[Python] Python intellisense / type sensing seems broken on function overload

I have the following overloaded function:

# **************************************************************************************


@overload
def get_mount(
name: Literal["SimulatedAltAzMountDeviceInterface"],
params: SimulatedAltAzMountDeviceParameters,
**extras: Any,
) -> SimulatedAltAzMountDeviceInterface: ...
@overload
def get_mount(
name: Literal["BaseMountDeviceInterface"],
params: BaseMountDeviceParameters,
**extras: Any,
) -> BaseMountDeviceInterface: ...
def get_mount(
name,
params,
**extras: Any,
):
"""
Retrieve a mount factory by name from the MOUNT_FACTORY_REGISTRY.

Raises:
UnknownDeviceError: If the mount name is not found in the registry.
"""
if name == "SimulatedAltAzMountDeviceInterface":
return SimulatedAltAzMountDeviceInterface(params=params, **extras)

valid = ", ".join(sorted(MOUNT_FACTORY_REGISTRY))

raise UnknownDeviceError(
f"Mount '{name}' not found in registry. Valid choices: {valid}"
)


# **************************************************************************************


Yet when I call this as:

mount = get_mount(
name="SimulatedAltAzMountDeviceInterface",
params={
...
}
)

mount.connect()


The typings in VS Code come back as:



Somehow, even though the typesense seems to be working somewhat by correctly typing the params argument, the return type seems to fall back. What I have done wrong here ?

Reproduction: Pyright playground

Continue reading...