super() usage in multiple inheritance in python

super() will find the next method in the MRO sequence. This means that only one of the __init__ methods in your base classes is going to be called.

You can inspect the MRO (the Method Resolution Order) by looking at the __mro__ attribute of a class:

>>> D.__mro__
(<class '__main__.D'>, <class '__main__.B'>, <class '__main__.C'>, <class 'object'>)

so from D, the next class is B, followed by C and object. From D.__init__(), the super().__init__() expression will only call B.__init__(), and then because C.__init__() is never called, self.c is not set either.

You'll have to add more super() calls to your class implementations; it is safe to call object.__init__() with no arguments, so just use them everywhere here:

class B():
    def __init__(self):
        print("__init__ of B called")
        super().__init__()
        self.b = "B"

class C():
    def __init__(self):
        print("__init__ of C called")
        super().__init__()
        self.c = "C"

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

    def output(self):
        print(self.b, self.c)

Now B.__init__ will invoke C.__init__, and C.__init__ will call object.__init__, and calling D().output() works:

>>> d = D()
__init__ of D called
__init__ of B called
__init__ of C called
>>> d.output()
B C