Pandas Rolling window aggregation functions do not skip NaNs

Code Sample, a copy-pastable example if possible

df = pd.DataFrame({'a': [None, 2, 3]})

df.rolling(3)['a'].sum()
df.rolling(3)['a'].mean()

The result gives all NaNs (even for the 3rd row):

Out[2]:
0   NaN
1   NaN
2   NaN
Name: a, dtype: float64

Problem description

Differently from DataFrameGroupBy aggregation functions, where NaNs are skipped by default (skipna=True), this is not the case for Rolling aggregation functions. If there is a NaN in the rolling Window, aggregation functions on the rolling Window will give NaN as result. There does not even exist the option skipna for aggregation functions on a Rolling window.

Expected Output

The expected output would be the one skipping NaNs:

In[2]: df.rolling(3)['a'].sum()
Out[2]:
0    NaN
1    NaN
2    5.0
Name: a, dtype: float64

Comment From: jreback

This is min_periods, part of the .rolling doc-string

min_periods : int, default None
    Minimum number of observations in window required to have a value
    (otherwise result is NA). For a window that is specified by an offset,
    this will default to 1.

In [12]: df.rolling(3)['a'].sum()
Out[12]: 
0   NaN
1   NaN
2   NaN
Name: a, dtype: float64

In [13]: df.rolling(3, min_periods=1)['a'].sum()
Out[13]: 
0    NaN
1    2.0
2    5.0
Name: a, dtype: float64