[Numpy-discussion] NA masks in the next numpy release?
Stéfan van der Walt
Fri Oct 28 04:11:40 CDT 2011
On Thu, Oct 27, 2011 at 7:51 PM, Travis Oliphant <email@example.com> wrote:
> I understand exactly what it would take to add bit-patterns to NumPy. I
> also understand what Mark did and agree that it is possible to add Matthew's
> idea to the current code-base. I think it is worth exploring
Another data point: I've been spending some time on scikits-image
recently, and although masked values would be highly useful in that
context, the cost of doubling memory use (for uint8 images, e.g.) is
too high. Many users with large data sets (and I think almost all
researchers working on >2D data would be included here as well) may
have the same problem.
So, while I applaud the efforts made to include a masked array
implementation, I'd like to ask that:
1) We are mindful that any design decisions taken before the next
release should not *preclude* the implementation of bit-masks (with,
hopefully, a shared interface) and
2) that we make a concerted effort to implement the bitmask mode of
operation as soon as possible.
The NEP stated that both would be implemented, and I understand that
due to lack of time a pragmatic call had to be made--but that was, in
my opinion, one of its strong features.
More information about the NumPy-Discussion