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

Re: Debian needs more buildds. It has offers. They aren't being accepted.



On Sat, Feb 14, 2004 at 09:56:50AM +0100, Eduard Bloch wrote:
> #include <hallo.h>
> * Anthony Towns [Sat, Feb 14 2004, 05:41:51PM]:
> > On Fri, Feb 13, 2004 at 04:10:29PM +0100, Goswin von Brederlow wrote:
> > > We want kde 3.2.0 in sid so it is (tried to be) autobuild. 
> > No, we don't want KDE 3.2.0 until we *know* it can be autobuilt. Uploading
> > it to experimental and asking developers to build it on various
> > architectures tells us that to a good approximation, and allows us to keep
> > working on KDE 3.1.5 in the meantime.
> So? Wasn't you favorite Testing branch intended to be a place for
> "working versions"? 

Yes it is, but that doesn't help here because KDE and Qt don't meet the
appropriate definition of working.

And as it turns out, keeping testing working requires that we not upload
just anything into unstable -- in particular it causes significant
problems if we upload things that we're not able to fix in a timely
manner. That's why uploads of a major new versions, particularly of major
packages with lots of dependants, or complicated interdependencies should
be made to experimental first wherever possible.

> Since Testing seems to mutate to another outdated,
> deadlocked package set, I understand the wish to make Sid the distro for
> everydays work - but it is not how the things should be.

The goal has always been to have testing mostly working, and mostly up
to date with unstable. That's not possible if the versions in unstable
remain in a non-working state for a long time, and apparently it's more
common for that to happen than I or others originally expected. To deal
with that, we need somewhere to upload packages that are likely to be
broken for an extended period. That place is experimental, and KDE 3.2
is exactly that sort of package.

> And is that the reason for making autobuild-trials on other arches a
> real pain for maintainers with packages in experimental? I see no good
> reason for not running autobuilders on experimental.

In the long term, there's no reason at all. In the short term, it's
too difficult and there are complications. If you want to set up a
buildd for experimental on your own bat, you're entirely welcome to,
and indeed encouraged.

Cheers,
aj

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

             Linux.conf.au 2004 -- Because we could.
           http://conf.linux.org.au/ -- Jan 12-17, 2004

Attachment: signature.asc
Description: Digital signature


Reply to: