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

Re: For your consideration: Secondary Architectures in Fedora



On Wed, 2007-05-30 at 11:52 -0400, Jesse Keating wrote:
> On Wednesday 30 May 2007 10:59:52 David Woodhouse wrote:
> > Allowing partially-failed builds to make it through into the repo
> > without user intervention is insane. Failures should _always_ be
> > investigated.
> 
> If there is nobody around for that secondary arch to investigate why the build 
> failed on that secondary arch, does that secondary arch need to continue 
> existing?  

Converse question: if a package-monkey is too lazy to even _look_ at a
build or test failure and work out whether it's a generic problem which
just _happens_ to show up in one build but not another this week, and if
they're too lazy even to add the ExcludeArch: and file the corresponding
bug, does that monkey's package need to continue existing?

It's not as if we're asking _much_ of them.

> It is entirely up to the community wanting to support $ARCH to be 
> on the hook to fix problems that arise in $ARCH.  They should not just pass 
> that off on the package maintainer if said package maintainer isn't part of 
> the $ARCH movement already.

Of course, and we currently strike a balance which is working very well.
It involves package maintainers at least _looking_ at failures, and
filing a bug if they end up satisfied that it's an arch-specific issue
and using ExcludeArch. Then the people who care about the architecture
in question can watch the FE-ExcludeArch-$ARCH bug and handle anything
which comes up.

It's working _extremely_ well at the moment -- there really seems to be
no reason to change it. In the relatively rare case where a failure
really is an arch-specific issue and not just a package bug, it's easy
enough for the package maintainer to file the bug and exclude the
offending architecture.

Build failures should _always_ be investigated, at at _least_ a cursory
level, before letting the affected packages get into the repository.
To require anything less than that is completely batshit insane.

-- 
dwmw2


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