[00:33:41] <halfak>	 Hey yuvipanda, you back>?
[00:33:47] <yuvipanda>	 halfak: boom, yes.
[00:33:49] <yuvipanda>	 halfak: which host again?
[00:34:00] <halfak>	 labels.eqiad.wmflabs
[00:35:20] <yuvipanda>	 halfak: hmm, interesting
[00:35:22] * yuvipanda modifies things
[00:37:11] <yuvipanda>	 halfak: boom! fixed
[00:37:16] <yuvipanda>	 halfak: your problem was: it had to have the .ini extension
[00:37:21] <yuvipanda>	 it had no idea what to do with a .uwsgi extension
[00:37:22] <halfak>	 Oh!
[00:37:30] <yuvipanda>	 it supports ini, yaml, xml, etc
[00:37:37] <yuvipanda>	 and so was confused as to what you wanted it to do :)
[00:37:44] <halfak>	 Ha!  I should have know.  I ran into that with apache before too. 
[00:37:52] <halfak>	 Thanks!
[00:38:07] <yuvipanda>	 yw!
[00:38:18] <yuvipanda>	 halfak: rememeber we should puppetize all of this in the end :)
[00:38:28] <yuvipanda>	 or you get https://www.wikiwand.com/en/Software_rot
[00:38:39] <halfak>	 yuvipanda, indeed.  It's got to work first. 
[00:38:42] <yuvipanda>	 yeah
[00:39:06] <halfak>	 \o/  I get a [fail] 
[00:39:11] <halfak>	 Now to look for the logs
[00:39:16] <yuvipanda>	 halfak: yeah /var/log/uwsgi/app
[00:39:25] <yuvipanda>	 halfak: I don’t understand why you’re passing a param to —deamonize or what not.
[00:39:33] <yuvipanda>	 I don’t think you need deamonize at all
[00:40:30] <halfak>	 yuvipanda, was trying lots of things. 
[00:40:37] <yuvipanda>	 :D
[00:44:20] * halfak reads through logs trying to figure out what the error is. 
[00:44:37] <halfak>	 I don't see an error listed. 
[00:51:26] <halfak>	 I really with that upstart would print out an error message for every [fail]
[00:57:33] <halfak>	 OK.  So,  it appears to have [fail]'d, but it is running. 
[00:57:39] <halfak>	 And I can access it from the web. 
[00:57:44] <halfak>	 yuvipanda, ^ common problem?
[00:59:51] <yuvipanda>	 hahaha
[00:59:53] <yuvipanda>	 not sure...
[01:00:15] <yuvipanda>	 halfak: I think in general the uwsgi package’s init scripts are wonkyasfuck
[01:00:25] <yuvipanda>	 halfak: and in production / puppet we use our own custom one that’s much better
[01:01:13] <yuvipanda>	 halfak: I’m not sure how exactly to fix this for you atm
[02:21:40] <halfak>	 yuvipanda, thanks for taking a look.  Sorry I had to run earlier.  
[02:22:49] * halfak digs into npm stuff. 
[17:41:50] <halfak>	 guillom, just saw https://phabricator.wikimedia.org/T98077 \o/
[17:42:17] <guillom>	 halfak: Yup; I decided to follow your advice :)
[17:43:11] <guillom>	 halfak: I'll probably need to ask you a few questions to get started, but after that I'll bother you less often.
[17:43:42] <halfak>	 No worries.  Happy to help when you are ready. 
[17:43:50] <guillom>	 Thanks!
[17:44:08] <yuvipanda>	 halfak: ah, ok. did you find a way to fix it?
[17:44:51] <halfak>	 yuvipanda, I don't know why it says [fail], but the system is online and working OK. 
[17:45:03] <yuvipanda>	 halfak: hahaha
[17:45:03] <yuvipanda>	 ok
[17:45:18] <halfak>	 Scope it out: http://labels.wmflabs.org/gadget/
[17:46:14] <halfak>	 It's actually intended to be used as an on-wiki gadget, but I plan to maintain that stand-alone interface as well. See docs https://meta.wikimedia.org/wiki/Wiki_labels
[17:46:29] <yuvipanda>	 halfak: niccceee :D
[17:46:36] <yuvipanda>	 halfak: hmm, thiscouldbeontoollabsatsomepointmaybe
[17:47:10] <halfak>	 yuvipanda, likely, yes. 
[17:47:25] <yuvipanda>	 yeah, we’ve sweet uwsgi support there these days
[17:47:28] <halfak>	 But I have a bit of an aversion to shared computing spaces. 
[17:47:47] <halfak>	 If there's nice compartmentalization, then I'll be more likely to start my next project there. 
[17:49:41] <yuvipanda>	 halfak: I’m going to say it’s better than what it used to be, and we have nicer limits in place now, and increasing it over time :)
[17:50:11] <yuvipanda>	 halfak: and I’ll have numbers for it soon :)
[17:50:17] <halfak>	 :) 
[17:50:34] <yuvipanda>	 halfak: in fact, I already have numbers for it: http://p.catchpoint.com/ui/Entry/PD/V/A.RNP-Ov-jSUbDu8Jdg/ErLK
[17:50:45] <yuvipanda>	 the uwsgi uptime is fairly hight
[17:50:49] <yuvipanda>	 because it doesn’t depend on NFS :)
[17:50:55] <harej>	 what's with about wiki label wikiprojects?
[17:52:24] <halfak>	 harej, wat
[17:52:33] <harej>	 on the meta page
[17:52:46] <harej>	 "Wiki labels is a human computation system for use in "Wiki labels" WikiProjects."
[17:52:56] <halfak>	 Yes. 
[17:53:18] <halfak>	 The initial question you asked was not sensical. 
[17:53:30] <harej>	 what's this about*
[17:54:11] <halfak>	 Oh!  Well, you have a the link to the docs.  Now scope out enwiki project for it: https://en.wikipedia.org/wiki/Wikipedia:Labels
[17:54:45] <harej>	 I'm surprised WP:Labels wasn't already used as a page name
[17:54:52] <halfak>	 :)  me too
[17:55:01] <harej>	 Today I am creating ~480 redirects
[17:56:00] <harej>	 116 to go!
[17:57:19] <halfak>	 That's a lot.  how come?
[17:58:59] <yuvipanda>	 halfak: it’s because mysql sucks at case insensitivity :)
[17:59:01] <yuvipanda>	 (I think)
[18:00:09] <halfak>	 yuvipanda, normalize before DB?
[18:00:10] <yuvipanda>	 harej: ^
[18:00:13] <halfak>	 Seems reasonable if we are going to store titles as byte arrays
[18:03:47] <harej>	 yuvipanda: that's not it
[18:03:51] <harej>	 i solved the case insensitivity problem
[18:04:08] <harej>	 ...by normalizing, exactly as halfak said
[18:04:35] * halfak flexes his insight muscles
[18:04:36] <harej>	 the new problem is "wikipedians are inconsistent"
[18:04:51] * halfak has develops a headache
[18:05:37] <harej>	 Case in point:  https://www.irccloud.com/pastebin/BnAOdvLy
[18:07:18] <harej>	 hello SVentura!
[18:08:16] <harej>	 The goal is a robust way to associate WikiProject assessment categories with their WikiProject
[18:08:21] <harej>	 Key word: "robust"
[18:29:30] <Ironholds>	 oh, that reminds me. halfak, you'll like this
[18:29:41] <Ironholds>	 R bug report: "R's variable names are case sensitive! WTF!"
[18:29:47] <halfak>	 lol
[18:29:48] <halfak>	 YES
[18:29:50] <halfak>	 <3
[18:29:55] <halfak>	 Case sensitivity!
[18:30:16] <halfak>	 What programming language doesn't have case sensitive variable names?
[18:30:20] <halfak>	 SQL I guess. 
[18:30:28] <halfak>	 (which is a shame!)
[18:52:00] <yuvipanda>	 halfak: vb!
[18:54:40] <halfak>	 yuvipanda, blerg
[19:08:15] <Ironholds>	 halfak, "SAS and SPSS"
[19:08:23] <Ironholds>	 those are the environments we should be modelling ourselves on
[19:13:29] <halfak>	 blerge again
[19:14:03] <halfak>	 Anyone seen DarTar around today?
[19:14:22] <Ironholds>	 goddammit hive
[19:14:49] <guillom>	 halfak: nope. But his calendar is quite busy for the day.
[19:15:15] * halfak looks at all the busy days on Dario's calendar
[19:16:52] <ottomata>	 cluster iS BUUSSYYYY and lagging
[19:16:58] <ottomata>	 elley is never online!
[19:17:08] <ottomata>	 isn't there some WMF requirement of being on IRC when you are working? :)
[19:17:44] <halfak>	 ottomata, OMG want that
[19:18:02] <halfak>	 ottomata, you should make an IRC bot that won't let you submit queries unless you are on IRC. 
[19:18:11] <halfak>	 >:) 
[19:18:11] <ottomata>	 haha, yeah for sure
[19:18:14] <guillom>	 heh
[19:18:34] <guillom>	 There's such a requirement for SWAT deploys.
[19:18:39] <yuvipanda>	 w000000
[19:18:45] <yuvipanda>	 got a 1 year schengen visa!!!!!1
[19:18:50] <guillom>	 Not on IRC => Your patch doesn't get deployed.
[19:18:50] <harej>	 :D
[19:19:03] <guillom>	 yuvipanda: woot! Congrats!
[19:19:08] <yuvipanda>	 ty :D
[19:52:18] <harej>	 halfak, yuvipanda: I said I needed to create ~480 redirects, but it turns out that was out of 2,108
[19:52:37] <harej>	 as in, 480 edge cases out of 2,108 total cases
[20:10:16] <leila>	 hey Ironholds. Do you know if Wikidata item pageviews are available publicly? Something like http://stats.grok.se/
[20:16:34] <halfak>	 harej, are you making redirects for WikiProject categories? 
[20:16:40] <harej>	 no?
[20:17:18] <harej>	 i'm mapping categories to project pages, which in some cases required me to create redirects
[20:17:19] <halfak>	 OK.   I was starting to worry that I needed to account for category redirects in my analysis of categories. 
[20:17:27] <harej>	 not that hard
[20:17:28] <halfak>	 How's that?
[20:18:00] <harej>	 all category redirects are soft redirects and are sorted as such, so really you'd just need to join the categorylinks table
[20:18:30] <halfak>	 harej, wouldn't I need to join the 'redirect' table?
[20:18:42] <harej>	 categories don't have real redirects as far as i can tell
[20:18:46] <harej>	 only the soft ones
[20:20:20] <halfak>	 Oh.  How does one detect a soft redirect?
[20:21:12] <harej>	 So, based on a WikiProject assessment category, you can infer the name of its associated WikiProject programmatically. At least I was able to 77% of the time; the remaining 23% was because (a) the category naming scheme deviated substantially from the name of the project and (b) there was a lack of a suitable redirect. Through improvements I managed to
[20:21:12] <harej>	 reduce it down to 23% (i.e. 480 redirects) from a higher number, but I figured it was a good idea to actually solve the problem anyway. For example, redirects for "WikiProject Canadian military history" --> "WikiProject Military history/Canadian task force", things like that.
[20:21:31] <harej>	 To answer your question,
[20:21:49] <harej>	 All categories that are soft redirects are in the category Wikipedia soft redirected categories
[20:21:52] <harej>	 At least on English.
[20:22:28] <harej>	 select cl_from from categorylinks where cl_to = "Wikipedia_soft_redirected_categories";
[20:22:42] <harej>	 or,
[20:23:44] <Ironholds>	 leila, not to my knowledge.
[20:24:19] <harej>	 select page.page_namespace, page.page_title from page join categorylinks on page.page_id = categorylinks.cl_from where cl_to = "Wikipedia_soft_redirected_categories";
[20:24:23] <harej>	 In fact, I'mma run that query.
[20:27:35] <leila>	 thanks, Ironholds.
[20:30:10] <harej>	 halfak: http://quarry.wmflabs.org/query/3469
[20:31:53] * halfak runs to meeting, but will remember to read scrollback
[22:57:18] <lzia>	 halfak: do you know if we have statistics on how notifications are used? (what percentage of them are viewed, what percentage read, etc.?)
[23:35:58] <halfak>	 lzia, sorry was in meeting
[23:36:05] <halfak>	 I don't know if we have anything like that. 
[23:36:23] <halfak>	 There was an old study we ran examining the engagement level of newcomers around the deployment of echo. 
[23:36:27] <halfak>	 TLDR: no effect. 
[23:36:50] <harej>	 (echo is useful for other reasons of course)
[23:46:27] <lzia>	 Hi Nettrom. question for you: do you know if we store Babel template data somewhere in the databases?