Setting values with multiindex .loc

[jongmmm]

Code Sample

df = pd.DataFrame({'A':[1,2,3],'B':[4,5,6]})
df = df.stack()
df.loc[0] = [0,0]

This returns

In [27]: df
Out[27]:
0  A    0
   B    0
1  A    2
   B    5
2  A    3
   B    6
dtype: int64

Then, if I try

df.loc[0] = df.loc[1]

I get

In [29]: df
Out[29]:
0  A    NaN
   B    NaN
1  A    2.0
   B    5.0
2  A    3.0
   B    6.0
dtype: float64

Expected Output

It should either raise an exception/warning, or produce the same result with df.loc[0] = df.loc[1].values.

[jreback]

duplicate of #10440

a PR to fix is welcome. This should just work, but as lots of issues related to MultiIndex setting, so would welcome some help here.

[jreback]

cc @toobaz

[toobaz]

Yes, it is also related to #12313, which is basically the same problem when the MultiIndex is on the columns, rather than on the rows. I'm pretty sure we want to support this eventually, but might take time.