Hi, On Thu, Dec 15, 2011 at 10:33:30AM +0100, Bastien ROUCARIES wrote: > Hi, > > I strongly disagree about closing this bug. It is a can of worms.
It _could_ indeed be one. > At least you should open a bug for qt where this bug lie, and if it is > a spurious bug report due to some gcc bug [1], you should open a bug > on the debian pts linking to > http://gcc.gnu.org/bugzilla/show_bug.cgi?id=43976 > > Do you have tested if the program run on arm and mips ? No, I didn't tests it -- to be honest I don't even know how I could exhaustively check whether this would work, other than simply starting the program and loading a bib. According to my understanding of the problem it is only one if memory is allocated for a structure the pointer refers to and the actual type after casting would result in misaligned data access. However, I do not see an indication that this is actually happening. It is not a problem to do this type of casting just to pass pointers around, as long as the get cast again into the proper types before data access. If you want to see this tracked down, I'd very much appreciate if you perform the actions that you suggested above. However, to make those bug reports useful it would need some sort of comprehensible test case that is a lot less compless than the autogenerated code in kbibtex that causes it. Once again, I'm not saying that this might not a be bug . But it seems to me that it is currently not worth investing a significant amount of time, or asking others to do so. Of course this should not prevent any volunteer from making an attempt. Michael PS: As you seem to have a long-standing interest in kbibtex, have you thought about participating in the maintenance of the package? I'd be happy add you. -- Michael Hanke http://mih.voxindeserto.de -- To UNSUBSCRIBE, email to debian-bugs-rc-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org
