Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:25338
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
DomainKey-Status: good
DomainKey-Signature: a=rsa-sha1; q=dns; c=nofws;
        s=beta; d=gmail.com;
        h=received:message-id:date:from:to:subject:cc:in-reply-to:mime-version:content-type:content-transfer-encoding:content-disposition:references;
        b=a5aP4laNYGnInVOod4oD70N2XIHxLOy3ryyRHUiKDKElHZDIshE/xnFCZBn5Z/zlin+rQafPF7iCLliVoumPpJmz6653HTtFJHBH4HVtkSdVAILQXfNxExSqHfRM7xK+6BYufiDvNUqDCGzy/Em3e7W4tlmlTzeF33+LravWiR4=
Message-ID: <fe05d1540608130508n73b36be8qefc087f56898448c@mail.gmail.com>
Date: Sun, 13 Aug 2006 14:08:26 +0200
To: "Matt W" <php_lists@realplain.com>
Cc: internals@lists.php.net
In-Reply-To: <005301c6bec7$783e5c80$0201a8c0@pc1>
MIME-Version: 1.0
Content-Type: text/plain; charset=ISO-8859-1; format=flowed
Content-Transfer-Encoding: 7bit
Content-Disposition: inline
References: <005301c6bec7$783e5c80$0201a8c0@pc1>
Subject: Re: [PHP-DEV] [PATCH] array_count_values bug: mishandling of numbers with leading whitespace
From: pierre.php@gmail.com (Pierre)

Hello,

On 8/13/06, Matt W <php_lists@realplain.com> wrote:
> Hi there,
>
> I didn't know whether to make an official bug report or send to the list --
> so I'm trying this first. :-)

You already tried in the previous thread about is_numeric.

> var_dump(array_count_values(array('  001', 1, '  1  ', '1')));
>
> Expected result (and what PHP 4 gives):
> array(3) {
>   ["   001"]=>
>   int(1)
>   [1]=>
>   int(2)
>   ["  1  "]=>
>   int(1)
> }
>
> Actual result:
> array(2) {
>   [1]=>
>   int(3)
>   ["  1  "]=>
>   int(1)
> }
> Can this patch please be applied for 5.2's release?  To maintain what I'm
> sure is the correct behavior from PHP 4, the function needs to use the
> zend_[u_]symtable* functions, which take care of *correctly* handling
> numeric strings, as zend_hash_[find|update] did in 4.x.  If they would've
> been used in 5+ in the first place, there wouldn't have been bugs like
> #34723 (still present with leading whitespace, as you can see), #30833,
> #29808, etc. (I guess that's all, actually). :-)
> The updated code should be faster, too...

As I said in the previous discussion (why in the world have you
splitted it out?), I don't think it is reasonable or safe to apply
this patch in 5.2. Secondly it seems to wrong address the problem. You
work around the actual is_numeric "problems" by suppressing its usage.
 Doing so reintroduce the numeric string keys problem (what addresses
the removed code).

> http://realplain.com/php/array_count_values_bug.diff
> http://realplain.com/php/array_count_values_bug_5_2.diff

I think it is much more safer to first determine the general policy
for is_numeric and then address this problem. It is too late in the
game to change such behaviors in 5.x (I care little about 4.x
compatibility).

Cheers,
--Pierre