now you can still use column access previously... mat[2] = 1, 2, 3 needed to be converted into... mat[0][2] = 1 mat[1][2] = 2 mat[2][2] = 3 but with column access you can do... mat.col[2] = 1, 2, 3 Having 'row' attribute is a bit redundant since direct indexing on a matrix uses row but included for completeness.