Bleadperl v5.19.6-22-gebdc880 breaks GWILLIAMS/RDF-Trine-1.007.tar.gz #13442
Comments
From @andkgit bisect commit ebdc880 S_process_special_blocks() should use a new stack for BEGIN blocks. diagnostics It happens only when XML::LibXML is installed. If XML::LibXML is not http://www.cpantesters.org/cpan/report/917e0d8a-546a-11e3-9195-4aca9d5baa37 perl -V Summary of my perl5 (revision 5 version 19 subversion 7) configuration: Characteristics of this binary (from libperl): -- |
From @nwc10On Mon, Nov 25, 2013 at 06:48:57AM -0800, Andreas J. Koenig via RT wrote:
I'm not going to get a chance to look at this (heck, I didn't get time myself Attempt to reload XML/LibXML.pm aborted. Why is it now attempting a reload? Nicholas Clark |
|
The RT System itself - Status changed from 'new' to 'open' |
From @andkalso affected: GSHANK/HTML-FormHandler-0.40053.tar.gz sample fail report: http://www.cpantesters.org/cpan/report/d85bd3a2-5561-11e3-a024-97dd9d5baa37 Running the failing test verbosely: PERL_DL_NONLAZY=1 /home/sand/src/perl/repoperls/installed-perls/perl/v5.19.6-22-gebdc880/127e/bin/perl "-MExtUtils::Command::MM" "-MTest::Harness" "-e" "undef *Test::Harness::Switches; test_harness(1, 'blib/lib', 'blib/arch')" t/fields/repeatable.t Test Summary Report t/fields/repeatable.t (Wstat: 6 Tests: 3 Failed: 0) -- |
From @cpansproutOn Mon Nov 25 06:48:56 2013, andreas.koenig.7os6VVqR@franz.ak.mind.de wrote:
It is not failing for me under current bleadperl. Does it fail for you still? I wonder whether 901ee10 might have fixed it, but it would -- Father Chrysostomos |
From @andk"Father Chrysostomos via RT" <perlbug-followup@perl.org> writes:
Yes, but I had to determine the conditions under which to reproduce more - XML::SAX::Expat is not installed --or-- - XML::SAX::Expat is also installed and XML/SAX/ParserDetails.ini is After this finding I checked whether 5.19.6 is indeed passing when -- |
From @andkNow also affected: SREZIC/Tk-804.032.tar.gz http://www.cpantesters.org/cpan/report/776f4750-86ac-11e3-95c9-a5a95d7e0655 |
From @andkAndreas Koenig <andreas.koenig.7os6VVqR@franz.ak.mind.de> writes:
Program terminated with signal 6, Aborted. -- |
From @wolfsageWith ebdc880, I can reproduce the HTML::FormHandler test crash, and mhorsfall@tworivers:~/.cpanm/work/1390878740.27975/HTML-FormHandler-0.40055$ (Though it doesn't crash when run under valgrind. Neat.): mhorsfall@tworivers:~/.cpanm/work/1390878740.27975/HTML-FormHandler-0.40055$ With ea35f80 (ebdc880~1) I cannot reproduce the crash, and the With latest blead (b373396) I still get the crash with the same Each time, I built perl with grindperl, for example: grindperl --prefix=/home/mhorsfall/blead -j 5 --install -- Matthew Horsfall (alh) |
From perl5-porters@perl.orgMatthew Horsfall wrote:
Could you get a Perl backtrace at that point? E.g., set a breakpoint on XS_Data__Clone_clone and then use: gdb) p Perl_warn("") /* no threads */ Then edit the cited sourced file and put Carp::cluck() where Then re-run the whole thing through valgrind. The output of that would be much appreciated.
That line is an EXTEND. So somehow the entersub call above is using a |
From @wolfsageI think I've got this right: ok 1 - use HTML::FormHandler::Field::Repeatable; -- Matthew Horsfall (alh) |
From @wolfsageAnd that looks terrible in my email client. Attached. |
From @wolfsage==3474== Memcheck, a memory error detector |
From perl5-porters@perl.orgMatthew Horsfall wrote:
Thank you. Sorry for putting you to the trouble. The valgrind output
Data::Clone is calling the perl code that reallocated the stack, so I (I suspect its use of call_sv is buggy and that the commit that I am trying to reproduce this on dromedary. The bad news is that I |
From perl5-porters@perl.orgI wrote, in response to Matthew Horsfall:
It is likely that, since require() does not use the existing stack, Indeed, Data::Clone does not protect against stack reallocation during RDF::Trine does not depend on Data::Clone, so I do not know which of |
From @tonycozOn Wed, Jan 29, 2014 at 05:44:07AM -0000, Father Chrysostomos wrote:
You can use screen(1) to deal with dodgy connections. Tony |
From @khwilliamsonOn 01/28/2014 10:55 PM, Tony Cook wrote:
Another possible thing to try is ssh -oServerAliveInterval=150 users.perl5.git.perl.org |
From @wolfsageOn Wed, Jan 29, 2014 at 12:44 AM, Father Chrysostomos <sprout@cpan.org> wrote:
No problem. I'm trying to set up the RDF::Trine case now. -- Matthew Horsfall (alh) |
From @wolfsageOn Wed, Jan 29, 2014 at 8:57 AM, Matthew Horsfall (alh)
I believe I've reduced it to this: use strict; use Module::Load::Conditional qw(can_load); BEGIN { eval "use XML::LibXML;"; If you look at %INC, 'XML/LibXML.pm' => undef ! -- Matthew Horsfall (alh) |
From perl5-porters@perl.orgI wrote:
I was implying that it should do PUSHSTACK/POPSTACK, but actually it void where sv_clone can reallocate the stack. That line translates into: PL_stack_base[ax + (0)] = Data_Clone_sv_clone( (sv)); Does the C standard guarantee any execution order here? (According |
From @tonycozOn Thu, Jan 30, 2014 at 04:38:47AM -0000, Father Chrysostomos wrote:
It doesn't. Tony |
From perl5-porters@perl.orgMatthew Horsfall wrote:
Thank you. I was able to reproduce it with that. XML::LibXML::bootstrap does POPMARK twice. No wonder it screws up. |
From @cpansproutThank you to everyone who provided diagnostics and advice. I have submitted a patch to CPAN ticket #92606 and mentioned RDF::Trine in the Known Problems section of perl5200delta in commit 1a85989. -- Father Chrysostomos |
|
@cpansprout - Status changed from 'open' to 'resolved' |
From @eserte"Father Chrysostomos via RT" <perlbug-followup@perl.org> writes:
Unfortunately Tk was forgotten. -- Berlin Perl Mongers - http://berlin.pm.org |
Migrated from rt.perl.org#120626 (status was 'resolved')
Searchable as RT120626$
The text was updated successfully, but these errors were encountered: