dolfin team mailing list archive

["Garth N. Wells" <gnw20@xxxxxxxxx>]

Martin Sandve Alnæs wrote:
> 2007/12/12, Garth N. Wells <gnw20@xxxxxxxxx>:
> > Martin Sandve Alnæs wrote:
> > > Simply duplicating the following signatures from Assembler with
> > > ufc::form instead of Form in PyDolfin would be nice:
> > >
> > >
> > >         assemble(self, Form form) -> real
> > >
> > >         assemble(self, Form form, SubDomain sub_domain) -> real
> > >
> > >         assemble(self, Form form,
> > >             dolfin::MeshFunction<(dolfin::uint)> cell_domains,
> > >             dolfin::MeshFunction<(dolfin::uint)> exterior_facet_domains,
> > >             dolfin::MeshFunction<(dolfin::uint)> interior_facet_domains) -> real
> > >
> > >         assemble(self, GenericTensor A, Form form, bool reset_tensor=True)
> > >
> > >         assemble(self, GenericTensor A, Form form, SubDomain
> > > sub_domain, bool reset_tensor=True)
> > Sure. I'll add it soon.

It won't be able to do this as soon as I expected, so feel free to make 
the changes and I can commit them to the repository.

> > I'll also need to send a DofMap since a Form now has a DofMap.
> >
> > Garth
>
> Ok, but is that really necessary? Can't the Assembler simply make a
> DofMap from the ufc::dof_map, and cache it in the DofMapSet using the
> dof_map::signature()?

We don't want the Assembler creating dof maps, otherwise it's difficult 
to keep track of who is using which DofMap, especially if the DofMap 
changes. We have an interface now in which

a) A dolfin::Form carries a DofMapSet; or
b) the form is defined by ufc::form and a DofMapSet is provided.

Garth

> I'll consider making SyFi generate dolfin::Form subclasses too.
>
> --
> Martin
> _______________________________________________
> DOLFIN-dev mailing list
> DOLFIN-dev@xxxxxxxxxx
> http://www.fenics.org/mailman/listinfo/dolfin-dev