Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:58570
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: error (pb1.pair.com: domain nebm.ist.utl.pt from 193.136.128.21 cause and error)
Content-Type: text/plain; charset=utf-8; format=flowed; delsp=yes
To: internals@lists.php.net, "Anthony Ferrara" <ircmaxell@gmail.com>
References: <CAAyV7nHajQqw5tsQh=eZWhxtotQqNx03sd=FDTbcu10nUuNxtA@mail.gmail.com>
Date: Sun, 04 Mar 2012 14:52:14 +0100
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
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.wanfhc15idpuyk@damnation.nl.lo.geleia.net>
In-Reply-To: <CAAyV7nHajQqw5tsQh=eZWhxtotQqNx03sd=FDTbcu10nUuNxtA@mail.gmail.com>
User-Agent: Opera Mail/11.61 (Win32)
Subject: Re: [PHP-DEV] Patches for Review
From: glopes@nebm.ist.utl.pt ("Gustavo Lopes")

On Sun, 04 Mar 2012 14:29:49 +0100, Anthony Ferrara <ircmaxell@gmail.com>  
wrote:

> I had messaged about these patches before, but with the 5.4 release
> process happening I think it slipped through the cracks.  I have 3
> patches that are ready for inclusion...
>> [...]
>
> https://bugs.php.net/bug.php?id=60789
> Bringing pow() inline with docs in that it will attempt to return an
> integer (by casting one or more of the arguments to int) if possible.
> This fixes a precision loss that can occur with float arguments that
> are exactly integers on 64 bit platforms.
>
> [...]

This doesn't seem right. It's overly clever. It's like saying that

4.0 + 1

should return int(5) and not float(5.). Or perhaps a closer analogy would  
be:

if we're in a 64-bit platform (only then can integers have a larger  
precision than doubles);
if one of the operands is exactly an integer;
if the other operand is a float exactly representable as an integer  
(perhaps a large power of 2), but its accuracy is negative (where accuracy  
is the 15.9546 - log10(abs(x)), or, informally, the effective number of  
digits to the right of the decimal point);
if the result of the sum is exactly representable as a 64-bit integer;
then return such integer.

It's easy to see these sort of tricks would have to be applied all over  
the place (additions, products, and all other mathematical operations).

In sum, I think the rule that if a mathematical operation involves a  
float, the result should be a float ought to be kept.

-- 
Gustavo Lopes