Re: corrupt environment?

[Radey Shouman <Radey_Shouman@splashtech.com>]

roland.kaufmann@space.at writes:

> >>>>> "Radey" == Radey Shouman <Radey_Shouman@splashtech.com> writes:

[ ... ]

>       (letrec ((foo (lambda (arg)
> 		      (or arg (and (procedure? foo)
> 				   (foo 99)))))) 
> 	(define bar (foo #f))
> 	(foo #f))
> 	==> 99
> 
>       (letrec ((foo 77)
> 	       (bar #f)
> 	       (retfoo (lambda () foo)))
> 	(define baz (retfoo))
> 	(retfoo))
> 	==> 77

> Both Guile 1.2 and a recent (June 11, 1999) CVS version yield #f for
> both forms.

Thanks for checking.  This behavior is clearly a bug, as can be seen
by converting the above forms to LETRECs.

For example:

   (letrec ((foo (lambda (arg)
		   (or arg (and (procedure? foo)
				(foo 99)))))) 
     (define bar (foo #f))
     (foo #f))

Becomes:

   (letrec ((foo (lambda (arg)
		   (or arg (and (procedure? foo)
				(foo 99)))))) 
     (letrec ((bar (foo #f)))
       (foo #f)))

Guile essentially converts the original expression to:

   (letrec ((foo (lambda (arg)
		   (or arg (and (procedure? foo)
				(foo 99)))))
	    (bar (foo #f)))
     (foo #f))

Which is plainly not quite right.

I don't know what plans are laid for the future Guile macro-expander
and evaluator.  If it is to remain SCM-like for some time then the
approach used in the current SCM could be retro-fitted.

I would love to see a macro system in which internal DEFINE could be
written in a natural way.