caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
* [Caml-list] Y combinator and type-checking
@ 2004-07-02 17:36 Martin Berger
  2004-07-02 17:54 ` Brian Hurt
  0 siblings, 1 reply; 4+ messages in thread
From: Martin Berger @ 2004-07-02 17:36 UTC (permalink / raw)
  To: The Caml Trade

i just noticed something confusing: consider an implementation of the
Y combinator (for a CBV language):

     let rec y f = f (fun x -> (y f) x );;

let's see what ocaml says about this definition.

     # let rec y f = f (fun x -> (y f) x );;
     val y : (('a -> 'b) -> 'a -> 'b) -> 'a -> 'b = <fun>

now i annotate the definition above with the types just
inferred:

     let rec y  ( f : ( 'a -> 'b ) -> 'a -> 'b )
            : ( ( 'a -> 'b ) -> 'a -> 'b ) -> 'a -> 'b
       = f (fun x -> (y f) x );;

suddenly, ocaml complains:

     #let rec y  ( f : ( 'a -> 'b ) -> 'a -> 'b )
              : ( ( 'a -> 'b ) -> 'a -> 'b ) -> 'a -> 'b
         = f (fun x -> (y f) x );;

     This expression has type 'a -> 'b -> 'c but is here used with
     type (('a -> 'b -> 'c) -> 'a -> 'b -> 'c) -> 'a -> 'b -> 'c


why would annotating a program with seemingly correct types render
a program untypable? this is my main question.

here's a related observation: if we run the last program under "ocaml
-rectypes" instead of just "ocaml" we get

     # let rec y  ( f : ( 'a -> 'b ) -> 'a -> 'b )
              : ( ( 'a -> 'b ) -> 'a -> 'b ) -> 'a -> 'b
         = f (fun x -> (y f) x );;
     val y :
     (((('a -> ('a -> 'b as 'b)) -> 'a -> 'b as 'a) -> 'b) -> 'a
     -> 'b) -> 'a -> 'b = <fun>

so now the typechecker gives a thumbs-up, but the inferred and annotated
types differ. why?

I hope this is not a well-known beginners issue!

martin

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners


^ permalink raw reply	[flat|nested] 4+ messages in thread
* RE: [Caml-list] Why type infenere enter infinte loop here?
@ 2004-07-06 23:41 Kevin S. Millikin
  0 siblings, 0 replies; 4+ messages in thread
From: Kevin S. Millikin @ 2004-07-06 23:41 UTC (permalink / raw)
  To: 'Andy Yang', caml-list

On Tuesday, July 06, 2004 5:25 PM, Andy Yang [SMTP:yyu08@yahoo.com] wrote:
> Hi, all
>
> I am relatively new in Ocaml. Perhaps this problem is
> trivial. With the following interactive process, ocaml
> cannot give out f2's type. Could you tell me why it
> cannot?
>
>
> # let rec y f = f (fun x ->(y f) x);;
> val y : (('a -> 'b) -> 'a -> 'b) -> 'a -> 'b = <fun>
> # let g h x = h x;;
> val g : ('a -> 'b) -> 'a -> 'b = <fun>
> # let f1 = y g;;
> val f1 : '_a -> 'b = <fun>
> # let k x = x * x + 1;;
> val k : int -> int = <fun>
> # let f2 = f1 k;;
>

For the same reason that it does not print the type of

# while true do () done;;

y is a fixpoint combinator.  g is the application combinator (Curry's B 
combinator).  f1 is the fixpoint of application, which does not 
terminating.  Your definition of f1 is the same as

# let f1 = y (fun h x -> h x);;

which is the same as

# let rec f1 = fun x -> f1 x;;

An application of f1 will not terminate.  The typechecker is perfectly h  
appy to determine that your f2 has type 'a -> 'b, but it won't print that 
type until it finishes computing the value.

-- Kevin

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners


^ permalink raw reply	[flat|nested] 4+ messages in thread

end of thread, other threads:[~2004-07-06 23:42 UTC | newest]

Thread overview: 4+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2004-07-02 17:36 [Caml-list] Y combinator and type-checking Martin Berger
2004-07-02 17:54 ` Brian Hurt
2004-07-06 23:24   ` [Caml-list] Why type infenere enter infinte loop here? Andy Yang
2004-07-06 23:41 Kevin S. Millikin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).