Re: is too an application

[Matthew Flatt <mflatt@cs.utah.edu>]

Quoting "Greg Pettyjohn":
> OK, that helps, but suppose I want to do something like this:
> 
> 
> (define t%
>   (class object%
>     (public t1 t2)
>     (define t1
>       (lambda x
>         (printf "t1 called with args: ~a~n" x)))
>     
>     (define t2
>       (lambda x
>         (printf "t2 called with args: ~a~n" x)
>         (let ([y (cons 'foo x)])
>           (send this t1 . y))))
>     (super-instantiate ())))
> 
> (define t (instantiate t% ()))
> (send t t2 'a 'b 'c 'd)
> 
> How can I rewrite t2 without using the embedded let?

When designing the `.'-based send form, it seemed like a bad idea, but
I couldn't pin down why. Now I see clearly.

The problem is that
  (send this t1 . (cons 'foo x))
and
  (send this t1 'foo x)
are equivalent at the reader level.

Requiring the expression after `.' not to have parentheses is weird.
Looks like we need an `apply-send' form, instead.

Matthew