Allowable subclasses for a function that accepts base class with a custom metaclass

[BaconPancakes]

I'm using a library that uses Pydantic.

Pydantic's BaseModel has a custom metaclass. The library has a function that takes BaseModel as an argument. When I subclass BaseModel into MySubclass, this subclass is not assignable into that function that accepts BaseModel.

Is this intended behavior?

Example that can demonstrate this without Pydantic at all:

Code sample in pyright playground

from abc import ABCMeta


class MyMeta(ABCMeta):
    ...

class MyBaseCls(metaclass=MyMeta):
    ...

class MySubclass(MyBaseCls):
    ...

def my_fun(some_cls: MyBaseCls):
    ...

my_fun(MySubclass)
# Argument of type "type[MySubclass]" cannot be assigned to parameter "some_cls" of type "MyBaseCls" in function "my_fun"
#   "type[MyMeta]" is not assignable to "type[MyBaseCls]"

[erictraut]

The annotation you're using for the some_cls parameter indicates that the corresponding argument value must be an instance of MyBaseCls. I think what you intend here is that the value must be a class object that is a subclass of MyBaseCls. If my assumption is correct, then you should change the annotation to type[MyBaseCls]. This eliminates the error.

Alternatively, if you want to accept any instance of the MyMeta metaclass, you could change the parameter's annotation to MyMeta.

[BaconPancakes]

Thanks - I missed that and you're absolutely correct. I wrongly assumed the library I am calling is type annotated correctly.