Brian Nelson a écrit :
Isn't an ICE always a compiler bug?
Yes, I was just asking if it was a new bug or a pre-existing one.
Thanks for the information...I guess that's bad news. It seems there are almost 2 monthes of changes in the gcc 4.0 branch between 4.0.3-1 and 4.0.3-3, so it's quite possible that it's indeed a GCC regression. When looking at the gcc-4.0 bugs reported in Debian's BTS for > 4.0.3-1, I can't see much that would explain this. I'm completely clueless about building, but somebody should start by figuring which GCC regression is blocking this. Given the fact that branch 4.0 looks very inactive, maybe the easiest thing would be to identify the changeset which introduced this regression.In any case, nothing in the Qt code changed between -1 and -2, so it's certainly a GCC regression.