Python dataclass inheritance with mutable default arguments returns zero-value

Question

I am trying to calculate a radius-like quantity from 3 lists containing Cartesian coordinates x, y & z. Below is my minimal code example to reproduce the issue I am facing; The child-class calculates the radius-quantity but returns a zero-value. What is the reason for this, and how can it be fixed?

Script:

# -*- coding: utf-8 -*-

from dataclasses import dataclass, field
from typing import List


@dataclass
class LoadHalo:
    x: List = field(default_factory=list)
    y: List = field(default_factory=list)
    z: List = field(default_factory=list)

    def __post_init__(self):
        self.x = [1, 2, 3]
        self.y = [1, 3, 5]
        self.z = [1, 4, 7]


@dataclass
class BinHalo(LoadHalo):
    r: List = field(default_factory=list)

    def __post_init__(self):
        self.r = self.modulus(self.x, self.y, self.z)

    def modulus(self, *args):
        """Modulus of vector of arbitrary size."""
        return sum([i ** 2 for containers in args for i in containers]) ** .5


halo = BinHalo()
print(f"halo.x: {halo.x}")
print(f"halo.r: {halo.r}")

Which outputs the following values for x and r:

halo.x: []
halo.r: 0.0

add super().__post_init__() before self.r = self.modulus(self.x, self.y, self.z) — eyllanesc
– eyllanesc, Commented Dec 3, 2019 at 9:52

juanpa.arrivillaga · Accepted Answer · 2019-12-03 09:59:21Z

4

Because you are overriding __post_init__, so the lists are still empty because they never got initialized with anything other than the empty list default. You must call the super-class method that you are overriding if you want its behavior as well.

What you want is the following:

def __post_init__(self):
    super().__post_init__()
    self.r = self.modulus(self.x, self.y, self.z)

Note a couple of things with your type-hints: you probably wanted float for the r field in the subclass, and also for the return type of modulus:

def modulus(self, *args) -> float:
    ...

also, you should initialize the default lists with float objects so you can write:

x: List[float]

For your list fields, since presumably you want to use floating point math.

So all-in all, I'd define everything like:

from dataclasses import dataclass, field
from typing import List


@dataclass
class LoadHalo:
    x: List[float] = field(default_factory=list)
    y: List[float] = field(default_factory=list)
    z: List[float] = field(default_factory=list)

    def __post_init__(self) -> None:
        self.x = [1.0, 2.0, 3.0]
        self.y = [1.0, 3.0, 5.0]
        self.z = [1.0, 4.0, 7.0]


@dataclass
class BinHalo(LoadHalo):
    r: float = 0.0 # or whatever is suitable

    def __post_init__(self) -> None:
        super().__post_init__()
        self.r = self.modulus(self.x, self.y, self.z)

    def modulus(self, *args: List[float]) -> float:
        """Modulus of vector of arbitrary size."""
        return sum([i ** 2 for containers in args for i in containers]) ** .5

edited Dec 3, 2019 at 9:59

answered Dec 3, 2019 at 9:52

juanpa.arrivillaga

97.6k14 gold badges141 silver badges190 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

Gustav Rasmussen Over a year ago

Very clear working example. Just what I was looking for. Thank you @juanpa.arrivillaga

juanpa.arrivillaga Over a year ago

@GustavRasmussen no worries, you should consider using mypy to type-check your code.

Collectives™ on Stack Overflow

Python dataclass inheritance with mutable default arguments returns zero-value

1 Answer 1

2 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Related