Is there a reason why in LinearAlgebra transposes and adjoints of triangular matrices are not triangular matrices themselves? For example why transpose of an UpperTriangular matrix is not a LowerTriangular matrix that wraps transpose of the original matrix.
StaticArrays does commute these two wrappers as the issue was raised here: https://github.com/JuliaArrays/StaticArrays.jl/pull/837#discussion_r506040427 . Essentially all algorithms that work on triangular matrices could also work on transposes and adjoints of them but not commuting these two wrappers makes it harder to dispatch on matrices with triangular structure.
As it turns out, this is what it used to be and got changed in #25364. In that gigantic PR, this was just a small detail and I believe there was no deeper reason to flip the order of the two lazy wrappers. I started changing it, so hopefully will have a PR soon.
Most helpful comment
As it turns out, this is what it used to be and got changed in #25364. In that gigantic PR, this was just a small detail and I believe there was no deeper reason to flip the order of the two lazy wrappers. I started changing it, so hopefully will have a PR soon.