On Wed, Nov 28, 2007 at 11:33:48PM +0100, Moritz Muehlenhoff wrote:
> > > I've checked KDE and Gnome and the security maintenance overhead is
> > > acceptable. So, if we have someone managing the release and the necessary
> > > resources and interest in the Gnome and/or KDE camp, this can become
> > > reality.
> > 
> >  Ok; I thought KDE and GNOME had been ranked out, as I read in an IRC
> >  meeting log of the project.
> 
> Dann didn't want additional work to test the new drivers also against the
> new desktop packages, a desktop update can still be handled independantly.

My point is more that etch 1/2's goal is to enable new hardware while
minimizing risk to existing etch users. This would be done primarily
by adding new kernel packages. Any other packages that might need to
get updated (wireless-tools, X drivers, etc) will need careful review
to ensure a low risk of regressions. For something as large as
KDE/GNOME, I don't think is feasible to either add a second optional
version, or to show that there is a minimal risk of regressions.

> I don't have the time myself, I'm only trying to get my point across that
> a lack of security support shouldn't block this. (Which is typically the
> main reason raised why Debian doesn't provide such an effort until now).

Its one reason, but I think the risk of breaking existing stable users
is an equal factor. I think Debian stable releases have a pretty good
reputation for being stable, and I don't want to lose that.

Granted, there is the class of users that always want the latest and
greatest and are willing to live with the extra maintenance that
requires - I suspect running testing is a good answer for a lot of
these users. There are of course others that want something less fluid
with a higher level of support - from Debian and third party
vendors. To satisfy these users, I think the only good answer would be
frequent releases - either a shorter timeframe between stable
releases, or something extra, like periodic snapshots of testing.

-- 
dann frazier



-- 
To unsubscribe, send mail to [EMAIL PROTECTED]

Répondre à