Neil Cerutti <[EMAIL PROTECTED]> wrote:
> In C++ they are used most often for factory functions, since they
> conveniently have access to the class's private members, and
> don't want or need an existing instance. Python seems to have
> adopted this use-case (ConfigParser, for example), but without a
> need for it (code organization?).
What staticmethod does ConfigParser have? Can't recall one offhand.
I think Python more commonly uses classmethod, rather than staticmethod,
for factories (i.e. a la Smalltalk, not a la C++). In that case the
advantage wrt a function _is_ there: when you subclass, you can get
instances of the new class, rather than the base class, from the
factory:
>>> class zap(dict): pass
...
>>> z = zap.fromkeys(range(4))
>>> z
{0: None, 1: None, 2: None, 3: None}
>>> type(z)
<class '__main__.zap'>
>>>
If dict_fromkeys was a plain function (or if fromkeys was a staticmethod
rather than a classmethod of dict) then z would be an instance of dict,
rather than one of zap (unless you also specifically passed the desired
class, kind of cumbersome).
Alex
--
http://mail.python.org/mailman/listinfo/python-list
