[Digikam-devel] Is this a digikam problem?

[Heiner Lamprecht]

Hi Gilles,

On Friday 25 August 2006 13:47, Gilles Caulier wrote:
> On Friday 25 August 2006 13:34, Heiner Lamprecht wrote:
> > Hi Gilles,
> >
> > On Friday 25 August 2006 12:42, Gilles Caulier wrote:
> > > Heiner,
> > >
> > > The crash arrive when you use an image plugin with the screen
> > > color management option enabled. Right ?
> >
> > Yes.
> >
> > On console, I get some warning regularly:
> >
> > digikam: intent: 0
> > digikam: input profile: default no alpha
> > lcms: Error #12288; Input profile is operating on wrong
> > colorspace digikam: [bool
> > Digikam::IccTransform::apply(Digikam::DImg&)] LCMS internal
> > error: cannot create a color transform instance
>
> This is important ! When screen color management is disabled,
> these errors disapear ?

Yes, they don't appear anymore.

> Hum, sound like http://bugs.kde.org/show_bug.cgi?id=132649
>
> Are you running digiKam in gdb to get this backtrace ?

No.  I only configured it with --enable-debug=full.  And that's what 
the KDE crashhandler shows me :-)

> Can you try to get a valgrind backtrace using the command line
> describe to the end of HACKING file from svn ?

Maybe tonight.

During the last hour, I hopefully made a big step forward:  I 
disabled hyperthreading in the BIOS.  And since than, digikam seems 
to run stable.  I'm not 100% sure, because the crashes were hard to 
reproduce.  But I was able to use (and stress) digikam for a much 
longer period then before :-)  Both with CMS disabled and enabled.

Now I will first prepare some printouts (what I wanted to do since 
Monday ...).  If digikam runs stable through this, I'll retry the 
same under 64bit.  When I'm sure, that the hyperthreading was the 
reason, I'll switch it on again to get some more debugging info.


    Heiner

-- 
    heiner at heiner-lamprecht dot net    GnuPG - Key: 9859E373
  Fingerprint:  3770 7947 F917 94EF 8717 BADB 0139 7554 9859 E373