Last modified: 2010-05-15 15:33:22 UTC
Installed apps: - Apache 2.0.50 - PHP 5.0.2 - Turck MMCache 2.4.7-CVS on a Mandrake 10.1 I get this error on "unlistedspecialpages" class call: [client 172.19.10.50] PHP Fatal error: Class 'unlistedspecialpage' not found in /var/www/html/mediawiki-1.4.0/includes/SpecialPage.php on line 26, referer: http://intranet.lan.lord-of-rings/mediawiki/index.php/User:172.19.10.50 Once Turck MMCache is disabled, everything works fine.
Same Problem here. A workaround *should* be to set mmcache.filter = "!*SpecialPage.php" in 99_mmcache.ini, but for some Reasons it doesn't work. I hope someone can fix this soon as mmcache is *really* good in speeding up mediawikis...
Also, have you tried eAccelerator? I've the impression that Turck MMCache isn't really well tested with PHP 5, and it's no longer being maintained since the last release in late 2003.
I tried installing php5-eAccelerator using Mandrake 10.2 wich fails to load: PHP Warning: PHP Startup: Unable to load dynamic library '/usr/lib/php5/extensions/eaccelerator.so' - /usr/lib/php5/extensions/eaccelerator.so: undefined symbol: php_sprintf in Unknown on line 0 Maybe self-compiling eAccelerator might help...
(In reply to comment #3) > I tried installing php5-eAccelerator using Mandrake 10.2 wich fails to load: I meant Madnrake 10.2 rpms actually
I wasn't able to install eAccellerator but I tried this on a Mandrake box with php4 and it works fine with php-mmcache, so I guess using eAccelerator *should* solve this problem for php5-users as this seems to be a problem with mmcache and php5.
Created attachment 559 [details] Replace mmcache with eAccelerator This patch replaces mmcache code with code for eAccelerator. Because mmcache wasn't maintained anymore we further develop it in the eAccelerator project. It's the code that's in mmcache cvs but with a lot of bugfixes at this moment. Mmcache has some basic php5 patches in cvs but eAccelerator is almost fully compatible with php5 (only the caching not yet encoding).
Why is that patch posted on this bug? Not only is that a separate bug report, it's actually already been done. Anyway, resolving as INVALID since the reported problem was caused by using an unsupported combination of versions.
