When restarting a checkpoint, we callSkip on the previous checkpoint, to trigger the effects of skipping, which includes resetting models and reharnessing.
The problem is, before we supported restarts, I put a sanity check on Task2CP3::Skip to prevent skipping it if it has already been completed. That prevents the harness and model resetting from happening.
I had checked that restarting 2/4 worked, but I had never tested completing 2/3 and then restarting 2/4.
Original report (archived issue) by Louise Poubel (Bitbucket: chapulina, GitHub: chapulina).
When restarting a checkpoint, we call
Skip
on the previous checkpoint, to trigger the effects of skipping, which includes resetting models and reharnessing.The problem is, before we supported restarts, I put a sanity check on
Task2CP3::Skip
to prevent skipping it if it has already been completed. That prevents the harness and model resetting from happening.I had checked that restarting 2/4 worked, but I had never tested completing 2/3 and then restarting 2/4.