PEP 681: Descriptor fields

[debonte]

Improves support for classes with fields that are descriptors. Setting a new dataclass_transform parameter named transform_descriptor_types to True indicates that __init__ parameters corresponding to descriptor fields have the type of the descriptor’s __set__ value parameter rather than the descriptor type.

In the following example, type checkers would understand that the quantity_on_hand parameter of the __init__ method on InventoryItem is of type int | None rather than Descriptor[int].

@dataclass_transform(transform_descriptor_types=True)
def decorator() -> Callable[[Type[T]], Type[T]]: ...
               
class Descriptor(Generic[_T]):
    def __get__(self, instance: object, owner: Any) -> Any: ...
    def __set__(self, instance: object, value: T | None): ...
               
@decorator
class InventoryItem:
    quantity_on_hand: Descriptor[int]

This change was driven by a need in SQLAlchemy as discussed at microsoft/pyright#2958.

A reference implementation is available in Pyright 1.1.222 and Pylance 2022.2.3.

[CAM-Gerlach]

Thanks. A few copyediting fixes and improvements.

[debonte]

@JelleZijlstra there was a Typing SIG email from Carl Meyer today that looks like it will result in some changes to transform_descriptor_types. Should we hold off on completing this PR?

[JelleZijlstra]

Up to you. There's no harm in having more PRs later. I'm happy to merge this now if you would like me to.

[CAM-Gerlach]

LGTM, thanks @debonte . This is good to merge if you are happy with it; my suggestion would be that it is probably best to just merge this now that we've all approved, so they go live now as a discrete commit, and then open a followup PR with additional changes. Then, its easier for both you and us to focus on the new changes, and keeping things in scope.

[debonte]

@JelleZijlstra, can you please merge this? I'll submit another PR to addresss Carl Meyer's feedback as CAM-Gerlach suggested above.