dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Sun, Jan 31, 2010 at 12:16:04PM +0000, Garth N. Wells wrote:
>
>
> Anders Logg wrote:
> > On Sun, Jan 31, 2010 at 11:59:40AM +0000, Garth N. Wells wrote:
> >>
> >> Anders Logg wrote:
> >>> On Sat, Jan 30, 2010 at 08:47:39PM +0000, Garth N. Wells wrote:
> >>>> Quite a few of the ufl files in the DOLFIN demos are failing with the
> >>>> new FFC when the flag '-l dolfin' is used.
> >>>>
> >>>> Garth
> >>> I have made some fixes and more demos are compiling now. I don't know
> >>> yet which demos are still failing. compileforms currently stalls on
> >>> compiling the HyperElasticity.ufl. Don't know why yet.
> >>>
> >>> The DOLFIN wrapper code generation utils are in serious need to
> >>> cleanup and simplification.
> >>>
> >> Yes. I've been looking at it. It needs to be broken up - it assumes
> >> forms, so it broke for elements.
> >
> > We should get rid of the UFCFormNames class and have a simpler
> > interface. It should essentially be something like
> >
> >   generate_dolfin_wrappers(forms, elements)
> >
> > where forms is a list of tuples (classname, rank).
> >
>
> I think that functionals may work (my mistake).

Yes, but only after a special hack I added. It's not that nice.

> Elements are broken. I'm working on a fix. It won't be pretty, but
> hopefully it will work. We can sort it out properly once FFC is tested
> and fully functional.

ok.

--
Anders

Attachment: signature.asc
Description: Digital signature