From mboxrd@z Thu Jan  1 00:00:00 1970
Received: from mail4-relais-sop.national.inria.fr (mail4-relais-sop.national.inria.fr [192.134.164.105])
	by walapai.inria.fr (8.13.6/8.13.6) with ESMTP id p7E6Kqg1022442
	for <caml-list@sympa-roc.inria.fr>; Sun, 14 Aug 2011 08:20:52 +0200
X-IronPort-Anti-Spam-Filtered: true
X-IronPort-Anti-Spam-Result: AiICAOZoR07UGyoFkWdsb2JhbABBmHCPCBQBAQEBCQsLBxQDIoFAAQEFJxNPC0YUiE6zEIVoXwSYLItd
X-IronPort-AV: E=Sophos;i="4.67,369,1309730400"; 
   d="scan'208";a="105423880"
Received: from smtp5-g21.free.fr ([212.27.42.5])
  by mail4-smtp-sop.national.inria.fr with ESMTP; 14 Aug 2011 08:20:44 +0200
Received: from yeeloong (unknown [82.67.194.89])
	by smtp5-g21.free.fr (Postfix) with SMTP id 7C10AD48054
	for <caml-list@inria.fr>; Sun, 14 Aug 2011 08:20:37 +0200 (CEST)
Received: by yeeloong (sSMTP sendmail emulation); Sun, 14 Aug 2011 08:19:59 +0200
Date: Sun, 14 Aug 2011 08:19:59 +0200
From: rixed@happyleptic.org
To: caml-list@inria.fr
Message-ID: <20110814061959.GC32098@yeeloong.happyleptic.org>
References: <20110806125021.GB10154@ombreroze.happyleptic.org>
 <4E3D3A89.90007@inria.fr>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <4E3D3A89.90007@inria.fr>
User-Agent: Mutt/1.5.21 (2010-09-15)
Subject: Re: [Caml-list] typer strangeness (3.12.0)

Thank you for your answers, but you aimed too high above my
understanding :-)

-[ Sat, Aug 06, 2011 at 02:58:49PM +0200, Fabrice Le Fessant ]----
> The type constraint that you specified does not constraint the
> polymorphism of the type. To declare a polymorphic constraint, you must
> use (with OCaml >= 3.12.0) :
> 
> let pipe : 'a 'b 'c. ('a, 'b) parzer -> ('c, 'a) parzer -> ('c, 'b) parzer =

Never saw this notation before. I checked in the manual but beside in
the language grammar (in the definition of a poly-typexpr) I could find
more explanation.
Is it explained somewhere why this notation was required and why this:

let pipe : ('a, 'b) parzer -> ('c, 'a) parzer -> ('c, 'b) parzer =

is not enough? What's adding the "'a 'b 'c. " exactly ? I understand
that it forbids ocaml to consider that the various 'a (for instance) may
be differents 'a, but I don't understand why ocaml migh do that in the
first place (especially to produce a _less general_ type than the one
given) ? If it's not a bug, then what's the purpose ?

As for my second question, I received no answer (was: why changing the
pattern match for something that should be equivalent changes the infered
type). If you suspect this can be a bug then I can look for a simple
reproductible case.