Re: [Kawa-commonlisp-dev] Multiple value binding

[Jamison Hope]

On Aug 3, 2012, at 1:16 PM, Helmut Eller wrote:

> On Fri, Aug 03 2012, Charles Turner wrote:
>
> > Should I go for the rough solution and continue testing
> > DESTRUCTURING-BIND or try to pursue the approach outlined by Jamison?
>
> I'd go with the rough solution, because changing calling conventions
> tends to be tricky.

Agreed. Let's circumvent the subtleties for now. Your rough solution
will continue to work once this gets addressed, right? It'll just
appear to be unidiomatic to a Lisper. Maybe put a TODO comment
mentioning that you're temporarily working around an implementation
bug and to revisit once it's fixed.

My sketch was incomplete, anyway. Consider:

(multiple-value-bind (a b) (values 1 2 3) (list b a)) ;; => (2 1)
(multiple-value-bind (a b) (values 1)     (list b a)) ;; => (NIL 1)

What it really needs to do is coerce N return values to the M values
expected by the continuation, where all of these scenarios are legal:
M == N
M < N
M > N

> FWIW, I think that (let ((x (values))) x) should be an error in Scheme
> and that the various RNRSs are ridicoulsy underspecified.

They are of course underspecified by design. I will avoid making any
political statements about the de facto Scheme standards on a
Common Lisp mailing list. :-)

Jamie

--
Jamison Hope
The PTR Group
www.theptrgroup.com