BUG: regresssion in 3.0rc: cannot save df with MultiIndex to HDF

[gdementen]

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

>>> import pandas as pd
>>> pd.__version__
'3.0.0rc0'
>>> index = pd.MultiIndex.from_tuples([('a', 'x'), ('b', 'y')], names=['level1', 'level2'])
>>> df = pd.DataFrame({'value': [1, 2]}, index=index)
>>> index.dtypes
level1    str
level2    str
dtype: object
>>> df.to_hdf('test.h5', key='test')
NotImplementedError: Saving a MultiIndex with an extension dtype is not supported.

While on Pandas 2.3.3, it worked:

>>> import pandas as pd
>>> pd.__version__
'2.3.3'
>>> index = pd.MultiIndex.from_tuples([('a', 'x'), ('b', 'y')], names=['level1', 'level2'])
>>> df = pd.DataFrame({'value': [1, 2]}, index=index)
>>> index.dtypes
level1    object
level2    object
dtype: object
>>> df.to_hdf('test.h5', key='test')
>>> # no error

Issue Description

Saving a dataframe with a MultiIndex to HDF does not work anymore

Expected Behavior

Saving is possible 😉

Installed Versions

Details

INSTALLED VERSIONS

commit : 1a3230d
python : 3.13.7
python-bits : 64
OS : Windows
OS-release : 11
Version : 10.0.26200
machine : AMD64
processor : Intel64 Family 6 Model 154 Stepping 4, GenuineIntel
byteorder : little
LC_ALL : None
LANG : None
LOCALE : English_Belgium.1252

pandas : 3.0.0rc0
numpy : 2.3.5
dateutil : 2.9.0.post0
pip : 25.3
Cython : None
sphinx : None
IPython : 9.7.0
adbc-driver-postgresql: None
adbc-driver-sqlite : None
bs4 : None
bottleneck : 1.4.2
fastparquet : None
fsspec : None
html5lib : None
hypothesis : None
gcsfs : None
jinja2 : None
lxml.etree : None
matplotlib : 3.10.6
numba : None
numexpr : 2.14.1
odfpy : None
openpyxl : 3.1.5
psycopg2 : None
pymysql : None
pyarrow : 21.0.0
pyiceberg : None
pyreadstat : None
pytest : 8.4.2
python-calamine : None
pytz : 2025.2
pyxlsb : None
s3fs : None
scipy : None
sqlalchemy : None
tables : 3.10.2
tabulate : None
xarray : None
xlrd : None
xlsxwriter : None
zstandard : None
qtpy : 2.4.3
pyqt5 : None

[gdementen]

FWIW, I used this little function as a workaround within my saving code. I know it is suboptimal but at least it works 😉.

def ensure_non_string_index(index):
    if (isinstance(index, pd.MultiIndex) and
            any(isinstance(dt, pd.StringDtype) for dt in index.dtypes)):
        obj_levels = [pd.Index(level_labels, dtype=object)
                      for level_labels in index.levels]
        index = pd.MultiIndex(obj_levels, index.codes, names=index.names)
    return index

I use it like this (because my function creates the dataframe itself, so I can safely replace the indexes inplace)

df.index = ensure_non_string_index(df.index)
df.columns = ensure_non_string_index(df.columns)

[jorisvandenbossche]

@gdementen thanks for the report! Yes, we certainly should ensure this keeps working, at least specifically for strings

[sanchit122006]

I want to ensure that anyone working on this ?
Happy to contribute.

[gdementen]

Wow, that was fast. Thanks !