Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:100084
Mailing-List: contact internals-help@lists.php.net; run by ezmlm
Received-SPF: softfail (pb1.pair.com: domain golemon.com does not designate 74.125.82.52 as permitted sender)
MIME-Version: 1.0
Sender: php@golemon.com
In-Reply-To: <422D7E05-864E-40EB-AEEB-81DFED80CA72@gmail.com>
References: <a9bfbaf5-6c48-ee69-62ab-b59daef3328e@rhsoft.net>
 <28093140-0D0D-4157-AAA3-E8F1A570E67A@gmail.com> <ebde09b9-c883-36ef-3727-fac2ec009e6c@rhsoft.net>
 <422D7E05-864E-40EB-AEEB-81DFED80CA72@gmail.com>
Date: Fri, 28 Jul 2017 10:45:29 -0400
Message-ID: <CAESVnVoGGitKwy+tK55jXSii=hKF8rFDjMDGGO3SNi=XDTezag@mail.gmail.com>
To: Rowan Collins <rowan.collins@gmail.com>
Cc: PHP internals <internals@lists.php.net>, "lists@rhsoft.net" <lists@rhsoft.net>
Content-Type: text/plain; charset="UTF-8"
Subject: Re: [PHP-DEV] session_start() should not reset $_SESSIOn if it's not empty
From: pollita@php.net (Sara Golemon)

On Fri, Jul 28, 2017 at 10:11 AM, Rowan Collins <rowan.collins@gmail.com> wrote:
> On 28 July 2017 14:37:10 BST, "lists@rhsoft.net" <lists@rhsoft.net> wrote:
>>(Arguably, all the other superglobals should be read only for the same
>>reason, but that would be a huge break now.)
>>
>>make them readonly would break my whole codebase
>
> Yes, I only meant that as an absolutely hypothetical "if I had a time machine and could design it a different way...", I realise it would break everything to change it now.
>
>
ftr; I'd vote in favor of several BC breaking things to do with
autoglobals, among them:

* Make them objects (though ArrayAccess based for less hostile BC breakage)
* Make most of them read-only (offsetGet(), but no offsetSet)
* Make $_SESSION[...] access produce an error or auto-start the session

I've seen too many codebases abuse GPCER vars as a generic storage
location because "globals are bad, but this is good because it doesn't
include the word global".  As a performance issue, the runtime has to
assume autoglobals are inherently volatile and could change on a whim
at any moment (much like $http_response_headers).  Restricting their
mutability would be a win.  The request globals could probably also be
optimized fairly significantly.

If anyone agrees, I'm willing to RFC it.  If not, I'll continue living
with it. :D

-Sara