PHP documentation on $_REQUEST says it includes cookies by default,
but PHP documentation on request_order actually says it does not.
The request_order documentation is the right documentation.
Allows site to auto-detect language based on HTTP Accept-Language header.
Uses a lightly modified version of the parser from
http://www.thefutureoftheweb.com/blog/use-accept-language-header.
Also cleans up unnecessary direct use of $_COOKIE and $_GET, which
are both covered by $_REQUEST, and uses a single language fallback
instead of a default + fallback.
As mentioned in f3c2203e2f,
the language switching was always 1 behind. As in, the
language would be the last language you selected.
The first problem is that some of the smarty config variables
were being set after the call to _smarty->config_load(). Since
the smarty instance is cached, it was using the last set of variables.
I don't know why the production web site isn't throwing errors. My
personal set up wouldn't load until I could set _smarty->compile_dir
the first time. Perhaps the language update was done on the fly, so
smarty was still cached.
The next potential problem is persistence with $_SESSION.
The session is never initialized with session_start(). However,
the production php.ini may have session auto start enabled.
The site code accesses a lot of uninitialized variables, which fills
the screen with notices on my system. Disabling notices by default is
probably best for the current code base
Without this fix we wouldn't be able to display the exception if MenuModel was the cause since it's used in the Controller, which ExceptionsPage inherits from.
svn path=/web/trunk/; revision=40948
* For each news item there is now a dedicate page showing just that item
* The title bar of each news items links to its dedicated page
* The Atom and RSS feeds make use of this new item specific URL
* The Atom feed now has the scummvm front page as an alternate link
svn path=/web/trunk/; revision=24983
* Removed the 'Posted by' text from the <author> field
* Drop fake email adress in the ATOM feeds (the same is not possible for the RSS feeds, though)
* Only add an author field if an author was specified in the first place
* Drop the CDATA encoding for the <title> field
svn path=/web/trunk/; revision=24982
Shots", "ScreenShots" and "ScreenShot Viewer", but someone else will have to
take care of rs-top.png. And of course, I may have missed some.
svn path=/web/trunk/; revision=22484