14 years agoPossible undefined behaviour, spotted by gcc 4.5.0 and HP's updated compiler.
Nicholas Clark [Mon, 19 Apr 2010 14:35:59 +0000]
Possible undefined behaviour, spotted by gcc 4.5.0 and HP's updated compiler.

14 years agoRevert "Documentation tweaks to Text::Tabs"
Gisle Aas [Sun, 18 Apr 2010 22:04:41 +0000]
Revert "Documentation tweaks to Text::Tabs"

This change was reported upstream in <https://rt.cpan.org/Ticket/Display.html?id=56699>.
Committing it directly to blead was a mistake.

This reverts commit f6e2320c520b8d610a864cd42aece1f0df2db895.

14 years agoLet the cmdVERSION usage message expand $0
Gisle Aas [Sun, 18 Apr 2010 20:36:46 +0000]
Let the cmdVERSION usage message expand $0

It makes more sense than literally printing "usage: $0 ..."

14 years agoDocumentation tweaks to Text::Tabs
Gisle Aas [Sun, 18 Apr 2010 20:58:02 +0000]
Documentation tweaks to Text::Tabs

14 years agoBring Porting/Maintainers in sync with reality regarding Scalar-List-Utils
Steffen Mueller [Sun, 18 Apr 2010 15:45:05 +0000]
Bring Porting/Maintainers in sync with reality regarding Scalar-List-Utils

14 years agoUpdate to MIME-Base64 3.09
Gisle Aas [Tue, 2 Feb 2010 22:22:15 +0000]
Update to MIME-Base64 3.09

Fixes issue where the Quoted-Printable encoder would sometimes output
lines that were 77 characters long.  The max line length should be 76.

This reverts revert 9ef40646096384d9f65acffab1b9a2df07d03a0c.

14 years agoEscape patch strings before embedding them in patchlevel.h
Gisle Aas [Sat, 27 Mar 2010 06:55:51 +0000]
Escape patch strings before embedding them in patchlevel.h

14 years agoPassing regexp vars as argument is unsafe
Gisle Aas [Sun, 18 Apr 2010 15:13:46 +0000]
Passing regexp vars as argument is unsafe

In this case $_[0] is unexpectedly modified by matching a regular
expression inside the function, making it report bad file names.

14 years agoAvoid "Use of uninitialized value $base" for the CPANPLUS .hidden file
Gisle Aas [Sun, 18 Apr 2010 15:12:24 +0000]
Avoid "Use of uninitialized value $base" for the CPANPLUS .hidden file

14 years agoDo not clobber %ENV in EU::MM test
Steffen Mueller [Sun, 18 Apr 2010 14:52:44 +0000]
Do not clobber %ENV in EU::MM test

175046355f07701 introduced an error in the ExtUtils::MakeMaker
prerequisites tests in that it clobbers %ENV instead of just locally
overriding $ENV{PERL_CORE}.

14 years agoEU::MakeMaker shouldn't warn about prereqs in core
Steffen Mueller [Sun, 18 Apr 2010 14:24:20 +0000]
EU::MakeMaker shouldn't warn about prereqs in core

This silences warnings about missing prerequisites when part of a perl
core build. Triggered by #74442.

14 years agoMake abstract_from more robust
Robin Barker [Thu, 15 Apr 2010 20:14:23 +0000]
Make abstract_from more robust

A few distributions in the perl source failure for abstract_from
Rather than fix the individual files I have made the parsing process
more robust.

14 years agoMove PERL_ASYNC_CHECK() from POPBLOCK() to the kill case of Perl_apply().
Nicholas Clark [Sun, 18 Apr 2010 12:24:20 +0000]
Move PERL_ASYNC_CHECK() from POPBLOCK() to the kill case of Perl_apply().

This ensures that (safe) signals sent to the same process are still dispatched
within the same statement (as before), without overloading the semantics of
block popping.

14 years agoNote the revised signal dispatch and performance gain in perl5130delta.pod.
Nicholas Clark [Sat, 17 Apr 2010 20:56:23 +0000]
Note the revised signal dispatch and performance gain in perl5130delta.pod.

14 years agoDispatch signals in infinite loops such as 1 while 1;
Nicholas Clark [Sat, 17 Apr 2010 20:24:21 +0000]
Dispatch signals in infinite loops such as 1 while 1;

With the move of PERL_ASYNC_CHECK() out from the runloop to control ops,
infinite loops became truely infinite, as their optree has no control ops.
Hence add a PERL_ASYNC_CHECK() to pp_unstack to ensure signals will be
dispatched.

Bug noticed by Jerry Hedden.

14 years agoFix RT #74290 - regression for labels immediately before string evals.
Nicholas Clark [Sat, 17 Apr 2010 19:11:49 +0000]
Fix RT #74290 - regression for labels immediately before string evals.

Fix location identified by Father Chrysostomos, who also offered a patch, but
this patch is more efficient, as it avoids any allocation. Test code based on
his test example.

14 years agoIn Perl_get_debug_opts(), restructure the message into fewer string literals.
Nicholas Clark [Sat, 17 Apr 2010 13:44:36 +0000]
In  Perl_get_debug_opts(), restructure the message into fewer string literals.

This reduces the object code size slightly.

14 years agoIn S_usage(), restructure the storage of the help message into fewer literals.
Nicholas Clark [Sat, 17 Apr 2010 12:13:39 +0000]
In S_usage(), restructure the storage of the help message into fewer literals.

This reduces the object code size slightly.

14 years agoAlso guess $ENV{EMAIL}, if present, as the originating address
Alex Vandiver [Sat, 17 Apr 2010 03:54:11 +0000]
Also guess $ENV{EMAIL}, if present, as the originating address

Signed-off-by: H.Merijn Brand <h.m.brand@xs4all.nl>

14 years ago[PATCH] [5.12] Allow build on aix 4.2 (sigaction and IPv6)
H.Merijn Brand [Fri, 16 Apr 2010 15:47:15 +0000]
[PATCH] [5.12] Allow build on aix 4.2 (sigaction and IPv6)

Based on a patch by AUGUSTE-ETIENNE José <jose.auguste-etienne@cgss-guyane.fr>

The attached patches are necessary to work around aix 4.2 lack of support for IPv6,
and limited support for POSIX sigaction()

The hints/aix_4.sh patch solves the following build failure:

"Socket.xs", line 468.16: 1506-007 (S) "struct in6_addr" is undefined.

The ext/POSIX/t/sigaction.t patch solves the failing test reported a looong time ago:

http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2006-01/msg01124.html

After testing a simple C program using POSIX sigaction() with SA_SIGINFO,
I found that it worked on aix 4.3 but failed on aix 4.2.

I think it's safe to skip the SA_SIGINFO test on the aix 4.2 platform.

perl 5.12.0 builds fine on the following platforms / C compiler :

aix 4.2.1 / vac 5.0.2.7 (with attached patches)
aix 4.3.2 / gcc 2.95.3
aix 4.3.3 / vac 4.4.0.3
aix 4.3.3 / vac 5.0.28
aix 5.2 / GNUpro gcc 2.9.aix51.020209

The atatched patches should be applied to blead and maint.

14 years agoMerge branch 'blead' of perl5.git.perl.org:/gitroot/perl into blead
H.Merijn Brand [Fri, 16 Apr 2010 12:03:05 +0000]
Merge branch 'blead' of perl5.git.perl.org:/gitroot/perl into blead

14 years agoAdd new Data::Dumper test file to MANIFEST
Rafael Garcia-Suarez [Thu, 15 Apr 2010 20:40:32 +0000]
Add new Data::Dumper test file to MANIFEST

14 years agoUpgrade Data::Dumper to 2.126 CPAN release
Steffen Mueller [Thu, 15 Apr 2010 19:57:15 +0000]
Upgrade Data::Dumper to 2.126 CPAN release

No significant changes, just synchronization to the exact release.

14 years agoFix Data::Dumper's Fix Terse(1) + Indent(2)
Michael G. Schwern [Thu, 15 Apr 2010 19:29:43 +0000]
Fix Data::Dumper's Fix Terse(1) + Indent(2)

RT #73604: When $Data::Dumper::Terse is true, the indentation is thrown
off. It appears to be acting as if the $VAR1 = is still there.

14 years agoFix a test failure in perl5db.t when Term::ReadLine::Gnu is installed
Niko Tyni [Sat, 27 Mar 2010 19:38:34 +0000]
Fix a test failure in perl5db.t when Term::ReadLine::Gnu is installed

With Term::ReadLine::Gnu on @INC, the debugger output will include
terminal escape sequences even when empty otherwise. This makes
lib/perl5db.t test 7 ("proxy constant subroutines") fail.

Most of the other tests are already run with ReadLine support disabled, so
bring this one in line too.

14 years agoPoint submitters to perlbug@perl.org, not perl5-porters@perl.org
Ævar Arnfjörð Bjarmason [Thu, 15 Apr 2010 18:06:24 +0000]
Point submitters to perlbug@perl.org, not perl5-porters@perl.org

We usually want patches in RT. Patches should only go to perl5-porters
if they're intended purely for discussion.

14 years agoSet the legacy process name with prctl() on assignment to $0 on Linux
Ævar Arnfjörð Bjarmason [Thu, 15 Apr 2010 17:12:04 +0000]
Set the legacy process name with prctl() on assignment to $0 on Linux

Ever since perl 4.000 we've only set the POSIX process name via
argv[0]. Unfortunately on Linux the POSIX name isn't used by utilities
like top(1), ps(1) and killall(1).

Now when we set C<$0 = "hello"> both C<qx[ps h $$]> (POSIX) and
C<qx[ps hc $$]> (legacy) will say "hello", instead of the latter being
"perl" as was previously the case.

See also the March 9 2010 thread "Why doesn't assignment to $0 on
Linux also call prctl()?" on perl5-porters.

14 years agoBump $VERSION of IPC::Open3 to 1.06
Rafael Garcia-Suarez [Thu, 15 Apr 2010 14:46:40 +0000]
Bump $VERSION of IPC::Open3 to 1.06

14 years agoopen3 errors in child croak parent RT#72016
Eric Brine [Mon, 18 Jan 2010 18:21:20 +0000]
open3 errors in child croak parent RT#72016

Errors in open3 no longer appear to originate from the executed command on forking systems.

14 years agoAdd TODO test for RT#72016
Eric Brine [Mon, 18 Jan 2010 04:44:14 +0000]
Add TODO test for RT#72016

14 years agoBump threads version to 1.77_01
Steffen Mueller [Thu, 15 Apr 2010 14:42:48 +0000]
Bump threads version to 1.77_01

This is due to the POD fix of b91a79b929f6eca75f18124340d2f0f89b9588a9.

14 years agoUpgrade to threads-shared-1.33
Steffen Mueller [Thu, 15 Apr 2010 13:24:57 +0000]
Upgrade to threads-shared-1.33

14 years agoUpgrade dist/threads to 1.77 from CPAN
Steffen Mueller [Thu, 15 Apr 2010 09:12:04 +0000]
Upgrade dist/threads to 1.77 from CPAN

While doing so fix a tiny, little POD error to prevent breakage of core
tests. (blead is upstream for threads.pm)

14 years agoSimplify the perldelta howto by using Git features
Ævar Arnfjörð Bjarmason [Thu, 15 Apr 2010 13:40:33 +0000]
Simplify the perldelta howto by using Git features

`git-log' can generate a list of authors without perl's help and
`git-diff` can produce a diffstat without diffstat(1).

14 years agoFix code before statement error introduced by f410a2119920dd04.
Nicholas Clark [Thu, 15 Apr 2010 13:03:08 +0000]
Fix code before statement error introduced by f410a2119920dd04.

14 years agoAvoid redundant reload of PL_op in the runloop.
Hugo van der Sanden [Thu, 15 Apr 2010 12:05:38 +0000]
Avoid redundant reload of PL_op in the runloop.

(gcc, at least, generates suboptimal code without this dirty great hint)

14 years agoMove PERL_ASYNC_CHECK() from the runloop to control flow OPs.
Nicholas Clark [Thu, 15 Apr 2010 10:37:53 +0000]
Move PERL_ASYNC_CHECK() from the runloop to control flow OPs.

For the typical code this gives a 5% speedup, and removes the cost of "safe
signals". Tight looping code will show less gains, but should never be slower.

Subtle bugs might remain - there might be constructions that enter the runloop
(where signals used to be dispatched) but don't contain any PERL_ASYNC_CHECK()
calls themselves.

14 years agomark two magic.t tests as TODO
David Mitchell [Thu, 15 Apr 2010 09:29:56 +0000]
mark two magic.t tests as TODO

fixing a bool cast bug exposes a hidden bug in @+ scope.
Mark as TODO for now.

14 years agouse cBOOL for bool casts
David Mitchell [Thu, 15 Apr 2010 09:20:50 +0000]
use cBOOL for bool casts

bool b = (bool)some_int

doesn't necessarily do what you think. In some builds, bool is defined as
char, and that cast's behaviour is thus undefined. So this line in mg.c:

    const bool was_temp = (bool)SvTEMP(sv);

was actually setting was_temp to false even when the SVs_TEMP flag was set.
Fix this by replacing all the (bool) casts with a new cBOOL() cast macro
that (hopefully) does the right thing.

14 years agoFix a minor perl5114delta error.
Florian Ragwitz [Wed, 20 Jan 2010 16:12:34 +0000]
Fix a minor perl5114delta error.

14 years agoRemove 'delete &function' from perltodo.
Florian Ragwitz [Thu, 18 Mar 2010 14:13:40 +0000]
Remove 'delete &function' from perltodo.

It wouldn't add any new feature we don't already have and be inconsistent with
the behaviour of other things.

14 years ago[perl #74372] := warning not in perldiag
Rafael Garcia-Suarez [Thu, 15 Apr 2010 08:54:32 +0000]
[perl #74372] := warning not in perldiag

Missing warning description noticed by Zefram

14 years agoPATCH: [perl #72998] regex looping
Karl Williamson [Wed, 14 Apr 2010 03:25:36 +0000]
PATCH: [perl #72998] regex looping

If a character folds to multiple ones in case-insensitive matching,
it should not match just one of those, or the regular expression can
loop.  For example, \N{LATIN SMALL LIGATURE FF} folds to 'ff', and so
    "\N{LATIN SMALL LIGATURE FF}" =~ /f+/i
should match.  Prior to this patch, this function returned that there is
a match, but left the matching string  pointer at the beginning of the
"\N{LATIN SMALL LIGATURE FF}" because it doesn't make sense to match
just half a character, and at this level it doesn't know about the '+'.
This leaves things in an inconsistent state, with the reporting of a
match, but the input pointer unchanged, the result of which is a loop.

I don't know how to fix this so that it correctly matches, and there are
semantic issues with doing so.  For example, if
    "\N{LATIN SMALL LIGATURE FF}" =~ /ff/i
matches, then one would think that so should
    "\N{LATIN SMALL LIGATURE FF}" =~ /(f)(f)/i
But $1 and $2 don't really make sense here, since they both refer to the
half of the same character.

So this patch just returns failure if only a partial character is
matched.  That leaves things consistent, and solves the problem of
looping, so that Perl doesn't hang on such a construct, but leaves the
ultimate solution for another day.

14 years agoUpgrade to CGI.pm 3.49
Rafael Garcia-Suarez [Thu, 15 Apr 2010 07:35:07 +0000]
Upgrade to CGI.pm 3.49

14 years agoBetween parens is the unit name
H.Merijn Brand [Wed, 14 Apr 2010 14:14:16 +0000]
Between parens is the unit name

Strip trailing blank. Manual fix after metaunit fix
3e6d782f8c2ac2b94d9f217b0a91989b144feca4

14 years agoMerge branches 'blead' and 'blead' of perl5.git.perl.org:/gitroot/perl into blead
H.Merijn Brand [Wed, 14 Apr 2010 13:59:22 +0000]
Merge branches 'blead' and 'blead' of perl5.git.perl.org:/gitroot/perl into blead

14 years agoMerge branches 'blead' and 'blead' of perl5.git.perl.org:/gitroot/perl into blead
H.Merijn Brand [Wed, 14 Apr 2010 13:46:42 +0000]
Merge branches 'blead' and 'blead' of perl5.git.perl.org:/gitroot/perl into blead

14 years agoAdd the new maint policy to perlpolicy.pod
Jesse Vincent [Wed, 14 Apr 2010 13:29:31 +0000]
Add the new maint policy to perlpolicy.pod

14 years ago[perl #74038] c2ph (and pstruct) use the deprecated 'getopts.pl' library
Zefram [Wed, 14 Apr 2010 12:05:57 +0000]
[perl #74038] c2ph (and pstruct) use the deprecated 'getopts.pl' library

14 years agoChanges to CPANPLUS and CPANPLUS::Dist::Build tests
Chris Williams [Wed, 14 Apr 2010 10:44:41 +0000]
Changes to CPANPLUS and CPANPLUS::Dist::Build tests

  Add the utils/ dir to $ENV{PATH} if running under PERL_CORE for the
  CPANPLUS/CPANPLUS-Dist-Build tests.

  Both modules' testsuites require access to the cpanp-run-perl script
  and the version in utils/ is already executable versus the version
  in cpan/CPANPLUS/bin

14 years agoadd step to update RT version numbers in release_managers_guide
David Mitchell [Wed, 14 Apr 2010 09:36:45 +0000]
add step to update RT version numbers in release_managers_guide

14 years ago[perl #74006] 5.12.0-RC stuffing bug
Zefram [Wed, 14 Apr 2010 07:29:15 +0000]
[perl #74006] 5.12.0-RC stuffing bug

There's a small bug in lex_stuff_pvn() that causes spurious syntax errors
in an obscure situation.  It happens if stuffing is performed on the
last line of a file, and the line ends with a statement that lacks its
terminating semicolon.  Attached patch fixes and adds test.

14 years agoAllow extension building on older (pre 7.3-2) VMS systems.
Craig A. Berry [Wed, 14 Apr 2010 02:55:58 +0000]
Allow extension building on older (pre 7.3-2) VMS systems.

DCL symbol length was limited to 1K up until about seven years or
so ago, but there was no particularly deep reason to prevent those
older systems from configuring and building Perl.

14 years agoClarify HAS_PRCTL_SET_NAME definition in Porting/Glossary.
Craig A. Berry [Wed, 14 Apr 2010 00:51:31 +0000]
Clarify HAS_PRCTL_SET_NAME definition in Porting/Glossary.

14 years agorevert "revert perl -h changes"
David Golden [Wed, 14 Apr 2010 01:36:01 +0000]
revert "revert perl -h changes"

Reapplies commit 482d21b3db741a7f6b59279ab7ad289307e2186b
that was reverted in commit bb87c82accf79d3e15da7fc9b646232a4c2bd47c

14 years agoRevert "Revert "* Fixed typo in toke.c docs, identified by Zefram""
Jesse Vincent [Tue, 13 Apr 2010 20:47:49 +0000]
Revert "Revert "* Fixed typo in toke.c docs, identified by Zefram""

This reverts commit 06164d6c3ad67ed7ba18030ae378f46f482a29af.

14 years agoAdd emacs/vim hints
Josh ben Jore [Tue, 13 Apr 2010 14:53:12 +0000]
Add emacs/vim hints

14 years agoRe-flow some errantly formatted pod
Josh ben Jore [Tue, 13 Apr 2010 14:52:05 +0000]
Re-flow some errantly formatted pod

14 years agoDetabify perldtrace.d
Josh ben Jore [Tue, 13 Apr 2010 06:48:25 +0000]
Detabify perldtrace.d

14 years ago[perl #73560] [PATCH] Fix typos in Pod::Functions
Nga Tang Chan [Tue, 13 Apr 2010 16:10:35 +0000]
[perl #73560] [PATCH] Fix typos in Pod::Functions

(also, bump $VERSION)

14 years agoReintroduce a lot of missing modules into perlmodlib.pod
Rafael Garcia-Suarez [Tue, 13 Apr 2010 16:03:18 +0000]
Reintroduce a lot of missing modules into perlmodlib.pod

That was due to the great move of dual-life modules into cpan/ and
dist/. This fixes the bug: [perl #74332] lots of modules missing from
perlmodlib.

14 years ago[perl #73712] ‘Variable is not imported’ cannot be suppressed
Father Chrysostomos [Tue, 13 Apr 2010 15:20:19 +0000]
[perl #73712] ‘Variable is not imported’ cannot be suppressed

The message ‘Variable "%s" is not imported’ cannot be suppressed, even
with -X (local $SIG{__WARN__}=sub{} is what I have to use):

perl -Xle '$foo;use strict; eval q/$foo/ or die "---$@---"'
Variable "$foo" is not imported at (eval 1) line 2.
---Global symbol "$foo" requires explicit package name at (eval 1) line 2.
--- at -e line 1.

This is because we have what appears to the user to be a multi-line
error message. It is in fact a warning ‘Variable...’ followed by an
error ‘Global symbol...’.

The attached patch assigns a warning category to the warning.

14 years agoFix the abstract in Module::CoreList's META.yml
Chris Williams [Tue, 13 Apr 2010 15:08:23 +0000]
Fix the abstract in Module::CoreList's META.yml

14 years ago* eliminated OP_SETSTATE also in comments
Slaven Rezic [Mon, 29 Mar 2010 20:19:37 +0000]
* eliminated OP_SETSTATE also in comments

14 years agoAdd d_prctl* for all other OS's
H.Merijn Brand [Tue, 13 Apr 2010 13:22:53 +0000]
Add d_prctl* for all other OS's

This was missing from c796e3db23c597b99f07485542338844e61a6a69

14 years agoProbe for prctl () and check id PR_SET_NAME is supported
H.Merijn Brand [Mon, 29 Mar 2010 08:16:44 +0000]
Probe for prctl () and check id PR_SET_NAME is supported

14 years agoNote how to deal with broken dbm.h on OpenSUSE
H.Merijn Brand [Wed, 24 Mar 2010 07:58:29 +0000]
Note how to deal with broken dbm.h on OpenSUSE

14 years agoAllow -Duse64bitint without promoting to use64bitall on HP-UX
H.Merijn Brand [Mon, 15 Mar 2010 15:33:22 +0000]
Allow -Duse64bitint without promoting to use64bitall on HP-UX

Fix for http://rt.perl.org/rt3/Public/Bug/Display.html?id=17736

14 years agofix a2p example
Frank Wiegand [Thu, 8 Apr 2010 20:16:30 +0000]
fix a2p example

14 years agofix minor casting issue
David Mitchell [Tue, 13 Apr 2010 10:00:30 +0000]
fix minor casting issue

14 years agoMerge commit 'origin/davem/post-5.12' into blead
David Mitchell [Tue, 13 Apr 2010 10:00:09 +0000]
Merge commit 'origin/davem/post-5.12' into blead

14 years agoBefore 96a8704c makedepend.SH used shell quote syntax that modern
Ævar Arnfjörð Bjarmason [Tue, 13 Apr 2010 09:48:20 +0000]
Before 96a8704c makedepend.SH used shell quote syntax that modern
shells barf on. Document how to work around this in the example bisect
script.

14 years agoWhen your bisect starts to go into perl-5.9.* territory `make -j4'
Ævar Arnfjörð Bjarmason [Tue, 13 Apr 2010 09:46:48 +0000]
When your bisect starts to go into perl-5.9.* territory `make -j4'
fails on Config.pm due to some race condition. This'll turn what would
otherwise be a bad/good report into skip.

14 years agoFix test comment (noticed by Alex Vandiver)
Rafael Garcia-Suarez [Tue, 13 Apr 2010 09:45:11 +0000]
Fix test comment (noticed by Alex Vandiver)

14 years ago[perl #74198] Typo in perlop(1)
Rafael Garcia-Suarez [Tue, 13 Apr 2010 09:42:12 +0000]
[perl #74198] Typo in perlop(1)

reported by rrt@sc3d.org

14 years agowhen(scalar) without argument should be a syntax error [perl #74114]
Rafael Garcia-Suarez [Fri, 9 Apr 2010 10:25:54 +0000]
when(scalar) without argument should be a syntax error [perl #74114]

14 years agoIgnore Devel/DProf test files in the git repository
Rafael Garcia-Suarez [Fri, 9 Apr 2010 10:29:04 +0000]
Ignore Devel/DProf test files in the git repository

14 years agoAdd the 5.13 deprecation stanza to Module::Corelist to make tests happy
Jesse Vincent [Tue, 13 Apr 2010 03:52:50 +0000]
Add the 5.13 deprecation stanza to Module::Corelist to make tests happy

14 years agofeature.pm now knows about 5.13
Jesse Vincent [Tue, 13 Apr 2010 03:32:12 +0000]
feature.pm now knows about 5.13

14 years agoBump to 5.13.0
Jesse Vincent [Tue, 13 Apr 2010 03:15:56 +0000]
Bump to 5.13.0

14 years agoCreate a perl5130delta.pod
Jesse Vincent [Tue, 13 Apr 2010 03:03:05 +0000]
Create a perl5130delta.pod

14 years agoRemoving the RC marker from patchlevel.h
Jesse Vincent [Mon, 12 Apr 2010 13:01:09 +0000]
Removing the RC marker from patchlevel.h

14 years agoRevert "* Fixed typo in toke.c docs, identified by Zefram"
Jesse Vincent [Mon, 12 Apr 2010 12:43:28 +0000]
Revert "* Fixed typo in toke.c docs, identified by Zefram"

The commit was good, but we're in freeze  for 5.12.0. I'd be happy to
see this hit blead again after 5.12.0 is tagged.

This reverts commit 675ac12c19e6fe00eff6e604a7d637bf621997ef.

14 years agoremove 'enable taint if modify gid/uid' feature
David Mitchell [Sun, 11 Apr 2010 22:45:29 +0000]
remove 'enable taint if modify gid/uid' feature

If at runtime you modify any of any the id variables $<, $>, $(, $),
such that the id and effective id differ, perl used to enable tainting,
even if -T wasn't specified at startup.

This commit removes that feature. See

    http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2010-03/mail9.html
    RT #67260

14 years agoRT 8857: premature free in local of tied element
David Mitchell [Sun, 11 Apr 2010 14:52:43 +0000]
RT 8857: premature free in local of tied element

[The original bug report concerned local($_) remained tied, but while
looking at it, Nicholas found some related code that popped up premature
free errors. This commit fixes the freeing issue rather than the issue of
the original bug report ]

Background:

    local $a[0]

does, approximately:

    svp = av_fetch(av);
    SAVE(av,*svp);
    sv = newSV();
    *svp = sv;

This used to leak when av was tied, as the new sv only got embedded in
*svp, which for tied arrays is a temporary placeholder rather than
somewhere within AvARRAY. This leak was fixed in 2002 by adding the
following:

    if (SvTIED_mg(sv, PERL_MAGIC_tiedelem))
sv_2mortal(sv);

which worked, except for the following:

    sub f { local $_[0] }
    f($_) for ($tied[0]);

Here, @_ is a real array not a tied one, yet its first element is a
PERL_MAGIC_tiedelem which trigged the test above. So the sv got
mortalised *and* stored in the array, so got freed twice. The fix is
to test the *array/hash* for tied-ness rather than the element.

14 years agocreate SV leak test infrastructure
David Mitchell [Sun, 11 Apr 2010 14:23:17 +0000]
create SV leak test infrastructure

Add an sv_count() function to XS::APItest to allow access to PL_sv_count,
then add new test file t/op/svleak.t that allows you yo run a code
fragment a few times and test whether the number of allocated SVs has
increased

14 years ago* Fixed typo in toke.c docs, identified by Zefram
brian d foy [Sun, 11 Apr 2010 12:15:27 +0000]
* Fixed typo in toke.c docs, identified by Zefram

14 years agoadjust release dates due to rc5
Jesse Vincent [Sat, 10 Apr 2010 03:20:36 +0000]
adjust release dates due to rc5

14 years agoAdd bug to known issues for perl-v5.12.0
Josh ben Jore [Fri, 9 Apr 2010 20:44:09 +0000]
Add bug to known issues for perl-v5.12.0

14 years ago[perl #74168] Assertion failure when emitting a stricture error message
Rafael Garcia-Suarez [Fri, 9 Apr 2010 12:19:00 +0000]
[perl #74168] Assertion failure when emitting a stricture error message

14 years agoRC5 due to a regression reported by <mons@veda.park.rambler.ru>
Jesse Vincent [Sat, 10 Apr 2010 03:17:44 +0000]
RC5 due to a regression reported by <mons@veda.park.rambler.ru>

14 years agofix RT 23810: eval and tied methods
David Mitchell [Thu, 8 Apr 2010 12:16:56 +0000]
fix RT 23810: eval and tied methods

Something like the following ended up corrupted:
    sub FETCH { eval 'BEGIN{syntax err}' }
The croak on error popped back the context stack etc to the EVAL pushed by
entereval, but the corresponding JUMPENV_PUSH(3) unwound all the way to the
outer perl_run, losing all the mg_get() related parts of the C stack.

It turns out that the run-time parts of pp_entereval were protected with
a new JUMPENV level, but the compile-time parts weren't. Add this.

14 years agoExplicitly set the location of cpanp-run-perl when under PERL_CORE
Jos Boumans [Tue, 6 Apr 2010 21:19:08 +0000]
Explicitly set the location of cpanp-run-perl when under PERL_CORE

    CPANPLUS::Config checks 3 specific scenarios first
    when looking for cpanp-run-perl: parallel to cpanp,
    parallel to CPANPLUS.pm, or installed into a custom
    prefix like /tmp/foo. Only *THEN* does it check the
    the path.
    If the perl core is extracted to a directory that has
    cpanp-run-perl installed the same amount of 'uplevels'
    as the /tmp/foo prefix, we'll pull in the wrong script
    by accident.
    Since we set the path to cpanp-run-perl explicitily
    at the top of this script, it's best to update the config
    ourselves with a path lookup, rather than rely on its
    heuristics. Thanks to David Wheeler, Josh Jore and Vincent
    Pit for helping to track this down.

Effects both CPANPLUS and CPANPLUS-Dist-Build

14 years agomore mods to -Dl debugging output
David Mitchell [Tue, 6 Apr 2010 19:53:45 +0000]
more mods to -Dl debugging output

14 years agoSkip make_path group/user tests for darwin - it's nobody and nogroup ids are -1
Josh ben Jore [Mon, 5 Apr 2010 23:36:10 +0000]
Skip make_path group/user tests for darwin - it's nobody and nogroup ids are -1

14 years agoUpdate perldelta for the removal of the deprecation warnings on perl4-era libraries.
Jesse Vincent [Mon, 5 Apr 2010 13:23:22 +0000]
Update perldelta for the removal of the deprecation warnings on perl4-era libraries.

14 years agoRemove perl4-library deprecation warnings
Steffen Mueller [Mon, 5 Apr 2010 13:15:38 +0000]
Remove perl4-library deprecation warnings

After the recent, long discussion about this topic and demonstrated
cases where the deprecation warnings may actually *break* things, we
agreed on IRC to push the "active" deprecation warnings to after 5.12.

--Steffen

From f1b2d650aa126e06fc270dd0a44b8a6bf0da6e2c Mon Sep 17 00:00:00 2001
From: Steffen Mueller <smueller@cpan.org>
Date: Mon, 5 Apr 2010 15:12:41 +0200
Subject: [PATCH] Remove deprecation warnings from perl4-era tools

The libraries still have deprecation warnings in their comments but the
mandatory run-time warnings are disabled until after 5.12.

14 years agoMost of the REGEXP-massaging modules have now been updated.
Jesse Vincent [Mon, 5 Apr 2010 13:05:19 +0000]
Most of the REGEXP-massaging modules have now been updated.
Moved the single known-failing example to the errata page.

14 years agoWe're gonna have a fourth RC.
Jesse Vincent [Mon, 5 Apr 2010 13:06:58 +0000]
We're gonna have a fourth RC.

14 years agoThere's gonna be an RC3
Jesse Vincent [Fri, 2 Apr 2010 19:22:22 +0000]
There's gonna be an RC3