jupyter
Autograding submissions in parallel
It would be really nice to be able to run nbgrader autograde in parallel (also probably nbgrader feedback). This would require making sure that the execute preprocessors don't try to start on the same ports (otherwise we get zmq binding errors).
Anybody familiar with the Source Code care to give a few pointers in the right direction? What are the challenges? It's kinda sad that this is not a feature yet, especially after having being proposed more than 8 years ago. Auto-grading hundreds of assignments in serial can take forever (days), when there are long-running single-threaded tasks in the assignment. Sending the auto-grading tasks to separate workers should have huge performance gains on today's hardware (or on a Compute Cluster).
