BUG: setitem using `loc` not aligning on index?

[MarcoGorelli]

Pandas version checks

• I have checked that this issue has not already been reported.

• I have confirmed this bug exists on the latest version of pandas.

• I have confirmed this bug exists on the main branch of pandas.

Reproducible Example

In [5]: df = DataFrame({"a": [1, 2], "b": [3, 4]})

In [6]: other = pd.Series([200, 999], index=[1, 0])

In [7]: df
Out[7]:
   a  b
0  1  3
1  2  4

In [8]: other
Out[8]:
1    200
0    999
dtype: int64

In [9]: df.loc[:, 'a'] = other

In [10]: df
Out[10]:
     a  b
0  200  3
1  999  4

Issue Description

When setting a column using .loc, and value is a Series, then the index is ignored

Expected Behavior

I think operations like this one usually align automatically on the index?

@phofl @jbrockmendel do you know about this one?

Installed Versions

INSTALLED VERSIONS

commit : 2a953cf
python : 3.10.12.final.0
python-bits : 64
OS : Linux
OS-release : 5.10.102.1-microsoft-standard-WSL2
Version : #1 SMP Wed Mar 2 00:30:59 UTC 2022
machine : x86_64
processor : x86_64
byteorder : little
LC_ALL : None
LANG : en_GB.UTF-8
LOCALE : en_GB.UTF-8

pandas : 2.1.3
numpy : 1.25.1
pytz : 2023.3
dateutil : 2.8.2
setuptools : 67.6.1
pip : 23.1.2
Cython : None
pytest : 7.3.1
hypothesis : 6.82.4
sphinx : None
blosc : None
feather : None
xlsxwriter : None
lxml.etree : None
html5lib : None
pymysql : None
psycopg2 : None
jinja2 : 3.1.2
IPython : 8.16.1
pandas_datareader : None
bs4 : 4.12.2
bottleneck : None
dataframe-api-compat: None
fastparquet : 2023.7.0
fsspec : 2023.6.0
gcsfs : None
matplotlib : 3.7.1
numba : 0.58.1
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
pyarrow : 12.0.1
pyreadstat : None
pyxlsb : None
s3fs : None
scipy : 1.11.0
sqlalchemy : 2.0.20
tables : None
tabulate : None
xarray : None
xlrd : None
zstandard : None
tzdata : 2023.3
qtpy : None
pyqt5 : None

[MarcoGorelli]

It is respected, though, if doing the assignment horizontally:

In [15]: df = pd.DataFrame({'a': [1,1,2], 'b': [4,5,6]})

In [16]: df.loc[0, :] = pd.Series({'b': 999, 'a': 888})

In [17]: df
Out[17]:
     a    b
0  888  999
1    1    5
2    2    6

[MarcoGorelli]

hey, the non-inplace version does actually respect the index 😄

In [25]: df = DataFrame({"a": [1, 2], "b": [3, 4]})

In [26]: other = pd.Series([200.5, 999], index=[1, 0])

In [27]: df.loc[:, 'a'] = other

In [28]: df
Out[28]:
       a  b
0  999.0  3
1  200.5  4

[phofl]

Can you check with df

df = DataFrame({"a": [1, 2], "b": [3, 4.5]})

we have code paths for single block and multi block.

You are correct, all of those should align

[MarcoGorelli]

that one works!

In [38]: df = DataFrame({"a": [1, 2], "b": [3, 4.5]})

In [39]: other = pd.Series([200, 999], index=[1, 0])

In [40]:  df.loc[:, 'a'] = other

In [41]: df
Out[41]:
     a    b
0  999  3.0
1  200  4.5

[phofl]

Yeah that's what I thought, the single block code path is the culprit (as usual...)

[jbrockmendel]

possibly related: #51386, #37516.

my default opinion is "if its ambiguous, let's deprecate allowing it"

[phofl]

No it's not ambiguous, it should work in our current setup

[phofl]

FWIW my PR that was a precursor for @MarcoGorelli PR about the warning should have fixed this (could you add a test in your other pr @MarcoGorelli ?)

[MarcoGorelli]

thanks - looks fixed on main, so have added 'needs tests'