splitting off nvidia-graphics-drivers-legacy-340xx
vcheng at debian.org
Fri Aug 7 07:17:08 UTC 2015
On Sun, Aug 2, 2015 at 5:14 PM, Andreas Beckmann <anbe at debian.org> wrote:
> I started picking some bits back into trunk targeting sid today. The
> first bits (to branches/340) are non-intrusive bugfixes that I plan to
> include in a possible security update for jessie (we needed such for
> squeeze and wheezy ... so maybe jessie too), the second part to trunk is
> more intrusive and won't hit jessie (except via backports).
> The next bits I'll do are
> * nvidia-detect updates
> * renaming the nvidia-uvm.ko module
> These shouldn't be cherry-picked into experimental until they are
> finished and working. Especially for renaming the kernel module I could
> need some help for testing that this works properly ...
> Once all this is working and in sid, it will be easy to finally split
> off the new 340xx legacy package (and upload newer releases to sid).
Awesome, thanks for working on this!
I'm unsure whether I'll be able to find some time to help with testing
in the near future, but it looks like Luca's already several steps
> Regarding conftest.h: there shall be exactly one version being used for
> all packages (current and legacy, sid and unstable), otherwise
> maintaining it becomes even more hell :-)
Ack, that makes sense. I wasn't entirely sure what changes in
branches/352 were worth backporting to other branches / legacy
packages, so up to now I've just avoided backporting anything, but I
can see how conftest.h would be a prime candidate for being
> Regarding patches: I usually don't refresh patches as long as they still
> apply, that reduces the noise in the diff between branches (including
> current vs. legacy).
> The packages I uploaded today (source-only uploads are sufficient for
> n-g-d) are currently stuck as BD-Uninstallable due to the ongoing
> libstdc++ transition, that should resolve itself over the next days.
I've done source-only uploads for n-g-d in sid, but for experimental
uploads I've been doing i386+amd64 binary uploads because I've noticed
that packages that are both non-free+experimental seem to take forever
to be built. I frankly don't know how the buildds prioritize packages,
but e.g. 346.59-1 which you uploaded on May 12 was stuck as
Needs-Build for a whole month on i386 and amd64 (and the armhf build
failed, so you can't find any binary packages for 346.59-1 at all on
snapshot.d.o ) prior to 346.72-1 being uploaded. I wouldn't be
surprised to see 352.30-1 still unbuilt for i386 and amd64 by the time
September rolls around.
More information about the pkg-nvidia-devel