From mboxrd@z Thu Jan  1 00:00:00 1970
Return-Path: <robert.fischer@SmokejumperIT.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 discorde.inria.fr (discorde.inria.fr [192.93.2.38])
	by yquem.inria.fr (Postfix) with ESMTP id 61FD1BC0A
	for <caml-list@yquem.inria.fr>; Mon, 21 May 2007 15:04:43 +0200 (CEST)
Received: from looneymail-a3.g.dreamhost.com (d071c805.dreamhost.com [208.113.200.5])
	by discorde.inria.fr (8.13.6/8.13.6) with ESMTP id l4LD4gXJ002198
	for <caml-list@inria.fr>; Mon, 21 May 2007 15:04:43 +0200
Received: from [192.168.101.113] (unknown [66.84.219.53])
	by looneymail-a3.g.dreamhost.com (Postfix) with ESMTP id 351A927B5C
	for <caml-list@inria.fr>; Mon, 21 May 2007 06:04:43 -0700 (PDT)
Message-ID: <465198E5.3010004@SmokejumperIT.com>
Date: Mon, 21 May 2007 08:04:37 -0500
From: Robert Fischer <robert.fischer@SmokejumperIT.com>
Organization: Smokejumper Consulting
User-Agent: Thunderbird 1.5.0.10 (Windows/20070221)
MIME-Version: 1.0
To: caml-list@inria.fr
Subject: Re: [Caml-list] tail rec
References: <1179543365.26755.33.camel@rosella.wigram> <4651973F.9010503@janestcapital.com>
In-Reply-To: <4651973F.9010503@janestcapital.com>
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
X-Miltered: at discorde with ID 465198EA.002 by Joe's j-chkmail (http://j-chkmail . ensmp . fr)!
X-Spam: no; 0.00; recursion:01 recursive:01 recursion:01 recursive:01 tailcall:01 tailcall:01 rec:01 rec:01 caml-list:01 tail:01 tail:01 let:03 let:03 guess:04 brian:04 

I'm with Brian on this, although it the "tailrec" specification needs to 
be optional.  I use tail recursion by accident a lot when I'm coding 
("Oh, hey, I guess this is tail recursive.  Nifty!"), and I would be 
really annoyed to have to think about whether my recursion is tail 
recursive or not every time I use it.

~~ Robert.
>
> 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