[Zope3-dev] Re: Interface declaration API

[Guido van Rossum]

> Guido van Rossum <guido@python.org> wrote:
> 
> >Um, forget that mail.  I had decided not to send it, and hit the wrong
> >button. :-(
> 
> Good.  Because the example you gave would certainly surprise me
> if that's the way it worked.
> 
> --RDM

Really?  Yet I believe this was the intent of Jim's mutable interface
specifications.  Here's my example again:

  x = implementedBy(Foo)
  print IBar in x		# False
  addImplements(Foo, IBar)
  print IBar in x		# True

x is the interface spec for class Foo.

At the first print statement, Foo doesn't implement IBar, so x doesn't
contain IBar.  At the second print statement, Foo does implement IBar,
so x (which still is the interface specification object that describes
all Foo's interfaces) does contain IBar.  At least that's how I
understand Jim's proposal of a few days ago -- it may have evolved
since then, I'm behind in catching up on that discussion (which in
part was taken off-line to a smaller audience).

--Guido van Rossum (home page: http://www.python.org/~guido/)