With Datastore, we're writing the stat dict directly to the SDK (rather than pickling it first). This is why party members are no longer a set (and therefore need explicit duplicate prevention).
json.pickle is overkill... we can save all the literal data with json.dumps, and doing so will keep us from accidentally trying to persist pickled objects to datastore.
With Datastore, we're writing the stat dict directly to the SDK (rather than pickling it first). This is why party members are no longer a
set(and therefore need explicit duplicate prevention).json.pickleis overkill... we can save all the literal data with json.dumps, and doing so will keep us from accidentally trying to persist pickled objects to datastore.