[Numpy-discussion] Change default order to Fortran order

[Charles R Harris]

On Mon, Aug 3, 2015 at 10:24 AM, Matthew Brett <matthew.brett at gmail.com>
wrote:

> On Mon, Aug 3, 2015 at 5:01 PM, Sturla Molden <sturla.molden at gmail.com>
> wrote:
> > Matthew Brett <matthew.brett at gmail.com> wrote:
> >
> >> Sure, but to avoid confusion, maybe move the discussion of image
> >> indexing order to another thread?
> >>
> >> I think this thread is about memory layout, which is a different issue.
> >
> > It is actually a bit convoluted and not completely orthogonal. Memory
> > layout does not matter for 2d ndexing, i.e. (x,y) vs. (row, column), if
> you
> > are careful when iterating, but it does matter for Nd indexing. There is
> a
> > reason to prefer (x,y,z,t,r) in column major order or (recording, time,
> > slice, row, column) in row major order. Otherwise you can get very
> > inefficient memory traversals. Then if you work with visualization
> > libraries that expects (x,y,z) and column major order, e.g. ITK, VTK and
> > OpenGL, this is really what you want to use. And the choise of indexing
> > (x,y,z) cannot be seen as independent of the memory layout. Remember, it
> is
> > not just a matter of mapping coordinates to pixels. The data sets are so
> > large in MRI processing that memory layout does matter.
>
> I completely agree that memory layout can affect performance, and this
> can be important.
>
> On the other hand, I think you agree that the relationship of axis
> order to memory layout is just a matter of mapping coordinates to
> pixels.
>
> So you can change the axis ordering without changing the memory layout
> and the memory layout without changing the axis ordering.
>
> Of course you could argue that it would be simpler to fuse the two
> issues, and enforce one memory layout - say Fortran.    The result
> might well be easier think about, but it wouldn't be much like numpy,
> and it would have lots of performance and memory disadvantages.
>
> Cheers,
>

I would also strongly suggest that once you have decided on a convention
that you thoroughly document it somewhere. That will not only help you, but
anyone who later needs to maintain the code will bless you rather than d*nm
you to eternal torture by the seven demons of Ipsos.

Chuck
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.python.org/pipermail/numpy-discussion/attachments/20150803/5bd14d70/attachment.html>