Re: "Current" solution for generalized set!

[Maciej Stachowiak <mstachow@mit.edu>]

mdj@nada.kth.se writes:
> Maciej Stachowiak <mstachow@MIT.EDU> writes:
> 
> 
> > > (In any case, note that make-procedure-with-setter be a primitive
> > >  procedure in the real implemenation.  This makes it easy to create
> > >  procedure-with-getters from C code.)
> > 
> > Yes, thinking about it your closure-based
> 

I hate when that happens.

> Yes...?  :)  (I often do this myself...)
> 

I was going to say that it should work with primitive procedures too,
since it has no intrinsic need to hack the original closure, this is
just an optimization. In fact if you apply your unmemoization protocol
to the getter case as well, it needn't even introduce any costs in
efficiency.

> I've made an initial proposal to him now about a hook-free
> unmemoization protocol which doesn't introduce any visible change in
> semantics.  However, since this protocol will be accessible on the C
> level, it will be possible to some nice optimizations.
> 
> Later, we may add the user-visible hooks.
> 
> If he'll accept the proposal, I'll first commit it on a CVS branch so
> that we can evaluate the effects on execution speed.  I have good
> hopes that it won't change noticeably.
> 

Cool.

 - Maciej