Currently, in C, the default semantic action is implemented by being
always run before running the actual user semantic action. As a
consequence, when the user action is run, $$ is already set as $1.
In C++ with variants, we don't do that, since we cannot manipulate the
semantic value without knowing its exact type. When variants are
enabled, the only guarantee is that $$ is default contructed and ready
to the used.
Some users still would like the default action to be run with
variants. Frank Heckenbach's parser in
C++17 (http://lists.gnu.org/archive/html/bug-bison/2018-04/msg00011.html)
provides this feature, but relying on std::variant's dynamic typing,
which we forbid in lalr1.cc.
The simplest seems to be actually generating the default semantic
action (in all languages/skeletons). This makes the pre-action (that
sets $$ to $1) useless. But... maybe some users depend on this, in
spite of the comments that clearly warn againt this. So let's not
turn this off just yet.
* src/reader.c (grammar_rule_check_and_complete): Rename as...
(grammar_rule_check_and_complete): this.
Install the default semantic action when applicable.
* examples/variant-11.yy, examples/variant.yy, tests/calc.at:
Exercise the default semantic action, even with variants.
The code was already using midrule only, never mid_rule. This is
simpler to remember, and matches a similar change we made from
look-ahead to lookahead.
* NEWS, doc/bison.texi, src/reader.c, src/scan-code.h, src/scan-code.l
* tests/actions.at, tests/c++.at, tests/existing.at: here.
Suggested by Paul Eggert.
* src/reader.c (find_start_symbol): Don't check 'res', we know it is
not null. That suffices to avoid the GCC warnings.
* bootstrap.conf: We don't need 'assume', which doesn't exist anyway.
Commit 3df32101e7 introduced invalid C
code. Caught by GCC 7.3.0.
* bootstrap.conf (gnulib_modules): We need assume.
* src/reader.c (find_start_symbol): Fix the signature (too much C++,
sorry...).
Prefer 'assume' to 'assert', so that we don't have these warnings even
when NDEBUG is defined.
Make sure that we cannot apply a type to the (main) action of a rule.
* src/reader.c (grammar_rule_check): Issue the warning.
* tests/input.at (Cannot type action): Check the warning.
Prompted on Piotr Marcińczyk's message:
http://lists.gnu.org/archive/html/bug-bison/2017-06/msg00000.html.
See also http://lists.gnu.org/archive/html/bug-bison/2018-06/msg00001.html.
Because their type is unknown to Bison, the values of midrule actions are
not treated like the others: they don't have %printer and %destructor
support. In addition, in C++, (Bison) variants cannot work properly.
Typed midrule actions address these issues. Instead of:
exp: { $<ival>$ = 1; } { $<ival>$ = 2; } { $$ = $<ival>1 + $<ival>2; }
write:
exp: <ival>{ $$ = 1; } <ival>{ $$ = 2; } { $$ = $1 + $2; }
* src/scan-code.h, src/scan-code.l (code_props): Add a `type` field to
record the declared type of an action.
(code_props_rule_action_init): Add a type argument.
* src/parse-gram.y: Accept an optional type tag for actions.
* src/reader.h, src/reader.c (grammar_current_rule_action_append): Add
a type argument.
(grammar_midrule_action): When a mid-rule is typed, pass its type to
the defined dummy non terminal symbol.
grammar_current_rule_action_append was used in two different places:
for actual action (`{...}`), and for predicates (`%?{...}`). Let's
split this in two different functions.
* src/reader.h, src/reader.c (grammar_current_rule_predicate_append): New.
Extracted from...
(grammar_current_rule_action_append): here.
Remove arguments that don't apply.
Adjust dependencies.
* origin/maint:
build: don't try to generate docs when cross-compiling
package: fix a reporter's name
%union: fix the support for named %union
package: bump to 2015
flex: don't trust YY_USER_INIT
yacc.c: fix broken union when api.value.type=union and %defines are used
doc: fix missing xref
gnulib: update
location: remove some ugly debugging code traces
build: use abort to pacify compiler errors
package: bump to 2014
doc: specify documentation encoding
Rather than having duplicate info in the symbol and the alias that has
to be resolved later on, both the symbol and the alias have a common
pointer to a separate structure containing this info.
* src/symtab.h (sym_content): New structure.
* src/symtab.c (sym_content_new, sym_content_free, symbol_free): New
* src/AnnotationList.c, src/conflicts.c, src/gram.c, src/gram.h,
* src/graphviz.c, src/ielr.c, src/output.c, src/parse-gram.y, src/print.c
* src/print-xml.c, src/print_graph.c, src/reader.c, src/reduce.c,
* src/state.h, src/symlist.c, src/symtab.c, src/symtab.h, src/tables.c:
Adjust.
* tests/input.at: Fix expectations (order changes).
When reporting a duplicate directive on a rule, point to its first
occurrence:
one.y:11.10-15: error: only one %empty allowed per rule
%empty {} %empty
^^^^^^
one.y:11.3-8: previous declaration
%empty {} %empty
^^^^^^
And consistently discard the second one.
* src/complain.h, src/complain.c (duplicate_directive): New.
* src/reader.c: Use it where appropriate.
* src/symlist.h, src/symlist.c (symbol_list): Add a dprec_location member.
* tests/actions.at: Adjust expected output.
* src/complain.h, src/complain.c (warning_is_unset): New.
* src/reader.c (grammar_current_rule_empty_set): If enabled -Wempty-rule,
if not disabled.
* tests/actions.at (Implicitly empty rule): Check this feature.
Also check that -Wno-empty-rule does disable this warning.
Provide a means to explicitly denote empty right-hand sides of rules:
instead of
exp: { ... }
allow
exp: %empty { ... }
Make sure that %empty is properly used.
With help from Joel E. Denny and Gabriel Rassoul.
http://lists.gnu.org/archive/html/bison-patches/2013-01/msg00142.html
* src/reader.h, src/reader.c (grammar_current_rule_empty_set): New.
* src/parse-gram.y (%empty): New token.
Use it.
* src/scan-gram.l (%empty): Scan it.
* src/reader.c (grammar_rule_check): Check that %empty is properly used.
* tests/actions.at (Invalid uses of %empty, Valid uses of %empty): New.
* src/complain.c: Space changes.
* src/reader.c: Comment changes.
Avoid && in assertions.
* src/location.c: Move comments to...
* src/location.h: here.
* src/symlist.h, src/symlist.c: Create a pseudo section for members
that apply to the rule.
* src/reader.c (packgram): Improve readability.
The parser calls grammar_current_rule_end at the end of every rhs,
which adds a NULL to separate the rules. So there is no need to
check whether "p" is non-null before proceeding.
* origin/maint:
news: prepare for forthcoming release
doc: explain how mid-rule actions are translated
error: use better locations for unused midrule values
doc: various minor improvements and fixes
tests: ignore more useless compiler warnings
tests: be robust to C being compiled with a C++11 compiler
build: beware of Clang++ not supporting POSIXLY_CORRECT
maint: post-release administrivia
version 2.6.90
build: fix syntax-check error.
cpp: simplify the Flex version checking macro
news: improve the carets example and fix a typo
cpp: improve the Flex version checking macro
carets: improve the code
maint: update news
build: keep -Wmissing-declarations and -Wmissing-prototypes for modern GCCs
build: drop -Wcast-qual
gnulib: update
Conflicts:
NEWS
doc/Makefile.am
doc/bison.texi
gnulib
src/reader.c
tests/actions.at
tests/atlocal.in
tests/input.at
On
%%
exp: {;} {$$;} { $$ = $1; }
instead of reporting (with -fcaret -Wmidrule-value)
midrule.y:2.6-8: warning: unset value: $$ [-Wmidrule-values]
exp: {;} {$$;} { $$ = $1; }
^^^
midrule.y:2.6-27: warning: unused value: $2 [-Wmidrule-values]
exp: {;} {$$;} { $$ = $1; }
^^^^^^^^^^^^^^^^^^^^^^
report
midrule.y:2.6-8: warning: unset value: $$
exp: {;} {$$;} { $$ = $1; }
^^^
midrule.y:2.10-14: warning: unused value: $2
exp: {;} {$$;} { $$ = $1; }
^^^^^
* src/reader.c (grammar_rule_check): When warning about the value of a
midrule action, use the location of the midrule action instead of the
location of the rule.
the location of the part of the rule.
* tests/actions.at (Default %printer and %destructor for mid-rule values):
Adjust expectations
* tests/input.at (Unused values with default %destructor): Ditto.
(AT_CHECK_UNUSED_VALUES): Ditto.
And use -fcaret.
Treating warnings as errors caused Bison to exit earlier than needed, making it
hide warnings that would have been printed had -Werror not been set.
Also, fix a bug that caused some context information of errors to not be
shown.
* src/complain.c (complaint_issued): Rename as...
(complaint_status): This, and change its type from boolean to
* src/complain.h (err_status): This, new enumeration.
* src/main.c (main): Adjust (only finish early if an actual complaint was
risen, not a mere warning treated an error).
* src/reader.c: Adjust.
* src/complain.c (complain_at_indent): Rename as...
(complaint_indent): This, and take the location as a pointer.
* src/complain.h, src/muscle-tab.c, src/reader.c, src/scan-code.l,
src/symtab.c: Adjust.
Signed-off-by: Akim Demaille <akim@lrde.epita.fr>
This used to be the format of the error report:
input.y:6.5-10: result type clash on merge function 'merge': [...]
input.y:2.4-9: previous declaration
In order to distinguish the actual error from the context provided, we
rather this new output:
input.y:6.5-10: result type clash on merge function 'merge': [...]
input.y:2.4-9: previous declaration
Another patch will introduce an "error: " prefix to all non-indented
lines, giving yet better readability to the reports.
* src/complain.h (SUB_INDENT): Move to here.
* src/reader.c (record_merge_function_type): Use complain_at_indent to
output with increased indentation level.
* src/scan-code.l (SUB_INDENT): Remove from here.
* tests/glr-regression.at: Apply this change.
Signed-off-by: Akim Demaille <akim@lrde.epita.fr>
There is too much code duplication between %printer and %destructor.
We used to have two functions for each action: the first one for
destructors, the second one for printers. Factor using a
'code_props_type', and an array of code_props instead of two
members.
* src/symlist.h, src/symlist.c (symbol_list_destructor_set)
(symbol_list_printer_set): Fuse into...
(symbol_list_code_props_set): this.
* src/symtab.h, src/symtab.c (default_tagged_destructor)
(default_tagged_printer): Fuse into...
(default_tagged_code_props): this.
(default_tagless_destructor, default_tagless_printer)
(default_tagless_code_props): Likewise.
(code_props_type_string): new.
(symbol_destructor_set, symbol_destructor_get, semantic_type_destructor_set)
(default_tagged_destructor_set, default_tagless_destructor_set)
(symbol_printer_set, symbol_printer_get, semantic_type_printer_set)
(default_tagged_printer_set, default_tagless_printer_set): Replace by...
(symbol_code_props_set, symbol_code_props_get, semantic_type_code_props_set)
(default_tagged_code_props_set, default_tagless_code_props_set): these.
* src/parse-gram.y (grammar_declaration): Adjust.
* src/output.c (CODE_PROP, grammar_declaration): Ditto.
* src/reader.c (symbol_should_be_used): Ditto.
The previous commit, which turns into a warning what used to be an
error:
%printer {} foo;
%%
exp: '0';
has two shortcomings: the warning is way too long (foo is reported
to be useless later), and besides, it also turns into a warning much
more serious errors:
%printer {} foo;
%%
exp: foo;
Reduce the amount to warnings in the first case, restore the error in
the second.
* src/symtab.h (status): Add a new inital state: undeclared.
* src/symtab.c (symbol_new): Initialize to undeclared.
(symbol_class_set): Simplify the logic of the code that neutralize
the "redeclared" warning after the "redefined" one.
(symbol_check_defined): "undeclared" is also an error.
* src/reader.c (grammar_current_rule_symbol_append): Symbols appearing
in a rule are "needed".
* src/symlist.c (symbol_list_destructor_set, symbol_list_printer_set):
An unknown symbol appearing in a %printer/%destructor is "used".
* src/reduce.c (nonterminals_reduce): Do not report as "useless" symbols
that are not used (e.g., those that for instance appeared only in a
%printer).
* tests/input.at (Undeclared symbols used for a printer or destructor):
Improve the cover the cases described above.
We used to raise an error if a symbol appears only in a %printer or
%destructor. Make it a warning.
* src/symtab.h (status): New enum.
(symbol): Replace the binary "declared" with the three-state "status".
Adjust dependencies.
* src/symtab.c (symbol_check_defined): Needed symbols are an error,
whereas "used" are simply warnings.
* src/symlist.c (symbol_list_destructor_set, symbol_list_printer): Set
symbol status to 'used' when associated to destructors or printers.
* input.at (Undeclared symbols used for a printer or destructor): New.
* doc/bison.texinfo: Space change.
* src/system.h (STREQ, STRNEQ): New.
* src/files.c, src/ielr.c, src/lalr.c, src/muscle-tab.c,
* src/output.c, src/print.c, src/print_graph.c,
* src/reader.c, src/scan-skel.l, src/tables.c,
* src/uniqstr.c:
Use them.
* src/scan-gram.l: Do not use streq.h, use system.h's STREQ.
* cfg.mk: The documentation is an exception.
See <http://lists.gnu.org/archive/html/bug-bison/2012-01/msg00120.html>.
Use quotearg as often as possible instead of leaving the choice of
the quotes to the translators. Use shorter messages. Factor similar
messages to a single format, to make localization easier.
* src/files.c, src/getargs.c, src/muscle-tab.c, src/reader.c
* src/scan-code.l, src/scan-gram.l, src/symtab.c:
Use quote() or quotearg_colon() on printf arguments instead of
quotes in the format string.
* data/bison.m4: Keep sync with the changes in muscle-tab.c.
* tests/skeletons.at, tests/input.at, tests/regression.at: Adjust
expected messages.
(cherry picked from commit 4a9cd8f24a)
Conflicts:
src/muscle-tab.c
See <http://lists.gnu.org/archive/html/bug-bison/2012-01/msg00120.html>.
Use quotearg as often as possible instead of leaving the choice of
the quotes to the translators. Use shorter messages. Factor similar
messages to a single format, to make localization easier.
* src/files.c, src/getargs.c, src/muscle-tab.c, src/reader.c
* src/scan-code.l, src/scan-gram.l, src/symtab.c:
Use quote() or quotearg_colon() on printf arguments instead of
quotes in the format string.
* data/bison.m4: Keep sync with the changes in muscle-tab.c.
* tests/skeletons.at, tests/input.at, tests/regression.at: Adjust
expected messages.
