[NDTensors] [BUG] Iterating over GradedUnitRange yields unexpected type change

[ogauthe]

using NDTensors.Sectors: U1
using NDTensors.GradedAxes: gradedrange

g1 = gradedrange([U1(0) => 2, U1(1)=>1])
for s in g1
    println(s, " ", typeof(s))
end

outputs

NDTensors.LabelledNumbers.LabelledInteger{Int64, U1{Int64}}(1, U(1)[0]) NDTensors.LabelledNumbers.LabelledInteger{Int64, U1{Int64}}
2 Int64
3 Int64

with a change in iterator type beyond the first element. One obtains the same result with s in eachindex(g1) (I guess they are the same internally)

[emstoudenmire]

I dug into this a bit. Best I can tell, the iteration in BlockArrays (which GradedUnitRange is implemented through) internally calls to a function inc which produces the next value in the iteration by adding the integer 1, like this:

inc(state::Tuple{Integer}, start::Tuple{Integer}, stop::Tuple{Integer}) = (state[1]+1,)

(the above is line 411 of BlockArrays/src/blockindices.jl). Then for a labeled integer, adding +1 returns a regular integer. So I think that is the origin of the current behavior. (Just giving some background so we can consider the next step to take / fix.)

[mtfishman]

This will be fixed in a rewrite I will do of GradedAxes which will account for the new BlockArrays.jl v1.0 release.

[mtfishman]

It was easy enough to fix in #1468 without rewriting the GradedUnitRange type for now.