When all of the parts of a question are submitted at once, Numbas sets the SCORM key cmi.score.scaled several times in quick succession.
When these get to the LTI server, they each trigger a task to save the cached scaled_score on the Attempt model, and somehow they can get out of order.
I think it's reasonable to group these elements, and only send something like one per second at most.
When all of the parts of a question are submitted at once, Numbas sets the SCORM key
cmi.score.scaledseveral times in quick succession.When these get to the LTI server, they each trigger a task to save the cached
scaled_scoreon the Attempt model, and somehow they can get out of order.I think it's reasonable to group these elements, and only send something like one per second at most.