On Thu, Jun 03, 2004 at 12:26:52PM +0200, Christoph Hellwig wrote: > a) less maintaince overhead for the security team
I don't think this is the case; but like you said, lets hear from them before making this conclusion. > b) kernel-patch packages always work cross-plattform define "work" :) By some definition, I could make mine work by having the apply & unpatch scripts exit 0 on non-ia64; today, I make them ia64-only packages. If they're truly ia64 only, I don't think its important that they work elsewhere. > c) debian actually has one codebase and one set of bugs over all > architectures For assignment, if the bugs are against common code, then can be assigned to kernel-source, if they are arch specific, they'd be assigned to the arch kernel-image package. During analysis, I can think of situations where seeing the latest ppc bugs & ia64 bugs in the same place would be useful, but so is the view of just the ia64 bugs. When bugs are componentized, its easier to write an interface that joins them rather than one that tries to split them up based on heuristics. > d) better exposure of arch patches. yes, this is true. > if an architecture patch touches > common code it doesn't need review. I would think common pieces of arch packages would be the most important to review.