On 2 May 2013 09:45, Garth N. Wells <[email protected]> wrote: > On 2 May 2013 09:38, Johan Hake <[email protected]> wrote: >> On 05/02/2013 10:05 AM, Garth N. Wells wrote: >>> On 2 May 2013 07:56, Anders Logg <[email protected]> wrote: >>>> On Thu, May 02, 2013 at 08:33:07AM +0200, Johan Hake wrote: >>>>> Hello! >>>>> >>>>> DOLFIN next has been red for a while now and the reason, I guess, is the >>>>> xdmf fix from Chris. >>>>>
This is fixed now. I removed the demo that failed because it didn't really demonstrate anything. There is still one Mac buildbot failing due to the BLAS and compiler combination. It looks like the buildbot is trying to link to a multi-threaded ATLAS BLAS (which uses OpenMP), but DOLFIN is being compiled with clang (which doesn't support OpenMP). Solution is to use the native Mac BLAS framework. Garth >>>>> This branch was recently merged into master and made that red. I think >>>>> we should have next green before feature branches get merged with >>>>> master. That is the whole point with next? Now I have fix branches which >>>>> are merged with next and they are waiting for it to turn green. >>>>> >>>>> Johan >>>> >>>> Yes, I think that's the point. Master should never be red. >>>> >>> >>> Yes. >>> >>> The buildbot on 'next' has only just become available. It was not >>> available when I last checked before merging next into master. >> >> next should not be merged into master. The topic branches should be merged. >> > > Right. I'm still learning . . . . > > Garth > >>> Are there instructions somewhere for users on how to run their branch >>> through a buildbot? We could require this before accepting a merge >>> request. >> >> That would be nice. >> >> Johan >> >>> Garth >>> >>>> -- >>>> Anders >>>> _______________________________________________ >>>> fenics mailing list >>>> [email protected] >>>> http://fenicsproject.org/mailman/listinfo/fenics >> _______________________________________________ fenics mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics
