Re: Empty matrices don't have trailing ones?

[Jordi Gutiérrez Hermoso]

On 10 May 2012 17:11, John W. Eaton <address@hidden> wrote:
> On 10-May-2012, Jordi Gutiérrez Hermoso wrote:
>
> | On 10 May 2012 00:46, Jordi Gutiérrez Hermoso <address@hidden> wrote:
> | > I'm trying to figure out the rationale for having dim_vector::redim
> | > padding all-zero dim_vectors with zeros instead of ones.
> |
> | Responding to myself, looks like many tests break if dim_vector::redim
> | always pads with ones, but the two problems I've encountered like bug
> | #33216 and x = []; x(:,:,1) do get fixed by padding with ones. I won't
> | try to investigate the logic, so instead I attempted the attached
> | patch which disables zero-padding for the two cases in which I think
> | it doesn't make sense.
> |
> | I think this breaks ABI or API, and the bug it's patching is very
> | obscure and minor, so I am thinking about pushing this on the default
> | branch. Thoughts?
>
> How about the following change instead?  With it, I don't see any
> additional test failures.  Probably it needs a test for the reported
> bug and to have the commit message include the bug report number.

Thanks, I see you took care of all the paperwork too. So it was only
Array<T>::assign that cared about that zero-padding behaviour?
Interesting. I didn't attempt to figure out the logic, so thanks for
doing it for me.

- Jordi G. H.