Sheehan Olver

Correct: it should at minimum only allow taking in a ` AbstractQuasiVector `

TBF up until now I was the only user and I never passed in junk. So perhaps the real issue is that the user base has doubled and is now...

I think this was the same bug as `U \ diff(U)` and is now fixed on the right branches: ```julia julia> P \ Normalized(Q) \(ℵ₀×ℵ₀ LazyBandedMatrices.Bidiagonal{Float64, LazyArrays.BroadcastVector{Float64, typeof(/), Tuple{InfiniteArrays.InfStepRange{Float64, Float64},...

The issue is the following: we are multiplying `R` and `D` that look like: ```julia julia> R inv(ℵ₀×ℵ₀ BandedMatrix{Float64} with bandwidths (0, 2) with data vcat(((-).((Float64) ./ (ℵ₀-element InfiniteArrays.InfStepRange{Float64, Float64}...

Yes I started debugging and looks like `1 .* X` is the MWE. It should return: ```julia julia> BroadcastArray(*, 1, X) (Int64) .* (ℵ₀×ℵ₀ LazyBandedMatrices.Tridiagonal{Float64, ApplyArray{Float64, 1, typeof(vcat), Tuple{Float64, BroadcastVector{Float64,...

It looks like the issue is: ```julia julia> Base.BroadcastStyle(typeof(X)) LinearAlgebra.StructuredMatrixStyle{Tridiagonal{Float64, ApplyArray{Float64, 1, typeof(vcat), Tuple{Float64, BroadcastVector{Float64, typeof(/), Tuple{Float64, BroadcastVector{Float64, typeof(*), Tuple{InfiniteArrays.InfStepRange{Float64, Float64}, InfiniteArrays.InfStepRange{Float64, Float64}}}}}}}}}() ``` This should be returning some sort...

Can you add a description to ElTypeExtensions.jl README?

I think this is fine but perhaps we should move EltypeExtensions.jl to JuliaArrays first? It's better for it to be a community developed pacakge.

I think `elconvert` is not a great name. `convert_eltype` would be preferable.

I used th only because the old plans had bad complexity, now that’s fixed we should switch to the default it uses h-matrices/ FMM I think