Permanent objects?

[Cliff Wells]

On Wed, 2002-12-25 at 13:46, Erik Max Francis wrote:
> Kevin Altis wrote:
> 
> > Hmm, I think I tend to use
> > 
> >   if x == None:
> > and
> >   if x != None:
> > 
> > rather than
> > 
> >   if x is None:
> > and
> >   if not x is None:
> > 
> > I don't suppose it really matters or that one is better or clearer
> > than the
> > other?
> 
> If everybody plays together well, it shouldn't make a difference, but
> someone perverse could create a custom instance that overrides either
> __eq__ or __cmp__ so that tests equal to None.  I can't think of a case
> when it would appropriate for someone to do this, but someone _could_.

But then isn't this sort of like the case where a function tests the
type of object being passed to it (i.e. type(s) == type(""))?  There
have been arguments on this list in the past that it's bad to explicitly
test for a particular type, so wouldn't this fall in the same category? 
That is, if an object mimics a particular type, and can be used in place
of that type, why should a function exclude it if it isn't that type? 
For instance, if a function expects a file, but is passed a file-like
object, it shouldn't reject that object just because it isn't a real
file, doing so only limits the usefulness of the function.  It seems
that by that argument, x == None would be preferable to x is None.

I'd be interested in hearing thoughts on this.

-- 
Cliff Wells <clifford.wells at attbi.com>