Repository navigation
Derived dataclass with slots generates exception calling super().__init__() #137237
Closed as duplicate
Description
Bug report
Bug description:
The following code generates an exception where using a derived dataclass with slots. The expected behavior is for no exception to be raised for Derived3. The exception is:
Exception: super(type, obj): obj (instance of Derived3) is not an instance or subtype of type (Derived3).
from dataclasses import dataclass
# No dataclass
class Base1:
b11: str
def __init__(self, b11: str) -> None:
self.b11 = b11
class Derived1(Base1):
d11: str
def __init__(self, b11: str, d11: str) -> None:
super().__init__(b11)
self.d11 = d11
@dataclass(slots=True)
class Base2:
b21: str
def __init__(self, b21: str) -> None:
self.b21 = b21
# Derived dataclass without slots
@dataclass(slots=False)
class Derived2(Base2):
d21: str
def __init__(self, b21: str, d21: str) -> None:
super().__init__(b21)
self.d21 = d21
# Derived dataclass with slots
@dataclass(slots=True)
class Derived3(Base2):
d31: str
def __init__(self, b31: str, d31: str) -> None:
super().__init__(b31)
self.d31 = d31
# Code
s1 = 's1'
s2 = 's2'
D1 = Derived1(s1, s2)
print(f'D1: {str(D1)}')
D2 = Derived2(s1, s2)
print(f'D2: {str(D2)}')
try:
D3 = Derived3(s1, s2)
print(f'D3: {str(D3)}')
except Exception as e:
print(f'Exception: {str(e)}')The output I get is:
> python polymorphism.py
D1: <__main__.Derived1 object at 0x000001921C709160>
D2: Derived2(b21='s1', d21='s2')
Exception: super(type, obj): obj (instance of Derived3) is not an instance or subtype of type (Derived3).
PS: I am not sure why I get different string for the output of D1 and for D2, but I assume this is existing behavior (though weird).
CPython versions tested on:
3.13
Operating systems tested on:
Windows
Activity
StanFromIreland commentedon Jul 30, 2025
D2is a dataclass, which implements a nice__repr__for you.I am unable to reproduce on main, I presume this has been fixed.
tycarac commentedon Jul 30, 2025
Thanks for the quick response. I was testing with Python version 3.13.5
StanFromIreland commentedon Jul 30, 2025
Yes, this is now fixed and will be available in the next release.