From mboxrd@z Thu Jan  1 00:00:00 1970
Received: (from majordomo@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id GAA21077; Fri, 21 Dec 2001 06:29:21 +0100 (MET)
X-Authentication-Warning: pauillac.inria.fr: majordomo set sender to owner-caml-list@pauillac.inria.fr using -f
Received: from concorde.inria.fr (concorde.inria.fr [192.93.2.39]) by pauillac.inria.fr (8.7.6/8.7.3) with ESMTP id GAA20686 for <caml-list@pauillac.inria.fr>; Fri, 21 Dec 2001 06:29:20 +0100 (MET)
Received: from fledge.watson.org (fledge.watson.org [204.156.12.50])
	by concorde.inria.fr (8.11.1/8.11.1) with ESMTP id fBL5TJr17678
	for <caml-list@inria.fr>; Fri, 21 Dec 2001 06:29:19 +0100 (MET)
Received: from localhost (patrick@localhost)
	by fledge.watson.org (8.11.6/8.11.5) with ESMTP id fBL5T9s93402;
	Fri, 21 Dec 2001 00:29:09 -0500 (EST)
	(envelope-from patrick@watson.org)
Date: Fri, 21 Dec 2001 00:29:08 -0500 (EST)
From: Patrick M Doane <patrick@watson.org>
To: Jacques Garrigue <garrigue@kurims.kyoto-u.ac.jp>
cc: caml-list@inria.fr
Subject: Re: [Caml-list] Objects and private methods
In-Reply-To: <20011221140658Z.garrigue@kurims.kyoto-u.ac.jp>
Message-ID: <20011221001434.V91723-100000@fledge.watson.org>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Sender: owner-caml-list@pauillac.inria.fr
Precedence: bulk

Hello, thanks for the quick reply.

On Fri, 21 Dec 2001, Jacques Garrigue wrote:

> > I'm having some trouble getting the following type of structure to
> > type-check:
> >
> >   class type a = object method get_b : a end
> >
> >          and b = object method b : unit end
> >
> >   class a_impl =
> >   (object (self)
> >     method get_b = (self :> b)
> >     method private b = ()
> >   end : a)
> >   ;;
> >
>
> Due to the way privacy attributes are implemented in the type checker,
> a type annotation can force a private method to become public (but
> never the opposite). You can see it on the following example.
>
>   # class c = object (self : < b : 'a; .. >) method private b = 1 end;;
>   class c : object method b : int end
>
> This explains why the type checker doesn't complain about b not being
> public: it makes it public as needed. And since b is public, you
> cannot hide it.
> (This may seem strange from the point of view of privacy, but you
> should just understand private methods as "may be hidden in a
> subclass"; if unified with a public method it may no longe be hidden.)

Okay - and in this case the signature matching constraint is considered a
subclass?

> > The basic idea is that the class 'a_impl' implements 'b' privately. This
> > seems like something that should be possible to do.  I suspect that the
> > coercion of self to 'b' is causing the problem here.
>
> I don't understand very well what you mean by "implement privately":
> a and b do not seem to be related in your code.
> I will suppose that you meant
>   class type a = object method get_b : b end

Yes, that's right. I was distilling from a larger example and missed that.

> You cannot convert something from private to public through a coercion
> (this is different from forcing it with an annotation). This is
> because the values are actually different: a private method and a
> public method have different internal representations.

Are there any papers that I could read over that would give a better sense
for the internal representations? That might help me understand some of
these limitations better.

> On the other hand, you can hide a method through coercions.
>
>   class a_impl = object (self)
>     method get_b = (self :> b)
>     method b = ()
>   end
>   let new_a = (new a_impl :> a)

Thanks! What then are the advantages to using an annotation instead of a
coercion for signature matching? For example, with a hypothetical syntax
like this:

class a_impl :> a = object (self)
  method get_b = (self :> b)
  method b = ()
end

Patrick

-------------------
Bug reports: http://caml.inria.fr/bin/caml-bugs  FAQ: http://caml.inria.fr/FAQ/
To unsubscribe, mail caml-list-request@inria.fr  Archives: http://caml.inria.fr