On Wednesday, 16 November 2016 at 10:51, Peter Robinson wrote:
>> > >> Anyone else seeing errors like this?
>> > >>
>> > >> reading sources... [ 27%]
>> > >> api/astropy.modeling.functional_models.Sersic1D reading
>> > >> sources... [ 27%]
>> > >> api/astropy.modeling.functional_models.Sersic2D Sphinx
>> > >> Documentation subprocess failed with return code -11
>> > >>
>> > >> I'm seeing it on ppc64 with python-astropy - filed
>> > >>
https://bugzilla.redhat.com/show_bug.cgi?id=1394975
>> > >>
>> > >
>> > > Appears to be crashing in matplotlib. I've added some debug info
>> > > and I can reproduce on a ppc64 instance, but I really don't know
>> > > what's doing on.
>> > >
>> >
>> > Workaround is to do:
>> >
>> > export MPLBACKEND=cairo
>>
>> Interesting. The default backend is Tk. Could you file a bug with
>> matplotlib upstream? Also, feel free to apply for co-maintainership.
>
> is it possible to make the cairo backend default (for big endian
> arches at least)?
TBH I think cairo makes sense on all architectures.
I wouldn't want to make such big change of upstream default setting.
Upstream will hate us for "matplotlib works differently on Fedora"-type
bugs. This is definitely something to be decided upstream.
Regards,
Dominik
--
Fedora
http://fedoraproject.org/wiki/User:Rathann
RPMFusion
http://rpmfusion.org
"Faith manages."
-- Delenn to Lennier in Babylon 5:"Confessions and Lamentations"