dolfin team mailing list archive

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

On 22/02/11 17:49, Anders Logg wrote:
> On Tue, Feb 22, 2011 at 05:38:24PM +0000, Garth N. Wells wrote:
>>
>>
>> On 22/02/11 17:35, Johan Hake wrote:
>>> What is the signature?
>>>
>>> Have you tried using:
>>>
>>>   PY_SEQUENCE_OF_SCALARS_TO_VECTOR_OF_PRIMITIVES(TYPE,
>>>                                                  TYPE_UPPER,
>>>                                                  ARG_NAME,
>>> 						 TYPE_NAME,
>>>                                                  SEQ_LENGTH)
>>>
>>> Look at the end of std_vector_typemaps.i for example of its use.
>>>
>>> Then you can pass any iterable for the values.
>>>
>>> If you want to expose
>>>
>>>    add_cell(uint c, const std::vector<uint>& v)
>>>
>>
>> That's the one.
>>
>>> You should be able to use something like:
>>>
>>>   IN_TYPEMAP_STD_VECTOR_OF_PRIMITIVES(dolfin::uint, INT32,
>>>                                       v, NPY_UINT, uint, I)
>>>
>>>
>>
>> Where should I put it?
>>
>> Winding this back, is it desirable to wrap std::vector, or should we use
>> Array in the C++ interface?
> 
> Can we add a constructor to Array that takes a std::vector as
> argument? Then it would be unproblematic to have Array "everywhere" in
> the C++ interface.
> 

I think that this would obscure ownership, and eventually lead to trouble.

Garth

> --
> Anders