From mboxrd@z Thu Jan  1 00:00:00 1970
Received: (from weis@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id SAA18459 for caml-redistribution; Mon, 14 Jun 1999 18:23:36 +0200 (MET DST)
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 SAA22627 for <caml-list@pauillac.inria.fr>; Mon, 14 Jun 1999 18:17:24 +0200 (MET DST)
Received: from pauillac.inria.fr (pauillac.inria.fr [128.93.11.35])
	by concorde.inria.fr (8.8.7/8.8.7) with ESMTP id SAA10002;
	Mon, 14 Jun 1999 18:17:22 +0200 (MET DST)
Received: (from weis@localhost) by pauillac.inria.fr (8.7.6/8.7.3) id SAA23666; Mon, 14 Jun 1999 18:17:22 +0200 (MET DST)
From: Pierre Weis <Pierre.Weis@inria.fr>
Message-Id: <199906141617.SAA23666@pauillac.inria.fr>
Subject: Re: forward function definitions
To: luther@dpt-info.u-strasbg.fr
Date: Mon, 14 Jun 1999 18:17:22 +0200 (MET DST)
Cc: caml-list@inria.fr
In-Reply-To: <19990614125538.B19275@maxime.u-strasbg.fr> from "Sven LUTHER" at Jun 14, 99 12:55:38 pm
X-Mailer: ELM [version 2.4 PL24 ME8]
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Sender: weis

> On Wed, Jun 09, 1999 at 02:38:22PM -0400, Junbiao Zhang wrote:
> > Hi,
> > 
> > 	We're developing a fairly large system with multiple function
> > definitions. It's inevitable that some functions may call other functions
> > defined later(may not even be recursive calls as in the case of
> > Sys.Signal_handle). My question is: how do I solve such a problem which is
> > extremely trival in other languages?  Thanks. 
> 
> If the two functions are in different modules, just provide a interface file
> for them, and it should work ok but can cause some linker problems.
> 
> 2 clean way of doing this are :
> 
> * use the later called function as a parameter to the first called function :

That's a neat solution, although this trick works only when the
``later defined'' function is monomorphic, or at least is not
polymorphically called (the same problem arises with the reference
trick). Furthermore, this extra argument is a bit confusing, since, as
a forward, it should always be exactly the same value and nothing
helps the reader to understand this. Furthermore, the compiler has no
ways to warranty this semantics.

> * Or you could define all this in the form of a module functor :

In this case, you can polymorphically use a forward definition,
although is appears to me to be really a heavy loaded syntactic way of
defining forwards...

Friendly,

Pierre Weis

INRIA, Projet Cristal, Pierre.Weis@inria.fr, http://cristal.inria.fr/~weis/