Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:52518
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain nebm.ist.utl.pt from 193.136.128.22 cause and error)
Content-Type: text/plain; charset=utf-8; format=flowed; delsp=yes
To: internals@lists.php.net
References: <BA175FCA-75BB-43E6-BE2A-03D1997F6C15@roshambo.org>
Date: Fri, 27 May 2011 11:00:35 +0100
MIME-Version: 1.0
Content-Transfer-Encoding: Quoted-Printable
Organization: =?utf-8?Q?N=C3=BAcleo_de_Eng=2E_Biom=C3=A9di?=
 =?utf-8?Q?ca_do_I=2ES=2ET=2E?=
Message-ID: <op.vv4wq9pzidpuyk@clk-0081.clk-domain>
In-Reply-To: <BA175FCA-75BB-43E6-BE2A-03D1997F6C15@roshambo.org>
User-Agent: Opera Mail/11.11 (Win32)
Subject: Re: [PHP-DEV] INF behavior
From: glopes@nebm.ist.utl.pt ("Gustavo Lopes")

Em Fri, 27 May 2011 04:03:01 +0100, Philip Olson <philip@roshambo.org>  =

escreveu:

> Hello geeks,
>
> A geek is needed to clarify PHP bug #45712. This is an edge case but t=
he  =

> test (bug45712.phpt) contains code similar to the following:
>
> <?php
> $inf =3D pow(0, -2);
>
> var_dump($inf);          // float(INF)
> var_dump($inf =3D=3D  $inf); // bool(false)
> var_dump($inf =3D=3D=3D $inf); // bool(true)
> ?>
>
> That's how it's behaved since ~forever (AFAICT) and remains in  =

> 5.3.7-dev, but PHP 5.4.0-dev changes behavior so both now return true.=

>
> Is this is how we want it? And how should this be documented/explained=
?

What IEEE Std 754-2008 says is this:

=C2=AB5.11 Details of comparison predicates

5.11.0
For every supported arithmetic format, it shall be possible to compare o=
ne  =

floating-point datum to another
in that format (see  5.6.1). Additionally, floating-point  data   =

represented in different formats shall be
comparable as long as the operands=E2=80=99 formats have the same radix.=


Four mutually exclusive relations are possible: less than, equal, greate=
r  =

than, and unordered. The last case
arises when at least one operand is NaN. Every NaN shall compare unorder=
ed  =

with everything, including
itself. Comparisons shall ignore the sign of zero (so +0 =3D =E2=88=920)=
. Infinite  =

operands of the same sign shall
compare equal.

Languages define how the result of a comparison shall be delivered, in o=
ne  =

of two ways: either as a relation
identifying one of the four relations listed above, or as a true-false  =

response to a predicate that names the
specific comparison desired.

Table 5.1, Table 5.2, and Table 5.3 exhibit twenty-two functionally  =

distinct useful predicates and negations
with various ad-hoc and traditional names and symbols. Each predicate is=
  =

true if any of its indicated
relations is true. The relation =E2=80=9C?=E2=80=9D indicates an unorder=
ed relation. (...)

(...)


Table 5.1 - Required unordered-quiet predicate and negation

Unordered quiet predicate          | Unordered-quiet Negation
-----------------------------------|------------------------------------=
---
True Relations | Names             | True Relations | Names
---------------|-------------------|------------------------------------=
---
EQ             | compareQuietEqual | LT GT UN       | compareQuietNotEqu=
al
                |         =3D         |                |     ?<>, NOT(=3D=
), =E2=89=A0

Given this, I think the behavior of 5.4 is correct because  "[i]nfinite =
 =

operands of the same sign shall compare equal" and compareQuietEqual map=
s  =

EQ to true. 5.4 still gives false for NAN=3D=3DNAN, which is also correc=
t.

This is also what this C99 program gives:

#include<stdio.h>
#include<math.h>
void main() {
     printf("%d %d\n", INFINITY =3D=3D INFINITY, NAN =3D=3D NAN);
}

> ./a.out
1 0

-- =

Gustavo Lopes