Jishnu Bhattacharya

In particular, in the above example, you may obtain the index along the dropped dimension using ```julia julia> parentindices(v) (Base.Slice(OffsetArrays.IdOffsetRange(values=0:1, indices=0:1)), -10) ``` Do views work for you? Or do...

For the record I obtain the same run-times ```julia julia> @btime arr_sum($X); 6.528 μs (0 allocations: 0 bytes) julia> @btime arr_sum($XO); 6.535 μs (0 allocations: 0 bytes) ``` This is...

Yes, this isn't urgent, as it's only a performance improvement. Maybe a `IsOneTo` trait in `Base` will be useful, so that custom ranges might opt into that.

This is largely resolved now: ```julia julia> one(Diagonal(Fill(1.0, 2))) 2×2 Diagonal{Float64, Fill{Float64, 1, Tuple{Base.OneTo{Int64}}}}: 1.0 ⋅ ⋅ 1.0 julia> zero(Diagonal(Fill(1.0, 2))) 2×2 Diagonal{Float64, Fill{Float64, 1, Tuple{Base.OneTo{Int64}}}}: 0.0 ⋅ ⋅ 0.0...

Maybe `Ones` or `Zeros` would be better, so this may be improved. This was raised [here](https://github.com/JuliaArrays/FillArrays.jl/issues/173) as well.

I'm starting to think that we should not solve the matrix multiplications involving `adjoint` and `transpose` here. These should really be solved in `LinearAlgebra`. The reason some of these work...

Related: https://github.com/JuliaArrays/FillArrays.jl/issues/23, it would make sense to have both of these added

This does have some unexpected consequences ```julia julia> sosa3 = @view osa[Base.IdentityUnitRange(11:12), Base.IdentityUnitRange(11:12)] 2×2 view(::Array{Float64,2}, 1:2, 1:2) with eltype Float64: 1.0 5.0 2.0 6.0 julia> axes(sosa3) (Base.OneTo(2), Base.OneTo(2)) ``` This...

This might open up a few surprises, since in my use I often use an `OffsetArray` and its parent interchangeably, especially for linear algebra usage

This is potentially breaking if one defines a type that subtypes `Real` but behaves like an integer. Consider something like this: ```julia module CustomInt struct MyInt r = UnitRange(CustomInt.MyInt(2), CustomInt.MyInt(3))...