Tim Holy

Moving to Base actually seems like the right solution. Or rather, a mixture of Base (for core traits) and LinearAlgebra (which will presumably become a package eventually).

My conclusion from that discussion is that the best interface may be the one we have now, where you manually write out the trait depot.

Re IArrays: it seems like it would be better to add whatever is needed to StaticArrays. They have a good abstract hierarchy and you can get specific features by subclassing...

I very much agree with this. It would be best, I think, to have a basal traits package (probably in JuliaArrays) consisting of just the fallback definitions. It would be...

I've always imagined that multispectral images would be handled with custom "color" types. ```julia struct Hyper8{T}

We definitely should make more efforts to unite the ML world with JuliaImages. It's one of my priorities for the next six months.

That can definitely help (and indeed is key to our handling of, say, time), but the issue is that then algorithms have to be written specially for the case when...

It's a consequence of https://github.com/JuliaImages/ImageTransformations.jl/pull/143. It only affects points exactly at the edge, and it's due to ulp-level roundoff errors causing a point that was judged as being interior now...

You have done wonders for these packages. "Cleanups" can be done after 1.0 if they don't break the public API. If we grossly change the result of a white-point aware...

Yeah, we need to do something about this. I am pretty swamped with TTFX work until at least mid-October, unfortunately. If you'd be willing to figure out what to do...