class A: def __init__(self, n=[0]): self.data = n a = A() print a.data[0] #print 0 a.data[0] +=1 b = A() print a.data[0] #print 1, desired output is 0 In the case above, is there any way to provide a default argument with the mutable object (such as list or class) in __init__() class A, but b is not affected by the operation a?
You could try this:
class A: def __init__(self, n=None): if n is None: n = [0] self.data = n Which avoids the biggest problem you're facing here, that is, that's the same list for every single object of your type "A."
One possibility is:
class A: def __init__(self, n=None): if n is None: n = [0] self.data = n Also:
class A: def __init__(self, n=[0]): print id(n) self.data = n[:] print id(self.data) del n a = A() print a.data[0] #prints 0 a.data[0] +=1 print a.data[0] #prints 1 print b = A() print b.data[0] #prints desired output 0 The principle is that it creates another list. If a long list is passed as argument, there will be two long list in memory. So the inconvenience is that it creates another list.... That's why I delete n.
Don't think it's better, but it may give you comprehension of what happens
