Pandas Nan in groupby gives a wrong behavior for cumsum() of ints contrary to floats

Example : df = pd.DataFrame([[np.nan,0],[0,1]]) df.groupby(0)[1].cumsum()

out: 0 -1 1 1 which is incorrect

df = pd.DataFrame([[np.nan,0],[0.0,1.0]]) df.groupby(0)[1].cumsum()

out: 0 NaN 1 1.0 which is correct

Problem description

pandas version 0.22.0 when the aggregation key contains nan, the cumulative aggregation of int gives a wrong result on the contrary the aggregation of float works.

Comment From: mroeschke

This looks to work on main and I believe we have tests for this so closing