Re: [Caml-list] Confusing behaviour of type inference for polymorphic classes.

[Dmitri Boulytchev <dboulytchev@gmail.com>]

     Jeremy, Alain,

    the code snippet I presented was actually boiled down from the more 
complex example.
I don't have any particular interest to the concrete type t; I'm ok to 
give up
on "finitely regularizable" types but I definitely don't want to give up 
on GADTs
which might be essentially irregular.

 > As you've noted, you can work around the
 > problem using open recursion and typing the knot yourself, but you
 > lose the ability to subtype in the process.

   Well, having the solution with the polymorphic recursive function

     class ['a, 'b, 'ta, 'tb] proto_m f =
       object ... end

     let rec f : 'a 'b 'ta 'tb . unit -> ('a, 'b, 'ta, 'tb) #proto_m = 
fun () -> new proto_m f

     class ['a, 'b, 'ta, 'tb] m = ['a, 'b, 'ta, 'tb] proto_m f

   it is possible to inherit from proto_m:

     class ['a, 'b, 'ta, 'tb] m' f  =
       object
         inherit ['a, 'b, 'ta, 'tb] proto_m f as super
         method t fa fb s =
           match s with
           | A (a, bat) -> <some extra work>; super#t fa fb s
           | B  a       -> <some extra work>; super#t fa fb s
       end

     let rec g : 'a 'b 'ta 'tb . unit -> ('a, 'b, 'ta, 'tb) #proto_m = 
fun () -> new m' g

     class ['a, 'b, 'ta, 'tb] subm = ['a, 'b, 'ta, 'tb] m' g

   Here subm is a legitimate subclass of m with the desired behaviour 
(this property can not be
achieved with the recursive module though it is really much more elegant 
way to provide the proper
instantiation for class type parameters). So I don't see why I'm loosing 
the ability to subtype.
It seems to me that this workaround can be used to overcome the "regular 
class" restriction
completely in a syntax-conversion manner. I appreciate any comments.

   Dmitry.