John Max Skaller [Sunday 4 May 2003] :
 > Similarly, subtyping of polymorphic variants is invariant
 > in the arguments, when covariance is sound and more useful.
 > This is a real pain for me, since it destroys the main use
 > I hoped to obtain for polymorphic variants:
 > 
 > type x = [`A | `B of x]
 > type y = [`A | `B of y | `C]
 > 
 > x *is* a subtype of y, but the type system
 > disallows it, failing to recognise that every `B x
 > of type x is also a `B y of y. Typically I have
 > say a term for an expression and I want to eliminate
 > some cases by say a term reduction, but I can't restrict
 > the resultant type as desired because there's no way
 > to 'narrow' the type recursion.

# type x = [`A | `B of x] ;;
type x = [ `A | `B of x]
# type y = [`A | `B of y | `C] ;;
type y = [ `A | `B of y | `C]
# fun a -> (a :x :> y) ;;
- : x -> y = <fun>

Doesn't that mean that x is a subtype of y ?

-- 
   Olivier

-------------------
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