Closed vvvvalvalval closed 9 years ago
Hi @vvvvalvalval,
Thanks for the PR. I can understand why this benefits on CCS matrix: the CCSMatrix
constructor fills the matrix column-by-column using the given source, while the transpose()
method always does it in a row-by-row way.
There is new operations design concept which is already working for vectors. I was planing to keep doing the same thing for matrices. This performance issue should be gone with TransposeOperation()
. But for now we can have you fix merged. Could you please fix the identation (use 4 spaces instead a tab) so I can merge the fix?
Here you go!
Thanks @vvvvalvalval! :+1:
The tests pass on my machine.