Matt Bauman

> do we allow any monotonic sequence, or only increasing? I'm fine with purely increasing. I had initially restricted this to purely increasing, but we decided relatively early on that...

I'm not sure how much of my thinking here made it onto public venues… or still remains in my head for that matter (see my edit above). Here's a bit...

Very interesting! No, I've not given Sparse arrays any thought at all… but I'd be happy for AxisArrays to support such a use-case. Especially since I think they'll work out-of-the...

I think that is all very doable. Right now the axis selection is done by trait dispatch, so we could add a new "Sparse" trait that would find its indices...

Yeah, this is tough. Since the axis values aren't in the type domain, we can't do the same sort of clever compile-time lookup like we can with the axis names....

The type instability in your `Val` code is because you're not fully putting the axis values into the parameters of `B` - the parameter is simply `DataType`. Try using ```...

Oh interesting. I didn't think about that. This should become consistent once the Julia parser lowers colons as Colon() instead of calculating the range (which I believe is the plan)....

I'm coming to the conclusion that supporting _both_ by-axis-value and by-axis-index within the `A[]` is fraught with trouble. I think it's crucial to look at prior art here and decisions...

I find the function call interface to make much more sense for interpolations (where it feels like I'm actually executing a calculation to determine the value) vs. here (where it...

Ah, interesting! Thanks for the report. For my own purposes, I really have valued the immediate error there, especially since the nonscalar case (like `atindex(-0.2 .. 0.2, [1, 5])`) would...