dataframe rolling issue

[mattyd100]

I have a PintArray in a Dataframe column and I want to calculate the rolling sum and put the values in a new column. The resulting column is an array of floats instead of a PintArray. Is there a way that I can perform the calculation so that the result is a PintArray?

`import numpy as np
import pandas as pd
import pint
import pint_pandas

df = pd.DataFrame(pint_pandas.PintArray(np.arange(10), dtype='pint[in]'), columns = ['A'])
df['B'] = df['A'].rolling(2).sum()
print(df.dtypes)
`

[andrewgsavage]

you can do this for now
df['B'] = df['A'].rolling(2).sum().astype(df['A'].dtype)
but yea that's not ideal

[andrewgsavage]

This will need fixing in pandas, so open an issue there.