For your consideration: Secondary Architectures in Fedora

Chris Weyl cweyl at alumni.drew.edu
Tue May 29 23:00:45 UTC 2007


On 5/29/07, David Woodhouse <dwmw2 at infradead.org> wrote:
> On Tue, 2007-05-29 at 16:18 -0500, Tom "spot" Callaway wrote:
> > This is a very early draft of what I am planning on presenting to FESCo
> > in the near future:
> >
> > http://fedoraproject.org/wiki/TomCallaway/SecondaryArchitectures
> >
> > I am very interested in comments and suggested changes.
>
> "Build failures on secondary architectures are not fatal: if packages
> build for the primary architectures, then they will push to the
> repositories."
>
> I'm concerned by this -- we would need to be very careful here to ensure
> that this isn't a massively retrograde step for these architectures. The
> ExcludeArch tracker bugs have been _really_ useful, but there's not a
> lot of point in them if we're going to let builds silently fail without
> even having an ExcludeArch: in the spec file, let alone a bug filed.

Isn't this exactly the point of secondary architectures, tho?
Maintainers are already on the hook to ensure their packages are 1)
functioning or 2) tracked for the primary architectures, one of which
many have probably never directly used.  Relaxing the rules for
secondary architectures would allow support for those architectures to
be added automatically to the buildsys w/o imposing additional burdens
on our (mostly unpaid!) maintainers.  If it's really documentation or
tracking secondary arch build failures, I'm sure a koji report could
be rigged to do much the same.

In other words, by only failing a build when a primary arch fails, we
enable the inclusion of many other architectures for those who care
about them, without imposing additional burdens on all maintainers
(who may not care about them).

Otherwise, why bother making a distinction at all?

                                         -Chris

-- 
Chris Weyl
Ex astris, scientia




More information about the fedora-devel-list mailing list