future kernel module rpm situation (was: kernel-source vs. kernel-sourcecode (please revert))

[Panu Matilainen]

On Tue, 15 Jun 2004, seth vidal wrote:

> > It's on my todo-list, haven't done anything about it yet. Mind you I don't
> > think the exactarch-behavior is necessarily a win-win situation either
> > because it breaks when a package's arch is downgraded on purpose like we
> > all saw with this particular change. Having it as an option is/would be
> > nice though.
> 
> I agree, Matthias had a pretty good suggestion.
> 
> have a list of packages that should not change arch, if at all possible,
> and only treat those as special:
> 
> glibc kernel kernel-smp etc etc.

Yep, the extremely rare cases of those packages downgrading arch can 
be dealt as .. well, special cases. It was suggested just yesterday on 
apt-rpm list that support for pinning packages by architecture should be 
added, which would do the above, more or less. Considering the mess that 
pinning is, a list like 
ExactArchPkgs = ["glibc", "kernel" ... ] 
might be less intimidating :)

	- Panu -