[00:01:31] TimStarling: Per robla's request, here's instructions on how to lower the RL startup module cache timeout before deployment. You want to set $wgResourceLoaderMaxage['unversioned'] = array( 'server' => N, 'client' => N ); for the wiki(s) you're deploying 1.19 to (suggested values for N are 30 or 60, default is 300). You'll need to make the config change at least 15 minutes prior to the... [00:01:33] ...switchover to be sure it's been picked up everywhere, and of course you're encouraged to revert it once 1.19 is running fine [00:02:04] ok [00:02:11] when are we switching? [00:03:18] I don't know [00:03:22] That's up to RobLa [00:03:39] uh? what just happened [00:03:40] I'm in and out of the office (I ran an errand just now and I'm leaving again in 45) so he said I should hand this off to you [00:03:50] |<-- Krinkle has left freenode (Read error: Connection reset by peer) [00:04:04] well, I hear you again so it's fine now :) [00:05:13] so....here's where we're at [00:05:29] *very* cursory testing shows that our IRC output is not perfect, but better than before [00:06:03] there are some cases where the switch statement is incomplete, and we need to make sure that we're at least returning the old text [00:07:14] robla: the same text is not required per se. as long as the used text is located under the same message-key and uses the same values for $1 and $2, etc. still. actual phrasing may be different or even localized (which is the case in 1.18 as well). [00:07:48] sorry: s/old text/what was previously in 1.19/ [00:08:07] actually, I took a closer look at it, and it looks pretty logical now [00:08:46] I hadn't noticed the bit where we set $text=null, then do a null check at the end, and set default text. that seems pretty good [00:09:28] might be a few more parameters that need switching around [00:11:02] I wonder if we could've just copied the messages across, and fixed the parameters to match.. [00:11:53] Reedy: possibly, but this means we don't have a big Message*.php automation project on our hands [00:12:08] true [00:14:03] TimStarling: could you do a code review of what we've got so far? [00:14:29] easiest thing to do would be svn diff -r112043 in the logging directory and RecentChange.php [00:14:44] I think that picks up everything without picking up the noise [00:15:07] sure [00:15:16] are you guys in the middle of a deploy? [00:15:23] Ryan_Lane: will be very soon [00:15:27] ok... [00:15:29] (hopefully) [00:15:38] fixing one last blocker [00:15:42] there's a git change I just pushed that could possible cause issues with svn [00:15:55] want me to revert for now? [00:16:24] or test really quick, and revert if necessary? [00:16:25] Ryan_Lane: you can check if you're right about it screwing up svn [00:16:31] ok [00:16:40] it's been tested in labs... [00:16:51] but, I don't want to screw up your deploy, just in case [00:17:31] testing it. if you guys run into any issues, let me know and I'll roll back the change [00:17:46] Ryan_Lane: k...thanks for the heads up! [00:17:54] bah. seems I'm missing a cherry-pick necessary for this change to go through. [00:17:56] *Ryan_Lane fixes [00:21:12] it's funny to see $actionComment left in there, what's it for? [00:21:21] not problematic, just funny [00:21:36] *robla looks [00:23:06] oh, right...that's the one you were saying was only used for IRC [00:23:13] heh...oh well [00:23:23] yeah [00:23:35] I think it's passed to a hook, but presumably any hook that uses it will be broken also [00:23:45] for the same reasons that IRC was broken [00:24:24] anyway I assume this will all be ripped out before 1.20 and replaced with something nicer and less backwards-compatible, so it's no big deal [00:28:13] *ImSexyAndIKnowIt sodomizes Reedy [00:28:18] anyway, it looks fine in overview [00:28:37] TIMStarling prepare your anus [00:28:59] TimStarling: dare we deploy and iterate on it? [00:29:49] ok [00:30:06] TrevorParscal: ^ [00:30:30] I'll reduce the startup module expiry [00:30:30] Reedy: wtf? [00:30:38] Apparently you has ops [00:30:40] -ChanServ- You are not authorized to perform this operation. [00:30:51] ah [00:30:57] I may need to revert this change. [00:31:08] Roan, Casey and Guillom aren't about atm [00:31:15] TrevorParscal I will rip your balls off if you kick me [00:31:16] of course, puppet hasn't finished running???. [00:31:37] ImSexyAndIKnowIt identify yourself, or be kicked [00:31:58] I am yo mama that's who my identity is [00:32:21] no one in here has ops in the channel? [00:32:36] Hahaha [00:32:46] Trevor does also [00:32:49] chanserv does, let's just get them to do it [00:32:54] Hence pinging him [00:32:59] TrevorParscal: http://wikitech.wikimedia.org/view/Freenode#Becoming_ops_in_a_channel [00:33:06] trevor doesn't know shit about ops [00:33:07] http://wikitech.wikimedia.org/view/Freenode#Ban_someone_from_a_channel [00:33:20] /ban *!*@mobile-166-147-*.http://mycingular.net/?? [00:33:24] err [00:33:33] /mode +b *!*@mobile-166-147-*.http://mycingular.net/?? [00:33:35] can I have +voitA please? [00:33:49] TimStarling denied [00:33:50] damn it, why does puppet take so long to run? [00:34:01] Ask the puppetmaster [00:34:05] /chanserv flags #wikimedia-dev TimStarling +voitA [00:34:13] none of us has ops [00:34:28] I thought Trevor did [00:34:54] Oh shit [00:35:01] Here it comes [00:35:09] TrevorParscal is listed as a founder in /chanserv info [00:35:10] *Ryan_Lane rolls his eyes [00:35:23] GIVE IT TO ME [00:35:28] HARD AND RAW [00:35:41] fail [00:35:52] You failed [00:35:55] You're failing more [00:35:58] Kill yourself [00:36:05] /mode +b *!*@mobile-166-147-*.http://mycingular.net/?? [00:36:15] I think he's kind of funny [00:36:23] Reedy come at me bro [00:36:24] He really is a crappy troll [00:36:26] which is more amusing [00:36:35] in a 10-year-old troll kinda way [00:36:41] Tim would you like to go out for a beer sometime [00:36:57] of course if Trevor would just paste in that /chanserv flags command he would be kickbanned in a few seconds [00:37:38] He can't even set a ban correctly how can he deputize you [00:37:43] heh. my stupid client pastes this stuff incorrectly [00:38:07] boring trolls make me sad [00:38:23] It's kinda funny [00:38:44] Last week I had robh convinced I was on the ops team [00:39:01] well, I'm glad my change didn't break anything. it's finished. let me know if you guys have any issues with svn [00:39:14] I changed out all the LDAP libraries... [00:39:35] Fancy dishing out some more rights while you're at it? [00:40:03] oh, ^demon's gitweb change went in with this too [00:40:08] Reedy: which rights? [00:40:17] Roan giving more people op in here [00:40:28] oh [00:40:31] that's a good call [00:40:50] Heh. [00:41:14] Reedy: You're an op now [00:41:23] *maplebed created RT-2494. [00:41:44] Hmmm. [00:41:53] I know I've only been around for 9.5 years but really I think I can be trusted with ops access [00:42:00] Are RT ticket titles private too? [00:42:01] pretty please? [00:42:03] TimStarling: you seem to already be on the list [00:42:05] Or just the ticket contents? [00:42:11] I think that's too many privileges for you TimStarling [00:42:16] :D [00:42:26] [modified 5 minutes, 11 seconds ago] [00:42:41] Would a RT feed of ticket resolutions be possible/useful? [00:42:43] *AaronSchulz is not an op here, but is on #mediawiki [00:42:45] I may as well preemptively ban him from labs channel [00:43:17] Joan: summary: ops need more ops [00:43:42] TimStarling: Trevor just gave you op rights like 4 minutes before you asked :) [00:43:44] Reedy: ty [00:44:52] } elseif ( $entry->getSubtype() === 'patrol-auto' ) { [00:44:58] why is that still there? ;) [00:45:30] so he did [00:45:33] thank you TrevorParscal [00:46:02] sorry for my lack of IRC skills [00:47:14] xchat gives you a kickban command in user context menus, so not much skill is required [00:48:11] TimStarling: AaronSchulz: are you absorbed by deployment stuff or do you have time for the commit access review meeting right now? [00:48:30] RoanKattouw_away: added another gerrit change we want: http://code.google.com/p/gerrit/issues/detail?id=118 [00:48:32] /kick username [00:48:36] is usually enough [00:48:39] to the gerrit changes that matter page [00:48:53] Reedy: this one was coming back after kicks [00:49:06] suew [00:49:08] *sure [00:49:16] but it gets him out for a bit ;) [00:49:21] *Ryan_Lane nods [00:50:13] sumanah: I think we're planning on deploying to commons in 5 minutes or so [00:50:18] RL is broken on commons [00:50:22] OK. nm then [00:50:55] oops [00:51:05] that's what you get for trying to do three things at once [00:54:48] Merged the bug 34508 revisions into 1.19wmf1 [00:57:08] you want to deploy them? [00:57:11] brion: if you have spare time to help a bit with commit access review, I could use it [00:57:24] Yup will do [00:58:25] wassap [00:58:31] we now have a 30s expiry time for the startup module on commons, it just found its way through varnish [00:59:07] let's do some quick smoke testing of Reedy's changes on the exisitng wikis before going out to commons [01:00:16] I'm particularly interested to see if the right stuff is making it from irc://irc.wikimedia.org/#meta.wikimedia to irc://irc.freenode.net/#cvn-meta [01:00:36] Live [01:01:56] meta is almost dead [01:04:00] [[Special:Log/move]] move * RobLa-WMF * moved [[RobLa-WMF/Test123]] to [[User:RobLa-WMF/Test345]] [01:04:15] Seriously, who came up with the idea that it needed colour? [01:04:42] purty [01:05:00] all of our APIs need color [01:05:19] https://en.wikipedia.org/w/api.php [01:05:24] Blue and black, what more do you want? [01:06:30] alright, I'm happy that we haven't at least broken logging horribly, at least not at first investigation [01:06:32] it was a user request, I think [01:06:39] test [01:07:01] Seriously? [01:07:07] *Reedy wonders if we can kill it in 1.19 [01:07:12] anyway....should we finally deploy to commons? [01:07:27] yes, is Reedy appointed button presser today? [01:07:53] can be... Reedy, want to do the honors? [01:08:01] sure [01:08:32] chrismcmahon: you still around? [01:08:42] hi robla, been watching [01:08:54] what is commons? [01:09:12] https://commons.wikimedia.org [01:09:18] pokeswap: I think you want #mediawiki, this is mostly for wikimedia-related stuff [01:09:25] Done [01:09:27] pokeswap: Wikimedia Commons, https://commons.wikimedia.org/wiki/Main_Page but right now most of the people talking in this channel are busy with something urgent. [01:09:44] just wanted to know what it is [01:10:05] Commons mainpage is taking a while [01:10:26] can't connect to db just now on commons [01:10:28] (Can't contact the database server: Unknown error (10.0.6.32)) [01:10:40] hey everyone [01:11:18] TimStarling: 10.0.6.32 is db40? [01:11:27] nope [01:11:37] commons master has issues [01:11:38] db22 [01:11:42] commons main page is back [01:12:02] mostly MessageBlobStore::clear() I think [01:12:24] brion: I'm seeing that opensearchxmp api array_map error again. Stupid thing [01:12:32] no where conditions on that? [01:12:39] hmm [01:12:40] I'll just kill them [01:12:49] Reedy, got an err line or bt to show me? [01:13:15] brion: 57 Warning: array_map() [function.array-map]: An error occurred while invoking the map callback in /usr/local/apache/common-local/php-1.18/extensions/OpenSearchXml/ApiO [01:13:16] penSearchXml.php on line 0 [01:13:23] not very useful [01:14:24] ok this is kind of uncool [01:14:55] now we've got 100 threads trying to insert into msg_resource [01:15:22] let's assume that one of them finished already and that I can kill the rest, ok? [01:16:10] no luck, still loads of threads deleting [01:16:53] I'm going to revert to 1.18 [01:17:13] not sure if it will help [01:17:24] *AaronSchulz waits to chuckle at Special:Log [01:18:07] apparently we broke a disk on it [01:18:17] who knew that was possible? [01:19:21] ok, situation resolved, db22 is OK now [01:19:22] Reedy, any sign of related error msgs above it? exception for instance [01:19:29] nope [01:19:36] bah [01:20:10] I just ran into a merge account error on meta. File under MediaWiki:SpecialPages, right? [01:20:25] PROBLEM - RAID on db22 is CRITICAL: CRITICAL: Defunct disk drive count: 1 [01:22:33] sumanah: which two accounts? [01:22:40] robla: https://bugzilla.wikimedia.org/show_bug.cgi?id=34582 [01:23:05] robla: I tried to merge my meta login with my unified global login, got this error [01:23:12] "Call to undefined method SpecialMergeAccount::doUnattach" [01:23:17] hexmode: ^^ [01:23:29] this was about 5 min ago [01:24:32] // TODO: Give this some more thought [01:24:32] // TODO: Is TRUNCATE better? [01:24:42] so is there anyone here who knows this code? [01:25:03] TimStarling: are you looking at SUL or elsewhere? [01:25:20] I'm kind of focused on the thing that made the site go down at the moment [01:25:52] just looking for context for your comment... I guess I'll grep [01:26:01] MessageBlobStore::clear() [01:26:22] we can just switch to manualRecache [01:27:12] it may take an hour or so to write and test the scripts [01:28:27] also we should probably do a master switch [01:29:12] Roans comment (somewhat unsuprisingly) [01:29:44] TimStarling: might as well. you want to do that now, or do you want me to see if someone in ops can [01:30:04] Ashers out isn't he? [01:30:11] yeah [01:30:23] probably me then [01:30:42] domas is about [01:30:42] apparently asher did some work on switch.php, I don't know if he told anyone else how to use it while he was at it [01:31:06] Asher did leave instructions which someone else in ops might be able to use [01:31:07] He sent a mail around about it IIRC? [01:31:16] yeah, it's in the bug [01:31:16] the master switch is not urgent, it just has to be done at some point so that that server can be shut down and repaired [01:31:18] *robla looks up [01:31:28] It's on wikitech [01:31:38] http://wikitech.wikimedia.org/view/Switch_master [01:31:51] manualRecache is the top priority, after that's done, we can do the deployment [01:38:47] it's late everywhere except here [01:38:52] should we reschedule for tomorrow? [01:40:24] we've technically got until 7pm here in SF. we should probably not count on Reedy for much longer though [01:40:53] I'd say we keep going, and if Reedy is too exhausted to help at this point, he should drop off [01:41:07] TimStarling: ^ [01:41:48] ok [01:43:21] I should probably call it a night at 2am or so [01:43:26] AaronSchulz: can you take a look at https://bugzilla.wikimedia.org/34582 [01:43:29] Otherwise I'll be still up and it'll be 4am [01:44:46] robla: AaronSchulz, it's quite obvious why that one broke, ther isn't a doUnattch() method [01:46:16] that' [01:46:21] that is a factor, yes [01:47:29] Why would MergeAccount try to unattach something? Isn't it supposed to do the opposite? [01:47:29] TimStarling: https://svn.wikimedia.org/viewvc/mediawiki/trunk/extensions/CentralAuth/SpecialMergeAccount.php?view=markup&pathrev=24382 [01:47:34] I find that amusing too [01:47:36] 24382 [01:48:09] So apparently it never existed [01:48:58] Also, the invalidAction() goes nowhere [01:50:49] Reedy: read #wikimedia-tech [01:52:57] we can probably do this in the same loop as ExtensionMessages-$mwVerNum.php, right? [01:57:07] TimStarling: do what? [01:57:16] manualRecache [01:57:38] basically we need to change the sync scripts to run rebuildLocalisationCache.php [01:58:01] and set $wgLocalisationCacheConf['storeDirectory'] = "$IP/cache/l10n" [02:06:24] TimStarling: seems sane [02:07:59] how did you create ExtensionMessages-1.19.php before there were any 1.19 wikis? [02:09:35] did you just run mergeMessageFileList.php directly? or did you let it break on test2 and then run scap? [02:09:58] ( RoanKattouw_away is distracting him ) [02:10:27] ( now he's reading ... hi Aaron!) :) [02:10:57] TimStarling: it was Reedy that created it right? [02:11:09] *AaronSchulz didn't do that stuff this time around [02:11:12] I think so... [02:11:26] either way I think the script I'm writing will work [02:11:37] though breaking test2wiki sounds reasonable [02:12:40] I might have created it against 1.18 [02:12:57] as it's essentially the same [02:13:10] not really [02:13:25] the extension-list is the same [02:13:56] You're the only one that makes major changes to where message files are ;) [02:18:41] Updating LocalisationCache... [02:18:42] PHP Fatal error: Allowed memory size of 209715200 bytes exhausted (tried to allocate 108 bytes) in /home/wikipedia/common/php-1.18/extensions/LocalisationUpdate/LocalisationUpdate.class.php on line 37 [02:19:14] :O [02:19:26] easily fixed [02:19:45] well, if you can call any fix in three branches easy [02:37:25] fenari may be a bit toast [02:37:39] I am 70% certain it is not my fault [02:39:13] I like its memory graph: http://ganglia.wikimedia.org/latest/graph.php?h=fenari.wikimedia.org&m=load_one&r=month&s=by%20name&hc=4&mc=2&st=1329878298&g=mem_report&z=medium&c=Miscellaneous%20pmtpa [02:40:03] I saw "Passenger spawn server" in top [02:40:26] also LU just finished [02:40:49] as in the PHP part, it might be syncing [02:41:09] hmm [02:41:26] why does the memory graph say it's using so much swap [02:41:32] Swap: 10485752k total, 0k used, 10485752k free, 1414764k cached [02:42:21] it's using a negative amount according to ganglia [02:42:58] anyway it will need rebooting at this rate, I can't wait all day to get a root shell [02:43:07] It's an NFS problem looks like [02:43:16] I guess you managed to melt NFS [02:43:28] I badly need to tweak LU to not write incremental files to /home [02:43:36] but build them on /tmp and copy to /home after [02:43:46] Actually let me do that right now [02:44:21] maybe we should use a fileserver that doesn't melt just by being used a lot [02:44:21] Supposedly we're moving to a netapp [02:44:21] That's what Mark tells me [02:44:36] I've told him I will send him wine once that's done (provided it's not just as slow) [02:45:51] did you see how MessageBlobStore physically broke a disk earlier? [02:46:02] the timing was too close for it to be coincidence [02:46:25] I heard about it causing problems [02:46:31] I had no idea it physically broke a disk [02:46:37] That's disturbing [02:46:43] well, it overloaded the master, db22 [02:46:47] I'll comment it out in 1.19wmf1, we did the same in 1.17wmf1 and 1.18wmf1 [02:46:53] Ouch [02:46:54] and then nagios showed a broken disk alert on that server [02:48:28] *AaronSchulz struggles to get on hume or fenari [02:52:07] OK I just got kicked off of fenari, is it rebooting now? [04:38:09] hi Zaran, how are things going? [04:38:16] hey sumanah [04:38:20] I'm still at work :) [04:38:45] can we set up a meeting to talk about the things we discussed by email ? [04:38:49] I seem to be as well! [04:38:57] Zaran: sure, but this week is bad for me. Next week ok? [04:39:01] unless anything is urgent [04:39:09] sumanah : no next week is perfect [04:39:40] Zaran: if you prefer setting an appointment, then I'll need to look at my calendar [04:39:55] no no, don't worry [04:40:01] we'll probably find some time next week [04:40:19] ok! [04:40:47] so you're working late as well ? [04:41:20] I seem to be..... [04:41:33] Zaran: I had Monday off. So of course stuff piled up [04:41:42] I did take some breaks for dinner etc [04:41:49] sumanah : I see... [04:42:02] right now, I'm fueled by beer only [04:42:06] which is not very good [04:42:08] I just did something very exciting [04:42:13] I reorganized my TODO list [04:42:22] Zaran: is it at least good beer? [04:42:42] mirror pnd [04:43:00] http://www.mediawiki.org/wiki/User:Sumanah/TODO [04:43:07] is this the result of our work ? [04:44:02] oh, I'm on it [04:44:04] Zaran: my TODO list? I've been keeping it for months and months [04:44:05] woohoo ! [04:44:29] :-) [04:44:30] oh no it's not me, the spelling is different ;) [04:44:34] oh no! [04:44:45] I keep putting an L in [04:44:54] *sumanah fixes [04:45:06] hehe, thanks [04:51:37] Zaran: you might be interested in https://codechix-wikipedia.eventbrite.com/ or perhaps a few of your friends would be interested [04:54:57] sumanah : looks interesting, I'll send out the word [04:55:04] cool [04:55:52] I'm not sure I will be able to attend myself though [04:56:18] Understandable [06:58:10] hi ramkrsna [06:58:57] sumanah, hey [13:50:15] http://hi.wikipedia.beta.wmflabs.org no longer has gadgets. No gadgets tab in the prefs anymore (it was there yesterday) [13:50:21] Sid-G noticed this [13:50:46] :) [13:51:06] hexmode: petan: Reedy: ^^ I'm going to check some other beta sites to see if it's just Hindi [13:51:17] And the script file which used to contain twinkle does load, but without twinkle [13:52:05] http://de.wikipedia.beta.wmflabs.org/wiki/Spezial:Helferlein and http://en.wikipedia.beta.wmflabs.org/wiki/Special:Gadgets -- they're gone! why is this? [13:53:09] http://hi.wikipedia.beta.wmflabs.org/w/api.php [13:53:15] The API is also broken [13:53:21] yeah, that's weird [13:53:32] wow [13:53:45] *Reedy blames MaxSem [13:53:47] well, that means these sites are not useful at the moment for testing [13:54:14] *Sid-G sighs [13:54:39] Sid-G: thank you for reporting this [13:54:53] no problem :) [13:56:31] I've got a q: When 1.19 is deployed to wikipedia, will it look exactly like it does on the test sites? (Meaning if I'm seeing any untranslated interface messages on test, they need to be translated before deployment)? [13:57:13] live sites will have customised messages and stuff [13:57:25] if you're looking for untranslated messages, use translatewiki [13:57:47] by the way, Sid-G: you know about #mediawiki-i18n ? [13:58:06] API fixed, needs a svn up [13:58:10] Reedy: So 1.19 has some new messages over 1.18? [13:58:19] Most certainly [13:58:21] sumanah:yeah, but its mostly empty i think [13:58:44] *MaxSem blames Reedy for blaming too much [13:58:50] Reedy:k, thnx [13:58:51] Sid-G: I've been able to successfully talk with people in the i18n room [13:59:00] Sid-G: so, it may depend on the time of day [13:59:15] Sid-G: are you on the Hindi language support team? [13:59:16] hi looks mostly translated [13:59:17] sumanah: yeah, it does depend on that a lot [13:59:17] https://translatewiki.net/w/i.php?title=Special%3ATranslate&taction=translate&group=core&language=hi&limit=100&task=untranslated [13:59:30] sumanah: not officially [13:59:34] 62 messages that need updating [13:59:35] Sid-G: :-) [13:59:46] /are out of date/fuzzy [13:59:47] but, yeah I've done some translations [13:59:47] hmm [13:59:48] will check it out in a day or two [13:59:55] Sid-G: thanks, we'd appreciate it [14:00:37] Sid-G: have you seen https://translatewiki.net/wiki/Project:MediaWiki_localisation_in_the_50_most_spoken_languages ? [14:00:44] sumanah:Heh, no problem. :) [14:00:49] sumanah: long time ago, yeah [14:00:55] it's getting better [14:00:56] :-) [14:01:22] *Sid-G goes looking [14:01:29] We definitely need help with Sindhi, Urdu, Kannada, and some other languages, in case you know some folks [14:01:52] sorry, I dont [14:02:08] know folks I mean [14:02:51] sumanah: The WMIN chapter could help, tho [14:02:55] *though [14:04:16] Sid-G: https://translatewiki.net/wiki/Language_support_team in case you want to join for Hindi [14:05:21] Gadgets still work on trunk [14:05:39] interestingly the "Speaker" column on that page is translated as "loudspeaker" in Chinese. [14:05:41] sumanah:Well, is official support needed? (I mean, hi-wp ppl do get the work done) [14:05:41] and on test2 [14:06:07] au: wouldn't suprise me. Sometimes we get people just setting machine translations [14:06:21] yeah, that's what google translate returns as first hit as well. [14:06:25] lol [14:06:37] I'm not sure how we mark translations as "fixme" [14:07:06] I've seen worse than speaker/loudspeaker [14:07:15] atleast its close [14:07:17] Nikerabbit: siebrand: how does one mark a translation as FIXME? [14:07:49] https://translatewiki.net/w/i.php?title=Special%3ATranslate&taction=translate&group=core&language=hi&limit=100&task=untranslated [14:07:50] fail [14:07:54] sumanah: see mediawiki-i18n [14:07:55] Reedy: add !!FUZZY!! in front of it [14:07:59] siebrand Reedy: add !!FUZZY!! in front of it [14:08:08] thats ur answer [14:08:46] Sid-G: in volunteer-work communities, "they're getting the job done right now" is always a temporary state of affairs. So, right now, Hindi has a score of 90, but volunteers come and go, so it's worthwhile to join the team and help out, so if someone else has to leave, there are replacements available [14:08:58] one should think in terms of long-term sustainability [14:09:54] sumanah: I agree [14:09:55] I just dont remember my password right now :P [14:10:12] :P indeed :) [14:11:46] au: what's the message key? [14:12:15] Reedy: https://translatewiki.net/wiki/Thread:MediaWiki_talk:Translate-gs-speakers/zh-hant/_FIXME_%E4%BD%BF%E7%94%A8%E8%AA%9E%E8%A8%80%E4%BA%BA%E5%8F%A3 [14:12:47] er, sorry. {{int:translate-gs-speakers}} I mean [14:12:58] I've marked it as fixme [14:13:01] /fuzzy [14:13:09] Reedy++ much obliged [14:15:13] btw, when should I check the test site again for gadgets? 2hrs/5hrs/10hrs/tomorrow.... ? [14:15:35] No idea [14:15:56] who's in charge of the test sites? thought it was you, Reedy [14:16:02] Nope [14:16:05] I can't even login to labs [14:16:57] I suppose it falls to Ryan, then [14:17:09] Fixed [14:17:09] http://en.wikipedia.beta.wmflabs.org/wiki/Special:Gadgets [14:17:14] Cache failure by the looks of it [14:17:24] Test edit to http://en.wikipedia.beta.wmflabs.org/wiki/MediaWiki:Gadgets-definition and reverted [14:17:25] fixed [14:18:10] Reedy: http://hi.wikipedia.beta.wmflabs.org/wiki/%E0%A4%B5%E0%A4%BF%E0%A4%B6%E0%A5%87%E0%A4%B7:Gadgets (their Special:Gadgets page) is still empty [14:18:19] yes, not updated that page yet [14:19:04] Also fixed [14:19:31] Reedy: I'm pretty sure you're in charge of these test deployment sites, so you should probably learn to log in to labs [14:19:49] Mark handed over responsibility to you a few weeks ago, or at least meant to [14:20:03] Sid-G: check the Hindi site again? [14:20:06] login as: reedy [14:20:06] If you are having access problems, please see: https://labsconsole.wikimedia.org/wiki/Access#Accessing_public_and_private_instances [14:20:08] Not my fault [14:20:31] Not had the time, the effort, nor the inclination to ask anyone to do anything about it [14:24:14] sumanah:one sec [14:27:52] got the pref panel back [14:28:05] checking if the gadgets work [14:45:22] MaxSem: does it make sense to add a handler to ArticlePurge for the gadgets page too? [14:46:05] mmm [14:46:19] then it doesn't need to be a null/full edit [14:46:31] if problems with gadgets not being updated, why not [14:46:50] *are widespread*, why not [14:47:48] Maybe wait till we get it more live, and see if problems are occuring on a larger scale [14:47:51] are they widespread? [14:48:20] well, seemingly beta was all broken, and editing the page fixed it [14:48:43] not sure if that was just cache fail [14:51:04] whaddaya mean by "all broken"? [14:58:51] Reedy, ^^ [14:59:18] gadgets wouldn't show up in preferences/special:gadgets [15:00:26] anything in error log? might be class version not updated after a breaking change resulting in crap restored from cache [15:43:20] <^demon> sumanah: Thought you might like this http://code.google.com/p/gerrit/issues/detail?id=1241 [15:43:49] ^demon: I have broken out into an irrepressible grin. [15:44:34] <^demon> Just happened across that one [18:19:19] robla: I'm thinking of sending a short message to wikitech-l basically saying that we are declaring irc-output of mediawiki feature/bugfix frozen until it is superseded by https://www.mediawiki.org/wiki/Requests_for_comment/Structured_data_push_notification_support_for_recent_changes - what do you think? [18:19:35] yeah, that's fine [18:19:47] robla: wait does "fails" mean? The bot doesn't match it? [18:19:59] i.e. wont fixing bug 17305, and instead fixing those kind of things in the new system (also if the new system would use JSON, it's easy to add a property without breaking anything) [18:20:21] robla: ok [18:20:30] AaronSchulz: yup [18:31:49] !r 112128 [18:31:49] --elephant-- http://www.mediawiki.org/wiki/Special:Code/MediaWiki/112128 [18:31:58] robla: I assume we're focusing everything on 1.19 and skipping 20% today and tomorrow [18:31:59] AaronSchulz: could you review/mft [18:32:55] I can merge with the rest [18:33:19] *sumanah presumes similarly to Mark [18:38:29] hexmode: yeah, that's right (sorry...distracted) [19:03:22] Where are we upto now? [19:13:58] Reedy: so...here's where things stand. We're going to try out a fix Nikerabbit made that AaronSchulz just deployed [19:14:14] there's still a thumbnailing fix that RoanKattouw and AaronSchulz need to get to [19:14:24] I'm just going to have a quick poke and see if I can fix the commit notifications [19:14:31] it's starting to get beyond irritating [19:14:44] Reedy: I'm a little worried about that [19:15:01] RoanKattouw took a crack at it last night and brought down fenari [19:15:07] ... [19:15:12] NFS freakout [19:15:20] NFS isn't needed [19:15:22] :| [19:15:45] it's not getting from formey to cia.vc [19:15:46] He, I didn't bring it down [19:15:48] That was Tim [19:16:31] Ryan_Lane: what were the svn/git changes you made last night? [19:16:40] did it break something? [19:16:42] It was broken long before that [19:16:48] I changed out the ldap libraries [19:17:00] I checked to ensure nss was working properly, afterward [19:17:08] which is the only requirement the svn server has [19:17:14] *robla gets IRC log from yesterday [19:17:15] as above, you didn't break it with that change at least [19:17:17] the gerrit server has no system ldap requirement [19:17:33] (which is why I like that gerrit re-implements ssh :D ) [19:18:18] is this in regards to the nfs/fenari breakage? [19:19:12] yeah....here's my recollection from yesterday.... [19:19:20] Running cia_svn.py gives a stack strace [19:19:23] that doesn't sound great [19:19:30] [03:03:56] !log Manually started ircecho on fenari ; why doesn't this happen upon boot? Why didn't puppet start it? [19:19:47] shortly thereafter, fenari freaks out a second time [19:19:47] That's completely unrelated to commit notification to irc [19:19:48] ircecho is likely just because we don't have it forced enabled on fenari [19:20:07] I think at some point it was broken, because fenari was running hardy [19:20:16] so we disabled ircecho from starting [19:20:20] that just needs to be fixed [19:21:10] http://p.defau.lt/?M2vfcABmbWong5zdJOUTRQ [19:21:20] ircecho definitely did not cause the issue [19:21:28] it's very resource light [19:22:38] in fact, I rewrote it to use inotify [19:22:56] so, unless the kernel tells it a log file changed, it doesn't even bother to try to read it [19:23:22] robla: Nikerabbit: In case you're testing a log action on live meta-wiki and checking in cvn-meta yourself, note that the bot only purges i18n messages once a day. Ask in #cvn-bots if you want to purge it right away (only takes a second) [19:23:26] just got no CSS on officewiki, checking reports about CSS-free pages made recently [19:23:46] I just purged it for the bot in #cvn-meta [19:24:29] I think I know why it's broken [19:25:20] Krinkle: upload notifications seem to work [19:25:25] yeo [19:25:26] nice [19:26:04] Ryan_Lane: /puppet/files/ciabot_svn.py can you change line 114 from "lily.knams.wikimedia.org" to "smtp.pmtpa.wmnet" and commit and merge it please? [19:26:22] sure. gimme a sec [19:26:25] cheers [19:26:34] Krinkle: protect doesn't work still [19:26:59] indeed [19:28:08] robla---118: [[Special:Log/protect]] protect * RobLa * protected "[[File:2010-08-29 13.17.13.jpg]]" [edit=sysop] (indefinite) [move=sysop] (indefinite) [upload=sysop] (indefinite) [19:28:08] robla-trunk: [[Special:Log/protect]] protect * RobLa * protected "[[File:2010-08-29 13.17.08.jpg]]" [19:28:20] that's from my local test prior to Nikerabbit's fix [19:28:40] I think....hrm, maybe not [19:28:52] *robla reverts his local mod temporarily [19:30:10] looks like the order is reversed in 1.19 [19:31:52] Reedy: merged. where do you need a puppet force run? fenari? [19:31:58] formey please [19:32:02] RoanKattouw: The talk you have or will submit to Wikimania, am I on that? I'm planning to submit a talk about Countervandalism [19:32:08] ok [19:32:11] sec [19:32:17] I don't think you are, but we haven't discussed that yet [19:32:22] Well, wait, you wouldn't be [19:32:22] forcing it [19:32:25] It's visual editor [19:32:31] ok [19:32:41] (although I might be on that project in general, in the future) [19:32:59] but not enough to have useful things to say at Wikimania [19:44:34] brion: http://integration.mediawiki.org/testswarm/job/1492/ [19:44:59] looks like textSelection is partly broken for real, it's not emitting a DLL error for IE like it did before [19:45:07] also affecting IE10 and IE9 now [19:45:23] it might be a real one this time [19:45:27] a real bug [19:45:30] gosh [19:45:43] blugh [19:46:22] Reedy, RoanKattouw: is the api wonky in trunk right now? i'm trying something like this locally "?action=opensearch&limit=5&namespace=0&search=bananas&format=xml" and always getting just '["bananas",[]]' (it appears to be improperly handling 'format') [19:46:43] That's weird [19:46:44] Well [19:46:46] No [19:46:56] Extension:OpenSearchXML [19:47:04] oh. [19:47:31] yeah we never merged that did we [19:47:34] Krinkle: I can't tell the difference between the 1.18 and trunk versions of protect IRC log entries [19:47:44] robla: brb [19:47:46] dinnertime [19:48:48] we did it as an extension because microsoft paid us to do it but keep it under wraps until their ie8 beta launch or something :) [19:49:05] plus it's a bit hacky in how it does extraction of text and thumbnails [20:25:07] My mom is starting a habbit of calling me on my cellphone instead of googling. Yikes. [20:25:30] ! [20:25:31] I think there's an xkcd on that one ;P [20:26:03] It works if you know that the person on the other end of the line is likely to be at a computer [20:26:25] I've done it a few times to my mates when I've been out and about and 3g sucks [20:26:55] sure, but this is different. She's sitting at her computer browsing a website and wondering about something. [20:27:02] Oh [20:27:03] Awesome [20:27:24] in this case she's looking to buy a second hand iPod touch and came across an iPod classic, wondering if it can do Skype. [20:27:41] the answer is no, but that's ok. Mothers have an exempt :) [20:28:17] Suppose it's better then buying it wrongly [20:28:22] yep [20:47:36] maplebed: hey, I have a request for you -- sometime this week, could you pop into https://www.mediawiki.org/wiki/Roadmap and elaborate a little on what needs to happen with Swift in March & April? [20:47:54] sure. [20:48:00] Thanks maplebed [21:00:41] chrismcmahon: (and others). please test http://test2.wikipedia.org/wiki/Special:UploadWizard [21:01:10] robla: I just now uploaded this with no issues https://test2.wikipedia.org/wiki/File:Old_bike.jpg [21:03:06] (and got some proper error msgs, "file already exists", required fields, etc. [21:03:23] sumanah: random idea I thought you might find interesting or perhaps already exists but I don't know about: Wikimedia Tech Talk (like Google's) done for a small audience in the office and recorded and pushed online to various channels. We have a lot of knowledge to spread and lots of it is in our very office. Would also be a good practice for when going to conferences and we'd have an easy link to send to conferences to show what kind of stuff we [21:03:23] that they will get if they invite us to their conference. [21:04:34] I've seen several talks being recorded in the office already and watched them online. If done more organized (afaik now they're just sporadically done and uploaded to a Commons category) [21:04:42] show what kind of stuff we ....? [21:04:44] .. it could work [21:05:02] .[irc charlimit} what kind of stuff we got that they will get if they invite us to their conference. [21:05:34] My client is not showing me the limit was reached before or after sending. [21:05:47] it ate the "got" and then continued on another line(?) [21:09:26] chrismcmahon: could you do some quick IE testing there as well? [21:09:39] robla: I can look with IE7 [21:09:47] wonderful [21:10:02] (not that IE7 is a wonderful browser, but it's wonderful to test it) :) [21:14:44] robla: getting a syntax error and page freeze on the upload page in IE7 [21:14:59] RoanKattouw: ^ [21:15:10] oh, wait, it finally returned [21:16:12] chrismcmahon: what was the syntax error? [21:16:47] wow, you can use ie7 on Windows 3.1x? [21:16:47] one moment. I did ultimately succeed: http://test2.wikipedia.org/wiki/File:Ie7_icons.JPG [21:16:59] grr, ie5.x [21:17:06] yay [21:17:13] chrismcmahon: Did it show you thumbnails during the process? [21:19:57] running again. first off on Special:UploadWizard I see "Line 2/Char 1/Error: Syntax error/Code 0/URL: http://test2.wikipedia.org/wiki/Special:UploadWizard" [21:20:05] (this VM is a dog, too, sorry) [21:21:26] then see error "instActive is null or not an object" when choosing a file [21:22:19] no errors after that [21:22:33] and I see thumbnails [21:29:18] actually, I see a thumbnail after the upload but not on the step before, but that could be because the error msg. masks it. [21:30:32] slight layout glitch on the textbox for "Description", it appears to the right side of the language selectbox instead of below [22:03:01] AaronSchulz Reedy: delete/restore still fails. delete/delete is working again on irc [22:04:04] Krinkle: what fails about it? [22:10:53] Reedy: same thing that's been failing from the beginning, it's not detected as a restore and ignored as such [22:11:08] either the message key is different or the variable replacement variables are different [22:11:13] or the bot is broken suddently [22:11:17] case 'restore': [22:11:17] $text = wfMsgExt( 'undeletedarticle', $msgOpts, $target ); [22:11:21] will investigate [22:11:28] 'undeletedarticle' => 'restored "[[$1]]"', [22:12:07] Match udm = ((Project)Program.prjlist[rce.project]).rrestoreRegex.Match(rce.comment); [22:12:07] rce.title = Project.translateNamespace(rce.project, udm.Groups["item1"].Captures[0].Value); [22:12:09] item1 [22:12:18] I wonder what $text = $this->getActionText(); is for when it's set to null later [22:12:34] MediaWiki:Undeletedarticle [22:12:39] should match [22:12:44] Hm.. weird [22:13:04] rc-pmtpa: [[Special:Log/delete]] delete * PeterSymonds * deleted "[[User:Krinkle/Sandbox2]]": test [22:13:04] [9:58pm] rc-pmtpa: [[Special:Log/delete]] restore * PeterSymonds * restored "[[User:Krinkle/Sandbox2]]": 2 revisions restored: test done [22:17:23] AaronSchulz: seeing 1 Warning: Invalid argument supplied for foreach() in /usr/local/apache/common-local/wmf-config/swift.php on line 45 periodically [22:17:44] yeah, saw that before [22:19:18] chrismcmahon: anything look weird on commons? [22:19:40] robla: I've been zooming around commons in Chrome and nothing seems out of place. [22:26:18] anyone available to work on a javascript problem in upload wizard [22:26:19] ? [22:26:48] Reedy: fixed [22:27:12] *robla files bug first [22:40:17] aforementioned javascript problem: https://bugzilla.wikimedia.org/show_bug.cgi?id=34601 [23:09:44] brion: would you have time to look at bug 34601? [23:09:48] !b 34601 [23:09:48] --elephant-- https://bugzilla.wikimedia.org/show_bug.cgi?id=34601 [23:10:24] robla, i'll take a peek [23:10:30] thanks! [23:14:11] managed to repro, lemme see if i can diagnose [23:23:07] did the Commons update go all right, or is it still delaying tomorrow's deploy to the non-Wikipedia projects? [23:24:57] *sumanah is poised to update the Roadmap page [23:26:54] well it's definitely mispositioning the input [23:27:06] i'm not quite sure why but it may be related to css load order or something [23:29:34] brion: the new "improved" Commons is? [23:32:47] *AaronSchulz gives r111809 to roan [23:32:53] !r 111809 [23:32:53] --elephant-- http://www.mediawiki.org/wiki/Special:Code/MediaWiki/111809 [23:33:44] Oh, that's not needed anymore [23:33:46] But ... [23:33:54] well it's good for completeness [23:34:57] AaronSchulz: OKed [23:37:36] robla, ok i can't repro the problem on my local *install* so i'm not sure if this is going to help or not [23:37:47] but i suspect it's due to styles loading after we've started work or something [23:38:02] the input box gets mispositioned, and doesn't end up sitting over the button [23:38:13] forcing the styles to load immediately may help [23:39:55] TrevorParscal, hey do you know if an addModuleStyles should force styles to load sooner than an addModule? [23:40:14] RoanKattouw: ^ [23:40:42] Yes it does [23:40:43] yeah, addModuleStyles adds a link to them in the head  [23:40:46] yeah looks like that'll stick ext.uploadwizard's styles into the [23:40:46] dang you! [23:40:49] so this might work [23:40:53] addModuleStyles should be used for things that are in the HTML [23:41:04] addModule should be used for things that are added by JS [23:41:06] *TrevorParscal concurs [23:41:10] this is dynamically-generated HTML created by the JS [23:41:15] Oh [23:41:25] Well then it *should* not need to load earlier [23:41:25] but, i'm not sure they're getting run before the styles load consistently [23:41:26] that should not use addModuleStyles then [23:41:44] Maybe there are other styles that apply to it that come from other places? [23:41:55] Like, the styles and the script that adds the box need to be in the same module [23:42:48] looks like they're both in ext.uploadWizard if i read this correctly [23:42:56] any chance that the JS is getting run before the styles ? [23:43:31] robla: I'm updating https://www.mediawiki.org/wiki/MediaWiki_1.19/Roadmap to indicate that today's Commons deployment is DONE. right? [23:43:41] yup [23:46:23] brion: Only if the JS and CSS are in different modules [23:46:37] *RoanKattouw has to turn off his screens now, storytellers are filming at this desk [23:47:49] RoanKattouw_away is now known as RoanTheCelbrity [23:48:24] RoanTheCelbrity is now know as TrevorCantSpellCelebrity [23:48:38] Celbrity is 8 characters [23:48:52] TrevorParscal: are you in DOS? [23:50:00] TrevorCantSpellCelebrity is now known as Trevor~1 [23:50:38] Ha [23:50:48] ! r112166 [23:50:59] where's that bot bah [23:51:14] !r112166 [23:51:14] --elephant-- I don't know anything about "r112166". [23:51:17] heh [23:51:18] !r 112166 [23:51:18] --elephant-- http://www.mediawiki.org/wiki/Special:Code/MediaWiki/112166 [23:51:22] ah, the space is all [23:51:28] anybody willing to merge/deploy https://www.mediawiki.org/wiki/Special:Code/MediaWiki/112166 and see if it helps? [23:51:39] sure [23:51:43] i can't repro it on a local install (timing issues appear to be involved) so [23:51:44] thanks! [23:52:49] or brion can get shell again :p [23:52:56] nooooooooooooooo :) [23:55:40] It's live [23:57:54] robla, ok good sign: with the fix in a case that i think would have failed before (visible flash of unstyled list during initialization) didn't fail for me this time [23:57:57] Reedy, thx [23:59:28] ok, new status is up https://www.mediawiki.org/wiki/MediaWiki_1.19/Roadmap#Status [23:59:30] crap. still got it to fail once