Hooks, aspect-oriented programming, and design by contract

[gbreed at cix.compulink.co.uk]

Alex Martelli wrote:

> That's not compatible with changing the *metaclass*.  The same, it
> appears to me, for the rest of your advice.  Maybe there is a way to
> switch metaclasses dynamically and retroactively, but, if so, I still
> don't know about it.

I think my brain's about to explode.  A metaclass is a class that produces 
classes, right?  How would an object know what its metaclass is?  Surely 
it must be possible to produce the same effect as of a different 
metaclases being used.  Can you explain in words of one syllable what that 
would mean?


               Graham