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 XAA14560; Thu, 15 Jul 2004 23:15:17 +0200 (MET DST)
X-Authentication-Warning: pauillac.inria.fr: majordomo set sender to owner-caml-list@pauillac.inria.fr using -f
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 XAA14977 for <caml-list@pauillac.inria.fr>; Thu, 15 Jul 2004 23:15:16 +0200 (MET DST)
Received: from mproxy.gmail.com (rproxy.gmail.com [64.233.170.200])
	by nez-perce.inria.fr (8.12.10/8.12.10) with SMTP id i6FLFEEV016788
	for <caml-list@inria.fr>; Thu, 15 Jul 2004 23:15:15 +0200
Received: by mproxy.gmail.com with SMTP id d78so438018rnf
        for <caml-list@inria.fr>; Thu, 15 Jul 2004 14:15:14 -0700 (PDT)
Received: by 10.38.207.42 with SMTP id e42mr189901rng;
        Thu, 15 Jul 2004 14:15:14 -0700 (PDT)
Message-ID: <d849ad2a04071514153cd24c26@mail.gmail.com>
Date: Thu, 15 Jul 2004 17:15:14 -0400
From: John Prevost <j.prevost@gmail.com>
To: Ocaml Mailing List <caml-list@inria.fr>
Subject: Re: [Caml-list] Unboxing options, was RE: assertions or exceptions?
In-Reply-To: <200407152049.i6FKnnSd004687@psi-phi.mit.edu>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
References: <200407152049.i6FKnnSd004687@psi-phi.mit.edu>
X-Miltered: at nez-perce with ID 40F6F3E2.002 by Joe's j-chkmail (http://j-chkmail.ensmp.fr)!
X-Loop: caml-list@inria.fr
X-Spam: no; 0.00; prevost:01 prevost:01 caml-list:01 unboxing:01 2004:99 0400,:01 jfc:01 struct:01 val:01 val:01 pants:99 sig:01 assertions:01 match:02 silly:02 
Sender: owner-caml-list@pauillac.inria.fr
Precedence: bulk

On Thu, 15 Jul 2004 16:49:49 -0400, John Carr <jfc@mit.edu> wrote:
> Is there valid code (no Obj.magic) that cares that (Some None) and
> (None) are both represented by the same bit pattern?

module A =
  struct
    type t = string option
    let default = ref ""
    let set_default x = default := x
    let get_default = !default
    let create x = Some x
    let create_default () = None
    let get_value = function Some x -> x | None -> !default
  end : sig
    type t
    val set_default : string -> unit
    val get_default : unit -> string
    val create : string -> t
    val create_default : unit -> t
    val get_value : t -> string
  end

(* this comes from somewhere else... perhaps it reads a configuration
   file, and uses None if the value isn't available, and Some
(A.get_default ()) if
   it's a specific value meaning "the current default value". *)
val get_attribute : string -> A.t option

...
    match get_attribute "pants" with
      | Some x -> (* called when the value was in the file *)
      | None -> (* called when either the value wasn't in the file
*or* it's the default,
                   represented as Some None *)

In this case, the user of the module A doesn't even know that type t
is represented using an option type underneath--the user just knows
that the value could either be the default
value or a specific value.  The writer of A could of course choose to
make his own special type:

type t = Default | Value of string

But it's rather silly to say that "if you think someone might put your
type into an option, you should probably make your own different type
which is identical to option but doesn't do magic behind the scenes."

John.

-------------------
To unsubscribe, mail caml-list-request@inria.fr Archives: http://caml.inria.fr
Bug reports: http://caml.inria.fr/bin/caml-bugs FAQ: http://caml.inria.fr/FAQ/
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners