headers
George Greer | 1 May 2010 02:50

Smoke [5.12.0] v5.12.0-21-ga03bd48 FAIL(F) MSWin32 Win2000 SP4 (x86/2 cpu)

Automated smoke report for 5.12.0 patch a03bd4831688fb530e7c376bb9109d188d0fb11a v5.12.0-21-ga03bd48
perl-win2k: Intel(R) Core(TM) i7 CPU 920  <at>  2.67GHz(~2649 MHz) (x86/2 cpu)
    on        MSWin32 - Win2000 SP4
    using     cl version 14.00.50727.762
    smoketime 5 hours 5 minutes (average 38 minutes 12 seconds)

Summary: FAIL(F)

O = OK  F = Failure(s), extended report at the bottom
X = Failure(s) under TEST but not under harness
? = still running or test results not (yet) available
Build failures during:       - = unknown or N/A
c = Configure, m = make, M = make (after miniperl), t = make test-prep

v5.12.0-21-ga03bd48  Configuration (common) none
----------- ---------------------------------------------------------
O F         
F F         -Duselargefiles
O F         -Duseithreads
F F         -Duseithreads -Duselargefiles
| +--------- -DDEBUGGING
+----------- no debugging

Locally applied patches:
    uncommitted-changes
    SMOKEa03bd4831688fb530e7c376bb9109d188d0fb11a

Failures: (common-args) none
[default] -DDEBUGGING
../t/op/sselect.t...........................................FAILED
(Continue reading)

Permalink | Reply | 0 comments |
headers
Perl 5 commit summary | 1 May 2010 03:08

Perl 5 Commit Summary

Perl 5 commit summary, activity since Wednesday

Current branch blead
19 commits.  8 unique authors.  5 unique committers.
58 files changed, 416 insertions(+), 168 deletions(-)
Net code removal: Ricardo Signes (4 lines)
Thanks, applied:  Rafael Garcia-Suarez (4) David Golden (1)
Snapshot: http://perl5.git.perl.org/perl.git/snapshot/4d239afe17c8b494.tar.gz

  Upgrade to IPC::Cmd 0.58
  Rafael Garcia-Suarez        3 files changed, 100 insertions(+), 22 deletions
  http://perl5.git.perl.org/perl.git/commit/4d239afe17c8b494

  RT#73814 - unpack() didn't handle scalar context correctly for %32H and %32u
  Tony Cook                   2 files changed, 30 insertions(+), 13 deletions(
  http://perl5.git.perl.org/perl.git/commit/858fe5e170d95536

  Update title of "Supported Platforms" in perlport to avoid
  Rafael Garcia-Suarez        1 file changed, 3 insertions(+), 3 deletions(-)
  http://perl5.git.perl.org/perl.git/commit/7c35b6afcaba267d

  Re-run bump-perl-version to update a few missing spots
  Vincent Pit                 2 files changed, 10 insertions(+), 10 deletions(
  http://perl5.git.perl.org/perl.git/commit/b70b2c2a9fca4b1a

  Teach bump-perl-version about "libperl511.a"
  Vincent Pit                 1 file changed, 4 insertions(+), 4 deletions(-)
  http://perl5.git.perl.org/perl.git/commit/d4f05c0b45a10736

  Don't use a C++ keyword as a variable name ("new").
(Continue reading)

Permalink | Reply | 0 comments |
headers
Aristotle Pagaltzis | 1 May 2010 05:00
Picon
Picon
Gravatar

Re: *foo{SCALAR}

* Eric Brine <ikegami <at> adaelis.com> [2010-05-01 00:10]:
> On Fri, Apr 30, 2010 at 4:55 PM, Nicholas Clark <nick <at> ccl4.org> wrote:
> > (Given that when I changed the implementation to what we have
> > now, and didn't get it "right", so that sometimes
> > *foo{SCALAR} was undef, at least one module on CPAN did fail
> > tests. One of the YAML modules, IIRC.)
>
> I'm guessing a lot more modules are unable to do their job
> properly by keeping the existing behaviour than there are
> modules that would be broken by changing it.

Yes! Agree completely. I recall this behaviour stumping me on
several occasions in the past, though any specifics of what I was
trying to do are lost to the mists of time.

Furthermore, I will venture that in all instances of breakage, it
will be not because the broken modules functionally rely on this
behaviour, but because they go out of their way to cope with this
eccentricity. Fixing such breakage will boil down to a removal of
make-work code (which may still not be trivial, naturally).

Regards,
--

-- 
Aristotle Pagaltzis // <http://plasmasturm.org/>
Permalink | Reply | 1 comment |
headers
Alexandr Ciornii via RT | 1 May 2010 12:33
Picon
Favicon

[perl #7233] Data::Dumper Purity mode t/dumper.t fails if slightly rearranged

Similar results on 5.12.0.

On Wed Jul 04 06:11:31 2001, merlyn <at> stonehenge.com wrote:
>     use Data::Dumper;
>     my  <at> dogs = ( 'Fido', 'Wags' );
>     my %kennel = (
>       First => \$dogs[0],
>       Second =>  \$dogs[1],
>     );
>     $dogs[2] = \%kennel;
>     my $mutts = \%kennel;
> 
>     print "CORRECT:\n";
>     print Data::Dumper->new(
>       [\ <at> dogs, \%kennel, $mutts],
>       [qw($dogs $kennel $mutts)])->Purity(1)->Dump;
> 
>     print "WRONG:\n";
>     print Data::Dumper->new(
>       [\%kennel, \ <at> dogs, $mutts],
>       [qw($kennel $dogs $mutts)])->Purity(1)->Dump;
> 
> Output:
> 
>     CORRECT:
>     $dogs = [
>               'Fido',
>               'Wags',
>               {
>                 'First' => '',
(Continue reading)

Permalink | Reply | 0 comments |
headers
George Greer | 1 May 2010 12:47

Smoke [5.13.0] v5.13.0-124-g4d239af FAIL(XM) linux 2.6.31-20-generic [debian] (x86_64/8 cpu)

Automated smoke report for 5.13.0 patch 4d239afe17c8b494dcc4e900abe2587eb6affcc8 v5.13.0-124-g4d239af
zwei: Intel(R) Core(TM) i7 CPU 920  <at>  2.67GHz (GenuineIntel 2668MHz) (x86_64/8 cpu)
    on        linux - 2.6.31-20-generic [debian]
    using     gcc version 4.4.1
    smoketime 19 hours 49 minutes (average 37 minutes 10 seconds)

Summary: FAIL(XM)

O = OK  F = Failure(s), extended report at the bottom
X = Failure(s) under TEST but not under harness
? = still running or test results not (yet) available
Build failures during:       - = unknown or N/A
c = Configure, m = make, M = make (after miniperl), t = make test-prep

v5.13.0-124-g4d239af  Configuration (common) none
----------- ---------------------------------------------------------
O O O X O O -Dcc=gcc
O O O O O O -Dcc=gcc -Duse64bitint
O O O O O O -Dcc=gcc -Duselongdouble
O O O O O O -Dcc=gcc -Dusemorebits
O O O O O O -Dcc=gcc -Duseithreads
M - - O O O -Dcc=gcc -Duseithreads -Duse64bitint
O O O O X O -Dcc=gcc -Duseithreads -Duselongdouble
O O O O O O -Dcc=gcc -Duseithreads -Dusemorebits
O O O O O O -Dcc=g++
O O O O O O -Dcc=g++ -Duse64bitint
O O O O O O -Dcc=g++ -Duselongdouble
O O O O O O -Dcc=g++ -Dusemorebits
O O O O X O -Dcc=g++ -Duseithreads
O O O O O O -Dcc=g++ -Duseithreads -Duse64bitint
(Continue reading)

Permalink | Reply | 0 comments |
headers
Jesse Vincent | 1 May 2010 13:29
Gravatar

Re: on dual-lived versions in maint


On Tue, Apr 27, 2010 at 12:59:27PM +0100, Dave Mitchell wrote:
> On Tue, Apr 27, 2010 at 08:58:33AM +0200, Steffen Mueller wrote:
> > 2) Consideration of each case. If the available new release fixes the  
> > bug and isn't fraught with peril, pull it in. Otherwise ask upstream to  
> > make an intermediate release we can pull in or -- as a fallback -- do  
> > the cherry-picking & dev-versioning thing.
> 
> Yep, when I originally put forward my list of minimalist-maint
> suggestions, I always envisaged that we would, if possible, keep maint
> modules in sync with a CPAN release - but not at the risk of introducing
> new functionality and thus a new round of bugs. That's one of the things
> that delayed 5.10.1. So we try to persuade the CPAN author to produce a
> new release that has only the critical bug fixes and not a whole slew of
> new stuff too. If we can't do that, then we cherry-pick the bug fixes as a
> last resort.

( +1 )
Permalink | Reply | 0 comments |
headers
Zefram | 1 May 2010 18:59

Re: package block syntax

Attached patch is an updated version of my "package Foo { }"
implementation.  This fixes the SEGV in parser error recovery and adds
more tests.  No one has objected to the new syntax, and a few people
have expressed approval, so I think this can be applied.

This message provides a single patch incorporating both the original
implementation and these new changes, for ease of testing.  For
application to blead it might be preferred to apply the three patches
separately.  Whoever wants to apply it can negotiate with me about the
best form in which to supply the patches.  My git log has this:

commit 19dcffbd38634cd58401fcab66da7dcbc8fcca05
Author: Zefram <zefram <at> fysh.org>
Date:   Sat May 1 17:51:21 2010 +0100

    additional tests for package block syntax

    Test that __PACKAGE__ propagates into string eval correctly.  Test that
    __LINE__ is correct.  Test that goto into and out of package blocks
    works correctly.

commit bdf8b2094a086d18ae0d3970228a650e4e1b4783
Author: Zefram <zefram <at> fysh.org>
Date:   Sat May 1 17:30:26 2010 +0100

    fix SEGV with eval("package Foo {")

    OPs relating to the package name and version were subject to double
    freeing during error recovery from an incomplete package block.  Fixed by
    using the op_latefree mechanism to shift the op free time.
(Continue reading)

Permalink | Reply | 23 comments |
headers
Eric Brine | 1 May 2010 22:05
Gravatar

Re: [perl #74790] SelfLoader mixing sysseek() and tell()

On Thu, Apr 29, 2010 at 7:10 PM, Kevin Ryde <perlbug-followup <at> perl.org>wrote:

> # New Ticket Created by  Kevin Ryde
> # Please include the string:  [perl #74790]
> # in the subject line of all future correspondence about this issue.
> # <URL: http://rt.perl.org/rt3/Ticket/Display.html?id=74790 >
>

> While nosing around SelfLoader 1.17 I noticed it combines sysseek() with
> tell(). I wonder if that may be unfriendly to input stream layers which
> are doing a data transformation so that tell() doesn't correspond to the
> underlying file position.  Should it be plain seek() with tell()?
>

I'm not sure what the purpose of the sysseek is there, but since it's
followed by file handle duping, I suspect that dealing with the system file
handle (sysseek) instead of the Perl file handle (seek) is required. If so,
the solution would be to use change tell($fh) to sysseek($fh,0,1), not
changing sysseek to seek.
Permalink | Reply | 0 comments |
headers
Frank Wiegand | 1 May 2010 22:51
Picon
Gravatar

Re: [perl #74764] -E 'given( goto f ) { f: }' => crash

Am Donnerstag, den 29.04.2010, 01:36 -0700 schrieb Frank Wiegand:
> The following code crashes perl:
> 
>         % perl5.12.0 -E 'given ( goto f ) { f: }'
>         Use of "goto" to jump into a construct is deprecated at -e line 1.
>         perl5.12.0: pp_ctl.c:4005: Perl_pp_leavegiven: Assertion `((cx)->cx_u.cx_subst.sbu_type & 0xf)
== 3' failed.
>         zsh: abort      /opt/perl/perl-5.12.0-RC3/bin/perl5.12.0 -E 'given ( goto f ) { f: }
> 
> Yes, I see the deprecated warning.

This one is without the warning:

        % perl-5.12.0-RC3 -wE 'given( do { goto f } ) { f: }'
        perl-5.12.0-RC3: pp_ctl.c:4005: Perl_pp_leavegiven: Assertion `((cx)->cx_u.cx_subst.sbu_type &
0xf) == 3' failed.
        zsh: abort      LC_ALL=C perl-5.12.0-RC3 -wE 'given( do { goto f } ) { f: }'

> Yes, no one would do this.
> But perl should not crash, too.

Still true.

Frank
Permalink | Reply | 0 comments |
headers
George Greer | 1 May 2010 23:11

Smoke [5.13.0] v5.13.0-128-gc6bf6a6 FAIL(F) MSWin32 Win2000 SP4 (x86/2 cpu)

Automated smoke report for 5.13.0 patch c6bf6a65e4bf512beae58d59446046f4ca20372f v5.13.0-128-gc6bf6a6
perl-win2k: Intel(R) Core(TM) i7 CPU 920  <at>  2.67GHz(~2649 MHz) (x86/2 cpu)
    on        MSWin32 - Win2000 SP4
    using     cl version 14.00.50727.762
    smoketime 4 hours 21 minutes (average 32 minutes 40 seconds)

Summary: FAIL(F)

O = OK  F = Failure(s), extended report at the bottom
X = Failure(s) under TEST but not under harness
? = still running or test results not (yet) available
Build failures during:       - = unknown or N/A
c = Configure, m = make, M = make (after miniperl), t = make test-prep

v5.13.0-128-gc6bf6a6  Configuration (common) none
----------- ---------------------------------------------------------
O O         
O O         -Duselargefiles
F O         -Duseithreads
F F         -Duseithreads -Duselargefiles
| +--------- -DDEBUGGING
+----------- no debugging

Locally applied patches:
    uncommitted-changes
    SMOKEc6bf6a65e4bf512beae58d59446046f4ca20372f

Failures: (common-args) none
[default] -Duseithreads
[default] -Duseithreads -Duselargefiles
(Continue reading)

Permalink | Reply | 0 comments |

Gmane