On Tue, Jan 15, 2008 at 06:27:04PM +0100, Dirk Mueller wrote:
> > > I don't think that factory testing is the issue.
> > I don't follow your point, even with packages failing factory helps you
> > test and determine ahead of time how good an alpha would be.
> 
> Sure, my point is that the testing such a snapshot is not very meaningful 
> because it is just one arbitrary state during a rapidly changing time period. 
> 
> Also, as you know, factory syncing to the outside is staged ("slowed down") 
> to 
> not stress mirrors even more. It IMHO makes more sense to take care of 
> failing packages to ensure a consistent state for Alpha1 than testing one 
> arbitrary snapshot in time that can not be recreated. But it's your time 
> after all :) 

I fully agree here: I don't expect everthing in an alpha to work, but I
expect the alpha to be *consistent* at the rpm layer: A big don't should
be that there are still packages that are suse 10.3.1. If they don't build,
then remove them or fix them but don't let them be part of a release. The
same is true for packages that no longer have their dependencies
fulfilled (e.g. pdftk requires a lib that has been replaced by a newer
version). Again: Remove those packages or fix them, but don't ship them.
If you really want to ship a system with inconsistent dependencies then
please call it snapshot and not (alpha) release. A missing package is much
better than a package with missing dependencies.
Calling something an (alpha/beta) release should have a defined set of
criteria that it *must* fulfill and I'd like to see these criteria
defined somewhere. One criteria that comes to my mind for all types of
releases is that it must be possible to rebuild all the packages that
said release comprises of by using only packages from that release.

 ciao
   Joerg
-- 
Joerg Mayer                                           <[EMAIL PROTECTED]>
We are stuck with technology when what we really want is just stuff that
works. Some say that should read Microsoft instead of technology.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]

Reply via email to