Re: Announcing powerpc backport 2.6.12-6 kernel package for sarge

[Sven Luther <sven.luther@wanadoo.fr>]

On Thu, Sep 15, 2005 at 04:00:53PM +0900, Horms wrote:
> > I have done a second build, with the following changes, as 2.6.12-5.99.sarge2
> > Only the first change is relevant, URL as immediately above :)
> > 
> >    * Change kernel-source build dependancy to
> >      kernel-package (>= 8.135) [!powerpc] | kernel-package (>= 8.135.sarge1) [powerpc]
> >      so that this package can be build on an unmodified sarge install
> >      on non-powerpc
> >    * Add myself and Sven Luther as uploaders
> 
> I have withrawn the 2.6.12-5.99.sarge2 packages, and the URL above
> now only has the 2.6.12-5.99.sarge1 variant.
> 
> The reason for this is, despite the build dependancy, they were actually
> built with kernel-package 8.135.sarge1, so they are identical to the
> 2.6.12-5.99.sarge1 packages. Furthermore, I have been advised
> by Andres Salomon and Sven Luther that kernel-package 8.135 
> cannot build these packages because because of incompatibility
> with the defconfig target - I don't know more than that.

Well, defconfig is used to build some generic packages, docs and such, the fix
is rather orthogonal to anything else, small and localized, and has no chance
of breaking anything, the same goes for the powerpc/powerpc64 targets. I thus
propose that we do a kernel-package 8.135.sarge1 upload to
stable-proposed-updates, Manoj, do you have an opinion on this ? 

This way volatile will be able to build the packages without going outside
stable and stable-proposed-updates, and since volatile will be auto-built, it
is the best place for such packages if we don't want to target those 2.6.12
kernels at s-p-u themselves, which we probably don't want to do.

We could do it if the stable and stable-security team worked hand in hand with
the kernel team, and if the kernel team was ready to do sarge security stuff
for both 2.6.8 and the newer kernels, since i don't think we can purely and
simply remove the 2.6.8 packages from sarge :)

I also don't think how post-2.6.12 packages would be possible, unless we
reenable devfs on them or backport initramfs or something such.

Friendly,

Sven Luther