[nk-batman21]

With pandas 3, a pandas Series that has labels is no longer accepted as a location. That is the usual shape when building a map from a DataFrame:

df = pd.DataFrame({"lat": [45.0, 46.0], "lon": [3.0, 4.0]})
folium.Map(location=df[["lat", "lon"]].mean())
for _, row in df.iterrows():
    folium.Marker(row[["lat", "lon"]])

Both lines raise TypeError: Location should support indexing, like a list or a tuple does, ..., and so do PolyLine or Polygon built from DataFrame rows. validate_location reads location[0] and location[1], which on a Series with a string index only worked through the positional fallback of Series.__getitem__. pandas 2 printed a FutureWarning for that on every call, and pandas 3 removed it, so the lookup now raises KeyError.

This converts a Series to a list before validating it, like numpy arrays and DataFrames already are. test_validate_location already accepts pd.Series([5, 3]); I added the same values with a lat/lon index.

Ran: python -m pytest --ignore=tests/selenium --ignore=tests/playwright --ignore=tests/snapshots --ignore=tests/test_repr.py on Python 3.14 with pandas 3.1.0 and on Python 3.10 with pandas 2.3.3. The new case fails before the change on pandas 3 and passes after it on both, also with -W error::FutureWarning on pandas 2.
Checked: ruff, black and mypy folium.