Zachary P Christensen

I have nearly complete solutions worked for padding/resizing/permuting/etc. The remaining issue has really been arrays with different memory backends, which is the point of #130 and #140.

> There are a bunch of changes to Base Julia needed to change that. There are some methods that will have issues and we need to put together our own...

I should probably write something like "The State of ArrayInterface" thing to answer this properly. I've been doing a pretty terrible job of making stuff known and approachable for the...

BTW #141 is the PR I was referring to, so hopefully I can get started on this soon

I'm not sold on the terms at all but it's what I've seen used in several places. If we clearly need that functionality now and in the future it might...

Thanks @AriMKatz. If defining a generic version of `freeze` required explicit knowledge of a more complicated system of effects, then I think this might be the wrong place.

I guess my question really comes down to this: Given the the following code can we guarantee that `MVector{N,eltype(x}}()` and `freeze(out)` don't end up with two separate allocations. ```julia foo(op,...

We have a lot of downstream tests now and I'm consistently encountering failures that I'm pretty sure are unrelated to changes I'm making. Should we specifically be requesting a review...

What is `dense`. It looks to be an instance in some of those and maybe a global variable in others?

So the goal is to have a more structured hook for converting the source matrix?