nvidia-experimental package with expedited SRU process?

[Bryce Harrington]

On Tue, Sep 11, 2012 at 05:57:29AM +0200, Martin Pitt wrote:
> Bryce Harrington [2012-09-10 13:22 -0700]:
> > On Tue, Sep 04, 2012 at 08:45:35AM -0500, Mario Limonciello wrote:
> > > As a side note, I believe the ethos behind the "nvidia-current" name was to
> > > represent the presently stable nvidia driver version.  Appending the
> > > -updates to "nvidia-current" or -experimental to it may be convoluting that
> > > namespace and/or confusing the user.
> > > 
> > > Would just making it "nvidia-stable", "nvidia-updates", and
> > > "nvidia-experimental" perhaps make more sense?
> > 
> > Hi Mario, just to follow up on this question...  I've discussed this
> > with Alberto and we agree the 'nvidia-current-updates' package name is a
> > bit clumsy.  However since we're planning on doing quite a bit with
> > package backports we think the least disruptive time to redo package
> > names is during 14.04 LTS development.
> 
> nvidia-updates would also not fit the current schema. We have
> nvidia-series{,-updates}, where "series" is "96", "173", "current",
> or "experimental", and at least the first three can always come in an
> -update flavour. So nvidia-updates would presumably be
> nvidia-current-updates, but what do we do about nvidia-173-updates?

In practice the legacy drivers are not updated except to add new X
server ABI support, something unnecessary to backport (since we wouldn't
change the X ABI in a stable release).  IOW, I don't think we're likely
actually need an nvidia-173-updates.

That said, there's no reason NVIDIA *couldn't* put out an update for
-173 that has changes we'd like to backport, so it's certainly a
theoretical possibility.  I could see the argument either way.

In practice though, nvidia-current is going to be locked in to some
version number at release time (e.g. 280.xx in oneiric, 295.xx in
precise, etc.) and that version number won't change.  But
nvidia-current-updates will start at that version number at release and
change over time (e.g. 304.xx in precise).  So you could think of
'updates' as its own series, separate from -current.

> TBH I'd just keep them as they are.

Yeah maybe so; sounds like something we can hash through at some future
UDS.

Bryce