From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: weis
Received: (from weis@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id SAA25048 for caml-redistribution; Mon, 25 May 1998 18:36:51 +0200 (MET DST)
Received: from nez-perce.inria.fr (nez-perce.inria.fr [192.93.2.78]) by pauillac.inria.fr (8.7.6/8.7.3) with ESMTP id NAA29266 for <caml-list@pauillac.inria.fr>; Mon, 25 May 1998 13:46:14 +0200 (MET DST)
Received: from tcs.inf.tu-dresden.de (tcs.inf.tu-dresden.de [141.76.75.119])
	by nez-perce.inria.fr (8.8.7/8.8.7) with ESMTP id NAA21615
	for <caml-list@inria.fr>; Mon, 25 May 1998 13:46:12 +0200 (MET DST)
Received: from ithif18.inf.tu-dresden.de (ithif18 [141.76.75.118]) by tcs.inf.tu-dresden.de (8.8.3/8.8.3) with SMTP id NAA11282 for <caml-list@inria.fr>; Mon, 25 May 1998 13:45:37 +0200 (MET DST)
Received: by ithif18.inf.tu-dresden.de (SMI-8.6/SMI-SVR4)
	id NAA21664; Mon, 25 May 1998 13:46:19 +0200
Date: Mon, 25 May 1998 13:46:19 +0200
Message-Id: <199805251146.NAA21664@ithif18.inf.tu-dresden.de>
From: Hendrik Tews <tews@tcs.inf.tu-dresden.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
To: caml-list@inria.fr
Subject: co(ntra)-variant subtyping
In-Reply-To: <19980522112157L.garrigue@kurims.kyoto-u.ac.jp>
References: <199805201500.RAA19129@ithif18.inf.tu-dresden.de>
	<19980522112157L.garrigue@kurims.kyoto-u.ac.jp>
X-Mailer: VM 6.34 under Emacs 19.34.1
Sender: weis

Hi,
   
   Objective Caml's subtyping algorithm does not subtype under type
   constructors (but it does it under abbreviations), for the sake of
   efficiency. That is, eventhough colored_point :> point, you don't have
   colored_point option :> point option. This equation would be needed to
   check the subtyping in method move of example 2.
   
I must say I am really surprised.  

How about object types? Do they support co(ntra)-variant
subtype rules?

I would suggest to include a section about the subtype relation
in the ocaml documentation. Because IMHO anybody how knows
something about formal models for oo languages would expect
co(ntra)-variant subtype rules. 

In summary I think the lack of these subtype rules are a major
restriction of the ocaml language. Now it is not possible to mix
class types and (algebraic) data types in an application. Instead
a user has to decide beforehand what is more important for him:
algebraic data types and parametric polymorphism or object types
with code reuse and stepwise refinement.

Are there any plans to add the missing subtype rules to the type
system of ocaml?

Bye,

Hendrik