Recently I was debugging this simple PySpark code:

First, we get some DataFrame. Next, we collect it to dictionary. It doesn’t matter how you create the dictionary, actually it could be a set or list as well. Finally, we do some filtering with lambda using in operator.

I was running this in Python 3.7.3 and was getting this exception:

It looks like this is related to in operator as replacing lambda with something like list(someDict.keys()).count(x) >= 0 stopped exceptions.
This is due to new generators behavior.
Similar issue was here, here and here.