[12:38:48] hashar- Good morning (although I think it's afternoon where you are?) [12:44:02] hello [12:56:55] anomie: hey :-] [12:57:03] anomie: welcome aboard! [12:57:12] hashar- Thanks! [12:57:45] anomie: Rob asked me to find you something to do on beta. I am afraid I did not take the time to find something for you yet hehe [12:58:33] hashar- I'll just continue looking at the stuff on my Gerrit dashboard then. I haven't had time to look at the stuff Sumana has pushed to me for the past month or so. [13:10:09] anomie: please ping me whenever you need something :-] [13:10:24] hashar- I will [15:44:24] * Reedy glares at ori-l [15:58:32] out for now [15:58:38] will be back later this evening [16:39:26] chrismcmahon: zeljkof test2 should be useable again [16:40:00] Reedy: thanks, I have just noticed it is back [16:41:48] thanks Reedy [16:42:58] Change abandoned: Siebrand; "Dropping ownership." [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/26959 [16:43:03] Change abandoned: Siebrand; "Dropping ownership." [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/26960 [16:43:07] Change abandoned: Siebrand; "Dropping ownership." [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/26961 [16:43:11] Change abandoned: Siebrand; "Dropping ownership." [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/26962 [16:43:18] New patchset: Siebrand; "Add job for PageTriage extension" [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/26963 [16:43:18] Change abandoned: Siebrand; "Dropping ownership." [integration/jenkins] (master) - https://gerrit.wikimedia.org/r/26963 [17:45:05] kaldari, https://gerrit.wikimedia.org/r/#/c/15959/6/Hooks.php - are you sure about that $victimId = $oldRevision->getUser(); ? [17:45:37] Looking at Revision::getUser, it seems to handle whether the user's name was deleted or not, not whether they're anonymous or not [17:47:20] Krenair: I'll take a look. Thanks [17:50:06] bsitu, has WMF legal signed off on the text for the footer ( https://gerrit.wikimedia.org/r/#/c/30066/ ) ? [17:50:33] krenair: not yet [17:51:44] ok [17:51:59] i just -2 it [17:52:13] yep, I saw [18:00:45] RoanKattouw: https://gerrit.wikimedia.org/r/30627 [18:03:20] rmoen: Merged, fixes it for me [18:25:44] I really hate EchoBasicFormatter's constructor [19:17:07] anyone around who is working on parsoid in labs? [19:19:29] apergos: gwicke is [19:19:44] good cause [19:19:51] cronspam is coming from there once a minute [19:20:13] he really needs to fix that cause it's been going for days and every ops team member gets hit with it [19:20:30] is he an office person or off site? [19:21:03] Wait, [19:21:04] apergos: interesting- why is cron output sent to ops? [19:21:11] labs cronspam goes to you? [19:21:11] /data/project/parsoid/js/tests/client/runmany.sh 2 [19:21:25] /data/project/parsoid/js/tests/client/runmany.sh: not found [19:21:39] apergos: Which machine does that come from? [19:22:09] um [19:22:11] apergos: if it came from parsoid-spof, then it should be fixed now [19:22:22] root@parsoid-spof.pmtpa.wmflabs [19:22:50] labs cron output goes to ops because it's a feature down the road for everyone to get their own cronspam, but not yet implemented [19:23:03] remember that some things about labs are still under develpment [19:23:20] * gwicke would expect it to go to /dev/null [19:23:27] ah no [19:23:32] unless configured otherwise [19:23:32] it should go somewhere [19:24:04] because the fact that it produces output means that something is broken or that there's a notification that folks should pay attention to [19:24:04] unless deliberately set up otherwise [19:25:12] that seems to have got it, thanks [19:25:17] my mailbox thanks you too [19:25:53] apergos: I'm surprised you are not yet flooded with messages at this point [19:26:56] oh I am, but when I get over 1000 a day that's the point where heads start rolling :-P [19:27:17] ah, ok ;) [19:53:14] I have reports on VP/T that AFT is breaking the watchlist [19:57:00] same as a couple of days ago [19:57:00] 'undefined' is not a function (near '....stall( 'click', function(...') [19:57:22] ori-l: [19:58:00] thedj: stall wasn't to blame then and (i'd wager) not to blame now -- AFT wasn't declaring E3Experiments as a dependency [19:58:05] ^ mlitn [19:58:12] but taking a look anyhow to see if i can figure it out [19:59:27] robla: it's the 6th break of global JS this week or something. that's not good. [20:05:39] ticket 41505 [20:13:16] thedj: are there any reports other than Gary King's? [20:14:29] i see it too [20:15:09] can you give me an article to add to my watchlist that would trigger this? i'm not too familiar with AFT and find the description of the bug a bit confusing. [20:15:45] ori-l: hi, i'm here [20:15:48] just catching up [20:16:36] mlitn: ok, great. i'm going to get back to what i was doing earlier, but let me know if you need my help. [20:16:43] sure, thanks [20:16:50] ori-l: Help:Searching perhaps.... [20:17:42] thedj: yup, that does it [21:03:42] !google x x [21:03:42] https://www.google.com/search?q=x+x [21:50:31] Tim-away: ping [22:03:57] hi [22:06:13] preilly: did you see my pull request? [22:07:05] TimStarling: yeah, https://github.com/php/php-src/pull/223 [22:07:21] yes [22:07:27] TimStarling: did you see https://travis-ci.org/#!/php/php-src/builds/2969489 [22:07:35] yes, I was just looking at that [22:07:53] I don't think it is my fault [22:08:15] this one has at least some of the same failed tests: https://travis-ci.org/#!/php/php-src/builds/2894515 [22:08:17] TimStarling: yeah I would tend to agree [22:08:23] all of the recent php-src builds seem to have failed [22:09:18] binasher: take a look at https://github.com/php/php-src/pull/223 [22:09:53] TimStarling: yeah I think the travis config is wonky [22:12:17] my other PR (https://github.com/php/php-src/pull/222) was merged straight away, which was nice [22:12:46] but I guess it's easier to get PRs accepted in neglected corners of testing code ;) [22:14:03] TimStarling: yes indeed [22:17:32] TimStarling: I've just poked a few folks to try and take a look [22:17:41] thanks [22:17:41] TimStarling: I'll let you know if I hear anything [22:17:48] ok [22:20:20] is rrichards still active? felipe assigned the bug to him [22:20:47] TimStarling: I'm talking to Derick Rethans about it right now [22:24:06] excellent, my libxml2 patch was accepted: https://bugzilla.gnome.org/show_bug.cgi?id=687084 [22:25:25] TimStarling: I just pinged Rob Richards [22:27:21] TimStarling: http://git.gnome.org/browse/libxml2/commit/?id=0ad948ede2b5060a144c72e4e27c38d24a272ef4 nice! [22:27:37] yup [22:28:41] Hi j^ : just read your email [22:30:22] we're thinking maybe the best strategy now is: 1) Deploy to enwiki 2) deploy everywhere else except commons 3) deploy to commons [22:30:51] TimStarling: do you have any plans to revisit the ChronologyProtector Class? [22:31:12] robla: yes that might be better [22:31:28] no, is there a bug report open? [22:31:37] j^: ok....cool, I'll declare that to be the plan on wikitech-l [22:31:59] mdale: you around? [22:32:07] I am [22:32:13] w00t! [22:32:34] (03:30:20 PM) robla: we're thinking maybe the best strategy now is: 1) Deploy to enwiki 2) deploy everywhere else except commons 3) deploy to commons [22:32:44] mdale: make sense to you as well? [22:32:52] TimStarling: the /* DatabaseBase::getMasterPos */ show master status calls is the second worst slow query taking 8.16 % of the time and is 1.05 % of the queries [22:32:54] ... yea.. maybe start with not enwiki ;) .. but other then that yes [22:33:06] mdale: well, here's my logic: [22:33:07] TimStarling: based off of looking at https://ishmael.wikimedia.org/sample/?host=db63&hours=24&sort=time [22:33:08] there might be a better lower traffic wiki to start with [22:33:09] funny [22:33:22] why would that be slow? [22:33:52] I think we've got enough low traffic testing, so that now it's time to actually get enough traffic that we'll likely get feedback on any problem big or small [22:33:55] probably acquires a global lock [22:34:23] mdale: enwiki has the added benefit that there's not a communication barrier [22:34:38] TimStarling: it is probably locking related [22:34:58] robla: ok [22:35:01] mdale: (for all of us...clearly some of us wouldn't have a problem with, say dewiki or eswiki, but still) [22:35:03] TimStarling: but if we fix that call to something else it would help a lot [22:35:26] yea... maybe the foundation wiki for example? [22:36:28] mdale: I'm worried foundation wiki doesn't have a robust enough feedback mechanism [22:36:31] has chrismcmahon done a QA per-browser report? [22:36:56] preilly: are the numbers in brackets milliseconds? [22:37:27] robla: yes fair enough. [22:37:35] mdale: not really recently, but there was a bunch of testing over the summer [22:37:39] hi mdale I haven't done anything with TMH recently, should I? there's a re-deploy of EventLogging in about 30 minutes I'm going to be looking at [22:37:53] in IE8 especially [22:37:56] maybe cumulative milliseconds [22:38:03] TimStarling: I think this page https://ishmael.wikimedia.org/sample/more.php?host=db63&hours=24&checksum=5662166594118698801 shows Avg time (ms) [22:38:18] chrismcmahon: yea would not hurt to take a look at IE8 test2 pages. [22:38:19] chrismcmahon: it'd be good if you gave TMH a good look tomorrow, assuming you've got the cycles [22:38:32] chrismcmahon: today you should focus on the EventHandler stuff [22:39:03] tomorrow is pretty open after an early meeting [22:39:03] well, we can hack around it with some sort of memcached-based proxy [22:39:16] yea lets get chrismcmahon to do a pass, there is a minor close caption issue with multiple videos on the page that I will fix prior to chrismcmahon review tmw. [22:39:17] * chrismcmahon makes a note [22:39:30] xlnt [22:40:09] or I can ping domas, he likes to chase these sorts of locks [22:40:15] maybe he will have a patch for us already [22:41:56] TimStarling: that would be nice [22:43:26] TimStarling: do we still use this code for consistent ordering of events as seen by the user, despite replication? [22:43:33] why do we bother when we already have code to avoid hitting lagged slaves? [22:43:37] TimStarling: seems a bit heavy to me [22:43:40] preilly: yes [22:43:56] binasher: 30s is a lot of lag [22:44:20] when a user clicks "save" they expect to see the correct version of the article when they are redirected to the view page [22:44:30] and then if they go to their contributions, they expect to see the change they just made [22:44:34] even if it was less than 30 seconds ago [22:44:44] hitting save results in a parse though, right? [22:45:03] with the default user options [22:45:11] then the page view uses the user's option [22:46:05] IIRC [22:47:00] https://bugzilla.wikimedia.org/show_bug.cgi?id=5311 [22:48:27] it was reopened in 2009 when ChronologyProtector broke [22:57:16] TimStarling: have you been able to poke at https://gerrit.wikimedia.org/r/#/c/30155/ [22:57:17] ? [22:57:30] not yet [23:03:55] report.py has gone a bit haywire [23:04:08] looks like some counters have overflowed, total time is negative [23:39:01] AaronSchulz: what's the current state of row caching? [23:39:17] preilly: have you seen this ^^^ ? [23:39:29] binasher: way in the backburner [23:39:57] I'm kind of torn on the approach, I think that draft is to complicated, and a simpler way would be another layer over the db classes [23:40:26] the downside to that is that one wouldn't be able to use the regular db functions to change stufff...so I'm kind of torn on that [23:40:41] * AaronSchulz is working on some filebackend simplifications and changes to reduce HEAD requests now [23:40:51] *too [23:41:00] * AaronSchulz makes that typo waaaay too much [23:41:35] Huh, no that was correct usage of 'to', wasn't it? [23:41:35] it is pretty complicated [23:41:39] i'd like preilly to take a look [23:42:02] RoanKattouw: I mean the "too" in "to complicated" [23:42:16] oic [23:42:30] lol, that looks so wrong but works phonetically [23:42:45] preilly: https://gerrit.wikimedia.org/r/#/c/10349/ [23:43:18] binasher: I've almost given up on that approach, meh [23:44:29] Reedy, your HTTPS-Everywhere change doesn't appear to be working for me.. [23:44:49] binasher: I also got rid of the cache-on-write stuff for being too complicated, I think that's were another layer is needed [23:44:59] Aren't you supposed to edit the rule on line 37? You just added targets [23:45:09] so one could make a simple DAO base class for things simple enough where that stuff works well [23:45:40] Krenair: Yeah.. [23:46:40] Though, it'll likely need to change in the near future [23:46:49] maybe use ORMRow ;) [23:46:55] I had been putting off submitting this myself until the lang domains were working [23:47:32] "working" [23:47:40] Well, not getting the wrong SSL certificate [23:52:58] I'd imagine few people would be visiting via the language subdomains though