Type subclassing: bug or feature

[Bjorn Pettersen]

> From: Aahz [mailto:aahz at pythoncraft.com] 
> 
> Consider the following code:
> 
> class MyStr(str):
>     def contains(self, value):
>         return self.find(value) >= 0
> 
> s = MyStr("hello, world!")
> s = s.capitalize()
> if s.contains('Hello'):
>     print "Found it!"
> 
> It fails with an AttributeError when it calls s.contains(), because
> s.capitalize() returned a str instead of a MyStr.  Anyone 
> want to take a whack at defending this as the correct behavior?

Well, in e.g. C++ this would make perfect sense, since a base class
method can't know the real type of self/this. In Python we can do better
of course:

>>> class A:
...   def foo(self):
...     return self.__class__()
...
>>> class B(A):
...   pass
...
>>> b = B()
>>> b.foo()
<__main__.B instance at 0x00864CC0>

which version is correct is mostly a religious question, but personally
I think that if we _can_ do better than C++ we should <wink>.

File a bug and see what Guido says...

-- bjorn