BUG: union of MultiIndex throws exception for datetime and pd.Timestamp with identical values

[jwg4]

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

from datetime import date
import pandas as pd

mi_a = pd.MultiIndex.from_tuples([(date(2001, 1, 1), "foo")], names=["first", "second"])
mi_b = pd.MultiIndex.from_tuples([(pd.Timestamp(date(2001, 1, 1)), "asdf")], names=["first", "second"])

mi_a.union(mi_b)

Issue Description

The following exception is thrown:

InvalidIndexError                         Traceback (most recent call last)
...

InvalidIndexError: Reindexing only valid with uniquely valued Index objects

Expected Behavior

I would have expected the two values date(2001, 1, 1) and pd.Timestamp(date(2001, 1, 1)) to be treated as different values, which is how I believe pd.DataFrame.drop_duplicates acts. However treating the two values as identical could also be valid, but I don't think that the exception is.

Installed Versions

INSTALLED VERSIONS

commit : c888af6
python : 3.13.5
python-bits : 64
OS : Windows
OS-release : 10
Version : 10.0.19045
machine : AMD64
processor : Intel64 Family 6 Model 158 Stepping 13, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : English_United Kingdom.1252

pandas : 2.3.1
numpy : 2.3.1
pytz : 2025.2
dateutil : 2.9.0.post0
pip : 25.0
Cython : None
sphinx : None
IPython : 8.37.0
adbc-driver-postgresql: None
adbc-driver-sqlite : None
bs4 : 4.13.4
blosc : None
bottleneck : None
dataframe-api-compat : None
fastparquet : None
fsspec : None
html5lib : None
hypothesis : None
gcsfs : None
jinja2 : 3.1.6
lxml.etree : None
matplotlib : None
numba : None
numexpr : None
odfpy : None
openpyxl : None
pandas_gbq : None
psycopg2 : None
pymysql : None
pyarrow : None
pyreadstat : None
pytest : None
python-calamine : None
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : None
tabulate : None
xarray : None
xlrd : None
xlsxwriter : None
zstandard : None
tzdata : 2025.2
qtpy : None
pyqt5 : None

[jwg4]

This apparent bug also affects another basic operation which could be expected to succeed, DataFrame.combine_first when using frames with MultiIndex as described above:

from datetime import date
import pandas as pd

df_a = pd.DataFrame(
    [
        (date(2001, 1, 1), "foo", 11),
    ],
    columns=["a", "b", "c"]
)
df_a = df_a.set_index(["a", "b"])

df_b = pd.DataFrame(
    [
        (pd.Timestamp(date(2001, 1, 1)), "bar", 33),
    ],
    columns=["a", "b", "c"]
)
df_b = df_b.set_index(["a", "b"])

df_a.combine_first(df_b)

[rhshadrach]

Thanks for the report. pandas is converting the Index to a DatetimeIndex here:

pandas/pandas/core/indexes/base.py

Lines 6217 to 6221 in 7c2796d

	elif self.inferred_type == "date" and isinstance(other, ABCDatetimeIndex):
	try:
	return type(other)(self), other
	except OutOfBoundsDatetime:
	return self, other

Further investigations are welcome!

[simonjayhawkins]

Further investigations are welcome!

Probably a regression if not a design choice.

The code sample worked in pandas 1.5 but gave

FutureWarning: Comparison of Timestamp with datetime.date is deprecated in order to match the standard library behavior. In a future version these will be considered non-comparable. Use 'ts == pd.Timestamp(date)' or 'ts.date() == date' instead.