Skip to content

DOC: Inconsistent return type documentation for functions with "inplace" option #35409

Closed
@SycamoreLeaf

Description

@SycamoreLeaf

Location of the documentation

This proposal concerns all methods with the inplace keyword argument. Three examples:

1 (Return type described as "blah-blah-type or None if inplace=True) https://pandas.pydata.org/docs/dev/reference/api/pandas.CategoricalIndex.add_categories.html?highlight=inplace

2 (Return type specified, but doesn't mention None) https://pandas.pydata.org/docs/dev/reference/api/pandas.Series.interpolate.html?highlight=inplace

Documentation problem

Many methods in pandas have keyword argument inplace. When inplace=True is supplied, the method updates self and returns None. When inplace=False is supplied, the method does not affect self and returns a changed copy of self.

Some of these functions' return types are documented as "Returns blah-type if inplace=True". Others just say "Returns blah-type". This is confusing because it suggests inconsistencies in behavior that are not really there.

Suggested fix for documentation

Find all methods with the inplace keyword argument. Change the "returns" section of their docstrings to this form:

        Returns
        -------
        Series or DataFrame
            Returns the same object type as the caller, interpolated at
            some or all ``NaN`` values. (Or return `None` if
            `inplace=True`.)

Also change all the formal return type signatures to the form "--> Optional[Foo]" if they aren't all already like that.

Metadata

Metadata

Labels

Type

No type

Projects

No projects

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions