Re: "testing" improvements

[Martin Schulze <joey@infodrom.org>]

Jérôme Marant wrote:
> En réponse à Matt Zimmerman <mdz@debian.org>:
> 
> > > I'm sorry but most people that want recent versions of packages
> > > in a more stable state use testing, not unstable. These are facts,
> > > not theory. Since, testing is a pre-stable release it has to be
> > > tested even more than unstable, because what's going to be on CDs
> > > is testing not unstable.
> > 
> > What is your point exactly?  That people use testing to get newer
> > packages,
> > so we should force newer packages into testing despite its purpose?
> 
> Force packages into testing? You must have missed the point.
> I said let packages into testing for architectures where nothing
> prevent them to. Why would mipsel failures block x86 packages,
> for instance?

Maybe because it's sign that the package is probably broken and we
couldn't release before it builds on mipsel again.  The exception
is when maintainers decide that the package is not supposed for a
certain architecture and removes it from the Architecture: line.

I have to admit that I don't see the benefit of your proposal
except that some people get access to some packages in testing
earlier than others, depending on the architecture used.

Shouldn't we instead spend time on fixing the package on these
architectures for which it is broken instead?

Shouldn't we judge any architecture similar to another that we
support?  Please note that Debian is the only distribution that
runs on all of them, so I guess we should rather spend more time
on continuing the support instead of abandon certain architectures.

Regards,

	Joey

-- 
Still can't talk about what I can't talk about.  Sorry.  -- Bruce Schneier

Please always Cc to me when replying to me on the lists.