BUG: Inconsistency for iloc between np.array and pd.array

[phofl]

• 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.

---

Note: Please read this guide detailing how to provide the necessary information for us to reproduce your bug.

Code Sample, a copy-pastable example

We got some inconsistencies with iloc between setting a numpy array and a pandas array

df = pd.DataFrame(data={
    'col1': [1, 2, 3, 4],
    'col2': [3, 4, 5, 6],
    'col3': [6, 7, 8, 9],
})
rhs = pd.array([1, 2, 3])
df.iloc[[1, 2, 3]] = rhs

Problem description

This returns

   col1  col2  col3
0     1     3     6
1     1     1     1
2     2     2     2
3     3     3     3

When switching the rhs to a numpy array, this returns

rhs = np.array([1, 2, 3])

   col1  col2  col3
0     1     3     6
1     1     2     3
2     1     2     3
3     1     2     3

Same happens when using [[1, 2, 3], :] as indexer

Expected Output

We should be consistent here, if we want to change iloc for the numpy case I think we need to deprecate?

Is this already known?
cc @jbrockmendel @jorisvandenbossche

Output of pd.show_versions()

INSTALLED VERSIONS ------------------ commit : c8e077f python : 3.8.6.final.0 python-bits : 64 OS : Linux OS-release : 5.8.0-48-generic Version : #54~20.04.1-Ubuntu SMP Sat Mar 20 13:40:25 UTC 2021 machine : x86_64 processor : x86_64 byteorder : little LC_ALL : None LANG : en_US.UTF-8 LOCALE : en_US.UTF-8

pandas : 1.3.0.dev0+1320.gc8e077fa87.dirty
numpy : 1.20.0
pytz : 2020.5
dateutil : 2.8.1
pip : 20.3.3
setuptools : 49.6.0.post20210108
Cython : 0.29.21
pytest : 6.2.1
hypothesis : 6.0.2
sphinx : 3.4.3
blosc : None
feather : None
xlsxwriter : 1.3.7
lxml.etree : 4.6.2
html5lib : 1.1
pymysql : None
psycopg2 : None
jinja2 : 2.11.2
IPython : 7.19.0
pandas_datareader: None
bs4 : 4.9.3
bottleneck : 1.3.2
fsspec : 0.8.5
fastparquet : 0.5.0
gcsfs : 0.7.1
matplotlib : 3.3.3
numexpr : 2.7.2
odfpy : None
openpyxl : 3.0.6
pandas_gbq : None
pyarrow : 1.0.0
pyxlsb : None
s3fs : 0.4.2
scipy : 1.6.0
sqlalchemy : 1.3.22
tables : 3.6.1
tabulate : 0.8.7
xarray : 0.16.2
xlrd : 2.0.1
xlwt : 1.3.0
numba : 0.52.0
None

[jbrockmendel]

and a pandas array

just to be clear, rhs = pd.array([1, 2, 3]) gives an IntegerArray, not a PandasArray, right?

We should be consistent here, if we want to change iloc for the numpy case I think we need to deprecate?

I think the behavior with ndarray is "more correct" here. Do you disagree?

[phofl]

just to be clear, rhs = pd.array([1, 2, 3]) gives an IntegerArray, not a PandasArray, right?

Sorry, I meant the pandas array constructor. Yes this returns an IntegerArray.

I think the behavior with ndarray is "more correct" here. Do you disagree?

Yeah I think so too

[parkyo]

take

[parkyo]

So it looks like there is no issue. The reason why you got different outputs was because you modified df directly once with pandas array and then applied iloc with a numpy array to the modified dataframe. Try it by declaring two dataframes for numpy array and pandas array like below.

[phofl]

Your test is wrong.
Adding copy to the DataFrame assignment raises as expceted:

df = DataFrame(
    data={
        "col1": [1, 2, 3, 4],
        "col2": [3, 4, 5, 6],
        "col3": [6, 7, 8, 9],
    }
)
df_np = df.copy()
df_pd = df.copy()
np_arr = np.array([1, 2, 3])
pd_arr = pd.array([1, 2, 3])
df_np.iloc[[1, 2, 3]] = np_arr
df_pd.iloc[[1, 2, 3]] = pd_arr
tm.assert_frame_equal(df_np, df_pd)

[phofl]

@parkyo are you still working on this?