`__setitem__` should accept sequences containing None?

[anentropic]

Describe the bug
I have code a bit like:

In [1]: import pandas as pd
   ...:
   ...: df = pd.DataFrame({"flag": [True, False], "count": [1, 0]})
   ...:
   ...: df["result"] = [
   ...:     1 if flag and count > 0 else None
   ...:     for flag, count in zip(df["flag"], df["count"], strict=True)
   ...: ]

In [2]: df['result']
Out[2]:
0    1.0
1    NaN
Name: result, dtype: float64

The code works as intended.

Pylance complains:

No overloads for "__setitem__" match the provided arguments Pylance reportCallIssue
frame.pyi(849, 9): Overload 2 is the closest match
Argument of type "list[Any | None]" cannot be assigned to parameter "value" of type "Scalar | ArrayLike | NAType | NaTType | IndexOpsMixin[Any, Any] | Sequence[Scalar] | Sequence[Sequence[Scalar]] | DataFrame | Mapping[Hashable, Scalar | NAType | NaTType] | None" in function "__setitem__"
Type "list[Any | None]" is not assignable to type "Scalar | ArrayLike | NAType | NaTType | IndexOpsMixin[Any, Any] | Sequence[Scalar] | Sequence[Sequence[Scalar]] | DataFrame | Mapping[Hashable, Scalar | NAType | NaTType] | None"
"list[Any | None]" is not assignable to "str"
"list[Any | None]" is not assignable to "bytes"
"list[Any | None]" is not assignable to "date"
"list[Any | None]" is not assignable to "datetime"
"list[Any | None]" is not assignable to "timedelta"
"list[Any | None]" is not assignable to "datetime64[date | int | None]"
"list[Any | None]" is not assignable to "timedelta64[timedelta | int | None]"

Looking at the types in frame.pyi we have something like:

    @overload
    def __setitem__(
        self,
        idx: (
            int
            | IndexType
            | tuple[int, int]
            | tuple[IndexType, int]
            | tuple[IndexType, IndexType]
            | tuple[int, IndexType]
        ),
        value: (
            Scalar
            | IndexOpsMixin
            | Sequence[Scalar]
            | DataFrame
            | np_ndarray
            | NAType
            | NaTType
            | Mapping[Hashable, Scalar | NAType | NaTType]
            | None
        ),
    ) -> None: ...

I'm guessing if we have Mapping[Hashable, Scalar | NAType | NaTType] then Sequence[Hashable, Scalar | NAType | NaTType] should also allowed?

But it seems like both of these should also accept None?

Or None should be part of Scalar ?

Certainly Sequence[Scalar] seems too restrictive.

I had a bit of a look but I have no idea if/where such behaviour is documented in pandas. The closest I could find is https://pandas.pydata.org/docs/user_guide/10min.html#setting which is far from a rigorous description of what types are accepted on the RHS of assignment.

To Reproduce

1. see above
2. using pyright
3. see above

Please complete the following information:

• OS: macOS
• OS Version 14.6.1
• python version: 3.11
• version of type checker Pyright 1.1.407
• version of installed pandas-stubs: 2.3.3.251201

[cmp0xff]

Will resolve the partially-overlapping #1529 first and then this one

[Dr-Irv]

I think if we just change DataFrame.__setitem__() to accept Sequence[Scalar | NAType | NaTType | None] as value instead of Sequence[Scalar] in each of the 4 overloads, we should be OK on this one.