15 minutes ago, Neil Van Dyke wrote:
> Eli Barzilay wrote at 07/10/2011 02:25 AM:
> >> it should work as you originally expected, or it should raise an
> >> error to complain about an internal "define" referencing itself like
> >> that and ignoring the pre-existing binding from the argument.
> >>
> >
> > An error would be nice, but racket never did that. Same as `letrec'.
>
> So, I think that this should be an error, given the current
> semantics:
>
> (define X X)
>
> Maybe that's an extremely simple example of a more general error of
> attempting to evaluate an uninitialized variable. At least, I think
> that the ones that can be proven statically to always be
> uninitialized references should be compile errors. Or is there a
> good reason for the compiler to be more liberal?
I don't remember if there's a good reason for that. (And I'm sure
that others will have plenty to say about it in the morning.) I also
don't remember any implementation that does such a check, though some
will throw a runtime error.
> Separately, I also think that maybe the following should be an error,
> since someone doing this with internal-"define" instead of "let" is most
> likely a student, and a student probably doesn't mean to do this:
>
> (define (foo name)
> (define name ...)
> ...)
Yes, like I said -- such an error would be nice, if possible. (Not
only for students -- every once in a while I'll move some `let'
bindings into internal defines for convenience, and this is an easy
trap to fall into.)
--
((lambda (x) (x x)) (lambda (x) (x x))) Eli Barzilay:
http://barzilay.org/ Maze is Life!
_________________________________________________
For list-related administrative tasks:
http://lists.racket-lang.org/listinfo/users
