What to do with a noarch package that fails to build on ARM

[Andrew McNabb]

On Tue, Aug 20, 2013 at 03:28:13PM -0400, Stephen Gallagher wrote:
> > I've opened a report
> > with upstream, but neither the upstream maintainer nor I have
> > access to any ARM machines to figure out what's going on.
> 
> Well, you *can* use the ExcludeArch directive, but the obvious
> question to me is: Why is the test failing?

Good question.  Unfortunately, nobody has general-purpose ARM machines,
so it's impossible to expect upstream (or me) to be able to track it
down. :(

> If there's a bug in the
> test when run on ARM, can you just patch it to skip that test?

It probably is a legitimate problem on ARM, so skipping the test would
just be pretending that the package works when it really doesn't.

> Looking at
> http://kojipkgs.fedoraproject.org//work/tasks/4455/5834455/build.log
> it appears to be failing on a test to see if python-pexpect can send
> all control-codes to the client. It looks like it timed out. Is it
> possible that ARM is just slow to process it and that patching the
> test to increase the timeout would be sufficient?

I don't think so.  The time limit seems to be 30 seconds for an
operation that should only take a fraction of a second.  I could be
wrong, but I think it's a hang.

--
Andrew McNabb
http://www.mcnabbs.org/andrew/
PGP Fingerprint: 8A17 B57C 6879 1863 DE55  8012 AB4D 6098 8826 6868