[00:22:31] 03tparscal * r52877 10/trunk/phase3/skins/Vector.php: Resolved problem with classes being defined twice. [00:24:28] 03(FIXED) portlets have 2x class= when selected - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19573 +comment (10tparscal) [00:26:42] 03(mod) Internal server error on non-javascript sighting - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19562 (10JSchulz_4587) [00:26:56] 03dale * r52878 10/branches/new-upload/phase3/js2/mwEmbed/libEmbedVideo/embedVideo.js: cleanup of recommending firefox for any case where you have ogg video [00:36:00] 03aaron * r52879 10/trunk/phase3/includes/Skin.php: Fixed fatal error breakage (bug 19562) [00:36:21] 03(FIXED) Internal server error on non-javascript sighting - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19562 +comment (10JSchulz_4587) [00:37:28] *atglenn gives aa ron a cookie [00:37:37] (the flavor of your choice) [00:44:47] 03aaron * r52880 10/trunk/extensions/FlaggedRevs/FlaggedArticle.php: [00:44:47] *Use is_array() for tab check [00:44:47] *Fixed typo [01:00:13] 03aaron * r52881 10/trunk/extensions/FlaggedRevs/specialpages/RevisionReview_body.php: Treat unflagging of unflagged revision as success (bug 19545) [01:02:05] 03(mod) JS addPortletLink() should perhaps have vector specific code in order to be compatible - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19559 (10bawolff+wn) [01:03:10] 03aaron * r52882 10/trunk/extensions/FlaggedRevs/api/ApiReview.php: Short-circuited template stuff for API unapprove [01:14:59] 03(mod) "jump to" accessibility links jump to right place but with wrong names - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=17710 (10jidanni) [01:28:20] 03(NEW) Extension distributor not distributing UsabilityInitiative - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19575 enhancement; Normal; MediaWiki extensions: UsabilityInitiative; (Drilnoth) [01:55:49] Anyone around to help? I'm trying to configure my wiki with Short URLs, but the manual does not help my confusion. 8c [02:09:50] don't do it ;) [02:11:42] 03(mod) ogg files lost - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=5828 (10nospam) [02:34:27] what are some other ways to make mediawiki faster? Some reason if I refresh my page, its real slow. I have APC caching, should i install other caching methods too [02:48:31] RDove: file cache [02:49:13] !wg UseFileCache [02:49:13] --mwbot-- http://www.mediawiki.org/wiki/Manual:%24wgUseFileCache [02:49:46] squid or varnish would be faster than the file cache because it doesn't even have to touch your web server or php, but it's a bit of a pain to setup [02:50:49] can you have file cache and apc running @ one time? they different? [02:51:03] yeah, they're different [02:51:13] oh ok i think file caching disabled [02:51:13] squid/file cache cache rendered HTML [02:51:37] memcached could help too [02:52:00] if i dont specify a directory for file caching does it use a default or am i required to specify? [02:52:31] RDove: You are using the $wgMainCacheType (or one of its friends) configuration variable in addition to just enabling APC in the php configuration, right? [02:52:46] yeah [02:52:50] kk [02:52:52] and i feel like thats working well [02:53:59] my biggest complaint was if i browse through a few pages, every now and then the webbrowser got stuck on one [02:54:07] then i have to press f5 a few times to get it going again [02:54:35] i sometimes get that too for some reason [02:54:42] is $wgUseGzip = true; a good thing ? [02:55:19] wikipedia doesn't have that problem, wonder what is different [02:55:40] i have one server running this small wiki page... http://www.ronalddove.com is my page [02:55:49] i like wiki style for notes and stuff [03:10:11] 03(mod) JS addPortletLink() should perhaps have vector specific code in order to be compatible - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19559 (10roan.kattouw) [03:13:16] 03(mod) "jump to" accessibility links jump to right place but with wrong names - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=17710 (10roan.kattouw) [03:13:49] 03(mod) Extension distributor not distributing UsabilityInitiative - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19575 (10roan.kattouw) [03:24:49] Is there any reason for the use of the "name" attribute in tags in MediaWiki (Linker::makeHeadline) other than for Netscape 4.0 compatability? [03:27:59] witch hunter [03:28:22] witch! [03:28:29] *charitwo points at Splarka [03:28:54] 'cause I removed it and it actually made a noticeable difference on one of our pages, so. [03:29:31] (noticable as in size - admittedly, it was Russian, and the header names are all messed up in that due to restrictions on what you can put in an ID) [03:41:18] hmm, GreenReaper: IE 5-7 and Opera have some oddness about it [03:41:42] in rare cases, where a name="" of one object matches an id="" of another [03:42:14] is there a way to disable the table of contents for one page that has a bunch of = headings =? I am trying to format a news page with = DATA = but i don't want the navigation [03:42:21] i mean table of contents [03:42:31] __NOTOC__ [03:43:13] ty [03:43:54] Splarka: Can you see that being a factor in favour of not removing it? I mean, we shouldn't have a name matching ids in document the anyway, because then they would match a current name in the document too (the one currently being emitted for compatability) [03:44:22] not sure, since name="" attributes can be shared... [03:44:37] (in some circumstances anyway, and getElementsByName is an array) [03:44:47] *GreenReaper makes a ==top== just to cause trouble. [03:45:55] try something like
and
, and see which one getElementById('foo') returns in IE5-8 and Opera before 9.5 [03:46:13] though, I wonder what it would do with
and
[03:48:10] was wikipedia the first site to use mediawiki ever? is that what developers were using first to show off the work they did? [03:48:41] The software was developed for Wikipedia, yes. [03:48:53] but Wikipedia predated it [03:49:13] Yeah, this is just phase 3 (you don't want to see phase 1 and 2 :-). [03:49:15] http://en.wikipedia.org/wiki/UseModWiki [03:50:21] well, those were all retroactively named MediaWiki [03:50:24] sorta [03:50:58] if you say so [03:51:17] *Splarka snickers [03:51:37] (if people don't misinform the public, Tim gets too idle) [03:54:36] the fact that mediawiki is used on wikipedia contributed to mediawiki's popularity [03:54:47] but that wasn't the point, we just wanted to build an encyclopedia [03:54:56] cool [03:55:04] and if other people wanted to use the same software, that was fine [03:55:56] even when they come in and yell at you for not giving them support for CMS style features? [03:55:58] Then we started complaining about it. [04:03:56] you guys are talented for making the software, i enjoy playing with it [04:20:43] 03dale * r52883 10/trunk/extensions/MetavidWiki/ (2 files in 2 dirs): fixed inline play links for recent changes [04:21:36] 03dale * r52884 10/branches/new-upload/phase3/js2/mwEmbed/libEmbedVideo/ (embedVideo.js javaEmbed.js nativeEmbed.js): added oncanplaythrough play request for native [04:38:13] 03simetrical * r52885 10/trunk/phase3/includes/Skin.php: Fix weird calling convention for Skin::makeGlobalVariablesScript [04:52:35] 03mrzman * r52886 10/trunk/phase3/ (includes/DefaultSettings.php skins/common/wikibits.js): [04:52:35] Make importScriptURI() use a in all cases. [04:52:35] Previously it used document.createStyleSheet for IE (which appears to be the only browser that supports it) or @import for everything else [04:53:34] Splarka: ^ I forgot to mention you, noted it on codereview [04:54:11] 03(NEW) Remove the name attribute from header tags - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19576 enhancement; Normal; MediaWiki: Page rendering; (greenreaper) [04:58:13] MrZ-man: you also said "Script" instead of "Style", like I did... >_< [04:58:21] er, Stylesheet [04:58:32] :P [04:58:42] did you test it tho? I didn't ^_^ [04:58:52] yes, briefly [04:59:21] If it breaks, surely someone will complain. [04:59:23] oop brb [04:59:29] Hmm. Actually, I'm not entirely sure why the id isn't just on the h in this case. But I don't know whether other browser have issues with linking to IDs on elements. [05:00:42] its pretty much the same exact thing we use for importScriptURI, except with a instead of a [05:12:32] This was for FFx 2 so [05:12:52] would need oldMozilla. [05:13:33] http://hoaxblog.s3.amazonaws.com/jackodolia.jpg [05:15:24] 03(mod) Remove the name attribute from header tags - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19576 (10greenreaper) [05:36:04] 03(mod) Remove the name attribute from header tags - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19576 (10greenreaper) [07:31:33] 03(NEW) Save Button grey - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19577 enhancement; Normal; MediaWiki: Special pages; (markus.albert) [07:52:21] 03raymond * r52887 10/trunk/extensions/ (3 files in 2 dirs): [07:52:21] * Change 'delaydef_desc' to 'delaydef-desc' for consistency [07:52:21] * Add extension to translatewiki [08:12:49] 03tstarling * r52888 10/trunk/phase3/ (8 files in 4 dirs): (log message trimmed) [08:12:49] Implemented the PoolCounter feature and did some general refactoring in the areas that it touched. [08:12:49] * Renamed Article::outputFromWikitext() to Article::getOutputFromWikitext() [08:12:49] * Factored out cascade protection updates [08:12:50] * Removed recently-added Article::tryParserCache(): misnamed, can be done in one line of code in the caller. Deprecated OutputPage::tryParserCache(). [08:12:53] * Made some functions public instead of protected when they could be useful from hooks. [08:12:55] * In ParserCache, removed PHP 4-style ampersands [08:15:42] 03tstarling * r52889 10/trunk/extensions/PoolCounter/ (4 files): Client-side code for Domas's pool counter daemon (in python). [08:16:40] domas: I thought maybe we could put the server-side code there as well, along with any packaging stuff [08:22:09] 03siebrand * r52890 10/trunk/phase3/ (2 files in 2 dirs): [08:22:09] Follow-up to r52888: [08:22:09] * add new message to messages.inc [08:22:09] * rewording and trailing whitespace trim for 'view-pool-error' [08:30:02] TimStarling: probably :) [08:30:11] TimStarling: I wonder where master repo should be ;-) [08:30:54] with distributed VC it is somewhat more lax :) [08:38:57] 03siebrand * r52891 10/trunk/extensions/ (4 files in 2 dirs): [08:38:57] * PoolCounter update: extension credits, stylize.php, trailing whitespace removed, messages file header [08:38:57] * add support for PoolCounterClient to Translate [08:39:20] 03daniel * r52892 10/trunk/WikiWord/ (4 files in 3 dirs): analyzer test interface [08:39:32] TimStarling: PHP Strict Standards: Non-static method PoolCounter::factory() should not be called statically, assuming $this from incompatible context in /var/www/w/includes/Article.php on line 898 [08:42:26] 03tstarling * r52893 10/trunk/phase3/includes/PoolCounter.php: Added missing static keyword [08:42:58] domas: the client has a proposed "wait" command [08:43:12] TimStarling: I noticed, wanted to ask what it means [08:43:26] (and how to avoid race conditions there) [08:43:30] does it have to wait on stuff it has acquired? [08:43:51] does it wait for pool to shrink below a threshold, or does it wait for clean release? [08:44:05] TimStarling: thanks [08:44:26] it waits for the pool to shrink below a threshold, and also it has a server-side timeout [08:45:40] there's a problem there [08:46:02] more waiters = bigger pool [08:46:21] and if pool shrinks, should all waiters be signaled or just one [08:46:38] damn, I need to read a book about locking :) [08:47:01] mmm [08:47:13] I thought about release options [08:47:17] somebody know a great tutotrial for installation Extension:FlaggedRevs [08:47:21] 'broadcast' and 'signal' [08:48:01] this is why I somehow dropped this on monday and was looking for some inspiration afterwards :) [08:49:17] well, you can answer the problem of waiters being in the pool by just not having them be in it [08:49:42] right [08:49:45] as for signalling multiple waiters, I guess one will have to be signalled, then the process repeated after a timeout [08:50:35] same problems as a poolside cafe [08:50:44] no, no need for timeout [08:50:47] timstarling: the problem here is that if someone succeeded [08:50:52] the job may be not needed at all [08:50:57] so it may be useful to signal all [08:50:59] as long as the wait function acquires, then it's no problem [08:51:41] well, the way I've factored out Article.php currently, there's no way to recheck the parser cache [08:52:00] I guess I should fix that [08:53:30] say if the maximum is 5, and you have 5 in the pool, and some waiters [08:54:08] then when one of the clients in the pool releases, you can respond with a success value for one of the waiters, and simultaneously put it in the pool [08:54:16] so the pool count stays at 5 [08:55:08] well, and then... what to do if it was a failure [08:55:23] yeah, but waiters might have best result if they would be told that there was a success [08:55:35] which could cause stampede, if it wasn't [08:56:05] they could recheck the parser cache once the wait completes, then release their pool membership after a few milliseconds [08:56:26] not quite as fast as waking all of them [08:56:31] which would serialize [08:56:31] yes [08:56:45] could work, though [08:58:05] well, I can give release option [08:58:08] "broadcast" [08:58:14] and allow the client to determine [08:58:50] the other thing then [08:59:02] should "acquire" put into "wait" , in case of certain threshold? [08:59:41] not with the current client, because when acquire fails it checks the dirty cache [09:00:27] in theory we could broadcast some arbitrary data from release to all the waiters [09:00:28] right [09:00:33] like, say, the entire parser cache entry [09:00:46] well, I'm not worried about stampede hitting memcached [09:01:02] I'm worried about stampede consuming apaches [09:01:13] ok [09:01:47] memcached is cheaper than poolcounter too ;-) [09:02:16] we're going to need a limit on the number of waiters [09:02:45] it should start serving error messages if it gets too high, otherwise we'll DoS something [09:03:40] if there's 100 waiters, that's fine, if there's 10000, things will break [09:03:47] right [09:06:40] maybe instead of an error we should serve a refresh header, and a little animated GIF progress bar saying "please wait while the article is being rendered" [09:07:12] that would be kind of cute [09:12:10] http://www.ajaxload.info/cache/FF/FF/FF/59/61/92/8-0.gif [09:15:22] hehe :) [09:38:18] 03ialex * r52894 10/trunk/phase3/includes/api/ApiQueryAllmessages.php: follow-up r52858: remove unused "global $wgMessageCache;" [09:39:44] 03daniel * r52895 10/trunk/WikiWord/WikiWordBuilder/src/main/java/de/brightbyte/wikiword/ (7 files in 3 dirs): basic ResourceType detection for it, pt [10:30:19] 03(NEW) Vector could use more screen space - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19578 enhancement; Normal; MediaWiki: Vector Skin; (maxsem.wiki) [10:49:32] 03(mod) Database update for l10n_cache fails on non-EN installs - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19447 (10tstarling) [10:55:43] ugga [11:17:34] 03(mod) "jump to" accessibility links jump to right place but with wrong names - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=17710 (10jidanni) [11:22:56] 03(mod) "jump to" accessibility links jump to right place but with wrong names - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=17710 +comment (10jidanni) [11:24:49] 03(mod) "jump to" accessibility links jump to right place but with wrong names - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=17710 (10jidanni) [11:29:05] 03(mod) "jump to" accessibility links jump to right place but with wrong names - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=17710 (10jidanni) [11:34:04] 03(NEW) multi-thousand byte 1x1 pixel images. - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19579 trivial; Normal; MediaWiki: Vector Skin; (jidanni) [11:36:28] 04(REOPENED) vector skin hooks not documented in hooks.txt - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19522 +comment (10jidanni) [11:41:48] ugh [11:42:04] hey simpson, it wasn't use who put the link there [11:46:26] lol [11:47:09] 03(mod) Vector skin missing CSS for when background images are not displayed - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19538 +comment (10jidanni) [11:48:46] us* [11:50:24] so b#19579 is referring to spacer image files or something else? [11:52:18] PHP Strict Standards: Declaration of SkinStandard::getHeadScripts() should be compatible with that of Skin::getHeadScripts() in /var/www/w/includes/Skin.php on line 141 [11:54:50] p858snake: probably jidanni wonders why http://svn.wikimedia.org/viewvc/mediawiki/trunk/phase3/skins/vector/images/head-base.png?view=markup is 3569 bytes, and 1x1 transparent [11:55:00] when http://svn.wikimedia.org/viewvc/mediawiki/trunk/phase3/skins/vector/images/foot-base.png?view=markup is the same, and 124 bytes [11:55:34] Splarka: or he just likes to have his name appear as often as possible in the #mediawiki log... [11:55:47] but without ever being here? [11:56:02] Splarka: do you want to see your own spam? :) [11:56:17] I somehow would prefer single image to be used as spacer in multiple locations [11:56:25] 03avar * r52896 10/trunk/extensions/SlippyMap/SlippyMap.class.php: [11:56:25] * Remove global declarations that weren't used by anything [11:56:25] * Guard array access with isset() so the code doesn't spew errors under E_STRICT [11:56:25] * Remove boilerplate accessors that weren't used [11:56:27] yeah [11:56:30] that's make sense [11:56:33] isn't that what we have skins/common/ for? [11:56:35] 03nikerabbit * r52897 10/trunk/extensions/Translate/groups/ComplexMessages.php: Fix E_NOTICE [11:56:35] and png sucks for that [11:56:45] you can get a 1x1 transparent .gif at 35 bytes [11:56:49] much smaller than a png can ever be [11:57:13] 03avar * r52898 10/trunk/extensions/SlippyMap/SlippyMap.hooks.php: * Minor changes to make the code not spew errors under E_STRICT [11:57:22] possibly though there are planned replacements for these images [11:57:29] maybe they're just placeholders for future designs [11:57:54] siebrand: http://www.mediawiki.org/wiki/Special:Code/MediaWiki/52864 [11:58:00] 03avar * r52899 10/trunk/extensions/SlippyMap/SlippyMap.worldwind.php: * Remove global declarations that weren't used by anything [11:59:17] oop, 43 bytes to be transparent, 35 if not, heh [12:00:21] 03thomasv * r52900 10/trunk/extensions/ProofreadPage/ProofreadPage.php: minor fixes [12:01:49] why are we even still using spacer gif/png files? [12:02:08] p858snake: because it is 1997 apparently [12:02:23] *Splarka still guesses they are background placeholders to be replaced later... maybe? maybe not [12:02:59] They're not spacers I think [12:04:05] 03(mod) Check for empty section before printing title, else visible in text browsers - 10https://bugzilla.wikimedia.org/show_bug.cgi?id=19185 15enhancement->trivial; normal->lowest; +comment (10jidanni) [12:06:03] background-image: url(images/head-base.png); background-repeat: repeat-x; [12:06:17] same for foot base [12:06:23] so they do something, but it is transparent... [12:06:44] and that seems computationally wasteful to make the client repeat-x them for every pixel horizontally [12:08:00] meow [12:08:07] hmm, there is already a transparent placeholder gif in /common/images/: http://svn.wikimedia.org/viewvc/mediawiki/trunk/phase3/skins/common/images/Arr_.png?view=log [12:08:19] though it is 12x12 (for the little ajax nav arrows) [12:10:29] hmm, here is someone who reduced a .gif to 26 bytes for one color, 36 for transparent: http://cowburn.info/2009/03/15/tiny-little-gifs/ [12:11:29] and you can, in theory, make it any dimensions up to 16 bit, and stay at 36 bytes [12:11:56] I bet that is the kind of thing that excites GreenReaper (except for the fact that using spacer images is totally retro and wastes requests) [12:12:46] domas: wanna upload this thing to /skins/common/images ? http://cowburn.info/files/1px-trans.gif [12:13:06] *Splarka grins smarmily [12:18:20] data:image/gif,GIF89a%01%00%01%00%00%00%00!%F9%04%01%00%00%00%00%2C%00%00%00%00%01%00%01%00%00%01%01%01%00%3B [12:18:24] MrZ-man: ? [12:18:47] MrZ-man: i thought this was broken on old IE ? http://www.mediawiki.org/wiki/Special:Code/MediaWiki/52886 [12:21:18] thedj: it workes in IE5 (which is moot since IE5 totally breaks on most of wikibits.js anyway and would never be able to use it) [12:21:22] MrZ-man: oh no, that was with