[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

Re: Update on "upload of GNOME 2.6 to unstable" status

(Hrm, mutt's [L]ist-reply is dropping -gtk-gnome, suck)

On Tue, May 18, 2004 at 12:01:12AM +0200, Sebastien Bacher wrote:
> Le lun, 17/05/2004 ? 18:22 +1000, Anthony Towns a ?crit :
> > alpha, hppa and s390 should be there too. Skipping out m68k might be
> > okay, but you should certainly ask and try to get the core libraries
> > built there too.
> Ok, let me summarize, you want a build on all architectures and full
> tests with no bugs. Basically you are asking for the unstable -> testing
> conditions to upload something in unstable ... why ? 

No: I want you to do all you can to ensure there aren't going to be
problems before you upload to unstable. The differences are, eg, between
making sure Gnome 2.6 can be built everywhere, and actually having a
particular version of Gnome 2.6 built everywhere; between fixing RC bugs
you know about in advance, and spending a couple of weeks letting users
find new ones and fixing those too.

> Gnome 2.6 packages are really good, what's such a big deal to upload
> them in unstable ? 

If they're really that good, there's no big deal at all. This is how
you convince us they're really that good.

> We have not changed gnome2.4 packages for
> weeks/months, if you fear for sarge we just need to put a RC bug on
> atk/glib/gtk/pango from Gnome2.6 in unstable. 

There's no "just" about that. Doing that really does screw things up
pretty majorly: packages in unstable with RC bugs need to be fixed
_quickly_. Not immediately, maybe, but not after months and months either.

It's quite possible that that's not going to be a problem in this
particular case -- I'd put my money on it not being a problem, even in
spite of the Gnome RM's concerns expressed earlier.

But it often _is_ a problem, and it's not something we want to make a
habit of risking. The way we can avoid it is by running major changes
through experimental first; these things won't catch everything by any
means but they will catch a bunch of issues that do cause major problems

If you're really that confident about Gnome 2.6, that's fine: consider
this a practice run for Gnome 2.8 so that we can work out the kinks in
this process and get Gnome 2.8 into testing and out to our users soon
after it's released.

> Could please reconsider the situation?

I can reconsider it as often as you like, but I'm not a Magic 8-ball,
so the answer's not going to change.

> GTK+ has been built and tested
> on 8 architectures now without any problem, that's much than 95% of the
> unstable packages ...

Fine, get them into experimental, and work on the other three

If you get to the point where almost everything's done, and the remaining
1% is impossible for various reasons, come back and explain why each
particular remaining issue (missing builds, unfixed RC bugs) can't be done
("m68k would take days to build, and the machines are already overloaded,
and would spontaneously combust when they found out they had to rebuild
the same packages again for unstable", eg) and, unless someone has some
other approaches for you, you're done.


Anthony Towns <aj@humbug.org.au> <http://azure.humbug.org.au/~aj/>
Don't assume I speak for anyone but myself. GPG signed mail preferred.

``Like the ski resort of girls looking for husbands and husbands looking
  for girls, the situation is not as symmetrical as it might seem.''

Attachment: signature.asc
Description: Digital signature

Reply to: