On Thu, Aug 2, 2012 at 10:02 AM, Roy Stogner <roystgnr@ices.utexas.edu> wrote:

On Thu, 2 Aug 2012, Paul T. Bauman wrote:
Just to triple check: we currently tie the number of components to the
mesh_dimension() - this will return 2 for two-dimensional elements in 3
spatial dimensions, correct?

Correct.

Perfect, thanks. (FWIW, I did look at the code, but I wanted to triple check in case I missed something. Only thing worse than no comments are wrong comments, etc.).
 
At least for now.  Eventually we'd like to support meshes with mixed
element dimensionality, in which case a mixed 2D/3D mesh would have
mesh_dimension()==3... no clue what we'd do here in that scenario,
though.

Right. I have comments sprinkled in the library code where we are making this assumption. Almost all leverage FEInterface::n_vec_dim so we'll be able to concentrate there when we get around to removing this assumption, but there's a couple of places in build_solution_names and build_solution_vector where the FEType isn't available and I have to explicitly use the mesh_dimension.

Comments were updated and patch committed in r5912.

Thanks for the comments.

Best,

Paul