Is it all academic?
There's been lots of arguing about interpretations and meanings etc. but what does it all mean in practical terms? I'm not going to try and add to either side of the argument, I'm just curious about what will happen when one side or the other is proverd correct. As far as I can see, no matter who is right, we can continue to distribute these apps. There are 2/3 implications that I can see depending on who's arguments are correct:
1 - Mattias etc. is right and there is no conflict, in which case everyone can continue distributing KDE apps with no worries. This would probably be the most convenient result.
2 - Debian is right and there is a conflict in which case we have an interesting problem. There is no question about whether the QPL is valid as TT have not based their code on other people's work so they don't have a problem. The problem now is that hundreds of KDE apps were released with a license that is not valid. That is, the license contains terms that were broken by the author as soon as he declare the app to be GPLed and have been repeatedly broken by anyone who has since distributed it. I'm not sure where that leaves us in terms of contracts. I think (IANAL) that if a contract has an unenforcable term in it then that term becomes void but the rest of the contract remains in place. So we have 2 more possibilities:
2a - The KDE programs are still under the GPL but with several of the terms nullified in which case we can continue to distribute the apps as before just under a kind of weakened GPL.
2b - The whole license is invalidated (I'm pretty sure this isn't correct, but just in case...), in which case the apps were released without any license, does this make them public domain? I doubt it. Maybe it means that the authors are free to retrospectively choose a new license. In which case, assuming they still want to distribute their apps freely, they can (or must!) choose a license which will not conflict with the QPL. Either way the apps are once again free to use without any license worries.
1 - There is no conflict, keep doing what you're doing = happy days.
2a - There is a conflict, apps are forced into a weakened (less viral) version of the GPL = happy days (unless you're a big GPL fan, see below)
2b - There is a conflict, authors get to turn back time and release under a compatible license = happy days unless authors decide to be difficult. The only reason to believe this would happen is that section 7 of the GPL say that you may not release under the GPL unless you can satisfy it fully. If contract law works as in 2a then one could argue that section all of section 7 is invalid and never applied, including the bit about what happens if section 7 is invalidated (this is what you get when a lisp programmer write a license ;-). If not then the app was definitely never released under the GPL so I presume the original author can retroactively change the license to something which solves all problems.
Unless someone can explain why all of these 3 scenarios are incorrect and explain what the real situation is, the arguments as to whether the licenses conflict are academic. Whether they conflict or not, the software can be distributed safely. It was the original authors' faults for releasing the apps under contradictory licenses.
As an aside, there is reason to prefer that there is no conflict because, if 2a is the correct view then it may be possible to exploit the weakened GPL to use the source in a non-free way. If linking against proprietary libraries turns out to be OK then someone could write a library with their own proprietary extensions, keeping it closed source and then modify the original program to use this new proprietary library. In a way, they could turn the original code into a library and use all of it's features from their closed source code. Not a good result...