version: Python 3.10.6 (main, Nov 14 2022, 16:10:14) [GCC 11.3.0]
Setup:
class _Mixin:
def vars_to_dct(self, results=[]):
dct = self.to_dict()
for k in dct.keys():
results.append(k)
return results
@dataclass
class ClassA(_Mixin):
a: str = ""
b: str = ""
@dataclass
class ClassB(_Mixin):
c: str = ""
d: str = ""
e: str = ""
Instantiate ClassA and method by itself. This is fine:
a = ClassA()
assert ["a", "b"] == a.vars_to_dct()
Instantiate ClassB and method by itself. This is fine:
b = ClassB()
assert ["c", "d", "e"] == b.vars_to_dct()
Instantiate ClassA and ClassB together. This is not fine:
AssertionError
a = ClassA()
assert ["a", "b"] == a.vars_to_dct()
b = ClassB()
assert ["c", "d", "e"] == b.vars_to_dct()
What actually happens:
b = ClassB()
assert ["a", "b", "c", "d", "e"] == b.vars_to_dct()
Explicitly redeclaring the argument also works fine:
b = ClassB()
assert ["c", "d", "e"] == b.vars_to_dct(results=[])
Note: I have tried to apply the answers here to this behaviour:
- python why are the object properties leaking into one another
- How to avoid having class data shared among instances?
But the difference and my question is: I am explicitly declaring the argument to the method.
Why is the argument results=[] on the method not recognised when it's run again?
def vars_to_dct(self, results=[]):
Can someone confirm for me that it really is official Python behaviour to have variables declared in methods to be "static members"/"class variables"?
