ffc team mailing list archive

[Matthew Knepley <knepley@xxxxxxxxxxx>]

Anders Logg <logg@xxxxxxxxx> writes:

> ok, now I understand.
>
> Good to see that the abstractions I made (the format dictionaries)
> were almost general enough for you to actually add a new format.
>
> Have you looked at dofmap.py and pointmap.py? These generate code for
> mapping of dofs from the reference element. They are currently
> somewhat DOLFIN specific but I think they could also be generalized.

  I will look at them next. I have something similar called System now
which I am making compatible.

      Matt

> /Anders
>
> On Mon, Jun 06, 2005 at 08:14:58PM -0500, Matthew Knepley wrote:
>> 
>>        Alright, I think it should finally work. As to the
>> second question, the situation is different here. You do not
>> need DOLFIN or LaTeX installed to generate them because you
>> are using a certain text generation technology. The ASE format
>> is using another "generation" technology, namely ASTs instead
>> of text. It could equally well generate C code for DOLFIN.
>> 
>>    Hope that is clearer,
>> 
>>         Matt
>> 
>> There are still problems if I import the ase module:
>> 
>>   File "../ffc/compiler/compiler.py", line 23, in ?
>>     from ffc.format import ase
>>   File "../ffc/format/ase.py", line 43, in ?
>>     format = { "sum": sum,
>> AttributeError: 'NoneType' object has no attribute 'getVar'
>> 
>> I'll keep the import under the if-case in compiler.py until we know
>> how to do this.
>> 
>> What things from ASE are you using in ase.py? It should be possible to
>> compile without having access to the system you are compiling for (you
>> don't need to have DOLFIN or LaTeX installed to use -dolfin or -latex).
>> 
>> (As a side note, I should change -c++ to -dolfin.)
>> 
>> /Anders
>> 
>> 
>
> -- 
> Anders Logg
> Research Assistant Professor
> Toyota Technological Institute at Chicago
> http://www.tti-c.org/logg/
>
>
>

-- 
"Failure has a thousand explanations. Success doesn't need one" -- Sir Alec Guiness