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 KAA14438 for caml-redistribution; Wed, 26 Nov 1997 10:56:28 +0100 (MET)
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 MAA23851 for <caml-list@pauillac.inria.fr>; Tue, 25 Nov 1997 12:32:59 +0100 (MET)
Received: from trinquette.essi.fr (trinquette.essi.fr [157.169.3.104])
	by nez-perce.inria.fr (8.8.7/8.8.5) with ESMTP id MAA14016
	for <caml-list@inria.fr>; Tue, 25 Nov 1997 12:32:55 +0100 (MET)
Received: (from cuoq@localhost)
	by trinquette.essi.fr (8.8.8/8.8.7) id MAA03088
	for caml-list@inria.fr; Tue, 25 Nov 1997 12:30:44 +0100 (MET)
Date: Tue, 25 Nov 1997 12:30:44 +0100 (MET)
From: Cuoq Pascal <cuoq@essi.fr>
Message-Id: <199711251130.MAA03088@trinquette.essi.fr>
To: caml-list@inria.fr
Subject: Re: type recursifs et abreviations cyclique and Co
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
Content-MD5: oeaev+BEoeIBbpQ+ItGbEA==
Sender: weis


Jason Hickey wrote :

> 2.  The policy imposes a needless efficiency penalty on type
>     abstraction.  
>  [...]
>     ocaml will insert an extraneous boxing for each
>     occurrence of an item of type x in t.  Consider an unlabeled
>     abstract binary tree:
> 
>         type 'a t = ('a option) * ('a option)    (* abstract *)
>         ...
>         type node = X of node t
> 
>     Every node is boxed, with a space penalty that is
>     linear in the number of nodes.

It seems to me that this is an argument for adding an optimization to
ocaml, not to change the typechecking algorithm :)

Pascal