dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Thu, May 13, 2010 at 11:13:30AM -0700, Johan Hake wrote:
> I think it is as complete as it might get. It all boils down to if we want to
> move the rest of std::vector<Foo>, and Foo* interface to Array<Foo>. I am not
> sure we want that. Maybe for some methods in the LA interface, see below and
> its own Blueprint.
>
> On Thursday May 13 2010 10:27:33 Anders Logg wrote:
> >  * Cleanup (remove?) the std::vector typemaps.
> >     -> We cannot remove all std::vector typemaps.
> >          |-> We still use std::vector<Foo*> typemaps.They might be changed
> >          |   with Array, but it will not be easier.
> >          |-> std::vector<Foo>& argout typemaps are useful to pass data out
> >          |   from intersection detectors and like.
>
> I think we are fine with the few std::vector<Foo> typemaps we still have. It
> will be difficult to move both of these cases to Array<Foo>
>
> >   Things left to consider:
> >   * Consider which other foo* or std::vector<foo> methods should be using
> >     Array<foo>
> >     -> GenericFoo.{get,set} in the la interface?
> >     -> Expression(std::vector<uint> value_shape)?
> >     -> GenericMatrix.{get,set}row?
>
> The first two cases might benefit from a change. Any such change should be
> readily wrapped by the present Array typemap. We might need to do something in
> the python layer for the Expression typemap.
>
> I am not sure with the last issue. I remember that there were some benefits
> using std::vector for this.

I think we should use Array<> as much as possible in the public
interface, but have std::vector<> as much as possible internally for
exchanging data between classes.

I've added two new blueprints for further discussion:

https://blueprints.launchpad.net/dolfin/+spec/switch-from-std-vector-to-array
https://blueprints.launchpad.net/dolfin/+spec/add-std-vector-constructor-to-array

--
Anders

Attachment: signature.asc
Description: Digital signature