How does Python 3 super() work in this particular multiple inheritance situation? [duplicate]

Question

I have the following code:

class A:
    def __init__(self):
        print("A.__init__")

class B(A):
    def __init__(self):
        print("B.__init__")
        super().__init__()

class C(A):
    def __init__(self):
        print("C.__init__")
        super().__init__()


class D(B,C):
    def __init__(self):
        print("D.__init__")
        super().__init__()

When I instantiate an object out of class D,

d = D()

I got:

>>> d = D()
D.__init__
B.__init__
C.__init__
A.__init__

Why did class A's __init__() only got called once, and at the very end? Note both class B and class C called class A's __init__(), so I expected that class A's __init__() should be called twice.

I read both Python 3's documentation and Mr. Hettinger's blog, but didn't figure out the underlying mechanism. So how does the super() determine that it only needs to execute once? Thanks.

Answer 1

Python has a mechanism called Method Resolution Order (MRO) to find proper methods in class hierarchies. There are two rules in MRO:

1- dept-first left-to-right search

2- only the last occurrence of a class is considered

Now back to your example:

I. The search starts at D by adding it to the MRO list [D].

II. Class D is inherited from B and C. B is the left most parent and then comes C. So the MRO list becomes [D, B, C].

III. Then the search continues to a deeper level with is parent of B and A is added to the list. Now we have [D, B, C, A]. The search continues to parents of C with is again A. Since A is already in the list, nothing changes.

IV. The search goes to a deeper level (parents of A) and because A does not have any explicit parent, the search ends at this point.

So the final list is [D, B, C, A].

super() method uses this list to do its job and thats why A is called at the end