simple metaclass question

[Michael Hudson]

Carl Banks <imbosol at vt.edu> writes:

> Alex Martelli wrote:
> > Carl Banks wrote:
> >   ...
> >> 2. Because M is not a subclass of type, instances of M are not types.
> >>    Therefore, A is not a type.  (I'm not sure of this one, though.  I
> >>    think in Python an object can be a type only if it is a subclass of
> >>    type; I'm not sure if this is true in other languages with
> > 
> > I'm not exactly sure of what it is that you believe, but, for the record: 
> > objects CAN be types of other objects, in Python, without subclassing the 
> > built-in `type` -- for example:
> 
> Thanks.  I was thinking at the C level, where objects that serve as
> types have to have a certain fixed structure.  To create a metatype in
> Python (not C), it would have to be an instance of a class that
> subtypes type, right?

I don't know whether type(type(ob)) is always a subtype of type -- I
think so -- but certainly the thing you write 

class Bob:
    __metaclass__ = ...<---- here

doesn't have to be a subtype of type.  It can be handy to have a
function here, for instance -- but if this function doesn't return a
subtype of type, inheriting from the resulting class gets totally
bewildering (or at least, I managed to totally bewilder myself along
these lines).

> Gawd, it takes a lot to spin my head in circles, but this metaclass
> stuff is doing it.

Yeah, with metaclasses who needs drugs?  Or merry-go-rounds?

plenty-more-where-this-confusion-came-from-ly y'rs
M.

-- 
  > Touche! But this confirms you are composed of logic gates.
  Crud, I thought those were neurons in there.
                    -- Thomas F. Burdick, Kenny Tilton, comp.lang.lisp