From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mail1-relais-roc.national.inria.fr (mail1-relais-roc.national.inria.fr [192.134.164.82])
	by walapai.inria.fr (8.13.6/8.13.6) with ESMTP id p1IJodgW014279
	for <caml-list@sympa-roc.inria.fr>; Fri, 18 Feb 2011 20:50:39 +0100
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AgYHAL9bXk2BZgNHgWdsb2JhbACYHI4LFQEBFiIkuweFXgSMEQ
X-IronPort-AV: E=Sophos;i="4.62,188,1297033200"; 
   d="scan'208";a="100217942"
Received: from smarthost1.ircam.fr ([129.102.3.71])
  by mail1-smtp-roc.national.inria.fr with ESMTP; 18 Feb 2011 20:50:33 +0100
Received: from localhost (localhost.localdomain [127.0.0.1])
	by smarthost1.ircam.fr (Postfix) with ESMTP id 47EC326E00C;
	Fri, 18 Feb 2011 20:50:33 +0100 (CET)
X-Virus-Scanned: amavisd-new at ircam.fr
Received: from smarthost1.ircam.fr ([127.0.0.1])
	by localhost (smarthost1.ircam.fr [127.0.0.1]) (amavisd-new, port 10024)
	with ESMTP id 5TcPLSIuWiKW; Fri, 18 Feb 2011 20:50:33 +0100 (CET)
Received: from dns-ha1.ircam.fr (dns-ha1.ircam.fr [IPv6:2001:660:3004:2:1000:2ff:fed1:1])
	by smarthost1.ircam.fr (Postfix) with ESMTP id 3343026DE5A;
	Fri, 18 Feb 2011 20:50:33 +0100 (CET)
Received: from m1909.ircam.fr (m1909.ircam.fr [129.102.65.25])
	(Authenticated sender: giavitto)
	by dns-ha1.ircam.fr (Postfix) with ESMTP id 2A3E12FEBC;
	Fri, 18 Feb 2011 20:50:33 +0100 (CET)
Message-ID: <4D5ECD89.6060906@ircam.fr>
Date: Fri, 18 Feb 2011 20:50:33 +0100
From: Jean-Louis Giavitto <jean-louis.giavitto@ircam.fr>
Reply-To: jean-louis.giavitto@ircam.fr
Organization: UMR STMS 9912 - IRCAM & CNRS
User-Agent: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.6; fr; rv:1.9.2.13) Gecko/20101207 Lightning/1.0b2 Thunderbird/3.1.7
MIME-Version: 1.0
To: caml-list@inria.fr
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 8bit
Subject: [Caml-list] Contraintes dans la =?ISO-8859-1?Q?d=E9finition?= =?ISO-8859-1?Q?r=E9cursive?= de classes

Bonjour.

Je rencontre le problème suivant : une définition récursive de classe 
contraint les types paramètres de la classe, plus que je ne le souhaiterais.

Je ne comprend pas d'où viennent ces contraintes supplémentaires. Voici 
un exemple minimal :

    class virtual ['a, 'b] base =
    object
      method virtual m : 'a -> 'b derive
    end

    and virtual ['a] derive =
    object
      inherit ['a, 'a] base
    end


le typage résultant donne :

   class virtual ['a, -'b] base1 :
   object
      constraint 'b = 'a
      method virtual m : 'a -> 'a derive1
   end
   and virtual ['a] derive1 :
   object
      method virtual m : 'a -> 'a derive1
   end


ce qui ne me permet pas d'avoir un 'b arbitraire.
Je peux plus ou moins contourner ce problème de la manière suivante :

    class virtual ['a, 'b, 'c] pre_base =
    object
      method virtual m : 'a -> 'c
    end

    class virtual ['a] derive =
    object
      inherit ['a, 'a, 'a derive] pre_base
    end

    class virtual ['a, 'b] base =
    object
      inherit ['a, 'b, 'b derive] pre_base
    end



dont le type est :

   class virtual ['a, 'b, 'c] pre_base :
   object
      method virtual m : 'a -> 'c
   end

   class virtual ['a] derive :
   object
      method virtual m : 'a -> 'a derive
   end

   class virtual ['a, 'b] base :
   object
      method virtual m : 'a -> 'b pre_derive
   end



J'imagine bien qu'il y a une histoire de généralisation de variable mais 
je ne comprend pas ce qui change fondamentalement quand on définit dans 
le même class ... and ... end


Merci de vos éclaircissements.
Jean-Louis Giavitto.