Bug: super() call in @dataclass(slots=True) causes TypeError

[Gumminess]

When running this example you will get TypeError: super(type, obj): obj must be an instance or subtype of type

from dataclasses import dataclass
@dataclass(slots=True)
class Base:
    def method(self):
        print("hello")

@dataclass(slots=True)
class Derived(Base):
    def method(self):
        super().method()
        print("world")

Derived().method()

The solution is to call super(Derived, self) instead of super(), as explained here

The problem is that pyupgrade simplifies this to just plain super(), which is ok in most cases, but not here. As this is a standard behavior of a standard library, there should be a special case for this in pyupgrade.

[asottile]

there's a few duplicates. this isn't a pyupgrade problem but a problem with dataclasses

[bhperry]

This is a pyupgrade problem. Dataclass with slots creates a new class, because it's required for slots to work. Pyupgrade breaks the required format for datalass super with slots.

[asottile]

bet. python/cpython#90562