[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Stray Birds

   Date: Tue, 2 Mar 93 00:36:14 -0500
   From: "Guillermo J. Rozas" <gjr@martigny.ai.mit.edu>
   Reply-To: gjr@martigny.ai.mit.edu

      Date: Mon, 1 Mar 93 19:59:14 PST
      From: kend@newton.apple.com (Ken Dickey)

      At  7:56 PM 93/03/01 -0500, Guillermo J. Rozas wrote:
      >I haven't used the HL macros, but what is wrong with the following
      >(besides "exporting" define-values-helper) as a _reference_
      >implementation
      >
      >(define-syntax DEFINE-VALUES
      >  (syntax-rules
      >   ()
      >   ((define-values (<name> ...) <body> ...)
      >    (begin
      >      (define <name> #f) ...
      >      (call-with-values
      >       (lambda () <body> ...)
      >       (lambda all-the-values
      >         (define-values-helper all-the-values <name> ...)))))))

      My recollection is that the receiver in a call-with-values form has to
      accept the exact number of values returned by the thunk.  I certainly think
      that it *should* be the rule...  Anyone?

      -Ken                              kend@newton.apple.com


   It does accept them.  It accepts any number!

   I think the agreement was that the implicit application in
   call-with-values should be no more or no less strict than the
   application in a procedure call.

   Thus, since lambda expressions with symbols for argument lists
   generate procedures that take any number of arguments, the
   continuation created by call-with-values above should be able to
   accept any number of values.

The above is not the agreement on multiple values as I understood it.
I distinctly remember us stating that "it is an error" to supply too
many or too few values.  Since it is not required that an
implementation "signal an error", the above implementation is correct,
albeit probably not desirable.
--------------------
Morry Katz
katz@cs.stanford.edu
--------------------