The problem seems to be what Guillaume pointed to in an earlier comment, the portions where kpat searches for the images files in / don't appear in a working setup.
However, I have no clue what the difference between a working and a broken setup is.
I installed a clean Kubuntu 8.10 and installed kpat => works
I installed a clean Ubuntu 8.10 and installed kpat => works
Due to earlier comments I am lead to belive that it would not work in fresh installations, even though it does. So we can rule out a dependency issue.
Since it is not working with a new user we can also rule out a caching or configuration problem.
The only thing we know for sure is that the problem appears in rendering/opening the card decks, since the assert kicks in when the card deck pixmap is empty.
It appears the deck information get cached using the KDE system cache, a natural guess would be that the cache just didn't get generated, my testing indicates different though (also the strace shows clear documentation of successful ksycoca queries).
So since I am not sure how to processed I'd really appreciate some ideas (even very absurd ones ;-). Also talking to upstream could help a great deal.
Otherwise the last option we have is adding loads of kDebug messages to the affected code. That is going to be a pretty nice waste of resource though since we would have to be poking the code blinding because no one within the development team is affected by this issue.
I am seriously running out of ideas.
The problem seems to be what Guillaume pointed to in an earlier comment, the portions where kpat searches for the images files in / don't appear in a working setup.
However, I have no clue what the difference between a working and a broken setup is.
I installed a clean Kubuntu 8.10 and installed kpat => works
I installed a clean Ubuntu 8.10 and installed kpat => works
Due to earlier comments I am lead to belive that it would not work in fresh installations, even though it does. So we can rule out a dependency issue.
Since it is not working with a new user we can also rule out a caching or configuration problem.
The only thing we know for sure is that the problem appears in rendering/opening the card decks, since the assert kicks in when the card deck pixmap is empty.
It appears the deck information get cached using the KDE system cache, a natural guess would be that the cache just didn't get generated, my testing indicates different though (also the strace shows clear documentation of successful ksycoca queries).
So since I am not sure how to processed I'd really appreciate some ideas (even very absurd ones ;-). Also talking to upstream could help a great deal.
Otherwise the last option we have is adding loads of kDebug messages to the affected code. That is going to be a pretty nice waste of resource though since we would have to be poking the code blinding because no one within the development team is affected by this issue.