9 months agoUpdate META files v5.18.2
Ricardo Signes [Mon, 6 Jan 2014 23:13:39 +0000]
Update META files

9 months agodisarm the release candidate marker
Ricardo Signes [Mon, 6 Jan 2014 22:47:49 +0000]
disarm the release candidate marker

10 months agoadd haarg to AUTHORS v5.18.2-RC4
Ricardo Signes [Sat, 21 Dec 2013 23:25:54 +0000]
add haarg to AUTHORS

10 months agoskip RC tags when checking versions diffs
Graham Knop [Sat, 21 Dec 2013 23:08:12 +0000]
skip RC tags when checking versions diffs

10 months agobump to RC4
Ricardo Signes [Sat, 21 Dec 2013 21:12:36 +0000]
bump to RC4

10 months agoremove a now-fixed podcheck known problem
Ricardo Signes [Sat, 21 Dec 2013 20:51:38 +0000]
remove a now-fixed podcheck known problem

10 months agoperldelta for the rest of the changes since RC3
Ricardo Signes [Sat, 21 Dec 2013 20:38:07 +0000]
perldelta for the rest of the changes since RC3

10 months agoupdate CoreList for B::Concise update
Ricardo Signes [Sat, 21 Dec 2013 20:37:49 +0000]
update CoreList for B::Concise update

10 months agobump version on B::Concise
Ricardo Signes [Sat, 21 Dec 2013 20:32:09 +0000]
bump version on B::Concise

10 months agoUpdate README copyright to 2013
Smylers [Fri, 6 Sep 2013 05:02:41 +0000]
Update README copyright to 2013

For RT #119625.

(cherry picked from commit a271a376b9ff839eed6d1db3181e47e01d846591)

10 months agoupdate the super quick patch guide
Ricardo Signes [Fri, 20 Dec 2013 17:00:57 +0000]
update the super quick patch guide

This is an import of the blead perlhack.pod from blead,
with changes made by Smylers <Smylers@stripey.com> and
detailed in [perl #119599].

10 months agoRestore perlrepository.pod in stub form
Smylers [Tue, 3 Sep 2013 13:52:12 +0000]
Restore perlrepository.pod in stub form

Give Perl doc sites a sane ‘latest’ version to display, directing readers
to current information, rather than showing the Perl 5.12 version in
perpetuity.

And help anybody typing man perlrepository find where the docs have moved
to.

Suggested by Father Chrysostomos in:
http://www.nntp.perl.org/group/perl.perl5.porters/2013/09/msg207079.html

(cherry picked from commit 5edbc4ffd619bf985c0d95ab3981022eea5bcc99)

10 months agoFix segfault in filehandle duplication
Leon Timmermans [Sat, 7 Sep 2013 06:58:18 +0000]
Fix segfault in filehandle duplication

Previously PerlIOBase_dup didn't check if pushing the new layer succeeded
before (optionally) setting the utf8 flag. This could cause
segfaults-by-nullpointer.

(cherry picked from commit df8c7dee25da69fc88678b8949166e08fb686037)

10 months agoFix buffer overflow with overlong identifiers
Father Chrysostomos [Mon, 9 Sep 2013 07:35:38 +0000]
Fix buffer overflow with overlong identifiers

This was introduced by commit 32833930e32dc in 5.17.10.

$  ./perl -Ilib -e Foooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooooo
Identifier too long at -e line 1.
Segmentation fault: 11

(That was an amusing use of macro parentheses for the while condition,
at least while it lasted.)

(cherry picked from commit eaaaaa32882752f15fc0db1c73a1adbe34b49642)

Conflicts:
t/comp/parser.t

10 months agoDon’t assume targs are contiguous for ‘my $x; my $y’
Father Chrysostomos [Mon, 2 Sep 2013 15:29:50 +0000]
Don’t assume targs are contiguous for ‘my $x; my $y’

In commit 18c931a3, the padrange optimisation was prevented from mak-
ing this assumption for ‘my ($x,$y)’, but the assumption was still
there in the code that combines multiple statements into one.

This would lead to assertion failures (or, as of ce0d59f, crashes
under non-debugging builds) if a keyword plugin declined to handle
the second ‘my’, but only after creating a padop.

This fixes a regression from 5.16 affecting Devel::CallParser under
threaded builds.

(cherry picked from commit 7601007bbbe673d3791a2d77c692a0c377835430)

10 months agoUnbreak Concise glob output
Father Chrysostomos [Sun, 25 Aug 2013 14:07:37 +0000]
Unbreak Concise glob output

$ ./perl -MO=Concise -Ilib -le '; print <\n>'
8  <@> leave[1 ref] vKP/REFC ->(end)
1     <0> enter ->2
2     <;> nextstate(main 23 -e:1) v:{ ->3
7     <@> print vK ->8
3        <0> pushmark s ->4
6        <@> glob[t2] lK/1 ->7
-           <0> ex-pushmark s ->4
4           <$> const[PV "\n"] s ->5
Can't locate object method "NAME" via package "B::SPECIAL" at lib/B/Concise.pm line 707.
CHECK failed--call queue aborted.

e88567f2acf38fe5ed90a88569b808e82cd3eca1 is the first bad commit
commit e88567f2acf38fe5ed90a88569b808e82cd3eca1
Author: Father Chrysostomos <sprout@cpan.org>
Date:   Sun Nov 4 20:18:51 2012 -0800

    Stop the glob operator from leaking GVs

    It was adding GVs to the symbol table (via newGVgen), so they
    would never be freed, even after the op was freed, unless done so
    explicitly.

    There is no reason for these GVs to be exposed.

That means $gv->STASH returns a B::SPECIAL.  B::Concise was not pre-
pared to handle that.

(cherry picked from commit 7b696247e72c0b92f1fcee2a830851021827aa4b)

10 months agoBackport Module-CoreList changes for v5.19.7 for sanity
Chris 'BinGOs' Williams [Sat, 21 Dec 2013 16:53:58 +0000]
Backport Module-CoreList changes for v5.19.7 for sanity

10 months agoupdate patchlevel to RC3 v5.18.2-RC3
Ricardo Signes [Thu, 19 Dec 2013 14:08:18 +0000]
update patchlevel to RC3

10 months agoupdate tentative release date in corelist
Ricardo Signes [Thu, 19 Dec 2013 14:08:03 +0000]
update tentative release date in corelist

10 months agoupdate tentative release date in perlhist
Ricardo Signes [Thu, 19 Dec 2013 14:07:53 +0000]
update tentative release date in perlhist

10 months agoperldelta for 18ca6dc3
Ricardo Signes [Thu, 19 Dec 2013 14:04:07 +0000]
perldelta for 18ca6dc3

10 months agoUnbreak -bareword under strict+integer
Father Chrysostomos [Tue, 22 Oct 2013 12:36:38 +0000]
Unbreak -bareword under strict+integer

Commit 077da62ff9 was not supposed to change behaviour, but only
remove logic rendered unnecessary two commits earlier in 1c2b3fd6f10.

But the special stricture exception for negation was in the same func-
tion (S_op_integerize) which applied it to OP_NEGATE, but now needed
to apply it to OP_I_NEGATE, too.

(cherry picked from commit fcbc518d7ac7875b7f443e72caf15fd07ab023a6)

10 months agoadd perldelta for 947ba0f
Ricardo Signes [Thu, 19 Dec 2013 14:00:31 +0000]
add perldelta for 947ba0f

10 months agoPATCH: [perl #120799] 5.18 regression with [:^ascii] and \x80-\xFF
Karl Williamson [Mon, 16 Dec 2013 17:59:36 +0000]
PATCH: [perl #120799] 5.18 regression with [:^ascii] and \x80-\xFF

Posix classes generally match different sets of characters under /d
rules than otherwise.  This isn't true for [:ascii:], but the handling
for it is shared with the others, so it needs to use the same mechanism
to deal with that.  I forgot this in commit
bb9ee97444732c84b33c2f2432aa28e52e4651dc which created this regression.

Our tests for this only use regexes with a single element, and an
optimization added in 5.18 causes this bug to be bypassed.  These tests
should be enhanced to force both code paths, but not for this commit,
which should be suitable for a maintenance release.

(cherry picked from commit 46c10357a881cd92500e4ade81cbc8813e49e2cb)

10 months agoperldelta for cherry pick of 257dc59 v5.18.2-RC2
Ricardo Signes [Sat, 7 Dec 2013 02:28:04 +0000]
perldelta for cherry pick of 257dc59

10 months agoupdate versions and dates for a second 5.18.2 RC
Ricardo Signes [Sat, 7 Dec 2013 02:14:28 +0000]
update versions and dates for a second 5.18.2 RC

10 months ago[perl #120694] Fix ->SUPER::foo and AUTOLOAD
Father Chrysostomos [Thu, 5 Dec 2013 00:00:48 +0000]
[perl #120694] Fix ->SUPER::foo and AUTOLOAD

Commit aae438050a20 (5.17.4) broke ->SUPER::foo with AUTOLOAD by look-
ing up AUTOLOAD from the current package, rather than the current
package’s superclass.

Instead of keeping track of whether it was doing a SUPER lookup via a
::SUPER prefix on the package name, that commit changed method lookup
to pass a GV_SUPER flag around (to fix another bug) and to pass the
current stash, rather than __PACKAGE__::SUPER.  But it did not update
gv_autoload_pvn to pass that flag through to gv_fetchmeth_pvn when
actually looking up the method.

(cherry picked from commit 257dc59d7b864a6cf0ccc9179de1f3f0a797f4e0)

Conflicts:
t/op/method.t

10 months agoarm the "this is RC1" alert v5.18.2-RC1
Ricardo Signes [Mon, 2 Dec 2013 16:00:09 +0000]
arm the "this is RC1" alert

10 months agocorelist: push back 5.18.2 release date
Ricardo Signes [Sun, 1 Dec 2013 13:53:29 +0000]
corelist: push back 5.18.2 release date

10 months agopush back 5.18.2 release date
Ricardo Signes [Sun, 1 Dec 2013 13:53:19 +0000]
push back 5.18.2 release date

10 months agoupdate perldelta for File::Glob version bump
Ricardo Signes [Sun, 1 Dec 2013 02:40:22 +0000]
update perldelta for File::Glob version bump

10 months agobump File::Glob version for 5.18.2
Ricardo Signes [Sun, 1 Dec 2013 02:29:56 +0000]
bump File::Glob version for 5.18.2

10 months agoFix ext/File-Glob/t/threads.t filenames for VMS.
Craig A. Berry [Wed, 2 Oct 2013 02:43:13 +0000]
Fix ext/File-Glob/t/threads.t filenames for VMS.

As noted in

http://www.nntp.perl.org/group/perl.perl5.porters/2013/09/msg208134.html

zero-length extensions always get a trailing dot on VMS, and the
easiest workaround is to always use an explicit extension.

(cherry picked from commit 43ed1b742e2b7be9184e1fb35c0f68d15b87feed)

10 months agoFile::Glob: Dup glob state in CLONE()
Brian Fraser [Sat, 21 Sep 2013 06:19:52 +0000]
File::Glob: Dup glob state in CLONE()

This solves [perl #119897] and [perl #117823], and restores the
behavior of glob() in conjunction with threads of 5.14 and older.

Since 5.16, code that used glob() inside a thread had been
unintentionally sharing state between threads, which lead to things
like this crashing and failing assertions:

./perl -Ilib -Mthreads -e 'scalar glob("*"); threads->create(sub { glob("*") })->join();'

(cherry picked from commit facf34ef484d62d15b2da11ee03d01942a22ff15)

10 months ago[perl #114864] Make UNIVERSAL::DESTROY changes invalidate caches
Father Chrysostomos [Sun, 18 Aug 2013 06:58:47 +0000]
[perl #114864] Make UNIVERSAL::DESTROY changes invalidate caches

Commit 8c34e50d inadvertently caused DESTROY caches not to be
reset when UNIVERSAL::DESTROY changes.  Normally, a change to
a method will cause mro_method_changed_in to be called on all
subclasses, but mro.c cheats for UNIVERSAL and just does
++PL_sub_generation.  So clearing the DESTROY cache explicitly
in mro_method_changed_in is clearly not enough.

(cherry picked from commit c716b3beb77406159d18fd52251821fee641f9fc)

10 months agoreparse compile-time /(?{})/ in right scope
David Mitchell [Tue, 6 Aug 2013 15:34:50 +0000]
reparse compile-time /(?{})/ in right scope

When a compile-time regex like /...(?{ code-block }) .../
is compiled in the presence of constant and concat overloading,
this can cause (still at compile-time) for the pattern to be evaled and
re-compiled, in order to re-compile any code-blocks that got messed up
during the overloading and thus whose text no longer matches that which
the perl parser previously compiled.

When this happens, eval_sv() happens to be called when the perl parser is
still in compiling state; normally its called from running state.
This tickles an undiscovered bug in Perl_find_runcv_where(), which
finds the current cop sequence by looking at PL_curcop->cop_seq.
At compile time, we need to get it from PL_cop_seqmax instead.

(cherry picked from commit c3923c33af542d8764d5a1e4eb5d7b311f443b89)

10 months agoPATCH [perl #119713] Regex \8 and \9 after literals no longer work
Karl Williamson [Wed, 27 Nov 2013 03:43:44 +0000]
PATCH [perl #119713] Regex \8 and \9 after literals no longer work

Commit 726ee55d introduced a regression that has been fixed in blead by
commit f1e1b256.  However the later commit changed some buggy behavior
into errors instead of warnings, and so is contraindicated in a
maintenance release.  This current commit attempts to fix the regression
without changing other behavior.  It includes the pat.t tests from f1e1b256.

10 months agoadd tentative 5.18.2 release date
Ricardo Signes [Fri, 22 Nov 2013 02:40:15 +0000]
add tentative 5.18.2 release date

11 months agoFor SDBM_File, stop EU::MM from generating its default subdirs rule.
Nicholas Clark [Fri, 22 Nov 2013 16:43:52 +0000]
For SDBM_File, stop EU::MM from generating its default subdirs rule.

The default subdirs rule creates a race condition with the rule that
Makefile.PL explicitly adds to generate libsdbm.a, which can cause parallel
makes to fail.

Signed-off-by: Chris 'BinGOs' Williams <chris@bingosnet.co.uk>

11 months agoremove 5.19.7 tentative entries from corelist
Ricardo Signes [Thu, 21 Nov 2013 18:50:34 +0000]
remove 5.19.7 tentative entries from corelist

11 months agoadd new CoreList tests to manifest
Ricardo Signes [Thu, 21 Nov 2013 03:42:15 +0000]
add new CoreList tests to manifest

11 months agoadd the v5.18.2 changes to Module::CoreList
Ricardo Signes [Thu, 21 Nov 2013 03:27:13 +0000]
add the v5.18.2 changes to Module::CoreList

11 months agoimport Module-CoreList from blead
Ricardo Signes [Thu, 21 Nov 2013 03:22:11 +0000]
import Module-CoreList from blead

11 months agoupdate perldelta for 5.18.2
Ricardo Signes [Thu, 21 Nov 2013 03:18:10 +0000]
update perldelta for 5.18.2

11 months agobump version to v5.18.2
Ricardo Signes [Thu, 21 Nov 2013 01:58:25 +0000]
bump version to v5.18.2

11 months agoEnglish.pm: update perl version where perf fixed
David Mitchell [Wed, 24 Jul 2013 13:18:22 +0000]
English.pm: update perl version where perf fixed

It still said that the performance of $`, $&, $' was fixed in 5.18.
Update that to 5.20, since COW wasn't enabled by default in 5.18.

Conflicts:
lib/English.pm

11 months agoDisable new //p tests
David Mitchell [Tue, 30 Jul 2013 19:14:37 +0000]
Disable new //p tests

The previous commit fixed a regression in 5.18.0:

    my $pat = qr/a/;
    'aaaa' =~ /$pat/gp or die;
    print "MATCH=[${^MATCH}]\n";

which printed 'a' in 5.16.0, and undef in 5.18.0.
5.18.0 only broke the /g behaviour; the non-/g variant was already broken
and the previous commit didn't fix that for maint.

The new tests are testing for the non-/g variant, which still fail, so
disable these tests.

(Getting it to capture under non-/g works in 5.19.2 due to a major
reorganisation of the pp_match/pp_subst/regexec() code, which isn't
suitable for backporting to maint-5.18.)

11 months agoRT #118213: handle $r=qr/.../; /$r/p properly
David Mitchell [Tue, 30 Jul 2013 15:16:35 +0000]
RT #118213: handle $r=qr/.../; /$r/p properly

(cherry-picked from 5b0e71e9d506. Some of the new tests are unsuitable for
5.18.x and fail with this commit; they'll be disabled in the next commit)

In the case where a qr// regex is directly used by PMOP (rather than being
interpolated with some other stuff and a new regex created, such as
/a$r/p), then the PMf_KEEPCOPY flag will be set on the PMOP, but the
corresponding RXf_PMf_KEEPCOPY flag *won't* be set on the regex.

Since most of the regex handling for copying the string and extracting out
${^PREMATCH} etc is done based on the RXf_PMf_KEEPCOPY flag in the regex,
this is a bit of a problem.

Prior to 5.18.0 this wasn't so noticeable, since various other bugs around
//p handling meant that ${$PREMATCH} etc often accidentally got set
anyway. 5.18.0 fixed these bugs, and so as a side-effect, exposed the
PMOP verses regex flag issue. In particular, this stopped working in
5.18.0:

    my $pat = qr/a/;
    'aaaa' =~ /$pat/gp or die;
    print "MATCH=[${^MATCH}]\n";

(prints 'a' in 5.16.0, undef in 5.18.0).
The presence /g caused the engine to copy the string anyway by luck.

We can't just set the RXf_PMf_KEEPCOPY flag on the regex if we see the
PMf_KEEPCOPY flag on the PMOP, otherwise stuff like this will be wrong:

    $r = qr/..../;
    /$r/p;  # set RXf_PMf_KEEPCOPY on $r
    /$r/; # does a /p match by mistake

Since for 5.19.x onwards COW is enabled by default (and cheap copies are
always made regardless of /p), then this fix is mainly for PERL_NO_COW
builds and for backporting to 5.18.x. (Although it still applies to
strings that can't be COWed for whatever reason).

Since we can't set a flag in the rx, we fix this by:

1) when calling the regex engine (which may attempt to copy part or all of
the capture string), make sure we pass REXEC_COPY_STR, but neither of
REXEC_COPY_SKIP_PRE, REXEC_COPY_SKIP_POST when we call regexec() from
pp_match or pp_subst when the corresponding PMOP has PMf_KEEPCOPY set.

2) in Perl_reg_numbered_buff_fetch() etc, check for PMf_KEEPCOPY in
PL_curpm as well as for RXf_PMf_KEEPCOPY in the current rx before deciding
whether to process ${^PREMATCH} etc.

As well as adding new tests to t/re/reg_pmod.t, I also changed the
string to be matched against from being '12...' to '012...', to ensure that
the lengths of ${^PREMATCH}, ${^MATCH}, ${^POSTMATCH} would all be
different.

12 months agoperldelta for 89c16a74d5f.
Craig A. Berry [Sun, 29 Sep 2013 02:40:02 +0000]
perldelta for 89c16a74d5f.

12 months agoFix processing of PERL_ENV_TABLES.
Craig A. Berry [Sat, 7 Sep 2013 11:55:25 +0000]
Fix processing of PERL_ENV_TABLES.

In a35dcc95dd24524931e I "improved" string safety in vms/vms.c by
converting to my_strlcpy and my_strlcat, but mangled the length
argument to my_strlcat when adding the name of the logical name
table specified in PERL_ENV_TABLES. This caused the command string
to be truncated, so a command that, for example, should have been:

    $ Show Logical * /Table=LNM$JOB
    ...

actually became:

    $ Show Logical * /Table=
    %DCL-W-VALREQ, missing qualifier or keyword value - supply all required values

Plus it turns out the strings holding the names of the tables were
being stored in dynamic string descriptors and were not
NUL-terminated, but the strl* functions require NUL-terminated
arguments.  So change those to static string descriptors and
allocate the exact amount of storage needed including room for a
NUL.

This was a regression in 5.16.0, first reported a couple of days
ago by Mark Daniel on comp.os.vms:

Date: Fri, 06 Sep 2013 12:56:01 +0930
From: Mark Daniel <mark.daniel [AT] wasd.vsm.com.au>
Newsgroups: comp.os.vms
Message-ID: <52294b4a$0$2875$c3e8da3$76491128@news.astraweb.com>

TODO: Figure out how and where to test this.

13 months agodocument fixing of #119927 (localizing $\) in 5.18.0
Ricardo Signes [Mon, 23 Sep 2013 15:52:47 +0000]
document fixing of #119927 (localizing $\) in 5.18.0

13 months agoperldelta for 0e903f68ee
Tony Cook [Wed, 11 Sep 2013 00:10:12 +0000]
perldelta for 0e903f68ee

13 months ago[perl #119351] fix B::CV::GV to return a B::SPECIAL on NULL
Tony Cook [Tue, 10 Sep 2013 23:40:11 +0000]
[perl #119351] fix B::CV::GV to return a B::SPECIAL on NULL

Cherry pick of

31d073689e2b5fa63aa7d4f8aaf529f75eeb5d0e
5f7c1602dfa694a4a6761e9e4fc077ce794f7ff0
f244b085582eaec9f2cf7bfabeb686bf74167bf6

combined with a version bump.

14 months agoadd new perldelta
Ricardo Signes [Mon, 12 Aug 2013 14:50:38 +0000]
add new perldelta

14 months agoupdate META.* for 5.18.1 (stable) release v5.18.1
Ricardo Signes [Mon, 12 Aug 2013 02:55:40 +0000]
update META.* for 5.18.1 (stable) release

14 months agopatchlevel.h: remove RC3 indicator
Ricardo Signes [Mon, 12 Aug 2013 02:47:08 +0000]
patchlevel.h: remove RC3 indicator

This is it, folks!

14 months agoperlhist: update dates for 5.18.1 in light of RC3
Ricardo Signes [Fri, 9 Aug 2013 02:05:17 +0000]
perlhist: update dates for 5.18.1 in light of RC3

14 months agoupdate CoreList dates for 5.18.1 in light of RC3
Ricardo Signes [Fri, 9 Aug 2013 02:04:47 +0000]
update CoreList dates for 5.18.1 in light of RC3

14 months agoupdate patchlevel.h for v5.18.1-RC3
Ricardo Signes [Fri, 9 Aug 2013 02:04:27 +0000]
update patchlevel.h for v5.18.1-RC3

14 months agoSkip the perldoc test if 'man' is missing
Niko Tyni [Tue, 6 Aug 2013 13:57:00 +0000]
Skip the perldoc test if 'man' is missing

This fixes a test failure in minimal build environments without
a 'man' command.

14 months agoThe language messages variable is called LC_MESSAGES
Niko Tyni [Tue, 6 Aug 2013 13:45:59 +0000]
The language messages variable is called LC_MESSAGES

This probably doesn't matter as LC_ALL overrides everything else.

14 months agoperldelta: mention v5.16, not v5.17, as previous version
Ricardo Signes [Thu, 8 Aug 2013 11:52:09 +0000]
perldelta: mention v5.16, not v5.17, as previous version

Thanks for the report, SHIRAKATA Kentaro.

14 months agoModule::CoreList: update Config found in 5.18.1
Ricardo Signes [Thu, 8 Aug 2013 00:45:19 +0000]
Module::CoreList: update Config found in 5.18.1

14 months agoAdd missing versioned Config to Module::CoreList
Chris 'BinGOs' Williams [Wed, 7 Aug 2013 22:48:21 +0000]
Add missing versioned Config to Module::CoreList

(cherry picked from commit cc42383380dce0d2e3c98f9f467a21c936902a8d)

14 months agoadd Module::CoreList::Utils delta for 5.18.1
Ricardo Signes [Mon, 5 Aug 2013 13:31:35 +0000]
add Module::CoreList::Utils delta for 5.18.1

14 months agoupdate perlhist and patchlevel for v5.18.1-RC2
Ricardo Signes [Sun, 4 Aug 2013 02:08:56 +0000]
update perlhist and patchlevel for v5.18.1-RC2

14 months agoperlhist: add 5.18.1!
Ricardo Signes [Fri, 2 Aug 2013 03:06:51 +0000]
perlhist: add 5.18.1!

14 months agoRT #119125: continue /[#$x]/x not interpolating
David Mitchell [Sat, 3 Aug 2013 15:04:45 +0000]
RT #119125: continue /[#$x]/x not interpolating

the earlier fix for /[#](?{})/x, although correct, as a
side-effect fixed another long-standing bug where /[#$x]/x
didn't interpolate the var $x. Although fixing that is good,
it's too big a change for maint, so keep the old buggy behaviour
for now.

14 months agoperldelta: remove a leftover XXX section
Ricardo Signes [Sun, 4 Aug 2013 01:30:10 +0000]
perldelta: remove a leftover XXX section

14 months agoperldelta: fix module updates section
Ricardo Signes [Sun, 4 Aug 2013 01:29:58 +0000]
perldelta: fix module updates section

14 months agoupdate corelist with v5.18.1 changes
Ricardo Signes [Sun, 4 Aug 2013 01:24:38 +0000]
update corelist with v5.18.1 changes

...again.  This time, using the updated updater, fixing the
deprecations screwed up in 5.18.1 RC1.

14 months agoimport updates to Porting/corelist.pl
Ricardo Signes [Sun, 4 Aug 2013 01:09:54 +0000]
import updates to Porting/corelist.pl

14 months agoRevert "update corelist with v5.18.1 changes"
Ricardo Signes [Sun, 4 Aug 2013 01:08:02 +0000]
Revert "update corelist with v5.18.1 changes"

This reverts commit f97801705daa638dc14b1ce38cac3542064df879.

14 months agoUpdate META.json for RC1
Ricardo Signes [Fri, 2 Aug 2013 01:38:48 +0000]
Update META.json for RC1

14 months agobump version to v5.18.1-RC1
Ricardo Signes [Fri, 2 Aug 2013 01:37:49 +0000]
bump version to v5.18.1-RC1

14 months agoupdate corelist with v5.18.1 changes
Ricardo Signes [Fri, 2 Aug 2013 01:30:38 +0000]
update corelist with v5.18.1 changes

I am not following the instructions in the RMG, because the
procedure is sort of a mess.  I will make sure that blead gets into
the right state when this is done.  In the meantime, doing it this
way should require less weird moving of work back and forth.

14 months agoimport Module-Corelist 2.95 from blead
Ricardo Signes [Fri, 2 Aug 2013 01:11:33 +0000]
import Module-Corelist 2.95 from blead

14 months agoversion bump, because Digest::SHA changed
Ricardo Signes [Fri, 2 Aug 2013 01:05:04 +0000]
version bump, because Digest::SHA changed

14 months agoperlhist: import the latest perlhist from blead
Ricardo Signes [Fri, 2 Aug 2013 01:00:10 +0000]
perlhist: import the latest perlhist from blead

14 months agoperldelta: remove sections we will not need
Ricardo Signes [Thu, 1 Aug 2013 23:56:45 +0000]
perldelta: remove sections we will not need

14 months agoperldelta: perldelta for /[#]/ and /(?[...])/ changes
Ricardo Signes [Thu, 1 Aug 2013 23:06:02 +0000]
perldelta: perldelta for /[#]/ and /(?[...])/ changes

14 months agoperldelta: hints updates from 4b985f1 and 9de5f95
Ricardo Signes [Thu, 1 Aug 2013 23:05:30 +0000]
perldelta: hints updates from 4b985f1 and 9de5f95

14 months agoperldelta: Digest::SHA fix in 79fe145
Ricardo Signes [Thu, 1 Aug 2013 23:05:06 +0000]
perldelta: Digest::SHA fix in 79fe145

14 months agoperldelta: the easy pre-editing before final checks
Ricardo Signes [Thu, 1 Aug 2013 13:32:37 +0000]
perldelta: the easy pre-editing before final checks

14 months agoperldelta: add a missing "=" from a Pod directive
Ricardo Signes [Thu, 1 Aug 2013 13:23:10 +0000]
perldelta: add a missing "=" from a Pod directive

14 months agoDon’t leak when compiling /(?[\d\d])/
Father Chrysostomos [Thu, 6 Jun 2013 21:14:06 +0000]
Don’t leak when compiling /(?[\d\d])/

The ‘Operand with no preceding operator’ error was leaking the last
two operands.

(cherry picked from commit b573e7000fd9c1cfae30ae5fb328a25b9bf3870a)

14 months agoFree operand when encountering unmatched ')' in (?[])
Father Chrysostomos [Thu, 6 Jun 2013 21:07:21 +0000]
Free operand when encountering unmatched ')' in (?[])

I only need to free the operand (current), not the left-paren token
that turns out not to be a paren (lparen).

For lparen to leak, there would have to be two operands in a row on
the charclass parsing stack, which currently never happens.

(cherry picked from commit 4bc5d08976b7df23b63a56cc017a20ac5766fbbc)

14 months agoStop /(?[\p{...}])/ compilation from leaking
Father Chrysostomos [Thu, 6 Jun 2013 20:42:32 +0000]
Stop /(?[\p{...}])/ compilation from leaking

The swash returned by utf8_heavy.pl was not being freed in the code
path to handle returning character classes to the (?[...]) parser
(when ret_invlist is true).

(cherry picked from commit c80d037c54749655d40eac068936c5222ce9d8ee)

14 months agoStop (?[]) operators from leaking
Father Chrysostomos [Thu, 6 Jun 2013 15:47:31 +0000]
Stop (?[]) operators from leaking

When a (?[]) extended charclass is compiled, the various operands are
stored as inversion lists in separate SVs and then combined together
into new inversion lists.  The functions that take care of combining
inversion lists only ever free one operand, and sometimes free none.
Most of the operators in (?[]) were trusting the invlist functions to
free everything that was no longer needed, causing (?[]) compilation
to leak invlists.

(cherry picked from commit a84e671a269f736a404a62f21caacc8a431c2aca)

14 months agoDon’t leak the /(?[])/ parsing stack on error
Father Chrysostomos [Thu, 6 Jun 2013 04:40:24 +0000]
Don’t leak the /(?[])/ parsing stack on error

Instead of creating the parsing stack and then freeing it after pars-
ing the (?[...]) construct (leaking it whenever one of the various
errors scattered throughout the parsing code occurs), mortalise it to
begin with and let the mortals stack take care of it.

(cherry picked from commit 1e4f088863436a8019c7d864691903ffdafeefda)

14 months agoHandle /[#]/ and /[(?#]/ with code blocks
David Mitchell [Wed, 31 Jul 2013 21:41:17 +0000]
Handle /[#]/ and /[(?#]/ with code blocks

This is a regression in 5.18.0.

In something like /[#](?{})/x, the perl toker incorrectly sees the '#' as a
comment and skips the code block without parsing it.

(cherry picked from commit c30fc27b4df65a43710b25dd1d2a57d78ee2fe33)

14 months ago[perl #118525] make it work on threaded builds
Tony Cook [Tue, 30 Jul 2013 04:41:56 +0000]
[perl #118525] make it work on threaded builds

(cherry picked from commit 5016d3c45a66ce5b279c253e8a8c2a38a9bd8431)

14 months agoCvGV is no longer a simple struct member access
Tony Cook [Thu, 25 Jul 2013 02:09:00 +0000]
CvGV is no longer a simple struct member access

The same slot is also used for the NAME_HEK for lexical subs, so:

- split B::CV::GV out into its own function that uses the CvGV macro

- add B::CV::NAME_HEK so the name of a lexical sub can be fetched

14 months agoperldelta: deep recursion warnings (07b2687d2/#118521)
Father Chrysostomos [Sun, 28 Jul 2013 06:22:43 +0000]
perldelta: deep recursion warnings (07b2687d2/#118521)

14 months agoperldelta for bdbfc51a7b (undef constant my sub)
Father Chrysostomos [Sat, 13 Jul 2013 21:31:31 +0000]
perldelta for bdbfc51a7b (undef constant my sub)
(cherry picked from commit 07a522a526321341d2a83a852eda7f6fecb333c9)

14 months agoperldelta for #118305/88dbe4a
Father Chrysostomos [Sat, 13 Jul 2013 18:57:02 +0000]
perldelta for #118305/88dbe4a
(cherry picked from commit 81d3ed5a0fa49c09ac57f7edc0917f345ff6b160)

14 months agoperldelta for lexsub syntax errors (3a74e0e282c)
Father Chrysostomos [Mon, 10 Jun 2013 07:50:47 +0000]
perldelta for lexsub syntax errors (3a74e0e282c)
(cherry picked from commit 3221bc400f3da87f0d721c2bf3501051fac946ea)

14 months agoperldelta for two lexsub fixes
Father Chrysostomos [Mon, 10 Jun 2013 07:49:26 +0000]
perldelta for two lexsub fixes
(cherry picked from commit 81748522aff12edb368615e9512781d00a2d5d5b)

14 months agoStop undef &foo from crashing on lex subs
Father Chrysostomos [Sun, 23 Jun 2013 19:06:11 +0000]
Stop undef &foo from crashing on lex subs
(cherry picked from commit bdbfc51a7bc15a2f0a187c1ef09a16838a4c9915)

14 months agodon't crash on deep recursion warnings in lexical subs (#118521)
Lukas Mai [Tue, 18 Jun 2013 07:51:32 +0000]
don't crash on deep recursion warnings in lexical subs (#118521)
(cherry picked from commit 07b2687d22462e599adb759b7c0082fb12b3f33d)