TYP: misc return types #57430
TYP: misc return types #57430
Conversation
| @@ -11815,19 +11899,19 @@ def kurt( | |||
| product = prod | |||
|
|
|||
| @doc(make_doc("cummin", ndim=2)) | |||
| def cummin(self, axis: Axis | None = None, skipna: bool = True, *args, **kwargs): | |||
| def cummin(self, axis: Axis = 0, skipna: bool = True, *args, **kwargs) -> Self: | |||
There was a problem hiding this comment.
None is not documented and 0 seems to behave the same?!
| @property | ||
| def _constructor(self) -> Callable[..., Self]: | ||
| def _constructor(self) -> Callable[..., Self]: # type: ignore[override] |
There was a problem hiding this comment.
Should technically be type[Self] but the NDFrame constructor allows far fewer arguments than the Series/DataFrame constructor, which leads to many errors when _constructor()(...) is called inside NDFrame, which are at runtime only reached by sub-classes.
| @@ -118,6 +118,7 @@ | |||
| Concatenate: Any = None | |||
|
|
|||
| HashableT = TypeVar("HashableT", bound=Hashable) | |||
| HashableT2 = TypeVar("HashableT2", bound=Hashable) | |||
There was a problem hiding this comment.
Curious why we need a second one?
There was a problem hiding this comment.
needed for _from_nested_dict. When using the same TypeVar in a function signature, the type checkers bind these to the same type:
def (x: HashableT) -> HashableT: ...
x(None) # return None
x((1, 2, 3)) # return tuple[int, int, int]When we have a function that tries to bind two different sets of types, we need a new TypeVar
def (x: HashableT, y: HashableT2) -> dict[HashableT, dict[HashableT2, Any]]: ...
(pandas-stubs uses the name HashableT2)
|
Thanks @twoertwein |
No description provided.