[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: KDE SC 4.5.1 packages available

On Fri, Sep 10, 2010 at 11:04 PM, Modestas Vainius <modestas@vainius.eu> wrote:
> Hello,
> On penktadienis 10 Rugsėjis 2010 19:47:25 George Kiagiadakis wrote:
>> On Fri, Sep 10, 2010 at 5:54 PM, Zsolt Rizsanyi
>> <rizsanyi@users.sourceforge.net> wrote:
>> > On Fri, Sep 10, 2010 at 4:34 PM, Martin Steigerwald <Martin@lichtvoll.de>
> wrote:
>> >> Digikam again with glib dbg package installed. This crash might be
>> >> related to the marble crash as well. At least that marble widget
>> >> appears in this
>> >
>> >> backtrace, too:
>> > Hmm, I have found a cure. Checking digikam I have seen, that digikam
>> > depends on libmarblewidget4 (which is kde 4.4.5) and marble depends on
>> > libmarblewidget10 (which is kde 4.5.1).
>> > After removing libmarblewidget4 (including digikam) now marble does not
>> > crash.
>> Thanks, that must be the problem. I will investigate it.
>> libmarblewidget4 should not be installed on the system, it is replaced
>> by libmarblewidget10 now. Obviously digikam needs a rebuild and
>> probably something needs to conflict with libmarblewidget4, but not
>> sure what yet.
> There is no good solution for such a case except binNMU. The recent failed
> libjpeg transition was another example of how messy such things could get.
> Unfortunately, Breaks would complicate migration to testing unnecessarily.

Right, so in the digikam case, digikam links to libmarblewidget4 and
loads plugins which link to libmarblewidget10. In this case it is
reasonable to crash and I agree that the correct solution is a binNMU.
But what happens in the case of marble itself? It links against
libmarblewidget10 and the plugins as well, but when libmarblewidget4
is installed, it crashes and there is nothing mentioning
libmarblewidget4 in the backtrace. That doesn't make much sense to me.

Reply to: