Why such structure
class A: def __init__(self, a): self.a = a def p(self, b=self.a): print b
gives an error
NameError: name 'self' is not defined?
Default argument values are evaluated at function define-time, but
self is an argument only available at function call time. Thus arguments in the argument list cannot refer each other.
It’s a common pattern to default an argument to
None and add a test for that in code:
def p(self, b=None): if b is None: b = self.a print b
Update 2022: Python developers are now considering late-bound argument defaults for future Python versions.
For cases where you also wish to have the option of setting ‘b’ to None:
def p(self, **kwargs): b = kwargs.get('b', self.a) print b
NameError can also occur if you fail to define self inside a method signature. This error typically will appear as
TypeError, as there will be a mismatch between expected and given arguments. However, if you accept a variable number of arguments,
self will be
arg, and the variable
self will be undefined.
A minimal example.
class Obj: def foo(*args): print(self.bar)
>NameError: name ‘self’ is not defined
class Obj: def baz(self, *args): print(self.bar)