[Python-3000] let's get rid of unbound super methods

[Guido van Rossum]

Thanks for proposing this -- I've been scratching my head wondering
what the use of unbound super() would be. :-) I'm fine with killing it
-- perhaps someone can do a bit of research to try and find out if
there are any real-life uses (apart from various auto-super clones)?

--Guido

On 9/1/07, Michele Simionato <michele.simionato at gmail.com> wrote:
> So Python 3000a1 is out! Kudos to everybody involved!
> You did an incredible amount of work in a relatively short time! :-)
>
> Having said that, let me go to the point. This morning I downloaded
> the tarball and compiled everything without issues, then I
> started playing around. One of the first thing I looked at was the new
> super, since it is a matter that made me scratch my head a lot in the
> past. Basically I am happy with the implementation, especially about
> the new magic name __class__ inside the methods which is something I
> always wanted. So I am not here to ask for new features. I am actually
> here to ask for less features: specifically, I would like the unbound
> syntax for super to be removed. I am talking about this:
>
> >>> help(super)
> Help on class super in module __builtin__:
>
> class super(object)
>  |  super() -> same as super(__class__, <first argument>)
>  |  super(type) -> unbound super object
>  |  super(type, obj) -> bound super object; requires isinstance(obj, type)
>  |  super(type, type2) -> bound super object; requires issubclass(type2, type)
>
>
> The single argument syntax 'super(type)' is what I call the unbound syntax.
> I would like 'super(type)' to be removed from the valid signatures.
> AFAIK, the only use case for it was the implementation of the autosuper
> recipe in Guido's new style classes essay. That use case has disappeared
> nowadays, and I cannot think of other situations where may want to use
> that feature (you may think differently, if so, please speak).
> The other reason why I would like it to be removed (apart from the fact
> that it looks unneeded to me) is that is very difficult to explain to
> beginners. For instance in the past I lectured on Python, and in order
> to explain why unbound super objects can be useful I gave this example,
> which is basically Guido's autosuper recipe implemented by hand:
>
> class B(object):
>   def __repr__(self):
>       return '<instance of %s>' % self.__class__.__name__
>   #@classmethod
>   def cmeth(self):
>       print("B.meth called from %s" % self)
>
> class C(B):
>   #@classmethod
>   def cmeth(self):
>       print("C.meth called from %s" % self)
>       self.__super.cmeth()
>
> C._C__super = super(C)
>
> c = C()
>
> c.cmeth()
>
> Here everything works because the unbound super object is a descriptor
> and self.__super calls super(C).__get__(self, C) which corresponds
> to the bound method super(C, self) which is able to dispatch to .cmeth.
> However, if you uncomment the classmethod decorator, self.__super (where
> self is now the class C) will just return the unbound super object super(C)
> which is unable to dispatch to .cmeth. Now, try to explain that to a beginner!
> We can leave just as well without unbound super methods, so let's take
> the occasion of Python3k to remove this glitch.
>
>
>                            Michele Simionato
> _______________________________________________
> Python-3000 mailing list
> Python-3000 at python.org
> http://mail.python.org/mailman/listinfo/python-3000
> Unsubscribe: http://mail.python.org/mailman/options/python-3000/guido%40python.org
>


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