Double pointers used for rectangular arrays

[Garth Wells]

The UFC interface uses double pointers for rectangular arrays, e.g.

    virtual void tabulate_coordinates(double** coordinates, const cell& c) const = 0;

This is annoying because it restricts the data structures that can be efficiently used on the library size. For example, boost::multi_array and Armadillo are useful for storing multi-dimensional arrays, and can return plain pointers to the underlying array, but UFC forces double pointers on the user.

[Garth Wells (garth-wells)]

See also https://lists.launchpad.net/fenics-ufc/msg00962.html

[Garth Wells (garth-wells)]

We should look at this - it will allow the removal of some ugly copies in DOLFIN.

[Johan Hake (johan-hake)]

On Wednesday September 21 2011 09:00:16 Garth Wells wrote:
> We should look at this - it will allow the removal of some ugly copies
> in DOLFIN.

Agree, but at the moment are boost::multi_arrays not so pleasent for the SWIG
interface. Copying is enforced. Not sure this is a big deal for such a low
lever functionality, but it is sad that it should be needed, if the underlying
structure is just a c-array, which numpy wraps just fine.

Johan

[Garth Wells (garth-wells)]

On 21 September 2011 17:40, Johan Hake <email address hidden> wrote:
> On Wednesday September 21 2011 09:00:16 Garth Wells wrote:
>> We should look at this  - it will allow the removal of some ugly copies
>> in DOLFIN.
>
> Agree, but at the moment are boost::multi_arrays not so pleasent for the SWIG
> interface. Copying is enforced. Not sure this is a big deal for such a low
> lever functionality, but it is sad that it should be needed, if the underlying
> structure is just a c-array, which numpy wraps just fine.
>

I think that if this type of data is used intensively at the Python
level, then there will be bigger performance issues to worry about
than data copying.

The multi_array interface is somewhat secondary - the important point
is flattening the double pointer in the UFC interface.

Garth

> Johan
>
> --
> You received this bug notification because you are subscribed to the bug
> report.
> https://bugs.launchpad.net/bugs/787405
>
> Title:
>  Double pointers used for rectangular arrays
>
> To manage notifications about this bug go to:
> https://bugs.launchpad.net/ufc/+bug/787405/+subscriptions
>

[Anders Logg (logg)]

On Wed, Sep 21, 2011 at 04:54:29PM -0000, Garth Wells wrote:
> On 21 September 2011 17:40, Johan Hake <email address hidden> wrote:
> > On Wednesday September 21 2011 09:00:16 Garth Wells wrote:
> >> We should look at this  - it will allow the removal of some ugly copies
> >> in DOLFIN.
> >
> > Agree, but at the moment are boost::multi_arrays not so pleasent for the SWIG
> > interface. Copying is enforced. Not sure this is a big deal for such a low
> > lever functionality, but it is sad that it should be needed, if the underlying
> > structure is just a c-array, which numpy wraps just fine.
> >
>
> I think that if this type of data is used intensively at the Python
> level, then there will be bigger performance issues to worry about
> than data copying.
>
> The multi_array interface is somewhat secondary - the important point
> is flattening the double pointer in the UFC interface.

We should avoid using any other data structures than plain C arrays in
the UFC interface.

Or we should use STL data structures (or boost) throughout UFC, but
that sounds more like UFC >= 3.0.

--
Anders

> Garth
>
> > Johan
> >
> >
> > Title:
> >  Double pointers used for rectangular arrays
> >
> > To manage notifications about this bug go to:
> > https://bugs.launchpad.net/ufc/+bug/787405/+subscriptions
> >
>