x86_64 blocks i386?

[Thorsten Leemhuis]

Am Sonntag, den 20.03.2005, 19:52 +0100 schrieb Nils Philippsen:
> On Sun, 2005-03-20 at 00:41 +0100, Michael Schwendt wrote:
> > Since I only just noticed that the requested rebuilds for gpgme and
> > gpgme03 in FE Development were not published, because they failed on
> > x86_64, I got to ask:
> > 
> > When and where has it been decided that x86_64 blocks i386?
> > 
> > If that is the case, I see myself unable to support my packages (and
> > gpgme03 which I've maintained for a series of revisions) as I don't have
> > access to an AMD64 machine and no AMD64 running Rawhide either.
> 
> Let me address this from a different angle: IMO, If we want Extras to be
> of similar quality as Core, failure on one platform should block all the
> others.

I don't like the idea -- even as a x86_64 user. That slows down
development to much IMHO and is afaik one of the reasons for the debate
in debian to get the number of supported first-tier-archs down.

A better solution IMHO would be:  If a package fails for one arch put
the packages for the working archs on hold for [4|7|<something else>]
days (maybe in testing?). If the package get fixed for all archs only
push the new ones and remove the other.

For the cases where it a package not gets fixed there should be group of
"arch-maintainers" that fix other people packages. This maybe is a lot
of work for those "arch-maintainers" -- but for them it's often easier
to fix problems because they often know the common build problems and
can fix them easily. 

>  Otherwise people don't have enough incentive to fix bugs on
> platforms they don't care about as much as about "their" platform and
> versions/releases are bound to leapfrog between the platforms.

Well, to fix them they often need to have access to such a arch. That is
often a problem. 
-- 
Thorsten Leemhuis <fedora at leemhuis.info>