dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Wed, Oct 19, 2011 at 10:56:55AM -0700, Johan Hake wrote:
> On Wednesday October 19 2011 10:43:21 Anders Logg wrote:
> > On Tue, Oct 18, 2011 at 06:22:20AM -0700, Johan Hake wrote:
> > > On Monday October 17 2011 23:41:03 Anders Logg wrote:
> > > > On Mon, Oct 17, 2011 at 11:27:59PM -0700, Johan Hake wrote:
> > > > > On Monday October 17 2011 23:25:06 Anders Logg wrote:
> > > > > > This has already been fixed (including new unit tests) but is
> > > > > > likely sitting in my branch. Are these the same fixes I made or
> > > > > > did you make some other changes?
> > > > >
> > > > > Not sure. I manage to uncomment a unittest which was failing becuase
> > > > > the subdomains where not handled correctly for SystemAssemble
> > > >
> > > > Yes, but I fixed all that (I think):
> > > >
> > > > http://bazaar.launchpad.net/~dolfin-core/dolfin/logg/revision/6332
> > > > http://bazaar.launchpad.net/~dolfin-core/dolfin/logg/revision/6333
> > > > http://bazaar.launchpad.net/~dolfin-core/dolfin/logg/revision/6334
> > >
> > > Yes it looks like we did some over lapping work there :P. Not sure you
> > > got the global facet number correct, nor the correct domains from the
> > > mesh ;) Did you test it against the outcommented test?
> >
> > I compared the two codes and couldn't see any problem with the global
> > facet number.
>
> Btw: I see now that the global facet number is correctly implemented, and your
> solution _is_ more elegant. I keep forgeting that you can access MeshFunction
> data using the MeshEntity! Could you look into my fix for domain assemble in
> the OpenMPAssembler while at it? I used the MeshConnectivity there as I did in
> the SystemAssembler fix, which is unnessesary.

ok, I'll have a look. Currently working on some renaming of list_foo
functions in 1000 or so files in dolfin/la...

--
Anders