http://fedoraproject.org/wiki/TomCallaway/SecondaryArchitectures

[David Woodhouse]

On Wed, 2007-07-11 at 18:31 -0500, Tom "spot" Callaway wrote:
> You keep saying this, and I disagree whole-heartedly. My experience
> with sparc tells me this is absolutely not the case.

I posted the bug numbers to support my observations, and they supported
my qualitative recollection.

Perhaps it's just that PowerPC is in dramatically better shape than
SPARC, in general. Or maybe you've suffered a lot by being out of sync
with Fedora proper. I'd be interested to see your data and how you've
classified the bugs.

If your recollection _is_ indicative of what we'll get when we pull in
more esoteric architectures, even that isn't necessarily so much cause
for concern -- we'll be a lot better off if we can keep those
architectures in sync with the main Fedora repository, and we'll also be
able to improve on it a lot just by enforcing the 'SHOULD regularly
rebuild in mock' rule¹. That'll tend to catch what I suspect would be
the most common arch-specific failure mode, which would be GCC or other
toolchain issues. (With your multitude of arch-specific bugs, perhaps
you can confirm or deny that's what the most common such bug is likely
to be?)

Besides, the fact remains that the 'burden' it places on the package
maintainer is _trivial_. The minimal case is that they just file an
_empty_ bug for the ExcludeArch and ship the failed package anyway.

Obviously we expect better from our competent maintainers than that, but
the lowest common denominator wouldn't be _forced_ into anything even
approaching conscientiousness; only encouraged towards it.

-- 
dwmw2

¹ I'd like to stick to the RFC2119 definition of 'SHOULD' -- as in; you
  don't have to, but you'd better have a damn good reason why not.