`Index.name` and `rename_axis` have inconsistent types

[Dr-Irv]

Describe the bug

If you grab the name of an Index as Index.name, you should be able to use that value in rename_axis

To Reproduce

import pandas as pd

def indexchg(df: pd.DataFrame) -> pd.DataFrame:
    idx_name = df.index.name
    result = df.set_index("foo")
    result = result.reset_index().rename_axis(idx_name)
    return result

error from pyright

indexasym.py
  c:\Code\pandas-stubs\play\indexasym.py:6:38 - error: Argument of type "Hashable | None" cannot be assigned to parameter "mapper" of type "Scalar | ListLike | None" in function "rename_axis"
    Type "Hashable | None" is not assignable to type "Scalar | ListLike | None"
      Type "Hashable" is not assignable to type "Scalar | ListLike | None"
        "Hashable" is not assignable to "str"
        "Hashable" is not assignable to "bytes"
        "Hashable" is not assignable to "date"
        "Hashable" is not assignable to "datetime"
        "Hashable" is not assignable to "timedelta"
        "Hashable" is not assignable to "datetime64[Any]"
    ... (reportArgumentType)

System specifications

• OS and its version: Windows 11
• python version: 3.12
• type checker and its version: pyright 1.1.411
• version of installed pandas-stubs: version on main commit 2b7f9f41f103cfec140888f4dae19724c83ea580

Additional context

I'm not sure if the right way to handle this is to make Index.name return Scalar | None or to modify rename_axis() to accept Hashable . Open to either.

[cmp0xff]

If you grab the name of an Index as Index.name, you should be able to use that value in rename_axis

This may not be true. Consider the following:

import pandas as pd


def indexchg(df: pd.DataFrame) -> pd.DataFrame:
    idx_name = df.index.name
    result = df.reset_index().rename_axis(idx_name)  # crashes at runtime with the input below
    return result


df = pd.DataFrame([1], index=pd.Index([1], name=(1, 2, 3)))  # fine
indexchg(df)  # crashes

[Dr-Irv]

This may not be true. Consider the following:

Sounds like we should report this as a bug in pandas, right?

[cmp0xff]

Sounds like we should report this as a bug in pandas, right?

We could. I feel like pd.Index([1], name=(1, 2, 3)) isn't a good practice.

[Dr-Irv]

Sounds like we should report this as a bug in pandas, right?

We could. I feel like pd.Index([1], name=(1, 2, 3)) isn't a good practice.

I agree. But I think there are API calls that do create names for an index that are tuples, such that the index values are tuples as well. I know I've seen it...

[loicdiridollou]

Reported at pandas-dev/pandas#66656
Indeed a weird configuration to allow something at creation but not at the renaming step.