Currently we do not test for null values. We should do so, however, Pandas and NumPy have some limitations (like the fact that to get an integer array that can contain nulls in Pandas you have to use an extension type). Additionally, in NumPy, integer arrays simply do not support np.nan. Only Int64 arrays do.
- related to
-
INTPYTHON-113 NaNs are replaced by pandas with int64 minimum
- Closed