dolfin team mailing list archive

["Martin Sandve Alnæs" <martinal@xxxxxxxxx>]

2008/10/31 Anders Logg <logg@xxxxxxxxx>:
> On Fri, Oct 31, 2008 at 10:12:03AM +0000, Garth N. Wells wrote:
>>
>>
>> DOLFIN wrote:
>> > One or more new changesets pushed to the primary dolfin repository.
>> > A short summary of the last three changesets is included below.
>> >
>> > changeset:   5053:a37ad63c855383fa6aae8b15bf20720b1a96398f
>> > tag:         tip
>> > user:        Anders Logg <logg@xxxxxxxxx>
>> > date:        Thu Oct 30 18:53:36 2008 +0100
>> > files:       demo/pde/poisson/cpp/main.cpp dolfin/fem/DirichletBC.cpp dolfin/function/Constant.cpp dolfin/function/Constant.h dolfin/function/Function.cpp dolfin/function/Function.h
>> > description:
>> > Make Constants work (only scalar constructors added so far)
>> >
>>
>> It would make things simple if we stick to scalar constants and I would
>> be happy with that. Constant vector functions for which all components
>> are equal are easy too, but not all that useful. Anything else, like a
>> vector [0, 1] will required a FiniteElement to guarantee the correct
>> tabulation, right?
>>
>> Garth
>
> That's not a problem. The FunctionSpace for a Constant will be
> discovered automatically.
>
> If a Constant is used in a Form (a.c = c), then the FunctionSpace will
> be created automatically.
>
> If a Constant is used in a DirichletBC, then the FunctionSpace is
> known (but will not be assigned to the Constant since it is const).
>
> --
> Anders

If I understand this correctly, this sounds severely limiting and ruins
the simplicity of "a Function always has a FunctionSpace".
A Constant should be possible to use everywhere a Function
is possible to use, including external code.

--
Martin