Joey Hess <[EMAIL PROTECTED]> wrote:
> Are you using the QT debconf frontend? Do the problems occur with a
> different frontend?
>
> What does it look like if you export DEBCONF_DEBUG=. and reproduce the
> problem (complete log please)?
I was using the QT ("KDE") frontend. I once tried with dialog (or
rather whiptail), that produced a totally corrupted terminal, maybe
because of all the internal debconf errors.
And you are right, it does not happen with the gnome frontend. I attach
the log with DEBCONF_DEBUG='.', truncated at the point where the first
"incorrect nice value" message is displayed. At this point, we have
# debconf-show x11-common
x11-common/xwrapper/actual_allowed_users: console
* x11-common/xwrapper/nice_value: 0
* x11-common/xwrapper/nice_value/error:
x11-common/x11r6_bin_not_empty:
* x11-common/xwrapper/allowed_users: Console Users Only
* x11-common/upgrade_issues:
#
> In your original bug report, you showed the preinst dying of a segfault.
> That's a shell script dying of a segfault? Do you have any reason to
> believe that this is not a hardware problem?
>
> + return 0
> + '[' 1 -ne 0 ']'
>
> If I'm reading this correctly, db_set is returning 0, and its caller is
> getting a 1 back in $?. Barring a broken shell, I don't see how that's
> possible.
You make me feel nervous. Yes, this is the same machine that grew a
corrupted initramfs image. On the other hand, I didn't encounter any
other problems, although I was using it quite heavily, both for package
building (mostly not yet uploaded) and for number crunching with
Mathematica.
I have reinstalled libqt-perl, libsmokeqt1 and libqt3-mt, but this
didn't change anything. Looking again at the output when reinstalling
x11-common with the Qt frontend, this time without any additional
debugging, shows clearly that something is amiss on the Qt side:
# apt-get --reinstall install x11-common 2>&1 | tee reinstall.lg
Reading package lists...
Building dependency tree...
0 upgraded, 0 newly installed, 1 reinstalled, 0 to remove and 3 not upgraded.
Need to get 0B/334kB of archives.
After unpacking 0B of additional disk space will be used.
Do you want to continue [Y/n]?
DESTROY created new reference to dead object ' Qt::SpacerItem', <> line 1
during global destruction.
Preconfiguring packages ...
Use of uninitialized value in join or string at
/usr/share/perl5/Debconf/DbDriver/Stack.pm line 104, <GEN1> line 4.
The first error comes even before the "Preconfiguring packages ..."
message.
I'll try to run memtest tonight (but don't expect much from it, since on
my system, the errors are reproducible except for the segfault). Apart
from that, which debugging steps do you suggest? Should this be kept
open as a bug at all?
Regards, Frank
--
Dr. Frank Küster
Single Molecule Spectroscopy, Protein Folding @ Inst. f. Biochemie, Univ. Zürich
Debian Developer (teTeX/TeXLive)
