Zefram [Tue, 19 Oct 2010 20:16:11 +0000 (21:16 +0100)]
refactor and regularise label/statement grammar
Refactoring of the grammar around statements. New production <barestmt>
encompasses a statement without label. It includes all statement types,
including declarations, with no unnecessary intermediate non-terminals.
It generates an op tree for the statement's content, with no leading
state op. The <fullstmt> production has just one rule, consisting of
optional label followed by <barestmt>. It puts a state op on the front
of the statement's content ops.
To support the regular statement op structure, the op sequence for for(;;)
loops no longer has a second state op between the initialisation and
the loop. Instead, the unstack op type is slightly adapted to achieve
the stack clearing without a state op.
The newFOROP() constructor function no longer generates a state op,
that now being the job of the <fullstmt> production. Consequently it
no longer takes a parameter stating what label is to go in the state op.
This brings it in line with the other op constructors.
Nicholas Clark [Mon, 25 Oct 2010 19:21:50 +0000 (20:21 +0100)]
Merge Hash::Util::FieldHash::_test_uvar_{get,set,same} using ALIAS.
On this platform it reduces the object code size by over 1K.
Ideally these test functions would not be in the installed code, but I can't
see an easy way to avoid this.
Nicholas Clark [Mon, 25 Oct 2010 19:06:36 +0000 (20:06 +0100)]
In Hash::Util::FieldHash::_test_uvar_set, the second SvRV(sverf) is not needed.
Father Chrysostomos [Mon, 25 Oct 2010 17:52:42 +0000 (10:52 -0700)]
[perl #77688] tie $scalar can tie a handle
Make tie() distinguish between $scalar and *$scalar when $scalar holds
a glob copy, by checking the FAKE flag.
Father Chrysostomos [Mon, 25 Oct 2010 17:04:29 +0000 (10:04 -0700)]
Tests for [perl #77508] List assignment to fake globs fails
This was fixed by
2acc3314e31a9.
Father Chrysostomos [Mon, 25 Oct 2010 17:00:47 +0000 (10:00 -0700)]
[perl #77498] Assignment ignores magick when the RHS holds a glob
Jan Dubois [Mon, 25 Oct 2010 17:23:04 +0000 (10:23 -0700)]
For PROXYSUBS, tables may sometimes have no entries.
The reason being that all entries in the tables are guarded
with #ifdef, so if none of the symbols are defined, then we
still end up with an empty table. This actually happens in
the POSIX module on Windows for the "yes" table.
This reverts commit
c553cca0 and changes based on it.
Nicholas Clark [Mon, 25 Oct 2010 13:53:19 +0000 (14:53 +0100)]
In FastCalc.xs, use SV * as the return type for _new() and _zero()
Previously these were using AV *. Whilst the code always was managing
references correctly, to work around the deficiency of the standard typemap
entry for AV * it had to mortalise the array it created, meaning that it was
creating 2 entries per call on the mortals stack, not 1.
Tony Cook [Mon, 25 Oct 2010 11:59:59 +0000 (22:59 +1100)]
rt #72398 - get magic before downgrading in SvPVbyte()
Niko Tyni [Mon, 25 Oct 2010 11:37:49 +0000 (22:37 +1100)]
TODO test: SvPVbyte should handle get magic before checking the utf8 flag
When $1 had the utf8 flag set from a previous match, SvPVbyte
may croak with 'Wide character in subroutine entry' before
resetting the flag to its new value.
Add a support function and a TODO test for this in XS-APItest.
http://bugs.debian.org/376329
Nicholas Clark [Mon, 25 Oct 2010 09:09:09 +0000 (10:09 +0100)]
Convert Math::BigInt::FastCalc to XSLoader.
As the minimum perl version is 5.006, enable 'use warnings'.
Nicholas Clark [Mon, 25 Oct 2010 08:53:44 +0000 (09:53 +0100)]
In FastCalc.xs, inline the macros CONSTANT_OBJ and RETURN_MORTAL_BOOL.
With the previous commit, both are now only used in one place.
No need to use sv_2mortal() on the reset of boolSV(), as both PL_sv_no and
PL_sv_yes are immortals. [And special-cased within the implementation of
sv_2mortal() - not only is it a no-op, it's a non-free no-op :-)]
Nicholas Clark [Mon, 25 Oct 2010 08:24:06 +0000 (09:24 +0100)]
Reduce repetition in FastCalc.xs by using ALIASes.
_one, _two and _ten are aliases to _zero
_is_odd is an alias of _is_even
_is_one, _is_two, _is_ten are aliases of _is_zero
On this system this reduces the object code size by about 4.5K (about 20%).
Nicholas Clark [Mon, 25 Oct 2010 07:54:07 +0000 (08:54 +0100)]
Move some GV tests to gv.t from glob.t - the latter is for testing glob()
They were incorrectly added to glob.t by
010be86bae505fbe.
Nicholas Clark [Mon, 25 Oct 2010 07:51:55 +0000 (08:51 +0100)]
Stop sysio.t changing into t/op, and hence creating a temporary file there.
Creating a file there causes a race condition with (at least) glob.t, which
lists the contents of t/op by various means, and compares the results for
consistency.
Nicholas Clark [Mon, 25 Oct 2010 07:46:53 +0000 (08:46 +0100)]
Correct tests for B::{cstring,perlstring,cchar}
Fix a Perl precedence error that prevented several test cases for cchar() from
being run. As a result, fix those test cases.
Fix an ASCII assumption in the tests common to cstring() and perlstring().
(chr 127 is " on EBCDIC, which is most definitely printable, so won't be
replaced with octal there.)
Father Chrysostomos [Mon, 25 Oct 2010 06:45:49 +0000 (23:45 -0700)]
Rename stashes when they move around
This is yet another patch in preparation for [perl #75176] (I keep
saying that.).
It uses the recently-added functions hv_name_add and hv_name_delete, to add and remove names when mro_package_moved is called.
mro_package_moved’s calling convention needed to change to make this
work, which is the bulk of the patch.
Code that was calling mro_package_moved was also doing it sometimes
when it was unnecessary. If the stash being assigned over had no name,
then there was no possibiiity of its being in the symbol table.
This probably fixes [perl #77358] (isa warnings), though I have not
tested that yet.
One user-visible change this introduces is that a detached glob whose
stash loses its name will no longer stringify the same way (a bit like
a glob that loses its stash pointer; except that it becomes
*__ANON__::foo instead of "").
Father Chrysostomos [Mon, 25 Oct 2010 01:39:25 +0000 (18:39 -0700)]
Memory-management macros evaluate their arguments multiple times
Father Chrysostomos [Mon, 25 Oct 2010 01:14:47 +0000 (18:14 -0700)]
[perl #77496] tied gets scalars and globs confused
Make pp_tied use the SvFAKE flag to distinguish between tied scalars
and tied handles. This is now possible as of change
2acc3314.
This fixes the problem of tied($scalar) ignoring the tie if the last
thing returned or assigned happened to be a glob.
Tony Cook [Mon, 25 Oct 2010 00:04:23 +0000 (11:04 +1100)]
use compatible types in a conditional expression
With some compilers NULL is ((void *)0) which isn't type compatible
with the integer returned by hv_exists_ent().
Father Chrysostomos [Sun, 24 Oct 2010 22:50:23 +0000 (15:50 -0700)]
[perl #77810] Scalars vs globs
Stop *{} from returning globs with the SVf_FAKE flag on.
It removes three tests from t/op/gv.t (that I added) that test buggy
edge cases that can no longer occur.
It also modifies tests in t/io/defout.t to keep them passing. I am not
sure that test script serves any purpose any more.
Todd Rinaldo [Sun, 24 Oct 2010 23:01:31 +0000 (00:01 +0100)]
Update Locale-Maketext to CPAN version 1.17
[DELTA]
2010-10-20
* Release 1.17
Test release of 1.16_01, versioning even deprecated Guts modules.
This prevents CPAN upgrade circular heck. Thanks BinGOs
Tested version fix for CPAN by BinGOs. Bump to stable version and release to public
Signed-off-by: Chris 'BinGOs' Williams <chris@bingosnet.co.uk>
Chris 'BinGOs' Williams [Sun, 24 Oct 2010 21:02:21 +0000 (22:02 +0100)]
Update Maintainers.pl with the latest version of Module::CoreList
Nicholas Clark [Sun, 24 Oct 2010 18:13:10 +0000 (19:13 +0100)]
In B.xs, use newSVpvn_flags() instead of sv_newmortal() and sv_setpvn().
Zefram [Sun, 24 Oct 2010 13:57:21 +0000 (14:57 +0100)]
don't rely on ghost contexts being unmolested
Dying and returning from a format both relied on the state of a
just-popped context frame being preserved across a LEAVE. Don't rely
on it. Test using an operator ripped off from Scope::Cleanup, which makes
it easy to run arbitrary Perl code during cleanup, without isolating it
on a separate context stack as the DESTROY mechanism does.
Florian Ragwitz [Sun, 24 Oct 2010 00:18:25 +0000 (17:18 -0700)]
Math::BigRat has blead as its upstream
Father Chrysostomos [Sun, 24 Oct 2010 17:42:33 +0000 (10:42 -0700)]
Make mro_package_moved use a mortal in case mro_isa_changed_in3 croaks
Nicholas Clark [Sun, 24 Oct 2010 16:23:17 +0000 (17:23 +0100)]
Fix SV leak in Perl_emulate_cop_io(), present since
8b850bd54aa90bd3.
Nicholas Clark [Sun, 24 Oct 2010 16:02:25 +0000 (17:02 +0100)]
In B.xs, tidy up make_*_object().
All callers to make_temp_object, make_warnings_object and make_cop_io_object
pass in a new mortal, so remove the first argument from all 3 and generate the
mortal within them.
Allow a NULL first argument for make_sv_object - generate a new mortal in this
case.
Ideally we'd remove its first argument too, but currently the output typemap
causes code to be generated that first assigns a new mortal to ST(0), then
passes that to make_sv_object(), and it's not obvious how to trivially fix
that.
Nicholas Clark [Sun, 24 Oct 2010 15:00:23 +0000 (17:00 +0200)]
B::hash() should only work on byte sequences.
Nicholas Clark [Sun, 24 Oct 2010 14:23:05 +0000 (16:23 +0200)]
Refactor B::hash to use Perl_newSVpvf() instead of my_sprintf() and new_SVpvn*.
Nicholas Clark [Sun, 24 Oct 2010 14:05:55 +0000 (16:05 +0200)]
More tests for B::hash().
Nicholas Clark [Sun, 24 Oct 2010 13:51:05 +0000 (15:51 +0200)]
B::cchar can be implemented as an ALIAS of B::cstring.
Nicholas Clark [Sun, 24 Oct 2010 12:23:43 +0000 (14:23 +0200)]
Refactor B::cchar() to create a temporary directly, and use Perl_sv_catpvf().
Nicholas Clark [Sun, 24 Oct 2010 12:07:26 +0000 (14:07 +0200)]
Tests for B::cchar().
Nicholas Clark [Sun, 24 Oct 2010 11:50:52 +0000 (13:50 +0200)]
In B.xs, refactor cstring() to return a mortal, which simplifies the XS code.
Nicholas Clark [Sun, 24 Oct 2010 11:49:12 +0000 (13:49 +0200)]
Test NUL bytes with B::cstring() and B::perlstring().
Nicholas Clark [Sun, 24 Oct 2010 11:35:06 +0000 (13:35 +0200)]
B::perlstring can be implemented as an ALIAS of B::cstring.
Nicholas Clark [Sun, 24 Oct 2010 10:26:43 +0000 (12:26 +0200)]
In cstring() in B.xs, use Perl_sv_catpvf(), instead of a temporary buffer.
Nicholas Clark [Sun, 24 Oct 2010 10:13:23 +0000 (12:13 +0200)]
More comprehensive tests for B::cstring() and B::perlstring().
Nicholas Clark [Sun, 24 Oct 2010 08:59:08 +0000 (09:59 +0100)]
In pp_getlogin, use sv_setpv_mg() to avoid needing an explicit strlen() call.
It was using PUSHp() here, meaning local object code for a call to strlen(),
and for SvSETMAGIC().
Zefram [Sat, 16 Oct 2010 21:22:57 +0000 (22:22 +0100)]
permit labels to appear before declarations
Include <label> in productions before <decl> and <package_block>. This
means that labels can now appear at the beginning of all statement-like
things. There was no technical reason for the restriction of labels
to substantive statements, and that restriction in any case couldn't be
applied to PLUGSTMT-based plugged-in declarations.
Father Chrysostomos [Sun, 24 Oct 2010 01:55:29 +0000 (18:55 -0700)]
Perl_hv_name_add needs to set xhv_name_count in one other place.
Perl_hv_name_add was not setting the name count when upgrading a HEK*
to an array of HEK*s.
Tony Cook [Sun, 24 Oct 2010 00:37:25 +0000 (11:37 +1100)]
.gitignore some extra ext/B build products
Tony Cook [Sun, 24 Oct 2010 00:35:02 +0000 (11:35 +1100)]
remove ; so the conditional actually controls some code
Tony Cook [Sun, 24 Oct 2010 00:26:50 +0000 (11:26 +1100)]
make the definition of sv_catpv_flags match the declaration
This was warning on Win32
Father Chrysostomos [Sun, 24 Oct 2010 00:04:42 +0000 (17:04 -0700)]
sv_setsv_flags was also using a potentially freed HV
Florian Ragwitz [Sat, 23 Oct 2010 22:50:46 +0000 (15:50 -0700)]
Tels doesn't want to maintain math stuff anymore
Nicholas Clark [Sat, 23 Oct 2010 18:26:53 +0000 (19:26 +0100)]
Convert B to use ExtUtils::Constant::ProxySubs for its constants.
Previously it was using a Perl script to generate C code with pairs of
newCONSTSUB(stash,"Foo_BAR",newSViv(Foo_BAR));
av_push(export_ok,newSVpvn("Foo_BAR",7));
for each constant it exported from C to Perl.
Now it uses a loop to process a table. Object code is size reduced by about 42K
on this system (almost 20%)
Nicholas Clark [Sat, 23 Oct 2010 18:00:49 +0000 (19:00 +0100)]
For PROXYSUBS, add an option to push all constants' names onto an array.
Typically this would be 'EXPORT_OK', to add all constants to the @EXPORT_OK
array for this package.
Nicholas Clark [Sat, 23 Oct 2010 17:41:04 +0000 (18:41 +0100)]
For PROXYSUBS, allow an explicit XS_SUBNAME of undef to suppress its generation.
Nicholas Clark [Sat, 23 Oct 2010 17:06:42 +0000 (18:06 +0100)]
Change concise.t to use Storable to test stash/src rendering of constant subs.
This allows B to be refactored to change the implementation of its constants.
Storable strives to main compatibility with 5.004 (or so), so unlikely to be
refactored soon.
Father Chrysostomos [Sat, 23 Oct 2010 16:48:31 +0000 (09:48 -0700)]
S_glob_assign_glob was using a potentially freed HV
Shlomi Fish [Wed, 21 Jul 2010 18:31:50 +0000 (21:31 +0300)]
perlipc: Convert tabs to spaces.
Signed-off-by: David Golden <dagolden@cpan.org>
David Golden [Sat, 23 Oct 2010 11:13:49 +0000 (07:13 -0400)]
perlipc: fix bug in example
Father Chrysostomos [Sat, 23 Oct 2010 06:56:29 +0000 (23:56 -0700)]
Add functions for adding and deleting stash names
Nicholas Clark [Sat, 23 Oct 2010 06:36:53 +0000 (07:36 +0100)]
For PROXYSUBS, also avoid calling get_missing_hash() in the XS constant code.
(Follow on from
4639bd9c727433ad)
For modules such as File::Glob and I18N::Langinfo, which always define all
constants on all platforms, this saves about 700 bytes of object code.
Ian Goodacre [Sat, 23 Oct 2010 04:42:55 +0000 (17:42 +1300)]
Clarify and correct description of comma operator in scalar context
The guarantee that in scalar context the comma operator evaluates its
arguments in scalar context is overstated.
In perl 5.10.0
print "Scalar assignment:\n";
$x = ( context(1), context(2), context(3) );
print "Scalar assignment in sub:\n";
sub list { ( context(1), context(2), context(3) ) }
$x = list();
sub context {
if(wantarray) {
print "list context\n";
} elsif(defined(wantarray)) {
print "scalar context\n";
} else {
print "void context\n";
}
}
prints:
scalar assignment:
void context
void context
scalar context
Scalar assignment in sub:
scalar context
scalar context
scalar context
This leaves only the right argument of the last comma operator in a list as
the only one that might always be evaluated in scalar context.
The comments on the sample outputs were at best ambiguous if not misleading
or false, and also unnecessarily pejorative of perl4. The revised comments
less ambiguously refer to the last expression in the list (@y in the example)
rather than to the literal list that is the argument of the assignment
operator.
Ian Goodacre [Sat, 23 Oct 2010 03:46:02 +0000 (16:46 +1300)]
Corrected Perl5 example of Formatted output and significant digits
The following note says "Your results may vary...", so maybe some perls
did give the value in the example as it was, but perl v5.10.0 on Intel
Pentium based system gives the more accurate revised output.
I am also dubious about the values given for perl4 as the error is quite
large, but I don't have perl4 easily available to test.
Ian Goodacre [Sat, 23 Oct 2010 00:38:52 +0000 (13:38 +1300)]
Added missing variable name in example code
Nicholas Clark [Fri, 22 Oct 2010 20:10:15 +0000 (21:10 +0100)]
Remove cruft from B's Makefile.PL
$noecho is unused since
70c68212ed57a5a0 (in 2005)
$e and $exeout_flag are unused since
e8edd1e67bd80dbb (in 1999)
All the clean FILES apart from 'defsubs.h' are vestigial or duplicate defaults.
The rule for LIBS created by post_constants() seems superfluous.
Nicholas Clark [Fri, 22 Oct 2010 17:59:36 +0000 (19:59 +0200)]
For PROXYSUBS, get the hash from the fetch to pass to the subsequent store.
This saves the store having to (re)compute the hash.
Nicholas Clark [Fri, 22 Oct 2010 14:58:36 +0000 (15:58 +0100)]
For PROXYSUBS, loop with do {} not while, as there is always at least 1 entry.
Nicholas Clark [Fri, 22 Oct 2010 12:03:09 +0000 (14:03 +0200)]
For PROXYSUBS, avoid calling get_missing_hash() if all symbols are present.
For modules such as File::Glob and I18N::Langinfo, which always define all
constants on all platforms, this saves about 1K of object code, plus 1 (empty)
anonymous hash (per process or thread) in
%ExtUtils::Constant::ProxySubs::Missing.
Nicholas Clark [Fri, 22 Oct 2010 10:15:17 +0000 (12:15 +0200)]
For PROXYSUBS, give the notfound struct file scope.
All others remain within the scope of the BOOT block.
Nicholas Clark [Fri, 22 Oct 2010 08:36:44 +0000 (10:36 +0200)]
Minor refactor of ExtUtils::Constant::ProxySubs.
Use a lexical in place of repeated lc()'s in WriteConstants.
Require ExtUtils::Constant::ProxySubs early, to avoid leaving empty files if it
fails for any reason.
H.Merijn Brand [Fri, 22 Oct 2010 10:40:26 +0000 (12:40 +0200)]
Fix on doc patch from Moritz Lenz (from IRC)
Nicholas Clark [Fri, 22 Oct 2010 07:47:09 +0000 (08:47 +0100)]
Merge the implementation of B::CV::{ROOT,START}.
Nicholas Clark [Fri, 22 Oct 2010 07:37:29 +0000 (08:37 +0100)]
Note the U8-sized space in intrpvar.h created by
737c24fc85ea612b.
Nicholas Clark [Fri, 22 Oct 2010 07:24:18 +0000 (08:24 +0100)]
Better diagnostics for XS::APItest::test_cv_getset_call_checker().
Nicholas Clark [Fri, 22 Oct 2010 06:41:39 +0000 (07:41 +0100)]
S_tokeq()'s fast scan loop should terminate on \\ not \
As-was, it would drop out of the scanner into the backslashed-backslash
processing loop earlier than need be, and hence would be copying the octets
of strings (in place) as soon as any backslash had been seen. Now it defers
copying until copying is actually unavoidable.
Father Chrysostomos [Fri, 22 Oct 2010 04:49:33 +0000 (21:49 -0700)]
perldelta: fix typo and POD syntax
Father Chrysostomos [Fri, 22 Oct 2010 04:42:35 +0000 (21:42 -0700)]
perldelta for things I committed recently
Father Chrysostomos [Fri, 22 Oct 2010 04:41:47 +0000 (21:41 -0700)]
Bump re’s version
Father Chrysostomos [Fri, 22 Oct 2010 03:42:47 +0000 (20:42 -0700)]
Bump charnames’ version
Father Chrysostomos [Fri, 22 Oct 2010 01:18:42 +0000 (18:18 -0700)]
[perl #78488] Bleadperl
304474c3 breaks GFUJI/Test-LeakTrace-0.13.tar.gz
This commit restores an SvREFCNT_dec that was inadvertently removed.
c8bbf67 removed the SvREFCNT_dec(HeVAL(entry)), adding
SvREFCNT_dec(old_val) instead.
304474c3 reverted that block, but failed to restore the
SvREFCNT_dec(HeVAL(entry)).
The result was that the %INC entry created by do "file" was leaking.
(Y’know, giving out commit bits to just *anyone* who comes along
with a few patches seems dangerous. Has anybody been looking at what
I’m doing?)
Jan Dubois [Fri, 22 Oct 2010 01:00:14 +0000 (18:00 -0700)]
Fix calling conventions in malloc_ctl.h
The Perl_malloc() etc. functions are *also* declared in
proto.h, so the declarations need to match. The inclusion
of malloc_ctl.h into perl.h had to be moved down until
after the point where PERL_CALLCONV was completely defined.
Karl Williamson [Wed, 20 Oct 2010 19:21:04 +0000 (13:21 -0600)]
regcomp.c: /d not overriding /u
The setting of the charset regex modifiers was wrong. /d didn't
override /u nor /l, and similarly /u and /l didn't properly override
each other.
Karl Williamson [Sun, 17 Oct 2010 16:38:37 +0000 (10:38 -0600)]
regcomp.c: Fix typo in comment
Jerry D. Hedden [Thu, 21 Oct 2010 21:47:50 +0000 (14:47 -0700)]
Fix ext/XS-APItest/t/multicall.t warning
'make test' produces the following warning:
ext/XS-APItest/t/multicall.....................................Useless
use of private variable in void context at t/multicall.t line 37.
ok
The attached patch fixes this by adding a "no warnings 'void';"
statement to the test file.
Nicholas Clark [Thu, 21 Oct 2010 19:32:31 +0000 (20:32 +0100)]
MY_CXT_INIT happens to contain a variable declaration, so give it its own block
Without this, Glob.xs won't build on any compiler which enforces this part of
C89.
Nicholas Clark [Thu, 21 Oct 2010 16:58:42 +0000 (17:58 +0100)]
Sys::Hostname::ghname should use newSVpv(s, 0), not newSVpvn(s, strlen(s))
Whilst functionally identical, the latter will produce more object code.
Also, avoid the strlen() entirely for the case of HAS_PHOSTNAME, where we know
the length.
Nicholas Clark [Thu, 21 Oct 2010 16:43:31 +0000 (17:43 +0100)]
Avoid using #ifdef inside a function call that may itself actually be a macro.
A tweak to
20439bc77dfeec46. Hopefully this fixes the build on Win32.
Nicholas Clark [Thu, 21 Oct 2010 15:18:11 +0000 (16:18 +0100)]
buildtoc needs to preserve the mode of any file it updates.
Makefile.SH should be +x, and t/porting/exe-bit.t rightly gets upset when it is
not.
Father Chrysostomos [Thu, 21 Oct 2010 04:33:53 +0000 (21:33 -0700)]
Allow stashes to have multiple names
This commits modifies the HvAUX structure as follows: A new field is
added, named xhv_name_count, indicating the number of names. If it is
zero (the default and most common case), then xhv_name is a HEK * as
usual. If it is non-zero, then xhv_name actually holds a pointer to an
array of HEK*s, the first being the default or ‘canonical’ name.
This code is a little repetitious, but more refactorings are to come,
so it is too soon to turn these repetitions into macros.
This is yet another commit in preparation for fixing [perl #75176].
Basically, whenever a stash is deleted from its containing stash, if
it has an alias elsewhere, it needs to assume the new name (of that
alias; so it needs to know its other names already) and update isarev
entries. Forthcoming commits will do that.
David Golden [Mon, 13 Sep 2010 00:26:43 +0000 (20:26 -0400)]
Add single-term prototype
The C<+> prototype is a special alternative to C<$> that will act like
C<\[@%]> when given a literal array or hash variable, but will otherwise
force scalar context on the argument. This is useful for functions which
should accept either a literal array or an array reference as the argument:
sub smartpush (+@) {
my $aref = shift;
die "Not an array or arrayref" unless ref $aref eq 'ARRAY';
push @$aref, @_;
}
When using the C<+> prototype, your function must check that the argument
is of an acceptable type.
Father Chrysostomos [Wed, 20 Oct 2010 18:27:28 +0000 (11:27 -0700)]
Fix the test count in t/re/pat.t
Karl Williamson [Wed, 20 Oct 2010 17:11:13 +0000 (11:11 -0600)]
regexec.c: utf8 doesn't match /i nonutf8 self
This is a continuation of [perl #78464]. It fixes it also for the /i
flag. After this, a character should match itself in the regrepeat
function, even if one is in utf8 and the other isn't, for both /i and
not.
The solution is to move the code for handling /i into the non-i
structure so that the decisions about utf8 are all in one place. When
the string is in utf8, it uses the utf8-fold function.
This has the added effect of fixing a few cases where a utf8 string did
not match a fold in a non-utf8 pattern. I haven't added tests for
these, as it only fixes a few cases where this is a problem, and I'm
working on a comprehensive solution to the problem, accompanied by
extensive tests.
Father Chrysostomos [Wed, 20 Oct 2010 06:50:07 +0000 (23:50 -0700)]
Fix a nested package deletion bug
In mro_package_moved, I was calling mro_isa_changed_in, instead
of mro_package_moved, for a deleted package. So its subpackages
were ignored.
Exempli gratia, delete $::{'Cur::'} would call mro_isa_changed_in on
Cur::Cur, but ignore Cur::Cur::Cur.
I probably added this bug (or reinstated it, as it was in 5.13.5) in
d056e33c1.
Karl Williamson [Wed, 20 Oct 2010 16:20:29 +0000 (10:20 -0600)]
regexec.c: utf8 doesn't match non-utf8 self
Some regex patterns don't match a character with itself when the target
string is in utf8 and the pattern isn't, and the character is variant
under utf8. (This means only Latin1-range characters in the pattern are
affected.)
The solution is to test for this case and use the utf8 representation of
the pattern character for the comparison.
Karl Williamson [Mon, 18 Oct 2010 04:32:07 +0000 (22:32 -0600)]
regcomp.c: Get rid of compiler warning.
This patch should remove a compiler warning that is currently only
showing up in one compiler. It declares a debug-only variable to be
volatile, so should silence the warning that it is getting clobbered.
Since this variable is only used for debugging purposes, when DEBUGGING
is defined, performance should not be an issue.
Zefram [Mon, 4 Oct 2010 23:20:57 +0000 (00:20 +0100)]
mark cophh API as experimental
Zefram [Tue, 19 Oct 2010 16:31:46 +0000 (09:31 -0700)]
full API for cop hint hashes
Expose cop hint hashes as a type COPHH, with a cophh_* API which is a
macro layer over the refcounted_he_* API. The documentation for cophh_*
describes purely API-visible behaviour, whereas the refcounted_he_*
documentation describes the functions mainly in terms of the
implementation. Revise the cop_hints_* API, using the flags parameter
consistently and reimplementing in terms of cophh_*. Use the cophh_*
and cop_hints_* functions consistently where appropriate.
[Modified by the committer to update two calls to
Perl_refcounted_he_fetch recently added to newPMOP.]
Karl Williamson [Fri, 15 Oct 2010 01:36:04 +0000 (19:36 -0600)]
charnames::viacode returning less correct name
There are several cases where more than one name is valid for a code
point. This happens usually when the original name was published with a
typo in it. It's best for viacode to return the revised name, though
the original remains valid.
The names data is in a table generated by mktables exclusively for
charnames, including vianame (and its kin) and viacode. The fix is to
mktables to put the more correct name first in the table, so that it is
found first and returned by viacode().
When I originally designed this code, I thought the correct name should
come last in the tables, so someone looping and reading it could just
overwrite the less correct one with the more correct one.
But to save memory we have the same table shared by viacode and
vianame, and vianame has to recognize both names, so both entries
are needed. viacode could do an rindex to find the more correct name,
but experiments show that that was twice as slow as going the other
direction. Therefore, this patch is for speed.
If the tables for vianame and viacode were ever to be split, this patch
could be reverted, if desired, to put things back to the reverse order.
Karl Williamson [Thu, 14 Oct 2010 23:29:12 +0000 (17:29 -0600)]
mktables: Complement variable meaning for clarity
When I was designing this code, I struggled for a long time in how to
name the concept the $cdm variable meant. I was never very happy with
it. Coming back to it after a year, I realized immediately that the
complement was much easier to understand, so this patch does that and
rewrites the comments to match
Zefram [Fri, 15 Oct 2010 12:11:54 +0000 (13:11 +0100)]
function to parse Perl code block
New API function parse_block() parses a code block, including surrounding
braces. The block is a lexical scope, but not inherently a dynamic scope.
Zefram [Thu, 14 Oct 2010 16:02:36 +0000 (17:02 +0100)]
fix and test PL_expect in recdescent parsing
Set PL_expect at the start of parse_fullstmt() as well as at the start
of parse_stmtseq(). Test both.
Father Chrysostomos [Tue, 19 Oct 2010 00:59:50 +0000 (17:59 -0700)]
[perl #78072] use re '/xism';
Zefram [Thu, 14 Oct 2010 00:52:59 +0000 (01:52 +0100)]
handle bracket stack better in recdescent parsing
When recursing into the parser for recursive-descent parsing, put
a special entry on the bracket stack that generates a fake EOF if a
closing bracket belonging to an outer parser frame is seen. This keeps
the bracket stack balanced across a parse_stmtseq() frame, fixing
[perl #78222].
If a recursive-descent parser frame ends by yyunlex()ing an opening
bracket, pop its entry off the bracket stack and stash it in the
forced-token queue, to be revived when the token is re-lexed. This keeps
the bracket stack balanced across a parse_fullstmt() frame.
Father Chrysostomos [Tue, 19 Oct 2010 00:57:25 +0000 (17:57 -0700)]
[perl #33752] B::Deparse fails with utf8 and our
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
This commit makes B::Deparse support code like ‘our $ḟōō’. Currently,
an ‘our’ variable whose name is an octet sequence that does not consist
of (\w|::)+ can only be a UTF8 variable name.
When the pad is made to support UTF8 properly, this may need to
be changed.