Sven Luther says:
Well, my experience in packaging powerpc kernels for debian tells me that a single kernel source for all architectures may not always be in the best interest, it depends on how it is handled and everything, but i would be interested on how you plan to handle the powerpc/apus case, whose patches are possibly not easily integrable with the remaining of the powerpc patches, let alone the non-powerpc ones. Have you thought about such things, before taking such an agressive stance on this issue?
That might be true for unstable and/or experimental, but having dozens of kernel packages simply *does not* work for stable. Unlike the people you're criticizing for somehow stealing your thunder I'd actually advocate not allowing any kernel into stable that didn't build against a single central kernel package. Sure as hell your method will lead to some architecture releasing a kernel that's 4 revs old at release day and which nobody at all is interested in developing a security patch for when a vulnerability comes along 3 years down the road. Maintainability is harder than hacking shit together, but it's important if you actually want to see a debian stable. Alternatively, if finding a *maintainable* solution isn't a priority to you, you could simply make sure your kernel stays in unstable. Mike Stone