Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:18825
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Message-ID: <4326CAC3.3080708@lerdorf.com>
Date: Tue, 13 Sep 2005 05:49:07 -0700
User-Agent: Mozilla Thunderbird 1.0.6 (Macintosh/20050716)
MIME-Version: 1.0
To: pierre.php@gmail.com
CC: internals <internals@lists.php.net>
References: <432534E4.9020602@lerdorf.com> <fe05d154050912080829ded89@mail.gmail.com>
In-Reply-To: <fe05d154050912080829ded89@mail.gmail.com>
Content-Type: text/plain; charset=ISO-8859-1
Content-Transfer-Encoding: 7bit
Subject: Re: [PHP-DEV] ref fix revisited
From: rasmus@lerdorf.com (Rasmus Lerdorf)

Just got back from Sri Lanka and am catching up on email.  But in
general I don't agree with your statement that things that are quiet in
5.0.4 must be quiet in 5.0.5.  Quietly corrupting memory doesn't really
work.  Sometimes we need to break things slightly to fix things.  My
message was more about if there was a better way to fix it.  I looked at
the code and couldn't tell if it was somehow possible to catch this on
the write instead of on the call.  It sounds like that isn't possible.
The second approach to try to regain some backwards compatibility is to
make some of our internal functions aware of the fact that they are
being passed a ref to a temp var and if so pretend it wasn't passed by
reference.

-Rasmus

Pierre Joye wrote:
> Hi Rasmus,
> 
> You start a new thread for my question,  maybe you will get more
> answers or at least better ones.
> 
> If it is a reference or not, I do not care, it should not "act"
> differently. Understand that a quiet code in 5.0.4 must be quiet in
> 5.0.5.
> 
> If we add notices in 5.1.0 and then make them fatal in 5.2.0, I'm
> fine. People will have time to migrate and "fix" their code.
> 
> Regards,
> 
> --Pierre
>