failing tests on 32 vs 64 bits

[Cyrille Berger]

On Friday 09 April 2010, Boudewijn Rempt wrote:
> Coupled with the invalid read of sizeof(channelsize) in all compositeops
> according to memcheck, I think this means we have allowed some really bad
> pointer-size bugs to slip in.
It is worth to note that [1] is runned on a 64bits computer too, and does not 
show the same failures. I experienced hanging in the tests that disappear 
after clean build/clean install, maybe it will help for your problem too ?

If you already tried that, can you attach the test log for the pigment tests ?

[1] koffice.org/testresults

-- 
Cyrille Berger