rowvals of transposed matrices in 0.7

General Usage
1

Dear Julia pros, I am trying to migrate an existing application, that works just fine in 0.6.2, to 0.7 (binary installation on Linux 64bit) and I am having the following problem:
When calling SparseArrays.rowvals on the original, non-transposed matrix, everything works fine.
When calling rowvals on the transposed matrix, I am getting a LoadError.
This is the non-working example:

using LinearAlgebra
using SparseArrays

function do_something(amat::SparseMatrixCSC)
    ats = Transpose(amat)
    println(typeof(ats))
    rows = SparseArrays.rowvals(ats)
    return rows
end

toy_matrix = sparse(1.0I,3,3)
bla = do_something(toy_matrix)

It does not make any difference whether I call SparseArrays.rowvals or rowvals directly.

Any insight is welcome. Best regards, Frank

2

You can materialize the transpose/adjoint with copy.

julia> function do_something(amat::SparseMatrixCSC)
           ats = copy(amat')
           println(typeof(ats))
           rows = SparseArrays.rowvals(ats)
           return rows
       end
do_something (generic function with 1 method)

julia> bla = do_something(toy_matrix)
SparseMatrixCSC{Float64,Int64}
3-element Array{Int64,1}:
 1
 2
 3
3

Thank you very much. Indeed, this settles the issue.