On Nov 15, 2012, at 02:22, Daniel Hartwig wrote:
> On 14 November 2012 18:20, Daniel Llorens <daniel.llor...@bluewin.ch> wrote:
>>> When the doc. states keyword arguments do not contribute to the
>>> success of a match, it refers only to keyword arguments in the
>>> case-lambda clause, not at the call site. This makes sense, otherwise
>>> it would inhibit writing functions that detect keywords internally
>>> from their rest arguments.
>>
>> Do you mean something like this?
>>
>> (define* (f a #:key x) x)
>> (define (g . args) (apply f args))
>> (g 0) -> #f
>> (g 0 #:x #t) -> #t
>>
>> i.e. g must accept being called with 3 'arguments' so that it can forward
>> the keyword args.
>
> I was thinking of:
>
> (define f
> (case-lambda*
> ((a . rest)
> (if (memq #:x rest) …
>
> where #:x is picked up from inside rest.
>
> Based on the other error, I'd say that any case-lambda* with keyword
> arguments is matched with a rest argument instead, i.e. “a #:key x” is
> treated as “a . rest”.
Ok, I think I see how
(define f (case-lambda*
((x #:optional y) 1)
((x #:key y) 2)
((x y #:key z) 3)))
(f #:y 2)
works. x is #:y and y is 2.
However, this should go to (x y #:key z) and it doesn't:
(f 1 2 #:z 3) -> Odd length of keyword argument list
while given
(define g (case-lambda*
((a #:key x) 1)
((a b c #:key x) 3)))
this should match (a b c #:key x) and it doesn't.
(g 1 2 3) - > Invalid keyword
Or maybe we should say that (g 1 2 3) shouldn't match (a #:key x)? Either way,
I'll file these as a bug.
Thanks,
Daniel
