dolfin team mailing list archive

["Garth N. Wells" <g.n.wells@xxxxxxxxxx>]

Anders Logg wrote:
> On Wed, May 30, 2007 at 05:35:40PM +0200, Garth N. Wells wrote:
>
> [snip]
> > I don't expect that many new classes/files will be needed (with some 
> > luck only one for the mesh, perhaps a PAssembler and some tuning of the 
> > PETSc wrappers). Also, what is needed is closely related to existing 
> > data structures (Mesh, PETScXXX and Assembler). We can add a 
> > sub-directory later if we have too many files.
> >
> > Garth
>
> ok, that sounds good.
>
> Don't we also need one for MeshFunction?

This will be handled by MPIMeshCommunicator. It has member functions

  /// Broadcast MeshFunction to all processes
  void broadcast(const MeshFunction<unsigned int>& mesh_function);

  /// Receive MeshFunction
  void receive(MeshFunction<unsigned int>& mesh_function);

What is required are (protected) member functions belonging to Mesh and 
MeshFunction which return pointers to the underlying plain arrays which 
contain the data, and member functions which allow Mesh and 
MeshFunction<unsigned int> to be initialised using pointers to the 
underlying plain arrays.

Garth


> /Anders
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev