[Numpy-discussion] Release of NumPy

[Stéfan van der Walt]

On 17/04/2008, Travis E. Oliphant <oliphant at enthought.com> wrote:
> I have mixed feelings on all of this discussion.  I do want to make
>  sure, however, that two ideas are considered:
>
>  1) How will sparse matrices behave?  It seems to me that the matrix
>  object and sparse matrices should behave identically if at all possible.

That's a good question.  Sparse matrices are not containers -- they
are always 2-dimensional, so it brings us back to what a "row" slice
of a sparse matrix should be: probably a sparse matrix.  That's the
way matrices behave currently, but that behaviour differs from the
ndarray behaviour -- where we have N-d arrays containing (N-1)-d
arrays.

>  2) How much code will break if we change the current behavior.

None (for dense matrices), if we introduce vectors.  The only
difference in behaviour would be for x[0][0] which then becomes
equivalent to the ndarray syntax.

Stéfan