Eric Davies

This should add some `@inferred` checks to `vcat` and `hcat` to make sure `fastcat` is const-propagating `n`.

Seems to not infer on Julia 1.2 and earlier. Anyone else have any thoughts on whether that's okay?

I might be able to tackle this using `oneunit`. I'll try to solve this tomorrow.

I think that fits with https://github.com/JuliaArrays/AxisArrays.jl/pull/152

Do you mean `A[1] == A[:b] == 0` and `A[2] == A[:a] == 1`?

@vtjnash has given me a tentative "won't fix" but pointed me to some examples on how this issue was fixed in Base.

I'm not sure what the difference between `NamedAxisArray` and `NamedAxisDict` is. Are these for the data arrays or for axes?

It sounds like `NamedAxis*` are for mapping dimension names to dimension numbers only. In that case, shouldn't the wrapper not care how indexing works on the wrapped structure, and just...

In my mind AxisArrays has these basic levels of processing (let me know if I'm missing anything): --- A. Index into an AxisArray with integers/ranges: - like a normal AbstractArray...

I agree, that makes sense