the user can properly use it.
However, when quitting yyparse, yylhs is also reclaimed by the C++
compiler: the variable goes out of scope.
Instead of trying to be too smart, let the compiler do its job: reduce
the scope of yylhs to exactly the reduction. This way, whatever the
type of scope exit (regular, exception, return, goto...) this variable
will be properly reclaimed.
Reported by Paolo Simone Gasparello.
<http://lists.gnu.org/archive/html/bug-bison/2013-10/msg00003.html>
* data/lalr1.cc (yyparse): Reduce the scope of yylhs.
* tests/c++.at: We now pass this test.
When variant are enabled, the yylhs variable (the left-hand side of
the rule being reduced, i.e. $$ and @$) is explicitly destroyed when
YYERROR is called. This is because before running the user code, $$
is initialized, so that the user can properly use it.
However, when quitting yyparse, yylhs is also reclaimed by the C++
compiler: the variable goes out of scope.
This was not detected by the test suite because (i) the Object tracker
was too weak, and (ii) the problem does not show when there is error
recovery.
Reported by Paolo Simone Gasparello.
<http://lists.gnu.org/archive/html/bug-bison/2013-10/msg00003.html>
* tests/c++.at (Exception safety): Improve the objects logger to make
sure that we never destroy twice an object.
Also track copy-constructors.
Use a set instead of a list.
Display the logs before running the function body, this is more
useful in case of failure.
Generalize to track with and without error recovery.
* data/glr.c (yyLRgotoState): Name the symbol argument yysym, instead
of yylhs.
* data/lalr1.cc (yy_lr_goto_state_): Likewise.
* data/lalr1.java (yy_lr_goto_state_): New, modeled after the previous
two routines.
Use it.
Apply the same rules for aver as for assert: no side effects,
especially not important ones.
* src/AnnotationList.c, src/muscle-tab.c: Adjust aver uses to resist
to -DNDEBUG.
Building C++ parsers with -Wsuggest-attribute=const and
-Wsuggest-attribute=noreturn triggers warning in generated code.
* data/lalr1.cc: Call b4_attribute_define.
(debug_stream, debug_level): Flag as pure.
* tests/headers.at (Several parsers): There are now more YY macros
that "leak".
Some tests now fail when compiled with G++ 4.3 or 4.4 on MacPorts.
* tests/local.at (AT_SKIP_IF_EXCEPTION_SUPPORT_IS_POOR): New.
* tests/c++.at (Exception safety): Use it.
* configure.ac (ENABLE_YACC): New conditional.
(YACC_SCRIPT, YACC_LIBRARY): Remove.
* lib/local.mk, src/local.mk: Use the former instead of the latter.
* doc/local.mk: Use ENABLE_YACC to avoid installing yacc.1.
Currently "-Werror -Wno-error=foo" still turns "foo" warnings into errors.
Reported by Alexandre Duret-Lutz.
See http://lists.gnu.org/archive/html/bug-bison/2013-09/msg00015.html.
* src/complain.c (errority, errority_flag): New.
(complain_init): Initialize the latter.
(warning_argmatch): Extract the loop iterating on the flag's bits.
Set and unset errority_flag here.
(warnings_argmatch): -Wno-error is not the same as -Wno-error=everything:
we must remember if category foo was explicitly turned in an error/warning
via -W(no-)error=foo.
(warning_severity): Use errority_flag.
* tests/input.at (Symbols): Just check --yacc, not -Wyacc, that's the
job of tests on -W.
(-Werror is not affected by -Wnone and -Wall): Rename as...
(-Werror combinations): this.
Tests more combinations of -W, -W(no-)error, and -W(no-)error=foo.
* tests/local.at (AT_BISON_CHECK_WARNINGS): Don't expect -Werror
to turn runs that issue warnings into runs with errors, as the
warnings might be enforced as warnings by -Wno-error=foo, in which
case -Werror does not change anything.
* doc/bison.texi (Bison Options): Try to be clearer about how
-W(no-)error and -W(no-)error=foo interact.
We don't ship the *.txt files that are used to build the info
file.
Reported by Colin Daley.
* doc/figs/example.txt: New.
* doc/local.mk (bison.info): Depend on the txt files.
And ship them.
* tests/c++.at (Exception safety): In variant mode $$ is an instance
of Object. Assigning YY_NULL in C++98 is incorrect, but behaves ok,
as it assigns YY_NULL=0 using Object::operator= (char v). It is wrong
in C++11 as there is operator for "$$ = nullptr".
Again some issues with the fact that yylval is reported by GCC as
possibly not initialized in some cases. Here, the case at hand is the
%destructor.
I am still not convinced that it is worth going all the trouble of
using pragmas to disable temporarily some warnings, instead of just
initializing the looking symbol once for all, but that's what Paul
voted for, see
<http://lists.gnu.org/archive/html/bison-patches/2012-10/msg00050.html>.
* data/c.m4 (b4_attribute_define): Define
YY_IGNORE_MAYBE_UNINITIALIZED_BEGIN, YY_IGNORE_MAYBE_UNINITIALIZED_END,
YY_INITIAL_VALUE here, as we will need them in the generation of the
destructor function, which is defined in yacc.c before yyparse, which
was in charge of defining these macros.
* data/yacc.c (b4_declare_scanner_communication_variables): Simplify:
trying to factor the definitions of the case pure and impure is
too complex.
Actually, it is not even clear that this macro should really exist,
as even the calls are complex.
Be careful not to issue a lone ";", as this is a statement, and C90
forbids declarations after statements ; so write
"YY_INITIAL_VALUE(Decl;)", not "YY_INITIAL_VALUE(Decl);".
There are possible conflicts between gnulib replacement functions (in
<stdio.h>) and their C++ wrappers (in <stream>). Trying to address
these in configure seems too hard, and I don't know how to fix the issue
in gnulib. Cowardly avoid the problem by skipping C++ tests when this
happens.
Reported by Stefano Lattarini.
http://lists.gnu.org/archive/html/bug-bison/2013-06/msg00001.html
* tests/atlocal.in (BISON_CXX_WORKS): Also set it to "skip" if we can't
compile a simple program using <stream>.
* tests/local.at: Comment changes.
* data/lalr1.java (parse): here, when in push-pull is in "both" mode.
This breaks the test suite, for instance
make check TESTSUITEFLAGS='-d 388 BISON_USE_PUSH_FOR_PULL=1'.
More generally make maintainer-push-check.
Using %code for config.h is wrong, as some headers will already have
been included by Bison. In some cases, e.g., glibc's string.h, this
results in some declaration not being made for lack of definition of
_GNU_SOURCE, which is performed by config.h.
* src/parse-gram.y: here.
Some directives cannot be used several times (e.g., a given symbol may
only have a single printer). In case of repeated definitions, an
error is issued for the second definition, yet it is not discarded,
and becomes the definition used for the rest of the file.
This is not consistent with the idea that multiple definitions are not
allowed: discard any repeated directive.
* src/symtab.c (symbol_type_set, symbol_code_props_set)
(semantic_type_code_props_set, symbol_class_set, symbol_translation):
Discard repeated directives.
* tests/input.at (Default %printer and %destructor redeclared)
(Per-type %printer and %destructor redeclared): Update expectations.
* src/symtab.c (symbol_precedence_set): Use prec_location, not
location (which is the first occurrence of the symbol, possibly just
%token).
Also, as redefinitions are not allowed, keep the first values, not
the subsequent ones.
* tests/conflicts.at, tests/existing.at, tests/regression.at: Adjust.
-Wsign-compare was disabled for bison's own code, following gnulib's
approach. However, the generated parsers should not trigger such
warnings.
Reported by Efi Fogel.
http://lists.gnu.org/archive/html/help-bison/2013-04/msg00018.html
See also http://stackoverflow.com/questions/16101062 for the weird
"-(unsigned)i" piece of code.
* configure.ac (warn_tests): Enable -Wsign-compare.
* data/location.cc (position::add_): New.
(position::lines, position::columns): Use it.
* tests/actions.at (AT_CHECK_PRINTER_AND_DESTRUCTOR): Fix signedness issues.
The purpose of this symbol was only to factor function calls. As a
result the actions were indeed simpler, but the grammar was somewhat
uselessly obfuscated. Get rid of this symbol, but introduce functions
to simplify dependencies.
There is no (intended) changes of behavior here.
* src/parse-gram.y (strip_braces, translate_code(
(translate_code_braceless): New.
(braceless): Remove, use "{...}" instead, and one of the previous
functions depending on the context.
(STRING, "%{...%}", EPILOGUE): Declare as <code>, instead of <chars>,
the difference between both is useless (well, I couldn't make sense of
it, even after having read the initial commit that introduced them).
(%union): Remove the now useless "chars" type.
Adjust the printers.
* src/scan-gram.l: Adjust.
Akim Demaille