[Python-Dev] Problem with super() usage

[Guido van Rossum]

On 7/16/06, Greg Ewing <greg.ewing at canterbury.ac.nz> wrote:
> For about the third time in my life, I thought I might
> have found a use for cooperative super calls, but I've
> run into another problem with the concept.
>
> Consider:
>
>   class A(object):
>    def m(self):
>      print "A.m"
>
>   class B(object):
>    def m(self):
>     print "B.m"
>     super(B, self).m()
>
>   class C(B, A):
>    def m(self):
>     print "C.m"
>     super(C, self).m()
>
>   >>> c = C()
>   >>> c.m()
>   C.m
>   B.m
>   A.m
>
> Okay so far, but... what if I want to use class B on
> its own, or in combination with other classes that don't
> have an m() method?
>
>  >>> b = B()
>  >>> b.m()
> B.m
> Traceback (most recent call last):
>    File "<stdin>", line 1, in ?
>    File "<stdin>", line 4, in m
> AttributeError: 'super' object has no attribute 'm'
>
> In general, how is one supposed to use super() in a
> class which may or may not be at the end of the mro
> with respect to a particular method?

One isn't. In the world where cooperative multiple inheritance
originated (C++), this would be a static error. You can only use super
when you're overriding a method, not when you're defining a new
method.

To do what you want to do in such a world, you'd have to create a base
class with a dummy implementation and inherit from that. You can do
that in Python, too.

> The only thing I can think of is to write all my
> super()-using methods defensively like this:
>
>    def m(self):
>      ...
>      s = super(C, self)
>      if hasattr(s, 'm'):
>        s.m()
>
> which seems awfully tedious.

Definitely don't that.

-- 
--Guido van Rossum (home page: http://www.python.org/~guido/)