From: Justin Forbes on
gitlab.com
https://gitlab.com/cki-project/kernel-ark/-/issues/42#note_555112975
Ok, that's what I had assumed. Can we make the naming consistent
though, while still accomplishing that?
The naming is inline with package guidelines. rcX releases are in fact
upstream releases (you can download the tarball on
kernel.org). The git
snapshots are not releases, and are snapshots. Versioning follows for
each. As to the sub packages, producing only kernel-debug for a non
debug build will break a whole lot of the things we wish to test. The
existance of a package called 'kernel' is fairly critical. Inverting
the naming would just be confusing for everyone involved. Again, we
want those scripts which just rely on 'kernel' to test the debug kernel
frequently, but also the non debug kernel at least once a week (we have
had issues where things worked in a debug kernel, and broke in non
debug). As to the users who prefer to only run non-debug kernels, they
get those in Rawhide nodebug, or only do updates on Monday evening or
Tuesday morning.
I'm not sure what you mean by the no-debug repository?
https://dl.fedoraproject.org/pub/alt/rawhide-kernel-nodebug/ It is set
to ensure that people always get the latest debug kernel with a 'dnf
update' for people using the rawhide repository for other things, they
can exclude kernel in the rawhide repository config, for people using
stable kernel but running a rawhide kernel, stable versions will always
be lower. In either case, adding rawhide-nodebug as a separate
repository will keep the intended kernel installed. We added this
repository and model in 2012.