Re: [Caml-list] Exhaustiveness, currying with GADTs

[Jacques Le Normand <rathereasy@gmail.com>]

2) seems to be the value restriction. basically, only constructors,
composition of constructors and lambdas can generalize.

What warning are you getting for 1) ?

On Thu, Oct 11, 2012 at 10:30 AM, Markus Weissmann
<markus.weissmann@in.tum.de> wrote:
> Hello,
>
> while playing around with GADTs, I came across two issues I don't understand: The exhaustiveness check and currying functions.
> 1.) I get a warning about my match (in function "componentwise") not being exhaustive, while imho it obviously is:
>
> --------8<----------
>   type d2
>   type d3
>
>   type _ t =
>     | Vector2 : (float * float) -> d2 t
>     | Vector3 : (float * float * float) -> d3 t
>
>   let componentwise : type d . (float -> float -> float) -> d t -> d t -> d t = fun f v1 v2 ->
>     match (v1, v2) with
>     | (Vector2 (x1, y1), Vector2 (x2, y2)) -> Vector2 (f x1 x2, f y1  y1)
>     | (Vector3 (x1, y1, z1), Vector3 (x2, y2, z2)) -> Vector3 (f x1 x2, f y1 y1, f z1 z2)
> -------->8----------
>
> 2.) When I use "componentwise" for other functions, I cannot do this:
>
>   let add = componentwise (+.)
>
> ("contains type variables that cannot be generalized")
> Fair enough. But I also cannot do this:
>
>   let add : type d . d t -> d t -> d t = componentwise (+.)
>
> ("This definition has type 'd t -> 'd t -> 'd t which is less general than 'd0. 'd0 t -> 'd0 t -> 'd0 t")
> Only the "full" version works:
>
>   let add : type d . d t -> d t -> d t = fun v1 v2 -> componentwise (+.) v1 v2
>
>
> I would be very happy if someone could shed some light especially on 1.).
>
>
> Thanks & regards
> -Markus
>
> --
> Markus Weißmann, M.Sc.
> Technische Universität München
> Institut für Informatik
> Boltzmannstr. 3
> D-85748 Garching
> Germany
> http://wwwknoll.in.tum.de/
>
>
> --
> Caml-list mailing list.  Subscription management and archives:
> https://sympa.inria.fr/sympa/arc/caml-list
> Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
> Bug reports: http://caml.inria.fr/bin/caml-bugs