On Wed, Jun 18, 2014 at 01:49:54PM +0200, Anders Logg wrote: > On Wed, Jun 18, 2014 at 12:32:51PM +0100, Garth N. Wells wrote: > > > > > > On Wed, 18 Jun, 2014 at 9:48 AM, Anders Logg <[email protected]> wrote: > > >The next buildbot for DOLFIN is failing and my suspicion is the reason > > >is a merge in FFC that didn't make it into next (but is in master). > > > > > >Can next be reset to master? Or should master be merged into next > > >(for FFC)? > > > > master can be merged into next. > > ok I'll see if that helps
Did not work. Something is broken with the regression tests in FFC. Code works fine but the regression tests seem to be confused. For DOLFIN, next breaks when FFC is called to generate code for the new demo multimesh-stokes. The error indicates that an old version of FFC is used (before support was added for mixed elements in custom integrals). I'm about to leave for vacation and a week without stable internet connection so feel free to revert DOLFIN next to before the merge of logg/topic-vector-custom-integral if FFC cannot be sorted out. -- Anders _______________________________________________ fenics mailing list [email protected] http://fenicsproject.org/mailman/listinfo/fenics
