bluss

rand is a public dependency (we use its traits publicly) - we want to avoid the coupling, that's the reason for the ndarray-rand indirection. For example we don't want an...

ArrayRef is hopefully coming in 0.16. I don't know if these conflict, but I suggest that these changes have to stand back if they can't be realized in the ArrayRef...

While it's a breaking change, we wait with merging this until the 0.15.x releases are done

That sounds good, and I love a good illustration to go together with expressing an idea. One concern I have here is that it's probably still not an efficient way...

Between as_ptr, as_mut_ptr, and the `.axes()` iterator (each axis' length and stride), all the information of the raw representation is available, except for issues related to ownership, and they don't...

Sure, the window iterator is a good idea. If I understand correctly, chunks has a chunk size in each dimension? .axis_chunks_iter() currently covers it if you just want to chunk...

axis_chunks_iter gives views that have the same dimensionality (1d → 1d, 2d → 2d) though.

Right, the current axis_chunks_iter cuts stripes, not squares. So they are different in that way.

Not important, but we can try to do it

Go ahead. 1. Remember how `windows` for slice has no mutable variant? That needs to be true for us too. 2. Maybe I should tell about the trick that we...