Bruno Haible just added a default implementation of libtextstyle's
interface when the library is not available.
https://lists.gnu.org/archive/html/bison-patches/2019-03/msg00025.html
* gnulib: Update.
* bootstrap.conf: Replace libtextstyle with libtextstyle-optional.
* src/complain.c, src/getargs.c: Remove now useless cpp guards.
The timevar and bitset modules now use the c99 module which causes
$CXX to now include -std=gnu++11 when possible. Unfortunately, G++
4.7 does not implement [[noreturn]] in C++11 mode, so our tests of
glr.cc (which uses _Noreturn) fail with
input.cc:954:1: error: expected unqualified-id before '[' token
right before [[noreturn]]. 4.8 works fine.
* data/skeletons/c.m4 (b4_attribute_define): Do not use [[noreturn]]
with GCC 4.7.
See f8408562f8.
* tests/calc.at: Stop imitating the C API.
Prepare more tests to run in the future.
%verbose works as expected (what a surprise, it's unrelated to the
skeleton...).
The variable spec_defines_file denotes the name of the generated
header. Its name is derived from --defines/%defines, whose name in
turn is derived from the fact that the header, in Yacc, contained the
Not only does the header now contain a lot more than just the token
definitions, but we no longer even generate macros, but an enum...
Let's modernize our vocabulary.
* src/files.h, src/files.c (spec_defines_file): Rename as...
(spec_header_file): this.
Currently when --defines is used, we generate a header, and paste an
exact copy of it into the generated parser implementation file. Let's
provide a means to #include it instead.
We don't do it by default because of the Autotools' ylwrap. This
program wraps invocations of yacc (that uses a fixed output name:
y.tab.c, y.tab.h, y.output) to support a more modern naming
scheme (dir/foo.y -> dir/foo.tab.c, dir/foo.tab.h, etc.). It does
that by renaming the generated files, and then by running sed to
propagate these renamings inside the files themselves.
Unfortunately Automake's Makefiles uses Bison as if it were Yacc (with
--yacc or with -o y.tab.c) and invoke bison via ylwrap. As a
consequence, as far as Bison is concerned, the output files are
y.tab.c and y.tab.h, so it emits '#include "y.tab.h"'. So far, so
good. But now ylwrap processes this '#include "y.tab.h"' into
'#include "dir/foo.tab.h"', which is not guaranteed to always work.
So, let's do the Right Thing when the output file is not y.tab.c, in
which case the user should %define api.header.include. Binding this
behavior to --yacc is tempting, but we recently told people to stop
using --yacc (as it also enables the Yacc warnings), but rather to use
-o y.tab.c.
Yacc.c is the only skeleton concerned: all the others do include their
header.
* data/skeletons/yacc.c (b4_header_include_if): New.
(api.header.include): Provide a default value when the output is not
y.tab.c.
* src/parse-gram.y (api.header.include): Define.
Running with CC='gcc-mp-8 -fsanitize=undefined' revealed Undefined
Behaviors.
https://lists.gnu.org/archive/html/bison-patches/2019-03/msg00008.html
* src/state.c (errs_new): Don't call memcpy with NULL as source.
* src/location.c (add_column_width): Don't assume that the column
argument is nonnegative: the scanner sometimes "backtracks" (e.g., see
ROLLBACK_CURRENT_TOKEN and DEPRECATED) in which case we can have
negative column numbers (temporarily).
Found in test 3 (Invalid inputs).
* data/skeletons/yacc.c: here.
This is more logical for the time stamps, but it's also required by
following patches: the shared declarations are also in charge of
handling api.value.type=union. So far, they are run in the
implementation file in both cases (with or without header). But if we
run them only in the header, then the implementation file is emited
with incorrect support for api.value.type=union.
Arguably we should not have such dependencies. This is because we
have side-effects in our backend (redefining the symbols' type and
type_tag). In the future we should find a better solution for this,
without sacrificing the independence of the backend from bison
itself (i.e., I don't think we should handle api.value.type=union in
bison, leave it to m4).
Currently we generate things like:
#line 683 "src/parse-gram.y" /* yacc.c:316 */
The first part is of course very important: compilers point the users
to their grammar file rather than into the generated parser. The
second part points to the place in the skeletons that generated this
piece of code.
This dependency on the Bison skeletons generates lots of useless 'git
diff'. This location is useless for the regular user (who does not
care about the skeletons) and is actually not useful for Bison
developpers too (I never used this to locate the code in skeletons
that generated output). So disable it completely. If someone thinks
this was actually useful, a %define variable should be provided to
control the level of verbosity of '#line', in replacement of
--no-lines.
So now, generate:
#line 683 "src/parse-gram.y"
* data/skeletons/bison.m4 (b4_sync_end): Emit nothing.
* src/lalr.c (follows_print): Just print the symbol tag.
Take and print a title.
Indent the output.
Use it to print the various steps of the computation.
(lookahead_tokens_print): Fix a lie: the number displayed is not the
number of tokens.
Don't display states that don't even have reductions.
* src/relation.h, src/relation.c (relation_print): Accept and use a
title.
Don't print empty rows.
Indent the output.
Adjust dependencies.
* src/lalr.c (initialize_goto_follows): Print 'reads' in traces.
* configure.ac (DCFLAGS): Define.
* tests/atlocal.in: Receive it.
* data/skeletons/d.m4 (api.parser.class): Remove spurious YY.
* data/skeletons/lalr1.d (yylex): Return an int instead of a
YYTokenType, so that we can use characters as tokens.
* examples/d/calc.y: Adjust.
* tests/local.at: Initial support for D.
(AT_D_IF, AT_DATA_GRAMMAR(D), AT_YYERROR_DECLARE(d))
(AT_YYERROR_DECLARE_EXTERN(d), AT_YYERROR_DEFINE(d))
(AT_MAIN_DEFINE(d), AT_COMPILE_D, AT_LANG_COMPILE(d), AT_LANG_EXT(d)):
New.
* tests/calc.at: Initial support for D.
* tests/headers.at
* examples/d/calc.y: Exit with failure on errors.
Remove useless operators (=, !) meant for the test suite.
Add unary + for symmetry.
* examples/d/calc.test: Adjust expectations.
Commit db34f79889 renames the variable F
as goto_follows, but forgot to rename this function.
* src/lalr.c (initialize_F): Rename as...
(initialize_goto_follows): this.
I need to be able to read includes and goto_follows.
* src/relation.h, src/relation.c (relation_print): Provide a means to
pretty-print the nodes of the relation.
* src/lalr.c (goto_print, follows_print): New.
(set_goto_map): Use goto_print.
(build_relations): Show INCLUDES.
(compute_FOLLOWS): Rename as...
(compute_follows): this.
Show FOLLOWS.
* src/lalr.c (build_relations): Reduce the scopes.
Instead of keeping rp alive in two different loops, clarify the second
one by having an index on the path we traverse (i.e., use that index
to compute the source state _and_ the symbol that labels the
transition).
This allows to turn an obscure 'while'-loop in a clearer (IMHO)
'for'-loop. We also consume more variables (by introducing p instead
of making more side effects on length), but we're in 2019, I don't
think this matters. What does matter is that (IMHO again), this is
now clearer.
Also, use clearer names.
* tests/local.at (AT_MAIN_DEFINE(java)): Exit failure on failure.
(AT_PARSER_CHECK): If in Java, run AT_JAVA_PARSER_CHECK.
* tests/conflicts.at (AT_CONSISTENT_ERRORS_CHECK): Simplify.
Currently the caller must specify the ./ prefix to its command. Let's
avoid that: it will be nicer to read, make it easier to have a version
that works for Java and C/C++.
* tests/local.at (AT_PARSER_CHECK): Prefix the command with ./.
Adjust callers.
* tests/conflicts.at (AT_YYLEX_PROTOTYPE): Don't define it, leave that
task to AT_DATA_GRAMMAR.
But be honest: tell AT_BISON_OPTION_PUSHDEFS all the options we use.
Reported by Hans Åberg.
https://lists.gnu.org/archive/html/help-bison/2019-02/msg00064.html
* src/files.c (spec_graph_file): Use `*.gv` when 3.4 or better,
otherwise `*.dot`.
* src/parse-gram.y (handle_require): Pretend we are already 3.4.
* doc/bison.texi: Adjust.
* tests/local.at, tests/output.at: Exercise this.
Currently, we have errors:
./doc/bison.texi:13005: node `History' lacks menu item for `Yacc' despite being its Up target
* doc/bison.texi (History): Add the local menu.
While at it, add a few index entries.
* examples/c/lexcalc/parse.y: Formatting/comment changes.
(line): Don't return a value.
Print the result here, which avoids printing a value for lines with an
error.
(yyerror): Be sure to increment the pointed, not the pointer...
* examples/c/lexcalc/lexcalc.test: Check errors.
* examples/c/lexcalc/local.mk: Fix a dependency.
While hacking on the computation of the automaton, I had yystate being
equal to -1, and the parser loops. Let's catch this when
parser.assert is enabled.
* data/skeletons/yacc.c (YY_ASSERT): New.
Use it.
Not using the name YYASSERT, to make it clear that this is private.
glr.c should probably move to YY_ASSERT too.
Also, while at it, report 'Entering state...' even before growing the
stacks.
* examples/c/calc/local.mk, examples/c/lexcalc/local.mk,
* examples/c/mfcalc/local.mk, examples/c/rpcalc/local.mk:
Regenerate the files if dependencies have changed.
I don't think this style:
/* If buffer is full, make it bigger. */
if (i == length)
{
length *= 2;
symbuf = (char *) realloc (symbuf, length + 1);
}
/* Add this character to the buffer. */
symbuf[i++] = c;
/* Get another character. */
c = getchar ();
or more readable than
/* If buffer is full, make it bigger. */
if (i == length)
{
length *= 2;
symbuf = (char *) realloc (symbuf, length + 1);
}
/* Add this character to the buffer. */
symbuf[i++] = c;
/* Get another character. */
c = getchar ();
Actually, I think the latter is more readable, and helps with width
issues in the PDF. As a matter of fact, I would happily move to //
only for single line comments.
* doc/bison.texi: A single space before closing comments.
* doc/bison.texi: Prefer 'fun' to 'fnct'.
Reduce local variable scopes.
Prefer strdup to malloc + strcpy.
Avoid gratuitous casts.
Use simpler names (e.g., 'name' instead of 'fname').
Avoid uses of 0 for NULL.
Avoid using NULL when possible (e.g., 'p' instead of 'p != NULL').
Prefer union names to casts (e.g. 'yylval.VAR = s' instead of
'*((symrec**) &yylval) = s').
Give arguments a name in fun declarations.
Use our typedefs instead of duplicating them (func_t).
Stop promoting an explicit $$ = $1;, it should be implicit (Bison
might be able to eliminate useless chain rules).
Help a bit Texinfo by making smaller groups.
Rely on the C compiler to call function pointers (prefer
'$1->value.fun ($3)' to (*($1->value.fnctptr))($3)').
Currently we have no simple example: rpcalc in reverse Polish, mfcalc
has functions, and lexcalc is using lex.
* examples/c/calc/Makefile, examples/c/calc/calc.y,
* examples/c/calc/calc.test, examples/c/calc/local.mk: New.
* examples/extexi: Since we issue #lines only at the beginning of
@example, leave empty line when removing content (such as @comment
lines), otherwise the lines that follow have incorrect source line
location. This leaves ugly empty lines, but they are removed when you
tidy the output for the end user: sequences of \n are mapped to at
most two sucessive \n.
The format is inconsistent. For instance most sections are
indented (including "Terminals unused in grammar" for instance), but
the sections "Terminals, with rules where they appear" and
"Nonterminals, with rules where they appear" are not. Let's indent
them. Also, these two sections try to wrap the output to avoid lines
too long. Yet we don't do that in the rest of the file, for instance
when listing the lookaheads of an item.
For instance in the case of Bison's parse-gram.output we go from:
Terminals, with rules where they appear
"end of file" (0) 0
error (256) 28 88
"string" <char*> (258) 9 13 16 17 20 23 24 109 116
[...]
Nonterminals, with rules where they appear
$accept (58)
on left: 0
input (59)
on left: 1, on right: 0
prologue_declarations (60)
on left: 2 3, on right: 1 3
prologue_declaration (61)
on left: 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 22 23 24
25 26 27 28 29, on right: 3
[...]
to
Terminals, with rules where they appear
"end of file" (0) 0
error (256) 28 88
"string" <char*> (258) 9 13 16 17 20 23 24 109 116
[...]
Nonterminals, with rules where they appear
$accept (58)
on left: 0
input (59)
on left: 1
on right: 0
prologue_declarations (60)
on left: 2 3
on right: 1 3
prologue_declaration (61)
on left: 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 22 23 24 25 26 27 28 29
on right: 3
[...]
* src/print.c (END_TEST): Remove.
(print_terminal_symbols): Don't try to wrap the output.
(print_nonterminal_symbols): Likewise.
Make two different lines for occurrences on the left, and occurrence
on the rhs of the rules.
Indent by 4 and 8, not 3.
* src/reduce.c (reduce_output): Indent by 4, not 3.
* tests/conflicts.at, tests/existing.at, tests/reduce.at,
* tests/regression.at, tests/report.at:
Adjust.
This should not be used to generate parsers. My point is actually to
facilitate debugging (when tweaking the generation of the LR(0)
automaton for instance, not carying -yet- about lookaheads).
* src/reader.c (prepare_percent_define_front_end_variables): Add lr(0).
* src/conflicts.c (set_conflicts): Be robust to reds not having
lookaheads at all.
* src/ielr.c (LrType, lr_type_get): Adjust.
(ielr): Implement support for LR(0).
* src/lalr.c (lalr_free): Don't free LA when it's not computed.
* maint:
maint: post-release administrivia
version 3.3.2
style: minor fixes
NEWS: named constructors are preferable to symbol_type ctors
gram: fix handling of nterms in actions when some are unused
style: rename local variable
CI: update the ICC serial number for travis-ci.org
Since Bison 3.3, semantic values in rule actions (i.e., '$...') are
passed to the m4 backend as the symbol number. Unfortunately, when
there are unused symbols, the symbols are renumbered _after_ the
numbers were used in the rule actions. As a result, the evaluation of
the skeleton failed because it used non existing symbol numbers.
Which is the happy scenario: we could use numbers of other existing
symbols...
Reported by Balázs Scheidler.
http://lists.gnu.org/archive/html/bug-bison/2019-01/msg00044.html
Translating the rule actions after the symbol renumbering moves too
many parts in bison. Relying on the symbol identifiers is more
troublesome than it might first seem: some don't have an
identifier (tokens with only a literal string), some might have a
complex one (tokens with a literal string with characters special for
M4). Well, these are tokens, but nterms also have issues: "dummy"
nterms (for midrule actions) are named $@32 etc. which is risky for
M4.
Instead, let's simply give M4 the mapping between the old numbers and
the new ones. To avoid confusion between old and new numbers, always
emit pre-renumbering numbers as "orig NUM".
* data/README: Give details about "orig NUM".
* data/skeletons/bison.m4 (__b4_symbol, _b4_symbol): Resolve the
"orig NUM".
* src/output.c (prepare_symbol_definitions): Pass nterm_map to m4.
* src/reduce.h, src/reduce.c (nterm_map): Extract it from
nonterminals_reduce, to make it public.
(reduce_free): Free it.
* src/scan-code.l (handle_action_dollar): When referring to a nterm,
use "orig NUM".
* tests/reduce.at (Useless Parts): New, based Balázs Scheidler's
report.
A rule is a useless chain iff it's a chain (aka unit, or injection)
rule (i.e., the RHS has length 1), and it's useless (it has no used
defined semantic action).
* src/gram.h, src/gram.c (rule_useless_chain_p): New.
(grammar_dump): Report useless chain rules.
* tests/sets.at: Check the traces.
There are several places where we need to recover the rule from an
item, let's factor that into item_rule. We also want to print items
in a nice way: we do it when generating the *output file, but it is
also useful in debug messages.
* src/gram.h, src/gram.c (item_rule, item_print): New.
* src/print.c (print_core): Use them.
* src/state.h, src/state.c: Propagate constness.
The memory allocated by 'closure' (and some data such as 'fderives')
is used to computed a state's full itemset from its core. This is
needed during the construction of the LR(0) automaton, and the memory
is reclaimed immediately afterwards.
Unfortunately the reports (graph, text, xml) also need this
information when describing the states with their full itemsets. As a
consequence the memory was allocated again, fderives computed again
too, and more --trace reports are generated which only duplicate what
was already reported.
Stop that. It does mean that we release the memory later (hence the
peak memory usage is higher now), but I don't think that's a problem
today.
* src/lr0.c (generate_states): Don't call closure_free.
* src/state.c (states_free): Do it here.
(for symmetry with closure_new which is called in generate_states).
* src/print-graph.c, src/print-xml.c, src/print.c: You can now expect
the closure module to be functional.
This is more consistent with the other files.
* closure.h, closure.c (new_closure, free_closure): Rename as...
(closure_new, closure_free): this.
Adjust dependencies.
This will make it easier to add new elements (that might already be
part of shift_symbol) without having to worry about the size of
shift_symbol (which is currently a fixed size vector).
I could not measure any significant differences in performances in the
generation of LR(0) automaton (benched on gramamrs of Ruby, C, and C++).
* src/lr0.c (shift_symbol): Make it a bitset.
When debugging Bison itself, this is very handy, especially when
tweaking the frontend badly enough to break the backends. It can also
be used to check a grammar.
* src/getargs.h, src/getargs.c (feature_syntax_only): New.
(feature_args, feature_types): Adjust.
* src/main.c (main): Use it.
* src/symtab.h (status): Rename as...
(declaration_status): this, to avoid colliding with status, the
argument of 'usage'.
'status' seems a tad too general to be used only here.
Akim Demaille