Fading without Segfaults

First of all, this should fix imagefader segmentation faults:
http://uberbrain.info/last.fm.diff

I guess not everybody's having them, else there would be lots of complaining, but under ArchLinux, the imagefade would cause a segfault everytime. This fixes it. Its also quite a bit cleaner if you ask me, a lot more obvious what's going on exactly.

Second;
You need to call m_transport->close() after m_transport->get(), otherwise coverimage downloads will fail unless the previous image was downloaded from the same host. I didn't bother with a patch this time, because its only one additional line.

Additionally, I noticed that you emit CoverLoader::requestFinished() after recieving a done() signal. The done signal is emitted when all tasks scheduled for the HTTP object are done. Theoretically, setHost() can be added and completed before get() is scheduled, causing the done() signal to be emitted, and calling the coverLoaded() function before it actually is. In practice, this hasn't happened to me once, but it's still something to think about. It's a very easy fix. You just need to add one member variable that tracks the taskID of the setHost() call and check for it inside slotRequestFinished().

Enjoy.

Hmm, very interesting. Do you mind if I incorperate this into the Gentoo ebuild?
I'll be sure to give credit!

for Linux 2.6.10/Debian sarge
devas