On Wed, 4 Apr 2007, Alain Frisch wrote:
> In the present case, we have good solutions that don't require Obj.
> Unless a strong case is made that performance is not adequate, there is
> really no reason to use Obj.

I agree with the "don't use Obj" part, but I think the current solutions
are not that good. Encoding with records or objects is unclear, and
while I find the encoding with recursive modules much clearer, that's
still listed as an experimental feature. It also seems a bit heavy,
using a recursive module just to get polymorphic recursion. Providing a
type for the function and having the compiler use it is the right
solution, IMO. It's not as much performance (though that is important!)
as clarity, admittedly subjective, that I am judging the solutions by.

-- Brian