[Numpy-discussion] Contiguous flags not quite right yet.
Thu Mar 22 17:19:36 CDT 2007
Charles R Harris wrote:
> All three shapes are both C_CONTIGUOUS and F_CONTIGUOUS. I think
> ignoring all 1's in the shape would give the right results for
> otherwise contiguous arrays because in those positions the index can
> only take the value 0.
I've thought about this before too. But, every time I've tried to do
something like that with the flags, I get many errors in numpy and scipy
because of the way contiguousness is actually used. Somehow, I've never
gotten the right check to allow additional striding arrays to work
functions are what should be changed if anything is done.
More information about the Numpy-discussion