Re: Opaque syntax objects
Andre van Tonder 13 Aug 2005 14:53 UTC
Mike Sperber wrote:
> You mean in a global hash table? That's a hack around the lack of a
> field in the syntax objects.
But necessary if you want the generic car/cdr to be efficient, which is
a more important constraint. With this representation, you can use the
primitive car/cdr unmodified on syntax objects.
> To make it work efficiently, you'd have
> to bring weakness in---a lot of machinery to duplicate functionality
> that would trivially work if syntax objects were abstract and thus
> extensible.
I would differ on "trivially" ;-) Reading through the psyntax implementation,
I think complexity would be pretty much conserved no matter how you do it.
Cheers
Andre