Ian Baker [Wed, 14 Sep 2011 17:23:09 +0000 (17:23 +0000)]
Fixed incorrect userId reference
Aaron Schulz [Wed, 14 Sep 2011 17:15:07 +0000 (17:15 +0000)]
Added @todo items
Roan Kattouw [Wed, 14 Sep 2011 16:48:29 +0000 (16:48 +0000)]
Introduce Language::getMessageKeysFor() and use it in ApiQueryAllmessages
Sam Reed [Wed, 14 Sep 2011 15:07:20 +0000 (15:07 +0000)]
Whitespace/documentation
Roan Kattouw [Wed, 14 Sep 2011 13:43:06 +0000 (13:43 +0000)]
Revert r96715 per committer's (Chad's) CR comment
Sam Reed [Wed, 14 Sep 2011 12:32:22 +0000 (12:32 +0000)]
Followup r86304, guard against $title AND $wgTitle being null
Add profiling too
Niklas Laxström [Wed, 14 Sep 2011 11:28:03 +0000 (11:28 +0000)]
Revert part of r97046 which was not intented to be committed.
As side note, would be nice to have this configurable someway.
Niklas Laxström [Wed, 14 Sep 2011 11:23:58 +0000 (11:23 +0000)]
Revert r96798 - makes parser go infinite loop somewhere
Niklas Laxström [Wed, 14 Sep 2011 10:43:50 +0000 (10:43 +0000)]
Removed some more unused message made obsolete by r96847 (one was still referenced in r97044, but was already unused)
Niklas Laxström [Wed, 14 Sep 2011 10:39:58 +0000 (10:39 +0000)]
Added a bit defensive programming, as suggested by Aaron in r97040
Niklas Laxström [Wed, 14 Sep 2011 10:36:24 +0000 (10:36 +0000)]
RevisionDeleter:
* Fixed comment of getChanges and made it public
* Removed getLogMessage and expandMessageArray, now unused
LogFormatter:
* Cache parsed parameters locally, since this function is going to be called few times for every log entry
DefaultSettings:
* Could not find any use of suppress/file log
LogPage:
* Removed old formatting code, now unused
WikiPage:
* Converted (suppress|delete)/delete logs to new format
Added DeleteLogFormatter, handles part of suppression log as well.
Updated messages.
This also addresses bug 18578.
Niklas Laxström [Wed, 14 Sep 2011 10:00:30 +0000 (10:00 +0000)]
Fix variable name typo in r96847
Niklas Laxström [Wed, 14 Sep 2011 08:32:52 +0000 (08:32 +0000)]
Chrome seems to already support dir=auto, and I see no downsides using it for user comments.
Amir said that it is good.
Niklas Laxström [Wed, 14 Sep 2011 08:30:11 +0000 (08:30 +0000)]
There was a double space between log action text and user comment. Remove the space produced by linker.
Niklas Laxström [Wed, 14 Sep 2011 07:49:39 +0000 (07:49 +0000)]
Instead of throwing fatals, accept null titles in old log entries.
New log entries will still die with fatal few lines below, but that's good since that should never happen.
Niklas Laxström [Wed, 14 Sep 2011 07:47:32 +0000 (07:47 +0000)]
Drop JSON support. It was never used and causes problems like bug 30854 since PHP managed to screw up json_decode too.
It was not a good idea to autodetection of the format anyway, but it isn't such a problem with unserialize.
Chad Horohoe [Wed, 14 Sep 2011 03:31:53 +0000 (03:31 +0000)]
Partial revert r97035, followup r96930: make recentchanges tests pass again
Krinkle [Wed, 14 Sep 2011 01:37:13 +0000 (01:37 +0000)]
Renaming SpecialRecentchangesTest file to match conventions
* There should be no two files with the same name
* Test suite files should be named after the main file containing the code being tested followed by "Test"
Sean Colombo [Wed, 14 Sep 2011 00:42:36 +0000 (00:42 +0000)]
Merged wgWhatlinkshereLimit from Wikia. It defaults to the same value it was hardcoded to before.
Sean Colombo [Wed, 14 Sep 2011 00:27:48 +0000 (00:27 +0000)]
Merged Wikia change.
Sean Colombo [Wed, 14 Sep 2011 00:19:32 +0000 (00:19 +0000)]
Merged in mime-type from Wikia (for Sawfish Window Manager).
Aaron Schulz [Tue, 13 Sep 2011 23:24:34 +0000 (23:24 +0000)]
Possible fix for issue reported in r83762. I had a hard time confirming the problem/fix with profiling. Committing this now to get more attention.
Roan Kattouw [Tue, 13 Sep 2011 20:36:24 +0000 (20:36 +0000)]
Followup r96978, clean up code duplication by factoring out the code building load.php requests into ResourceLoader::makeLoaderURL() and makeLoaderQuery()
Aaron Schulz [Tue, 13 Sep 2011 20:26:41 +0000 (20:26 +0000)]
Removed parser hacks from r88740, r88741, r88780
Roan Kattouw [Tue, 13 Sep 2011 20:15:00 +0000 (20:15 +0000)]
Document the fact that some ResourceLoaderContext methods may return null
Raimond Spekking [Tue, 13 Sep 2011 19:32:43 +0000 (19:32 +0000)]
Localisation updates for core and extension messages from translatewiki.net
Siebrand Mazeland [Tue, 13 Sep 2011 17:44:25 +0000 (17:44 +0000)]
Consistency in wording of tool tips. Spotted by Gustrónico.
Roan Kattouw [Tue, 13 Sep 2011 17:13:53 +0000 (17:13 +0000)]
Fix the fixme on r88053: dependency handling was broken in debug mode in certain cases. More specifically, if A is a file module that depends on B, B is a wiki module that depends on C and C is a file module, the loading order is CBA (correct) in production mode but was BCA (wrong) in debug mode. Fixed this by URL-ifying scripts and styles for those modules in debug mode, as I said to on CR. What this means is that the initial debug=true request for a module will now always return arrays of URLs, never the JS or CSS itself. This was already the case for file modules (which returned arrays of URLs to the raw files), but not for other modules (which returned the JS and CSS itself). So for non-file modules, load.php?modules=foo&debug=true now returns some JS that instructs the loader to fetch the module's JS from load.php?modules=foo&debug=true&only=scripts and the CSS from ...&only=styles .
* Removed the magic behavior where ResourceLoaderModule::getScripts() and getStyles() could return an array of URLs where the documentation said they should return a JS/CSS string. Because I didn't restructure the calling code too much, the old magical behavior should still work.
* Instead, move this behavior to getScriptURLsForDebug() and getStyleURLsForDebug(). The default implementation constructs a single URL for a load.php request for the module with debug=true&only=scripts (or styles). The URL building code duplicates some things from OutputPage::makeResourceLoaderLink(), I'll clean that up later. ResourceLoaderFileModule overrides this method to return URLs to the raw files, using code that I removed from getScripts()/getStyles()
* Add ResourceLoaderModule::supportsURLLoading(), which returns true by default but may return false to indicate that a module does not support loading via a URL. This is needed to respect $this->debugRaw in ResourceLoaderFileModule (set to true for jquery and mediawiki), and obviously for the startup module as well, because we get bootstrapping problems otherwise (can't call mw.loader.implement() when the code for mw.loader isn't loaded yet)
Roan Kattouw [Tue, 13 Sep 2011 14:36:15 +0000 (14:36 +0000)]
Fix typo in error message
Roan Kattouw [Tue, 13 Sep 2011 13:55:43 +0000 (13:55 +0000)]
Address one of the fixmes on r88053: because addScript() is asynchronous, calling it repeatedly can lead to scripts loading out of order. The specific bug was that if you had a module with two script files, A and B (in that order; scripts are always loaded in the order in which they are specified for multi-file modules) and A is 40M while B is 1K, then B would be executed before A if the module is loaded after document ready (through e.g. mw.loader.using) and the files are not in the browser cache.
So instead of calling addScript() repeatedly in a loop, this commit calls addScript() recursively from its own callback, using a helper function called nestedAddScript(). This also means the if block that checks if the scripts array is empty can be removed, because nestedAddScript() handles empty arrays correctly.
Sam Reed [Tue, 13 Sep 2011 12:37:05 +0000 (12:37 +0000)]
Fix documentation, document return type
Alexandre Emsenhuber [Tue, 13 Sep 2011 12:15:23 +0000 (12:15 +0000)]
* Use local context instead of global variables
* Pass the context to the HTMLForm object
Tim Starling [Tue, 13 Sep 2011 11:35:04 +0000 (11:35 +0000)]
Move MessagesEnRTL.php to MessagesEn_rtl.php so that it can be accessed with the more sensible language code "en-rtl". This code is not in $wgLanguageNames and so cannot be selected in the UI. This change is to support a debugging extension which I will commit shortly.
Alexandre Emsenhuber [Tue, 13 Sep 2011 11:10:05 +0000 (11:10 +0000)]
Call Linker methods statically
Sam Reed [Tue, 13 Sep 2011 10:59:40 +0000 (10:59 +0000)]
Fix revert artefact of r96930
Ian Baker [Tue, 13 Sep 2011 00:53:30 +0000 (00:53 +0000)]
If the scaler URL is protocol-relative, things just break here. Sets a sane default.
Chad Horohoe [Tue, 13 Sep 2011 00:19:04 +0000 (00:19 +0000)]
Revert r87992 and followups r87998, r89028 (Support abstraction for 'NOT IN' SQL structure). Per discussion on CR and elsewhere...we're not 100% sold on the new format yet.
Changing the database api like this should be carefully thought out before we get stuck with it for 6 more years and end up hating it.
Krinkle [Mon, 12 Sep 2011 22:18:33 +0000 (22:18 +0000)]
Use a username different from the main BlockTest.php, otherwise concurrency errors occur (as seen on Jenkins builds)
Raimond Spekking [Mon, 12 Sep 2011 21:47:32 +0000 (21:47 +0000)]
Localisation updates for core and extension messages from translatewiki.net
Robin Pepermans [Mon, 12 Sep 2011 21:26:08 +0000 (21:26 +0000)]
* Some documentation
* Use Title->isMainPage() in Article->view()
Aaron Schulz [Mon, 12 Sep 2011 21:15:09 +0000 (21:15 +0000)]
Updated r86144 code due to reverts in r96889 (specifically of Parser.php change in r86072)
Robin Pepermans [Mon, 12 Sep 2011 20:08:23 +0000 (20:08 +0000)]
Remove unused variable and update code comment per r96888.
Aaron Schulz [Mon, 12 Sep 2011 19:16:22 +0000 (19:16 +0000)]
Reverted r86072, r86419 per CR. Lots of conflicts resolved here. Removes lineStart from r87346 code as well for preprocess parser tests.
Brion Vibber [Mon, 12 Sep 2011 19:09:51 +0000 (19:09 +0000)]
* (bug 30865) Fix regression breaking category pages when there's no local page content
Output in CategoryViewer::getHtml() was trying to close an outside div, then re-open another one (with different dir/lang settings). Switched it to instead open a child div with the desired properties, so there's no danger of surprise inconsistency.
Alexandre Emsenhuber [Mon, 12 Sep 2011 16:29:21 +0000 (16:29 +0000)]
Turn ParserOptions::setSkin() into a no-op since mSkin is no more used
Sam Reed [Mon, 12 Sep 2011 15:17:16 +0000 (15:17 +0000)]
Add some documentation
Swap $this->file->name for $this->file->getName()
Niklas Laxström [Mon, 12 Sep 2011 14:58:41 +0000 (14:58 +0000)]
Start producing new style move log entries
Sam Reed [Mon, 12 Sep 2011 14:49:03 +0000 (14:49 +0000)]
Move RELEASE-NOTES from r96373 from RELEASE-NOTES-1.18 to RELEASE-NOTES-1.19 after it was merged in r96845
Niklas Laxström [Mon, 12 Sep 2011 14:46:17 +0000 (14:46 +0000)]
Already removed this support for very old move log entries from plain changes list, now also from enhanced.
Raimond Spekking [Mon, 12 Sep 2011 12:46:20 +0000 (12:46 +0000)]
* Move <small> styling from message to shared.css
* Wrap complete section into a div
* Wrap "Other resolutions:" into a span and remove it from printing. Useless on paper.
* Add plural to 'show-big-image-other'
Roan Kattouw [Mon, 12 Sep 2011 10:18:26 +0000 (10:18 +0000)]
Followup r94435: don't barf on jobs inserted before this revision that don't have a table param
Alexandre Emsenhuber [Mon, 12 Sep 2011 09:25:00 +0000 (09:25 +0000)]
Added visibility markers to some Linker methods
Niklas Laxström [Mon, 12 Sep 2011 09:11:54 +0000 (09:11 +0000)]
* (bug 30864) Use bengali numerals for <ol> for Assamese
Alexandre Emsenhuber [Mon, 12 Sep 2011 08:49:14 +0000 (08:49 +0000)]
* Use local context instead of global variables (where possible)
* Made getFormFields() non-static since the only call to it is in that file
Brian Wolff [Mon, 12 Sep 2011 02:10:11 +0000 (02:10 +0000)]
(bug 28545) When using the uca-default collation, sortkey's starting with a
space (U+20) will now sort under an invisible header like in 1.16 rather than a U+6DE (ARABIC START OF RUB EL HIZB)
Robin Pepermans [Mon, 12 Sep 2011 01:39:11 +0000 (01:39 +0000)]
SpecialSearch: add two hooks, one to modify the power search box, one to change the message pointing to the page title searched for
WikimediaIncubator:
* use these hooks and SpecialSearchSetupEngine to improve search usability
* use a different message for the language code in preferences, per suggestion of Amir
Robin Pepermans [Sun, 11 Sep 2011 21:07:17 +0000 (21:07 +0000)]
* r96737: set OutputPage->addWikiText() as an interface message by default
* r96760: use 'noflip' for $flip in OutputPage->addInlineStyle() so it's always a string
* Add some parameter documentation
Alexandre Emsenhuber [Sun, 11 Sep 2011 20:44:08 +0000 (20:44 +0000)]
Pass the context to the HTMLForm object
Brian Wolff [Sun, 11 Sep 2011 20:37:22 +0000 (20:37 +0000)]
(follow-up r96765). Appearently my release notes for that revision were horrible ;)
Brian Wolff [Sun, 11 Sep 2011 20:32:58 +0000 (20:32 +0000)]
(follow-up r96767) Make {{DEFAULTSORT:..|noerror}} and noreplace i18n-izable.
Hopefully i did this right. I've never added a magic word before...
Raimond Spekking [Sun, 11 Sep 2011 20:12:21 +0000 (20:12 +0000)]
Localisation updates for core and extension messages from translatewiki.net
Raimond Spekking [Sun, 11 Sep 2011 19:46:33 +0000 (19:46 +0000)]
Per r96723: register new message key to maintenance script
Robin Pepermans [Sun, 11 Sep 2011 17:56:27 +0000 (17:56 +0000)]
(Bug 30364) LanguageConverter should depend on the page content language instead of the wiki content language. Remains largely the same, except for extensions using the PageContentLanguage hook like Translate and WikimediaIncubator.
* Made variant tabs hidden on special pages (has no or sometimes wrong/inconsistent effect there; plus it's mainly in the user language)
* Made redirects be in content language again (remove from Title->getPageLanguage())
Robin Pepermans [Sun, 11 Sep 2011 14:52:29 +0000 (14:52 +0000)]
fix r96712 copy paste error :(
Leo Koppelkamm [Sun, 11 Sep 2011 09:16:14 +0000 (09:16 +0000)]
Adress fixme on r92054: Style print
Alexandre Emsenhuber [Sun, 11 Sep 2011 08:43:30 +0000 (08:43 +0000)]
* Use local context instead of global variables
* Call Linker methods statically
* Added missing call to outputHeader()
Alexandre Emsenhuber [Sun, 11 Sep 2011 08:14:46 +0000 (08:14 +0000)]
* Follow-up r96420: don't show e-mail links to anon users
* Fix for r24612 (!) spotted while fixing the above: parenthesis arround the bitwise check are needed since ! operator has higher precedence than &
Brian Wolff [Sun, 11 Sep 2011 02:20:20 +0000 (02:20 +0000)]
(bug 16755) Add syntax to DEFAULTSORT - {{DEFAULTSORT:Foo|noerror}} and {{DEFAULTSORT:Bar|noreplace}}
to in both cases prevent the "A sortkey already was specified earlier in the page, yadda yadda" message and in the second case, also do not replace the sortkey
if one was already specified.
Brian Wolff [Sun, 11 Sep 2011 01:13:08 +0000 (01:13 +0000)]
(bug 30722) Add an identity collation that sorts things based on what the unicode code point is (aka pre-1.17 behaviour).
I'm tagging this 1.18 because the original bug was for iswiktionary wanting it, so it'd be nice to get it in 1.18.
Robin Pepermans [Sun, 11 Sep 2011 00:57:32 +0000 (00:57 +0000)]
Fix standard skin, which doesn't style ul/ol. The margin added in shared.css (see r91518 and previous) made standard have both margin and default browser css. This resets that.
Robin Pepermans [Sun, 11 Sep 2011 00:13:18 +0000 (00:13 +0000)]
Do r93269 in a better way (also fixes some rare cases). Use 'margin/padding' instead of 'margin-left/padding-left' so it doesn't have to rely on the direction.
Robin Pepermans [Sat, 10 Sep 2011 23:51:08 +0000 (23:51 +0000)]
* Per Aaron on r96261, use string 'flip' instead of boolean
* Apparently @noflip should only be set once, otherwise it behaves differently
Krinkle [Sat, 10 Sep 2011 22:33:11 +0000 (22:33 +0000)]
mediawiki.Title.test: Add test for space collapsing
* When I first passed "Foobar___.js" to mw.Title and saw that it was stripping away (some) spaces, I thought it was a bug (since extension is part of the title and should be kept as is). Anyway, as it turns out it's not stripping them away because they are between 'title' and 'extension', it's a general rule that there should not two or more spaces next to each other in a title. Adding test to assert and confirm this behavior.
Krinkle [Sat, 10 Sep 2011 22:29:57 +0000 (22:29 +0000)]
mediawiki.log.js: Add support for variadic arguments (just like the native console does)
* Use unshift() when calling window.console.log (that way logging an object will not result in "prefix> [object Object]" but in "prefix> { key: .., .. }" (which most consoles do, similar thing for logging elements for which most console API implementations offer special features)
* Use string concatenation when using our own fabricated log. Did not use unshift for both, because in the latter that would show a bad comma ("prefix:>, foo, bar"). The native API does not show a comma between two logged variables, they're just shown next to each other.
Krinkle [Sat, 10 Sep 2011 22:18:39 +0000 (22:18 +0000)]
mediawiki.log.js: Refactor
* Add @author info etc. to the file header
* Merge var statements
* remove single quotes around object keys where not needed
* Changing bottom padding on <body> to be relatively added instead of absolutely (body might have a bottom padding already!)
* Add return statement after reaching out to window.console. No need for the rest of the function to be executed in that case. outdenting the rest after the new return statement back one tab (thus removing the else-case)
Sam Reed [Sat, 10 Sep 2011 22:09:31 +0000 (22:09 +0000)]
Simplify some contexts
Alexandre Emsenhuber [Sat, 10 Sep 2011 21:14:28 +0000 (21:14 +0000)]
Call Linker::titleAttrib() statically
Robin Pepermans [Sat, 10 Sep 2011 20:05:13 +0000 (20:05 +0000)]
Correctly parse interface messages as such, so it doesn't get converted according to the content language (like r96257/r94395/r94279)
Also whitespace fix in MessagesEn for r96723
Alexandre Emsenhuber [Sat, 10 Sep 2011 19:51:10 +0000 (19:51 +0000)]
Fix comment
Sam Reed [Sat, 10 Sep 2011 18:51:56 +0000 (18:51 +0000)]
Revert r96732
Sam Reed [Sat, 10 Sep 2011 18:46:13 +0000 (18:46 +0000)]
Invalidate user cache upon password change
Should help (maybe fix?) Bug 30850 - maintenance/changePassword.php does not purge (APC) cache
Chad Horohoe [Sat, 10 Sep 2011 17:16:41 +0000 (17:16 +0000)]
Add @since to the Deferred stuff, ping r96716
Robin Pepermans [Sat, 10 Sep 2011 17:12:35 +0000 (17:12 +0000)]
* Change the message 'yourvariant' to a more descriptive one (might be a bit too long however), and remove it as an optional message. Also added a help message to explain it.
* Make getVariantname fallback to the language name, and use that in Preferences. Shouldn't change the names in practice, but is a more logical implementation imho.
* (Fix a typo in Language.php)
Rotem Liss [Sat, 10 Sep 2011 13:04:07 +0000 (13:04 +0000)]
Localization update for he.
Alexandre Emsenhuber [Sat, 10 Sep 2011 11:15:15 +0000 (11:15 +0000)]
* Use local context instead of global variables
* Call Linker methods statically
Chad Horohoe [Sat, 10 Sep 2011 07:17:10 +0000 (07:17 +0000)]
Moved global define into class constant, replaced a bunch of $fname with __METHOD__.
Daniel Friesen [Sat, 10 Sep 2011 07:15:51 +0000 (07:15 +0000)]
Add some other ServicesJson tests.
Chad Horohoe [Sat, 10 Sep 2011 06:50:30 +0000 (06:50 +0000)]
Refactor deferrable updates into classes & interfaces, also add helper method for the most common use case:
$wgDeferredUpdateList[] = new HTMLCacheUpdate( $title, "sometable" );
I kept b/c with $wgDeferredUpdateList for now, but seeing as only 3 exts in svn use it (FileSearch, FlaggedRevs and WikiScripts), I'd like to deprecate it pretty soon :)
Chad Horohoe [Sat, 10 Sep 2011 06:47:37 +0000 (06:47 +0000)]
Add a comment about something I noticed while spelunking the code
Chad Horohoe [Sat, 10 Sep 2011 05:24:46 +0000 (05:24 +0000)]
Add some basic tests to compare output of native json support and the Services_Json implementation, as I promised on wikitech-l. Could use plenty more test cases, but that should be trivial since I did it properly with data providers :)
Robin Pepermans [Sat, 10 Sep 2011 04:09:20 +0000 (04:09 +0000)]
Conversion script between Tifinagh and Latin for the Tachelhit language
Preferences: made getLanguageNames include those that don't have a Messages file
Aaron Schulz [Fri, 9 Sep 2011 21:25:21 +0000 (21:25 +0000)]
Renamed Rev_List/Rev_Item classes to RevisionListBase/RevisionItemBase
Raimond Spekking [Fri, 9 Sep 2011 20:33:58 +0000 (20:33 +0000)]
Localisation update for core and extension messages from translatewiki.net
Brian Wolff [Fri, 9 Sep 2011 20:13:09 +0000 (20:13 +0000)]
(bug 30640; follow-up r92279) Rotating images was making skewed images
This is Bryan's patch from bug 30640 with a couple minor related changes, plus some unit tests.
This also addresses an issue with preventing too-big images from being scaled from r92279.
I also noticed that image magick's rotation support is broken in 6.3.7 (the version I had installed locally. I've since upgraded) I'm not sure if we should be doing something about that...
I did test this without both image magick, and gd (although only very briefly with gd) both seemed to work well. I didn't test any other image scalars.
Brion Vibber [Fri, 9 Sep 2011 18:43:37 +0000 (18:43 +0000)]
Followup r87986: qunit test case for bug 30825
jQuery's $.append() and friends go through some funky code paths which sometimes manually load <script> node contents via AJAX and eval them. That failed entirely in IE 7 if given a protocol-relative link ($wgServer = '//commons.wikimedia.org').
The fix in r87986 bypasses that by using DOM append directly on a script node, and thus fixes bug 30825 without even knowing it. :)
This test serves as a regression test.
Daniel Friesen [Fri, 9 Sep 2011 15:41:30 +0000 (15:41 +0000)]
Kill off $wg{Request,User,Lang,Output} usage within skins/ and redundant $this->skin local vars.
Daniel Friesen [Fri, 9 Sep 2011 15:40:19 +0000 (15:40 +0000)]
Followup r96667, I forgot the stylesheet change.
Daniel Friesen [Fri, 9 Sep 2011 15:37:20 +0000 (15:37 +0000)]
Add a class to the normal and hidden categories blocks. If non-positioning styles are going to be applied then they should be applied on classes, not ids. It's bad enough that we include a catlinks id in catlinks and make the process to attempt to have two catlinks blocks rediculous.
Siebrand Mazeland [Fri, 9 Sep 2011 14:43:56 +0000 (14:43 +0000)]
Tweak 'tog-oldsig' as suggested by Gustrónico.
Robin Pepermans [Fri, 9 Sep 2011 12:49:40 +0000 (12:49 +0000)]
Replace localisation cache check removed in r91432 with another check