ffc team mailing list archive

[Martin Sandve Alnæs <martinal@xxxxxxxxx>]

On Thu, Mar 5, 2009 at 9:55 PM, Garth N. Wells <gnw20@xxxxxxxxx> wrote:
>
>
> Anders Logg wrote:
>> On Thu, Mar 05, 2009 at 08:39:03PM +0000, Garth N. Wells wrote:
>>> What remains to be done on the UFL and FFC sides to get the necessary
>>> data to generate the DOLFIN wrappers?
>>>
>>> Garth
>>
>> I don't know. I'm still trying to get the basic stuff going for the
>> tensor representation. Kristian probably has a better idea.
>>
>
> Kristian is unsure of on what side of the UFL/FFC divide the work should
> be done w.r.t. extracting coefficients.
>
> Garth

FormData already does that, and load_forms reads form
and coefficient names from a .ufl file. The Python utility module
I added to dolfin a while ago (dolfin_utils I think it's called)
generates DOLFIN C++ code from UFL FormData.

a = f*u*v*dx
fd = a.form_data() # guaranteed to only be computed once
fd.original_functions # list of the original Function objects in a

Martin