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 QAA11534; Tue, 14 Oct 2003 16:37:18 +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 QAA24125 for <caml-list@pauillac.inria.fr>; Tue, 14 Oct 2003 16:37:17 +0200 (MET DST)
Received: from smtp10.atl.mindspring.net (smtp10.atl.mindspring.net [207.69.200.246])
	by nez-perce.inria.fr (8.11.1/8.11.1) with ESMTP id h9EEbF100833
	for <caml-list@inria.fr>; Tue, 14 Oct 2003 16:37:16 +0200 (MET DST)
Received: from user-0ccekej.cable.mindspring.com ([24.199.81.211] helo=minsky-primus.homeip.net)
	by smtp10.atl.mindspring.net with smtp (Exim 3.33 #1)
	id 1A9QIU-0001p7-00
	for caml-list@inria.fr; Tue, 14 Oct 2003 10:37:14 -0400
Received: from 141.155.88.179
        (SquirrelMail authenticated user yminsky)
        by minsky-primus.homeip.net with HTTP;
        Tue, 14 Oct 2003 10:37:14 -0400 (EDT)
Message-ID: <51792.141.155.88.179.1066142234.squirrel@minsky-primus.homeip.net>
Date: Tue, 14 Oct 2003 10:37:14 -0400 (EDT)
Subject: [Caml-list] Weird behavior with nan's and min/max
From: "Yaron Minsky" <yminsky@cs.cornell.edu>
To: caml-list@inria.fr
User-Agent: SquirrelMail/1.4.2-1
MIME-Version: 1.0
Content-Type: text/plain;charset=iso-8859-1
Content-Transfer-Encoding: 8bit
X-Priority: 3
Importance: Normal
X-Loop: caml-list@inria.fr
X-Spam: no; 0.00; yaron:01 minsky:01 yminsky:01 cornell:01 val:01 val:01 bug:01 yaron:01 equality:01 polymorphic:01 polymorphic:01 float:02 float:02 harder:02 comparison:02 
Sender: owner-caml-list@pauillac.inria.fr
Precedence: bulk

min and max have pretty strange behavior when it comes to nan's.  Here's
an example:

# min 3. nan;;
- : float = 3.
# min  nan 3.;;
- : float = nan

When you think about it, the reason for this is clear.  comparisons
involving nan's always return false, so if you simply implement min as
follows:

   if x < y then x else y

the the result will depend on the order.

Now here's the weird bit.  I decided I wanted a polymorphic comparison
that wouldn't have this problem.  But this is a little harder than it
seems, since it turns out that specialized float version of equality is
different from the polymorphic version.  Here's the example:

# let raw_min = min
val raw_min : 'a -> 'a -> 'a = <fun>
# let min x y =
  if not (y = y) then y
  else if not (x = x) then x
  else raw_min x y
;;
        val min : 'a -> 'a -> 'a = <fun>
# let fmin (x:float) y =
  if not (y = y) then y
  else if not (x = x) then x
  else raw_min x y
;;
        val fmin : float -> float -> float = <fun>
# fmin 3. nan;;
- : float = nan
# fmin nan 3.;;
- : float = nan
# min nan 3.;;
- : float = nan
# min 3. nan;;
- : float = 3.

So, does this count as a bug?  Any ideas about how to fix this behavior in
a reasonably efficient way?

Yaron

-------------------
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