ERR: DataFrame.fillna with non-unique mapping

[dsaxton]

• I have checked that this issue has not already been reported.
• I have confirmed this bug exists on the latest version of pandas.
• (optional) I have confirmed this bug exists on the master branch of pandas.

---

import numpy as np
import pandas as pd

df = pd.DataFrame({"a": [1.0, np.nan]})
s = pd.Series([0.0, 2.0], index=["a", "a"])
df.fillna(s)  # works silently on master or 1.1.2
#      a
# 0  1.0
# 1  0.0

Problem description

DataFrame.fillna using a Series with a duplicated index is potentially ambiguous, but the filling works silently (apparently using the first value).

Expected Output

Maybe a warning that the operation is ambiguous?

[erfannariman]

Is this expected behaviour? Should it not raise when we try to fillna on non unique labels? Non-intuitive, especially the matching over different axis.

[dsaxton]

Is this expected behaviour?

I'm not actually sure, it's at least undocumented how it should behave. I think you could argue it should be an error, but that could potentially break code somewhere and may not be worth it.