On Monday 19 July 2010 22:23:42 Damyan Ivanov wrote:
tags 589687 - unreproducible
retitle 589687 pbuilder: Incorrectly propagates DISPLAY into chroot env
reassign 589687 pbuilder
severity 589687 normal
thanks
> -=| Daniel Schepler, Mon, Jul 19, 2010 at 04:00:03PM -0700 |=-
>
> > Source: padre
> > Version: 0.63.ds1-1
> > Severity: serious
> >
> > From my pbuilder build log:
> >
> > ...
> > PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e"
> > "test_harness(0, 'inc', 'blib/lib', 'blib/arch')" t/*.t No protocol
> > specified
> > Error: Unable to initialize gtk, is DISPLAY set properly?
> > No protocol specified
> > Error while autoloading 'Wx::wxVERSION_STRING' at t/01-load.t line 22
> >
> > # Failed test 'no warnings'
> > # at /usr/share/perl5/Test/NoWarnings.pm line 45.
> > # There were 2 warning(s)
> > # Previous test 0 ''
> > # Use of uninitialized value $Wx::_universal in concatenation (.)
> > or string at /usr/lib/perl5/Wx.pm line 160. # at /usr/lib/perl5/Wx.pm
> > line 160
> > # require Wx.pm called at (eval 13) line 2
> > # main::BEGIN() called at /usr/lib/perl5/Wx.pm line 0
> > # eval {...} called at /usr/lib/perl5/Wx.pm line 0
> > # eval 'package main;
> > # use Wx @{$args[0]};
> > # 1;
> > ......
>
> Seems to build fine here (up to date sid pbuilder on amd64):
>
> PERL_DL_NONLAZY=1 /usr/bin/perl "-MExtUtils::Command::MM" "-e"
> "test_harness(0, 'inc', 'blib/lib', 'blib/arch')" t/*.t
> t/01-load.t ................ skipped: Needs DISPLAY
> t/02-new.t ................. skipped: Needs DISPLAY
> t/03-wx.t .................. skipped: Needs DISPLAY
> t/04-config.t .............. ok
> t/05-project.t ............. skipped: Needs DISPLAY
> t/06-utils.t ............... skipped: Needs DISPLAY
> t/07-version.t ............. skipped: Needs DISPLAY
> t/08-style.t ............... ok
> t/14-warnings.t ............ ok
> t/15-locale.t .............. skipped: Needs DISPLAY
> t/50-docbrowser.t .......... skipped: Needs DISPLAY
> t/60-db.t .................. skipped: Needs DISPLAY
> t/70-document.t ............ skipped: Needs DISPLAY
> t/71-perl.t ................ skipped: Needs DISPLAY
> t/72-dialog-html.t ......... skipped: Needs DISPLAY
> t/73-display.t ............. skipped: Needs DISPLAY
> t/74-history-combobox.t .... skipped: Needs DISPLAY
> t/75-autocomplete.t ........ skipped: Needs DISPLAY
> t/80-newline.t ............. skipped: Needs DISPLAY
> t/81-search.t .............. skipped: Needs DISPLAY
> t/82-plugin-manager.t ...... skipped: Needs DISPLAY
> t/83-autosave.t ............ ok
> t/84-task.t ................ skipped: Needs DISPLAY
> t/85-task-manager.t ........ skipped: Needs DISPLAY
> t/86-service.t ............. skipped: Needs DISPLAY
> t/91-vi.t .................. skipped: Needs DISPLAY
> t/92-padre-file.t .......... ok
> t/93-padre-filename-win.t .. ok
> t/94-padre-file-remote.t ... ok
> All tests successful.
> Files=29, Tests=343, 5 wallclock secs ( 0.14 usr 0.06 sys + 1.15
> cusr 0.22 csys = 1.57 CPU)
> Result: PASS
>
> I am lowering the severity claiming the bug not reproducible for now.
> Can you submit the full build log so I can check for possible
> differences in the build environment?
OK, after some more investigation, it turns out I had DISPLAY set for some
reason in my root session. If I unset DISPLAY before running pbuilder then I
get a successful build. So it looks like the problem is that pbuilder is
carrying over the DISPLAY setting into the chroot environment without doing
anything like bind-mounting /tmp/.X11-unix or copying xauth tokens into the
chroot (which I wouldn't expect it to do in any case, I'd instead expect it to
unset DISPLAY in the chroot).
--
Daniel Schepler
--
To UNSUBSCRIBE, email to [email protected]
with a subject of "unsubscribe". Trouble? Contact [email protected]
