* put a @ in front of get_headers() should hopefully avoid errors where allow_url_fopen=0
* combined $wgPhpCliPath into the similar used $wgPhpCli var
* minor js & example file mwEmbed fixes
* Remove all $wgAllowRealName, check $wgHiddenPrefs instead
* Make $wgAllowUserSkin obsolete too, check $wgHiddenPrefs instead
* And also only call globals if really needed, and remove unused globals
This is a global search and replace of NS_IMAGE and NS_IMAGE_TALK with NS_FILE and NS_FILE_TALK respectively in all core files, excluding those already updated in step 1 (r44004).
* Switch from running fix_magic_quotes() on $_COOKIE and $_GET/$_POST to running it on $this->cookies and $this->data. Should keep us from interfering with other programs that might do the same (and/or trying to start up a second WebRequest object). This partially fixes bug 11558.
* Todo: Do similar things with $_SERVER/$_ENV and switch to a lazy-load style, rather than on every new WebRequest.
I don't really like this in general; the API isn't meant for the UI and there should be little to no call to link to it from body content.
Additionally, I believe we're trying to move all new parser functions to the convention of using the # prefix to avoid conflict with the template namespace.
* $wgCookiePrefix:
** Shared stuff should always come before local stuff when setting the cookie prefix (by my count, if you set $wgSharedDB and $wgDBprefix with the old order, even though your usertable is shared, your cookies would be local and logins wouldn't be shared right.
** Because the user table isn't always shared anymore, we only want shared code when $wgSharedTables has 'user' set inside of it.
** We now have $wgSharedPrefix, this should also be accounted for in the cookie prefix.
* updaders.inc: The logic for the $doUser test has changed since the conditions that the user table is shared has changed. Now when $shared is true we only want to $doUser when we have a shared database and 'user' is being shared, otherwise if $shared is false, we only want to $doUser when there is no shared database, or 'user' is not being shared.
Check MediaWiki.org's $wgSharedDB documentation for those on 1.12, I'll post a link there to a patch I maintain for 1.12.
* Wrote a tool to check the integrity of the autoloader class list, fixed some issues that came up.
* Start the autoloader before LocalSettings.php, so that when an extension writer thinks an inefficient one-file special page extension is the way to go, they don't have to use explicit includes to make the class inheritance work. Should continue to work with $IP set in LocalSettings.php as long as $IP is set before extensions are included.
* Allow configuration of the default limit and days links with the variables $wgRCLinkLimits and $wgRCLinkDays.
* When selecting a limit and a days on the special pages, if you use a number not within the default list it will be added to the display.
** The selected limit and days is bolded. (This can be considered the first step for bug 3327)
* Optional other configuration option $wgRCFilterByAge, when enabled the list of days will be filtered so that only links within what $wgRCMaxAge says will be stored in the database will be displayed. Note that this is disabled by default because MediaWiki doesn't quite seam to trim out old entries as quickly as $wgRCMaxAge says it should be and some people appear to use the extra 30days links to display older entries even though $wgRCMaxAge is set to 7days on said wiki.
This should also solve bug 9257.
* Wrote two concrete implementations. LBFactory_Simple is for general installations. LBFactory_Multi will replace the runtime configuration used on Wikimedia and allow load-balanced connections to any DB.
* Ported Special:Userrights, CentralAuth and OAI audit to the LBFactory system.
* Added ForeignDBViaLBRepo, a file repository which uses LBFactory.
* Removed $wgLoadBalancer and $wgAlternateMaster
* Improved the query group concept to allow failover and lag control
* Improved getReaderIndex(), it will now try all servers before waiting, instead of waiting after each.
* Removed the $fail parameter to getConnection(), obsolete.
* Removed the useless force() function.
* Abstracted the replication position interface to allow for future non-MySQL support.
* Rearranged Database.php. Added a few debugging features.
* Removed ancient benet-specific hack from waitForSlave.php
* Made it possible to configure the parser class being used, via $wgParserConf.
* Moved defines from the top of Parser.php to either class constants or Defines.php
* Added Parser_DiffTest, a differential parser class for regression testing
* Added Parser_OldPP, a parser class which operates like the parser before this commit. I made one breaking change: a bugfix to avoid losing whitespace when adding MWTEMPLATESECTION markers.
* Made internal tidy work with PHP 5
* Added the ability to supply a hook for template fetching via ParserOptions. This is handy for testing.
* Updated parserTests.txt to account for the various breaking changes I made. Removed a few parser tests that no longer test for anything useful.
*Allow blocking of users to hide names
*Implement revision deletion for images/deleted files/deleted revs
*Log deletion set off for now
*Add 'hidden' file dir
*Dissallow merging via undelete (which was inefficient and hard to reverse)
*Use restore points and diffs to special:undelete
*Add a special page to merge pages
*Get changeslist to use tables to avoid ugly formatting
*Add logs into RC for rebuildrecentchanges.php
*Add private logs
*List private logs at specialpages
*Tweak/add some deletion and merge messages
* Ported file delete/restore to the filerepo framework. Some user-visible changes in error reporting.
* $wgSaveDeletedFiles has been removed, the feature is now enabled unconditionally. Added a "deleted" directory for the default location, protected by a .htaccess file and the practical obscurity of content hashes.
* Fixed bug 2735: "Preview" shown in title bar for action=submit on special pages
* Removed "restore" links from the deletion log embedded in Special:Undelete
* Added img_sha1/oi_sha1 fields, preserved through upload, delete and restore
* Referenced the new oi_metadata etc. fields to preserve metadata across upload and delete/restore.
* Rename 'licence' to 'license' for consistency with the rest of the code.
* The preview code was assuming simple templates with no parameters and no subst:s, which failed on most of the selections used now on Commons. Now doing a full pre-save transform and parse on {{$license}}, just as will be done on the final save.
* When selecting 'None' again after another option, the preview pane is now cleared.
A fun 'todo' might be to also pass the currently selected filename, if any, to the license preview. Some of the templates in use attempt to use the current-page-name variables to include a self-link. At the moment using a hardcoded 'Image:Sample.jpg' for the virtual title on the preview rendering.
