Bug #95050 for DBI: 5.19.11 broke pipeone in t/85gofer.t

Fri Apr 25 13:12:16 2014 rurban [...] x-ray.at - Ticket created

Subject:

5.19.11 broke pipeone in t/85gofer.t

2 errors: 1. uninitialized $stderr_msg diff -bu ./lib/DBD/Gofer/Transport/pipeone.pm~ ./lib/DBD/Gofer/Transport/pipeone.pm --- ./lib/DBD/Gofer/Transport/pipeone.pm~ 2013-06-24 16:03:21.000000000 -0500 +++ ./lib/DBD/Gofer/Transport/pipeone.pm 2014-04-25 12:11:06.032580741 -0500 @@ -169,7 +169,7 @@ my ($pid, $rfh, $efh, $ios, $cmd) = @{$info}{qw(pid rfh efh ios cmd)}; my $frozen_response; - my $stderr_msg; + my $stderr_msg = ''; $self->read_response_from_fh( { $efh => { 2. not yet analyzed t/85gofer.t ..................... 1/? # Failed test 'pipeone: DBI connect('transport=pipeone;policy=pedantic;timeout=240;perl=/usr/local/bin/perl5.19.11 -Mblib=/home/rurban/Perl/.cpan-build/DBI-1.631-NcUpsJ/blib;dsn=DBI:DBM:f_dir=/home/rurban/Perl/.cpan-build/DBI-1.631-NcUpsJ/test_output_5423;f_lock=0;dbm_type=SDBM_File','',...) failed: DBD::Gofer::Transport::pipeone command (/usr/local/bin/perl5.19.11 -Mblib=/home/rurban/Perl/.cpan-build/DBI-1.631-NcUpsJ/blib -MDBI::Gofer::Transport::pipeone -e run_one_stdio) failed: at /home/rurban/Perl/.cpan-build/DBI-1.631-NcUpsJ/blib/lib/DBD/Gofer/Transport/pipeone.pm line 191. at t/85gofer.t line 155. # ' # at t/85gofer.t line 113.

Fri Apr 25 13:16:23 2014 rurban [...] x-ray.at - Correspondence added

From:

rurban [...] x-ray.at

with threaded perl only, debugging or not. works ok without threads

Fri Apr 25 13:40:47 2014 rurban [...] x-ray.at - Correspondence added

From:

rurban [...] x-ray.at

On Fri Apr 25 13:16:23 2014, rurban@x-ray.at wrote: Show quoted text

> with threaded perl only, debugging or not. > works ok without threads

found and fixed problem 2. I had a wrong Params::Util dependency even if 5.12.5/x86_64-linux-thread-debug is not in my @INC /usr/local/bin/perl5.19.11d: symbol lookup error: /usr/local/lib/perl5/site_perl/5.12.5/x86_64-linux-thread-debug/auto/Params/Util/Util.so: undefined symbol: Perl_Gthr_key_ptr So only pull request #11 is needed. https://github.com/perl5-dbi/dbi/pull/11

Fri Apr 25 14:07:41 2014 REHSACK [...] cpan.org - Correspondence added

Hi Reini, even if I agree that initialize before use is always a good idea, I cannot see where the use is which fails. Is it possible to provide a test case which shows where it fails? Best regards, Jens

Fri Apr 25 14:07:42 2014 The RT System itself - Status changed from 'new' to 'open'

Fri Apr 25 15:24:37 2014 rurban [...] x-ray.at - Correspondence added

From:

rurban [...] x-ray.at

On Fri Apr 25 14:07:41 2014, REHSACK wrote: Show quoted text

> Hi Reini, > > even if I agree that initialize before use is always a good idea, I > cannot see where the use is which fails. > Is it possible to provide a test case which shows where it fails?

That will be hard, as you need a failing subshell to catch the stderr_msg, and then you'll see the Uninitialized $stderr_msg warning. My failing cmd came from problem 2, which is basically something crazy going on with xsloading the wrong Params::Util version. I fixed Params::Util so I cannot repro it anymore, but the problem is generic

Fri Apr 25 16:14:15 2014 REHSACK [...] cpan.org - Correspondence added

How about https://github.com/perl5-dbi/dbi/commit/9893965795307b7f7f383d708b16df5b356c2844

Sun Sep 21 08:19:25 2014 TIMB [...] cpan.org - Correspondence added

I'm marking this resolved as I can't reproduce with perl-5.20.1-debug-thread or a perl-5.19.4-debug-64int. Feel free to reopen if it's still an issue in master.

Sun Sep 21 08:19:26 2014 TIMB [...] cpan.org - Status changed from 'open' to 'resolved'