develooper Front page | perl.perl5.porters | Postings from May 2009

Re: 5.10.0 regressions that need fixing

Thread Previous | Thread Next
Dave Mitchell
May 27, 2009 06:24
Re: 5.10.0 regressions that need fixing
Message ID:
On Mon, Feb 23, 2009 at 06:05:42PM +0000, Dave Mitchell wrote:
> On Mon, Dec 29, 2008 at 10:11:55PM +0000, Dave Mitchell wrote:
> > While ploughing through the backlog on my p5p mailbox (I've currently
> > worked my way up to 2nd Oct), I've been making a list of bug reports etc
> > that seem to indicate a regression between 5.8.8 and 5.10.0, and which
> > don't appear fixed yet in bleed and maint-5.10.x.
> > 
> > If anyone has some time spare to fix bugs, these would probably be good
> > ones to concentrate on. I'll email out a second list once I've processed
> > Oct-Dec.
> Here's a second list, Oct to mid-Dec:
> Agin, these are just my notes i made why reading through my inbox; its
> possible that some of the have been fixed since, or aren't actually
> regressions, or whatever.

And here's a third list, from Dec to May, plus I've also re-included the
first two lists, less the ones I know have been fixed (which wasn't many).

Note that these are probably regressions since 5.8.8 or 5.10.0, and are
therefore a very high priory for getting fixed before 5.10.1 is released.

I'll be working on them myself sometime soon, but I don't think its
possible for me to fix them all single-handedly.


5.10.x bugs / regressions

I emailed the chunk below to p5p, 29-Dec-08


Subject: [perl #50250] perl5.10.0: pos function is much slower with
        "progressive match" and unicode

    5.10 and bleed very slow on unicode pos
    30/12/08 Yves says he'll look into it

Date: Wed, 20 Feb 2008 19:30:34 +0100
Subject: Re: [perl #50114] bug in regcomp code leading to panic

    add to perldelta the fact (missing from 5.10.0 delta) that /m
    no longer propagates into a /...${qr}.../

Date: Mon, 07 Apr 2008 08:45:12 -0700
Subject: [perl #52552] pack "A*" and pack "a*" untaint data in 5.10.0

    both bleed and maint still have a taint bug not present in 5.8.8
    (confirmed still broken 23/4/09)

Date: Sun, 29 Jun 2008 00:38:27 -0700
Subject: [perl #56444] Perl 5.10 breaks \N{} in regex interpolated inside regex
        (with charnames)

    regression since 5.8.x
    30/12/08 Yves says its tricky to fix

Date: Fri, 22 Aug 2008 15:28:30 -0700
Subject: [perl #58280] Speed lost on /^(foo|bar|baz)$/ match

    ((?: open | close | read ) is about 10x slower in 5.10 and bleed
    compared with 5.8.8
    30/12/08 Yves implies not-a-bug

Date: Sat, 30 Aug 2008 09:58:42 -0700
Subject: [perl #58468] crash on exit with fork done in do FILE on Win32

    5.10.0 regression following a null PL_curcop under windows.
    I can't reproduce it on Linux.

Date: Tue, 02 Sep 2008 12:33:34 -0700
Subject: [perl #58530] Bus error with constant + overload + stash manipulation
        + bless

    appears to be a 5.8.8 regression. fails on maint + bleed
    5/1/09 - Ben Morrow provided a patch, not yet applied. Also implied it
	might be older breakage.

Date: Fri, 05 Sep 2008 13:01:34 -0700
Subject: [perl #58614] foreach @foo ( @bar ) causes segfault with

    coredump on bleed, maint. regression since 5.8.8
    I think its the pad-already-freed-ibn yy_stack_clear() issue

    possible 64-bit alignment issue?
    30/12/08 nicholas knows the prob, hasn't had time to fix yet.

Subject: [perl #59498] Segfault on weird but plain Perl in eval on 5.10.0 on

    crashes on 5.10.0 win32; okay on 5.8.8 or 5.10.0 unix
    bug still new

Date: Thu, 02 Oct 2008 03:10:46 -0700
Subject: [perl #59550] readline returns EBADF at eof in 5.10.0

    appears to be a regression since 5.8.8


I emailed the chunk below to p5p, 23-Feb-09

Date: Wed, 08 Oct 2008 14:26:57 -0700
Subject: [perl #59734] Segfault when using (?|) in regexp.

   this crashes on 5.10.0, bleed, but not 5.8.8:
   perl -wE '";" =~ /(?<a>(?|(?<b>;)))/;'
   (looks like segv when cleaning up after an error)

Date: Sat, 25 Oct 2008 10:13:14 +0100
Subject: Re: [perl #60034] (?>) causes wrongness on long string

    regression in 5.10.0 onwards.
    A fix has already been added to 5.8.9, but a different fix is needed
    for 5.10.x, bleed

	From: Zefram <>
	Date: Mon, 23 Feb 2009 22:21:41 +0000
	Subject: Re: 5.10.0 regressions that need fixing

From: Marcus Holland-Moritz <>
To: p5p <>
Date: Sun, 26 Oct 2008 23:17:20 +0100
Subject: [PATCH] Use of freed comppad array during clear_yystack()

    5.10.x clear_yystack does nasty stuff with freed pads.
    MHX has a patch in bleed (a8ba03fb20), but I need to think about it
    some more.

From: Michael G Schwern <>
Date: Fri, 24 Oct 2008 15:53:37 -0700
Subject: Possible mro bug/change between 5.8.8 and 5.10.0

    MRO breaking DBIx::Class::CDBICompat on 5.10.0 ???

Date: Thu, 30 Oct 2008 08:17:23 -0700
Subject: [perl #60232] method cache not updated when dynamic subclass loaded
        through Storable::thaw

    appears to be a regression in 5.10.0

Date: Wed, 29 Oct 2008 21:38:06 -0700
Subject: [perl #60220] mro::method_changed_in(..) ignores AUTOLOAD

    not clear whether this is a regression or not. There's also an offered
    patch which hasn't been applied

Subject: [perl #62056] Variable expansion in same regex as \N{} escape results
        in compilation error.

    in 5.10.0,maint,bleed, but not 5.8.8, this
        use charnames ':full';
    gives an error:
        Constant(\N{START OF HEADING}) unknown: (possibly a missing
	"use charnames ...") in regex;

Date: Mon, 10 Nov 2008 21:27:24 -0800
Subject: [perl #60472] Module Encode degrades in Perl 5.10

    Either Encode is doing something with $1 it shouldn't, or
    there's something up with $1

From: "H.Merijn Brand" <>
To: Perl5 Porters <>
Date: Wed, 12 Nov 2008 13:00:21 +0100
Subject: perl-5.10.x + bignum assertion

    with bignum, get an assertion failure compiling an assign.
    It appears to be a regression from 588, and doesn't fail in bleed
    due to the different way Sv_RVs are done now.

From: Nicholas Clark <>
To:, Tim Bunce <>
Subject: 5.8.9/taint/NYTProf regression
Message-ID: <>

    This may have been a regression in 5.8.9 and 5.10.x.
    probably fixed in 5.10.x fixed by 34962, but I need to review it
    sometime. That caused a regression in Text::Template
    as was fixed by 34987 - see

	From: "Andreas J. Koenig" <>
	Subject: 34974 breaks MJD/Text-Template-1.45.tar.gz
	Date: Tue, 02 Dec 2008 23:24:12 +0100
	Message-ID: <87myfetdv7.fsf@k75.linux.bogus>


I emailed the chunk below to p5p, 27-May-09

Date: Fri, 19 Dec 2008 05:47:50 -0800
Subject: [perl #61520] Segfault in debugger with tr// and UTF8

    assertion failure under debugger in 5.10, bleed, not 5.8.9

    looks like my 'make tr/// threadsafe by moving swash into pad' was

Subject: [perl #63854] Error parsing "[~"
apparently a 5.10.0 regression now fixed in bleed by df3467db, f50fa36f2a6ffc958fea

Subject: [perl #63886] 5.8.9 binary incompatibility: PL_perlio_mutex not
    5.8.9 broke (and maybe 5.10.0 too?) OP includes a patch (not yet

Subject: [perl #61976] Errno ($!) not evaluated to a error message string
        (5.10.0 in taint mode)

    Works in 5.8.8; broken in 5.8.9, 5.10.0, 5.10.x, blead

Subject: [perl #62666] UTF8 failure with sprintf ()

    A panic in 5.10.0, maint, bleed; but not in 5.8.8,9 

Subject: Re: [perl #63110] Memory leak with /ee and use feature

    A 5.10.0 memory leak when 'use feature qw(:5.10);' is included

Subject: [perl #62522] warnings::register regressions introduced in 5.10

From: Bram <>
Date: Wed, 04 Mar 2009 23:36:28 +0100
Subject: re engine

    bram points out a maint-5.10 and bleed regression from 5.10.0, and
    attached  a few patches which haven't been applied yet.

Subject: [perl #60508] Assertion rx->sublen >= (s - rx->subbeg) + i failed:
       	file "regcomp.c"

    looks like a regression in 5.10.0, maint, blead.

    actually exists in 5.8.x too, but looks like a good one to fix for

From: Moritz Lenz <>
Date: Sat, 05 Jan 2008 17:14:37 +0100
Subject: Bugs in extended regexp features

    Bug #1 in the email still exists (bug 2 is fixed):

    $ perl5.10.0 -Mre=eval -wlE 'my $re = qr/(\w+)/;"foo" =~ m/($re)(?{say $^N})/'
    Can't locate object method "say" via package "foo" (perhaps you forgot

Subject: [perl #49830] Encode from_to() does not return on invalid conversion

    a 5.10.0 regression apparently

Subject: Re: [perl #50528] Perl 5.10.0 segfaults on Format with my variables

    a closure/format segfault in 5.10

From: Moritz Lenz <>
Date: Mon, 28 Jan 2008 15:35:49 +0100
Subject: Slowdown in 5.10.0 regexes with atomic patterns

    +ve superlinear cache issue

From: "Jerry D. Hedden" <>
To: pp <>
Date: Thu, 8 May 2008 17:00:06 -0400
Subject: Leaked scalars

    a regression since 5.10.0 - present in B,M

Subject: [perl #22977] Bug in format/write

    Was a format/closure/newATTSUB bug in 5.8.8; in 5.10.0, maint, blead,
    it gets worse and becomes an assertion failure

Subject: Re: [perl #56738] Re: Some bugs in Perl regexp (core Perl issues)

    not sure if this is 5.10.x regression

From: Ævar Arnfjörð Bjarmason <>
To: Perl 5 Porters <>
Date: Sat, 19 Jul 2008 08:07:45 +0000
Subject: Perl debugger runs out of memory, hangs or segfaults on

    possibly a 5.10.0 regression

Subject: [perl #55466] Panic in regex processing (unknown regstclass 0)

    spamassassin under 5.10.0, possibly only big-endian systems.
    Don't know whether its a regression

Subject: [perl #59654] SIGSEGV compiling regexp in 5.10.0

    Appears to be a 5.10.0 regression. Still present in maint,blead

From: Torsten Schoenfeld <>
Date: Sun, 19 Oct 2008 13:33:47 +0200
Subject: blead and Glib's lazy-loader

    apprently blead breaks's lazy-loading (an XS wrapper for a
    gnome lib??) Don't know whether maint has broken it too.

Subject: [perl #65372] Regex fails when string is too long
Date: Mon, 04 May 2009 12:17:07 -0700

    looks to be another 5.10.0 regression involving
    regexes longer than 32767

From: Nicholas Clark <>
Message-ID: <>
Date: Fri, 17 Apr 2009 17:50:32 +0100
Subject: Re: shared hash keys and ithreads (was Re: Perl Critic and (honest)
        hash references)

    I need to look at this - don't know whether its 5.10.x specific yet.

From: Graham Barr <>
Message-ID: <>
Date: Mon, 11 May 2009 20:59:20 -0700
Subject: [perl #65582] 5.10.0 functionality change and segfault

    Fine in 5.8.8,5.8.9; assert fails in 5.10.0, maint, blead.

Subject: [perl #65754] ExtUtils::ParseXS fails basic.t test without dynamic

    Andy D pointed this out - I don't know whether its a regression

From: Torsten Schoenfeld <>
Message-ID: <>
Date: Sat, 23 May 2009 20:29:54 +0200
Subject: @ISA lookup issue in blead and maint-5.10

    a possible regression since 5.10.0


"I do not resent criticism, even when, for the sake of emphasis,
it parts for the time with reality".
    -- Winston Churchill, House of Commons, 22nd Jan 1941.

Thread Previous | Thread Next Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About