[Raymond Hettinger]
> >class Cache(dict):
> > def __init__(self, n, *args, **kwds):
> > self.n = n
> > self.queue = collections.deque()
> > dict.__init__(self, *args, **kwds)
[Bengt Richter]
> Minor comment: There is a potential name collision problem for keyword
> n=something,
> so what is considered best practice to avoid that? __n or such as the n arg?
One solution is to drop the *args and **kwds part of the __init__() API
for the subclass. For a cache class, it doesn't make sense that you
would know all of the values when the instance is first created. If
the need arises, the update() method will suffice:
class Cache(dict):
def __init__(self, n):
self.n = n
dict.__init__(self)
. . .
The __n form doesn't get name mangled so its only advantage is in being
a less likely key.
Raymond
--
http://mail.python.org/mailman/listinfo/python-list
