Change of rolling behavior in Pandas 1.0.0 vs 0.25.3

[icexelloss]

Code Sample, a copy-pastable example if possible

df = pd.DataFrame({
        'a': np.arange(10, dtype=float),
        'b': [3.0, np.NaN] * 5,
        'key': list('ddeefffggh'),
    })

expected = df.sort_values(['key', 'a']).assign(
    rolled=lambda d: d.groupby('key')
    .b.rolling(3, min_periods=1)
    .apply(lambda x: x.mean(), raw=True)
    .reset_index(level=0, drop=True)
)

Problem description

1.0.0 result:

     a    b key  rolled                                                                                                                                                                                                                                                                                                                                                                    
0  0.0  3.0   d     3.0                                                                                                                                                                                                                                                                                                                                                                    
1  1.0  NaN   d     3.0                                                                                                                                                                                                                                                                                                                                                                    
2  2.0  3.0   e     3.0                                                                                                                                                                                                                                                                                                                                                                    
3  3.0  NaN   e     3.0                                                                                                                                                                                                                                                                                                                                                                    
4  4.0  3.0   f     3.0                                                                                                                                                                                                                                                                                                                                                                    
5  5.0  NaN   f     3.0                                                                                                                                                                                                                                                                                                                                                                    
6  6.0  3.0   f     3.0                                                                                                                                                                                                                                                                                                                                                                    
7  7.0  NaN   g     NaN                                                                                                                                                                                                                                                                                                                                                                    
8  8.0  3.0   g     3.0                                                                                                                                                                                                                                                                                                                                                                    
9  9.0  NaN   h     NaN

0.25.3 behavior:

     a    b key  rolled
0  0.0  3.0   d     3.0
1  1.0  NaN   d     NaN
2  2.0  3.0   e     3.0
3  3.0  NaN   e     NaN
4  4.0  3.0   f     3.0
5  5.0  NaN   f     NaN
6  6.0  3.0   f     NaN
7  7.0  NaN   g     NaN
8  8.0  3.0   g     NaN
9  9.0  NaN   h     NaN

Note: We receive a lot of issues on our GitHub tracker, so it is very possible that your issue has been posted before. Please check first before submitting so that we do not have to handle and close duplicates!

Note: Many problems can be resolved by simply upgrading pandas to the latest version. Before submitting, please check if that solution works for you. If possible, you may want to check if master addresses this issue, but that is not necessary.

For documentation-related issues, you can check the latest versions of the docs on master here:

https://pandas-docs.github.io/pandas-docs-travis/

If the issue has not been resolved there, go ahead and file it in the issue tracker.

Expected Output

Output of pd.show_versions()

[paste the output of pd.show_versions() here below this line]

[jreback]

see #26996

[icexelloss]

I see. Thanks! @jreback