Closed sjperkins closed 3 years ago
This allows pickling of the graph which should in principle allow transmission of the graph to the scheduler. Haven't tried it out yet.
This looks good to me and thanks @sjperkins for figuring this out! That seems like a somewhat subtle thing that would have taken me ages to figure out. Are you happy to test your branch before we merge?
This looks good to me and thanks @sjperkins for figuring this out! That seems like a somewhat subtle thing that would have taken me ages to figure out. Are you happy to test your branch before we merge?
Yes, let me test a bit more.
I am not sure why the tests aren't triggering. Maybe because this is from a fork?
I can see them triggering on the PR but you may not see them because of the fork?
Yeah - looks like this is a known thing with actions. Apparently, because the PR originates from your fork, it won't trigger my actions for security reasons.
Tests pass on my fork and the graph serializes onto the CHPC. I think this would be a good point for a review.
Could you run a local test please?
I'm not sure why the tests aren't triggering on the master merge, but https://github.com/JSKenyon/QuartiCal/actions/runs/377569887 passes
It has something to do with forks I think. Will need to figure it out at some point.
namedtuple names in the global space need to be the same as their actual name in order to pickle.