Working on #15354 I realized that
A = rand(10, 10)
sA = sub(A, 1:10, 1:10)
sA'
works, but
As = sprand(10, 10, 0.3)
sAs = sub(As, 1:10, 1:10)
sAs'
does not. I would love to fix it, but I am wondering what the preferred way would be. Should I add a method in linalg/abstractsparse.jl, or?
Of course A' and As' works, but sA' seems to be redirected to the correct ctranspose, while sAs' does not get dispatched to the sparse method.
Edit: So I just realized that this goes for all matrix operations on subs of sparse matrices. Is this intentional? @andreasnoack