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

Re: Buildds slow/unuseable -> drop that arch



On Sat, Sep 04, 2004 at 03:00:23PM +1000, Hamish Moffatt wrote:

> > > Efforts to keep dead architecturs alive are noble, but not very
> > > rational from the practical point of view.
> > In some cases, fringe architectures have helped Debian to improve as a
> > whole (for example, when they switched to a new GCC version before the
> > main distribution and provided hints of what has to be changed).
> So it's a good idea for us to continue to support fringe architectures
> in unstable. Is it necessary for them to be released, and influence
> 'testing'?

I think it's important to differ between a temporary backlog (as this
happened on mips f.e.) and a huge and permanent backlog (as f.e. on arm,
when it wouldn't be dropping right now).
If an arch is permanently backlogged (<90% keeping up or so), then it is a
valid candidate for being dropped from release. 

Release Manager are usually not as that dumb to not see those kinds of
problems when a package is hold back by a single, backlogged arch. They
usually decide then for each package if it will do the transition into
testing or not. 

In fact, personally I considered arm as a valid candidate for being dropped
from release until it's backlog began to drop a week ago. If it continues to
drop, then let it in, if it raises again, then drop it. 

The backlog discussions are showing that there are problems with the
autobuilder network, either with the number of machines or the number of
buildd admins doing the work. These problems need to get solved soon (after
the release). 

-- 
Ciao...              // 
      Ingo         \X/



Reply to: