From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <bhurt@janestcapital.com>
X-Spam-Checker-Version: SpamAssassin 3.1.3 (2006-06-01) on yquem.inria.fr
X-Spam-Level: 
X-Spam-Status: No, score=0.0 required=5.0 tests=none autolearn=disabled 
	version=3.1.3
Received: from concorde.inria.fr (concorde.inria.fr [192.93.2.39])
	by yquem.inria.fr (Postfix) with ESMTP id 796CDBC0A
	for <caml-list@yquem.inria.fr>; Mon, 21 May 2007 14:57:39 +0200 (CEST)
Received: from smtp.janestcapital.com (www.janestcapital.com [66.155.124.107])
	by concorde.inria.fr (8.13.6/8.13.6) with ESMTP id l4LCvb9a018744
	for <caml-list@inria.fr>; Mon, 21 May 2007 14:57:39 +0200
Received: from [172.25.129.161] [38.96.172.125] by janestcapital.com with ESMTP
  (SMTPD-9.10) id A74A0248; Mon, 21 May 2007 08:57:46 -0400
Message-ID: <4651973F.9010503@janestcapital.com>
Date: Mon, 21 May 2007 08:57:35 -0400
From: Brian Hurt <bhurt@janestcapital.com>
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:1.7.2) Gecko/20040804 Netscape/7.2 (ax)
X-Accept-Language: en-us, en
MIME-Version: 1.0
To: skaller <skaller@users.sourceforge.net>
Cc: caml-list@inria.fr
Subject: Re: [Caml-list] tail rec
References: <1179543365.26755.33.camel@rosella.wigram>
In-Reply-To: <1179543365.26755.33.camel@rosella.wigram>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Miltered: at concorde with ID 46519741.004 by Joe's j-chkmail (http://j-chkmail . ensmp . fr)!
X-Spam: no; 0.00; compiler:01 ocaml:01 ocaml:01 non-trivial:01 recursive:01 tailcall:01 tailcall:01 wrote:01 rec:01 rec:01 caml-list:01 tail:01 tail:01 loops:02 attribute:03 

skaller wrote:

>I have a silly idea. Introduce a new construction:
>
>let tailrec f .. 
>
>This is the same as let rec except it checks every direct call to f
>is in tail position (and bombs out the compiler if not).
>
>  
>
I dislike this idea, even for it's stated purpose (to make it easier for 
people to learn Ocaml).  Wether a particular call is a tail call or not 
is not an attribute of the function, but an attribute of the call.  The 
same function can be called as both a tail call and a non-tail call.  If 
we're going to teach Ocaml, let's at least teach it right.

And this is a non-trivial complaint.  It's very common for me to write 
recursive loops where the call inside the function is a tail call, but 
the original call to the function is not a tail call.  So I couldn't write:
    let tailrec loop i s =
        if i > 0 then
            loop (i-1) (s+i)  (* proper tail call- it's a loop *)
        else s
    in
    let t = 2 * (loop n 0) in ... (* this is not a tail call *)

A better proposal would be one where the call itself is marked as a 
tailcall, and bombs out if that call isn't.  Maybe something like:

    let rec loop i s =
       if i > 0 then
          tailcall loop (i-1) (s+i) (* force this to be a tailcall *)
       else s
    in
    let t = 2 * (loop n 0) in ... (* not a tail call, and that's OK *)

At which point I'm neutral on the idea.  The rules for what is or isn't 
a tail call are not complicated.  On the other hand, it's a little bit 
of extra correctness for not much cost.

Brian