[17:33:45] brion: inez wanted to talk about tables and templates today - do you have a minute right now? [17:34:04] sure let's do it [17:56:13] omg, why does babel have hardcoded old presentational attributes and why is it wrapped in single column table? such things shouldn't be put on live wikis :-/ [17:57:42] and why it uses strong.selflink when it is definitely not selflink? [17:58:41] robla: http://hexm.de/7h # bugs from test2 [17:59:41] running a couple minutes late for the scrum [18:06:19] might have to get a new headset [18:06:24] robla ^^ [18:06:43] bug list is my only contribution [18:07:12] Reedy, could you join call? [18:07:24] I just did the skype test and it worked there [18:07:27] weird [18:07:30] OH, yeah [18:08:38] just hung up, did it get better? [18:08:45] hexmode: yup, it [18:08:47] did [18:08:52] :( [18:09:13] *hexmode blames asterisk [18:09:23] its always the software's fault [18:09:30] hexmode: can you try using Google Voice or something to dial in? [18:09:50] think so... gimme a sec [18:18:10] nothing is working for me today [18:24:30] http://etherpad.wikimedia.org/bWwBzsE4Wl <- TrevorParscal [18:46:01] robla, can't reproduce it locally on 1.18wmf1 (login bug pointing to IP address) [18:46:19] bleh [18:54:49] robla, could we push test to 1.18wmf1, see if it's replicable there also? [18:54:54] Then we've got a machine to debug on [18:55:06] otherwise, it's slow and painful changing a line or 2, pushing to all apaches... [18:55:18] Reedy: assuming AaronSchulz agrees it's a good idea, go for it [18:55:43] we should also make sure everyone doing deploys knows that's happening [18:55:54] *robla checks calendar [18:56:04] Roan did one earlier, but nothing else today [18:57:06] excellent, I say go for it [19:05:42] robla, confirmed on testwiki at least [19:07:01] gross...yup, I see it [19:07:40] Reedy: running memcache on your local test instance? [19:07:53] not on my wmf wiki [19:07:56] I am on my trunk wiki [19:08:07] I can enable it and test it in a few mins [19:08:44] weird...so it's only that one page [19:13:04] $wgMainCacheType = CACHE_MEMCACHED; [19:13:04] $wgParserCacheType = CACHE_MEMCACHED; # optional [19:13:04] $wgMessageCacheType = CACHE_MEMCACHED; # optional [19:13:04] $wgMemCachedServers = array( "127.0.0.1:11211" ); [19:13:04] $wgSessionsInMemcached = true; [19:13:08] still doesn't do it [19:19:35] back in a few and then i'll have a look what it's doing [19:36:36] array(6) { ["userpage"]=> array(4) { ["text"]=> string(5) "Reedy" ["href"]=> &string(26) "/wiki/User:109.224.134.228" ["class"]=> string(3) "new" ["active"]=> bool(false) } ["mytalk"]=> array(4) { ["text"]=> string(7) "My talk" ["href"]=> &string(31) "/wiki/User_talk:109.224.134.228" ["class"]=> string(3) "new" ["active"]=> bool(false) } ["preferences"]=> array(3) { ["text"]=> string(14) "My preferences" ["href"]=> string(25) "/wiki/Speci [19:36:37] al:Preferences" ["active"]=> bool(false) } ["watchlist"]=> array(3) { ["text"]=> string(12) "My watchlist" ["href"]=> string(23) "/wiki/Special:Watchlist" ["active"]=> bool(false) } ["mycontris"]=> array(3) { ["text"]=> string(16) "My contributions" ["href"]=> string(33) "/wiki/Special:Contributions/Reedy" ["active"]=> bool(false) } ["logout"]=> array(3) { ["text"]=> string(7) "Log out" ["href"]=> string(66) "/w/index.php?title=Special:Us [19:36:37] erLogout&returnto=Special%3AUserLogin" ["active"]=> bool(false) } } [19:36:47] whee! [19:36:55] So userpageUrlDetails is wrong [19:37:33] robla: I didn't see any mention of bugs on the etherpad. Did you talk about them? http://hexm.de/7h [19:38:07] $this->userpageUrlDetails = self::makeUrlDetails( $this->userpage ); [19:38:09] string(20) "User:109.224.134.228" [19:38:51] ah, so you are talking 'bout them! :) [19:42:09] table sorting (https://bugzilla.wikimedia.org/31060) would be js, so TrevorParscal, right? [19:44:14] alolita: could you ping TrevorParscal on that ^^ if you see him? (or point me to the dev who should handle it) [19:44:45] hexmode: will do [19:44:57] tyvm [19:45:14] hexmode: you're welcome [19:45:28] hexmode: howdy [19:45:42] TrevorParscal: hey! [19:46:02] did you see https://bugzilla.wikimedia.org/31060 ? [19:46:19] and are you the right person to ask to handle it? [19:46:26] robla, "fixed" for some version of fixed [19:47:02] Krinkle rewrote that afaik [19:47:05] I can take a peek [19:47:33] hrm... krinkle isn't around atm, so I guess I'll CC him on the bug [19:55:26] hexmode: yeah, we talked about the bugs....sorry, forgot to take notes [19:55:59] np, looks like these two are the only ones that really need help [19:56:14] which two? [19:56:36] robla, not sure how it worked on 1.17 [19:57:55] Reedy: uninitialized variable? [19:58:03] robla: https://bugzilla.wikimedia.org/31060 & https://bugzilla.wikimedia.org/31064 [19:58:12] and Reedy just got *64 [19:58:22] seemingly, I ended up doing a variant of what's in trunk [19:58:29] but I can't find where it's assigned in 1.17 [19:58:50] well, this'll be a good test ;) [19:59:15] hexmode: you're keeping track of 31051, right? [19:59:36] Trunk now has $this->userpage = $user->getUserPage()->getPrefixedText(); [19:59:52] sounds like there's a way to solve 31051 [20:00:08] TrevorParscal: do tell [20:00:12] AaronSchulz: ^ [20:00:59] robla: looks like krinkle got prioritized it w/o CCing me... *sigh* [20:01:01] on it now [20:01:04] Examples [20:01:04] Yes [20:01:06] I'm not sure I have anything more to add - I was just informing hexmode we have somewhat of a plan [20:01:10] Let's see if it's still in my scrollback [20:01:26] i'm looking at the global configs for RL right now though.. [20:01:40] ah, ok. AaronSchulz is working on it. AaronSchulz, it was Ryan_Lane that set up bits + RL originally [20:02:46] $wgResourceLoaderSources is how you configure load and api scripts [20:04:44] so...the job queue problems seem to be correlated to stuff that Roan pushed early this morning [20:05:11] correlated....no idea of causation [20:06:37] onlyjust before midday [20:06:40] robla: just so you know, the URLs for resources (in debug mode) use $wgScriptPath as a prefix [20:07:42] siebrand: we think maybe the Babel deployment had something to do with the problems we're seeing with the job queue [20:07:57] siebrand: what's the Babel stuff doing? [20:08:24] Makes userboxes [20:08:40] Personally, that seems co-incidental [20:08:46] hrm...ok [20:08:50] robla: then the job queue should contain a lot of user pages and categy pages. [20:08:58] robla: otherwise it's not connected, I thinl. [20:09:02] it's mainly enotif [20:09:11] yeah....probably not related [20:10:30] Annoying, -tech doesn't seem to be logged [20:14:32] The 404's don't seem to be there [20:16:58] ah [20:17:02] AaronSchulz, robla http://test2.wikipedia.org/w/resources/mediawiki/mediawiki.user.js [20:17:04] 404 [20:17:16] should exist in 1.18 [20:17:33] so it's still pointing at php-1.17 [20:19:41] $wgExtensionAssetsPath = "$urlprotocol//bits.wikimedia.org/w/extensions-" . $wmfVersionNumber; [20:19:41] $wgStyleSheetPath = "$urlprotocol//bits.wikimedia.org/skins-" . $wmfVersionNumber; [20:19:47] resources-version? [20:20:47] Ah [20:20:52] http://test2.wikipedia.org/w/resources-1.17/mediawiki/mediawiki.user.js 404 [20:20:56] http://test2.wikipedia.org/w/resources-1.18#/mediawiki/mediawiki.user.js [20:20:58] downloads [20:21:02] but it's looking at with no revsion [20:21:19] Probably not helped by 'resources/mediawiki/mediawiki.js' [20:30:36] "I will need to make the URL to the [20:30:37] resources directory configurable in MediaWiki" [20:32:20] robla, fixing the resources issue isn't going to be an easy fix [20:32:29] resources/ is hard coded all over the place in mw [20:32:57] Reedy: instead of doing resources-1.18/blah, can we do 1.18/resources/blah? [20:33:43] possibly [20:35:59] 'remoteBasePath' => $GLOBALS['wgStylePath'], [20:35:59] 'localBasePath' => $GLOBALS['wgStyleDirectory'], [20:36:05] I'm guessing that's similarily what it's doing... [20:38:47] (stoopid office wifi) [20:39:22] ok....so, the job queue is still a mess [20:51:56] Reedy: AaronSchulz: are you both working on https://bugzilla.wikimedia.org/31051 ? [20:52:57] Also, hexmode: I wonder if https://bugzilla.wikimedia.org/31064 is very high priority [20:53:05] I am lookign to see if anything obvious can be done [20:53:12] oops [20:53:38] hexmode: oops, I meant https://bugzilla.wikimedia.org/31060 [22:06:39] robla, do we have a gen eng meeting now? Or is my gmail calendar suffering from stuff not being removed? [22:08:23] Reedy: the latter [22:08:39] it's postponed until tomorrow [22:08:48] right [22:08:49] blahh [22:40:54] hexmode: you there? [22:42:36] robla: ?? [22:43:18] https://bugzilla.wikimedia.org/31060 <- I bumped this down to "high" priority [22:43:44] ...and honestly, I'm tempted to say "wontfix", but I guess I should let Trevor make that call [22:43:56] robla 1s [22:45:27] robla: I think my first take was same as yours but then I realized a lot of people could be relying on it [22:45:37] not that I'm right [22:45:45] I don't know how many are [22:45:54] Was just being cautious [22:46:56] yeah, let's push back on this one. I think Trevor may have implemented stock jquery sortable tables, which may not support it [22:47:20] I don't think we should support it if the upstream doesnt' support it, and it was never a documented feature [22:47:32] robla: fwiw, trevor said Krinkle did this work [22:48:01] Krinkle: you listening? [22:48:03] Yes [22:48:24] Why wont fix? There is interface elements using images from that path. [22:48:27] Krinkle: any opinion on #31860 ?? [22:48:35] Oh that one [22:48:40] I confused the bug ids [22:48:46] sorry 060 [22:49:01] Krinkle: yeah, actually, I don't think I want to distract you from helping with 31051 [22:49:15] np, carry on [22:49:17] :) [22:49:34] 31051 is our only blocker, I *think* [22:50:17] though there's some lucid image debugging that Tim and the ops folks are doing now in #wikimedia-operations that could delay things tonight [22:50:45] :/ [22:51:19] AaronSchulz: (moving back here from #mediawiki) [22:51:29] AaronSchulz: So yea, the -1.18 symlink appears to work fine now [22:51:48] So what, let's implement wgResourcesPath/Directory in trunk, merge and deploy to test ? [22:52:23] Seems like the obvious thing to do. It's a client-side accessed directory just like skins and extensions [22:52:42] the reason this didn't fail with the 1.17 deployment is because it was a new directory back then, only had one version [22:53:12] We need to have a /resources-1.17/ as well tho if we're gonna do this. [22:53:29] and keep /resources/ pointing to 1.17 for backwards compatibility [22:53:38] (since 1.17 is actually referring to that) [22:54:14] I ran a search on "/resources" in /trunk http://toolserver.org/~krinkle/wikimedia-svn-search/view.php?id=293&hash=0ba1fbf527a479463bb4db6decfdd33f [22:54:32] and on "resources/" http://toolserver.org/~krinkle/wikimedia-svn-search/view.php?id=294&hash=3252094e4d5b9dcedd8c8b9a907ab3ba [22:54:49] (slash position) [22:55:16] 250 line(s) matched in 24 different file(s). [22:55:16] 576 line(s) matched in 34 different file(s). [22:55:19] pfew [23:08:21] yeah...that's pretty hardcoded :) [23:09:08] just a bit ;) [23:09:20] any backport would conflict too [23:09:39] i suppose it'd be ok as a temporary [23:09:46] depending if say roan could fix it tomrorow [23:09:59] there's quite a few cases where it wouldn't be necessary, I think [23:10:00] What is ok as a temporary ? [23:10:04] branch only s&p [23:10:10] * s&r [23:10:24] replace with -1-18 ? [23:10:29] still ugh [23:10:31] Hm.. [23:10:34] Not sure what's gonna work [23:10:42] that* [23:10:56] does resources-1.18 work as a directory on-disk as well ? [23:11:15] it should [23:11:19] currently there is 1 path given that is used for both $IP + path and $wgScriptPath + path [23:12:09] If this is a rewrite or alias in apache, then it wouldn't work for PHP when reading a file from disk, right ? [23:12:22] oh, but it's not an alias. it's a symlink [23:12:22] at worse we create a symlink in the php-1.18 folder also from resources-1.18 to resources [23:12:55] robla: Indeed. Only needed for paths uses from the web. [23:12:59] used* [23:13:21] Anyway, I could fix this but I am off to bed now. need to get up early tomorrow and it's already midnight [23:13:23] seeya later [23:26:16] i'm seriously about to unsub from the AFT talk page.