5.0.5 RC2

php.internals

Zeev Suraski

21 years ago
Any outstanding things for this release beyond PEAR and XMLRPC? I'd like to roll RC2 as soon as possible. Thanks, Zeev

Sönke Ruempler

21 years ago
Hi, Zeev Suraski <mailto:zeev@zend.com> wrote on Wednesday, August 24, 2005 9:50 AM:
> Any outstanding things for this release beyond PEAR and XMLRPC? I'd > like to roll RC2 as soon as possible.
Some hint: Some applications get broken after upgrade from PHP 5.0.4 to 5.0.5RC1. For example the MediaWiki software: Fatal error: Only variables can be passed by reference in /xxx/www/wiki/includes/SpecialSpecialpages.php on line 68<: eeeeaeaeciaobsoeeeeee I thought these reference-`fixes' would only affect PHP 5.1.x and PHP 4.4.x branches? Because this is IMHO a BC break. Just to let you know :)
-- Regards.

Sebastian Bergmann

21 years ago
Zeev Suraski schrieb:
> Any outstanding things for this release beyond PEAR and XMLRPC? I'd > like to roll RC2 as soon as possible.
libpcre needs to be bumped.
-- Sebastian Bergmann http://www.sebastian-bergmann.de/ GnuPG Key: 0xB85B5D69 / 27A7 2B14 09E4 98CD 6277 0E5B 6867 C514 B85B 5D69

David Zülke

21 years ago
Yeah, http://www.securitytracker.com/alerts/2005/Aug/1014744.html - David

Andrei Zmievski

21 years ago
It's already been bumped. -Andrei On Aug 24, 2005, at 1:39 AM, Sebastian Bergmann wrote:

Jani Taskinen

21 years ago
Yeah, don't release it and push 5.1 instead. Nobody should be using 5.0 anymore after 5.1 is out.. --Jani On Wed, 24 Aug 2005, Zeev Suraski wrote:

David Zülke

21 years ago
Uh-huh, mister know-it-all... the only problem is that 5.1 isn't out yet. So why don't you just put a sock in it and help pushing 5.1 instead? - David

Greg Beaver

21 years ago
Zeev Suraski wrote:
> Any outstanding things for this release beyond PEAR and XMLRPC? I'd > like to roll RC2 as soon as possible.
PEAR is not an issue (that's what my message about finishing sync was all about) Greg

Zeev Suraski

21 years ago
At 11:33 24/08/2005, =?us-ascii?Q?Sonke_Ruempler?= wrote:
>Hi, > >Zeev Suraski <mailto:zeev@zend.com> wrote on Wednesday, August 24, 2005 9:50 >AM: > > > Any outstanding things for this release beyond PEAR and XMLRPC? I'd > > like to roll RC2 as soon as possible. > >Some hint: Some applications get broken after upgrade from PHP 5.0.4 to >5.0.5RC1. > >For example the MediaWiki software: > > Fatal error: Only variables can be passed by reference in > /xxx/www/wiki/includes/SpecialSpecialpages.php on line 68<: >eeeeaeaeciaobsoeeeeee
It appears to be due to a genuine fix for bug #31525. More of the same 'Only variables can be * by reference' that we're used to. Zeev

Sönke Ruempler

21 years ago
Hi Zeev, Zeev Suraski <mailto:zeev@zend.com> wrote on Thursday, August 25, 2005 7:37 PM:
>> Fatal error: Only variables can be passed by reference in >> /xxx/www/wiki/includes/SpecialSpecialpages.php on line 68<: >> eeeeaeaeciaobsoeeeeee > > It appears to be due to a genuine fix for bug #31525. More of the > same 'Only variables can be * by reference' that we're used to.
Uhm, if it were a E_NOTICE - no problem. But this breaks PHP 5.0.x and therefore many existing appliations (I assume mediawiki is one of the most popular and famous of PHP applications). If this new 'fatal error' behaviour gets into PHP 5.0.5, you'll have to warn the users that their applications could break, because something like nested array function calls is IMHO very common. Correct me if I'm wrong. Thx. soenke

Derick Rethans

21 years ago
On Fri, 26 Aug 2005, Sonke Ruempler wrote:
> Hi Zeev, > > Zeev Suraski <mailto:zeev@zend.com> wrote on Thursday, August 25, 2005 7:37 > PM: > > >> Fatal error: Only variables can be passed by reference in > >> /xxx/www/wiki/includes/SpecialSpecialpages.php on line 68<: > >> eeeeaeaeciaobsoeeeeee > > > > It appears to be due to a genuine fix for bug #31525. More of the > > same 'Only variables can be * by reference' that we're used to. > > Uhm, if it were a E_NOTICE - no problem. But this breaks PHP 5.0.x and > therefore many existing appliations (I assume mediawiki is one of the most > popular and famous of PHP applications). If this new 'fatal error' behaviour > gets into PHP 5.0.5, you'll have to warn the users that their applications > could break, because something like nested array function calls is IMHO very > common.
It's just PHP prevent your memory from getting corrupted. If you prefer that to happen, patch PHP yourself for your local installation. Derick

Pierre-Alain Joye

21 years ago
On Fri, 26 Aug 2005 09:12:51 +0200 (CEST) derick@php.net (Derick Rethans) wrote:
> On Fri, 26 Aug 2005, Sonke Ruempler wrote: > > > Hi Zeev, > > > > Zeev Suraski <mailto:zeev@zend.com> wrote on Thursday, August > > 25, 2005 7:37 PM: > > > > >> Fatal error: Only variables can be passed by reference in > > >> /xxx/www/wiki/includes/SpecialSpecialpages.php on line 68<: > > >> eeeeaeaeciaobsoeeeeee > > > > > > It appears to be due to a genuine fix for bug #31525. More > > > of the same 'Only variables can be * by reference' that we're > > > used to. > > > > Uhm, if it were a E_NOTICE - no problem. But this breaks PHP > > 5.0.x and therefore many existing appliations (I assume > > mediawiki is one of the most popular and famous of PHP > > applications). If this new 'fatal error' behaviour gets into > > PHP 5.0.5, you'll have to warn the users that their > > applications could break, because something like nested array > > function calls is IMHO very common. > > It's just PHP prevent your memory from getting corrupted. If you > prefer that to happen, patch PHP yourself for your local > installation.
Well, enough of this argument. This is a real problem. Having a fatal error instead of a) nothing or b) a notice is a BC break. Moving from nothing to a notice is acceptable, but having fatal erros is not. Btw, asking one to maintain a patched php to keep BC is insane. Is there any tests about this memory corruption problem? --Pierre

Derick Rethans

21 years ago
On Fri, 26 Aug 2005, Pierre-Alain Joye wrote:
> Is there any tests about this memory corruption problem?
Impossible, as it is an internal thing that doesn't really always show when zend memory manager is enabled AND valgrind is used. Although there should be one test for it with strange behavior. Derick