[13:25:33] Good morning (for me) science people. [13:25:54] hi halfak [13:27:41] How you doin' YuviPanda [13:27:42] ? [13:27:54] good [13:27:57] sad to leave Europe [13:29:46] Leaving today? [13:30:05] halfak: 6am tomo [13:30:10] Gotcha. [13:30:22] That's a rough time for a flight [13:30:34] It won't exactly help you switch timezones either. [13:30:42] * halfak likes late flights out of Europe [13:30:45] halfak: ya but I'm fucked anyway lol :P [13:30:57] halfak: but that's what you get for deferring planning [13:30:57] Fair point. [13:31:07] halfak: but deferring planning let me spend 10days in berlin, so I'm ok with that [13:31:08] Heh. Speaking of which I need to plan a flight. [13:31:21] Gotta go visit Mizzou on Oct 6th [13:31:37] Gotta talk about ORES or socio-technical manipulations or something. [13:31:50] heh [13:31:51] nice [13:32:01] I like the requirements for giving a colloquium. [13:32:06] Be interesting for 30 minutes [13:32:21] halfak: I'll do the next step towards productization (use debian packages only, move away from pip) next week [13:32:44] I'm going to spend the next 6 months being more interesting than before. less traveling [13:33:43] YuviPanda, sounds good. I'ma need to learn something about packaging in case you guys get busy so that I can do upgrades as necessary later. [13:34:01] yeah, packaging in general is kindof a mess but python is ok [13:34:30] python easier than other choices? [13:35:40] yeah [13:37:43] :) I'm glad for that. [13:39:13] halfak: yes, the only nontrivial one we had was the scikit-learn [13:39:42] Is that because of it's dependencies on c libs like libsvm? [13:40:55] halfak: yeah [13:41:14] halfak: and cython [13:41:37] Ahh. I didn't realize there was much C inside of sklearn that wasn't part of scipy/numpy [13:58:39] 10Quarry: Time limit on quarry queries - https://phabricator.wikimedia.org/T111779#1616299 (10Jarekt) 3NEW [14:00:14] 10Quarry: Time limit on quarry queries - https://phabricator.wikimedia.org/T111779#1616307 (10yuvipanda) Those aren't actually running for months, I think - those happen when for some reason there's an unhandled exception in the query results serializer, preventing it from updating the status accordingly... Nee... [15:28:31] o/ [15:29:13] hey guillom, bearloga :) [15:29:24] hello Ironholds [15:31:04] hello Ironholds guillom how were your Labrador weekends? https://www.facebook.com/AnimalWelfareLeague/photos/a.238868932598.138662.222023457598/10153324019617599/?type=1&permPage=1 [15:53:13] bearloga, I have both a groan and an awwww for you [15:53:28] It was labrador weekend for my pup. :) [15:53:37] (who is half lab, but whatever) [15:53:59] bearloga: Some was good (Turner exhibition, vacation planning) and some was bad (delibitating allergies, insomnia). Good overall, I guess :) You? [15:54:08] * YuviPanda worked on monday :P [15:54:31] and am taking the day off on wednesday [15:54:32] YuviPanda: Did you remember to remove the holiday in ADP? [15:54:34] so I can fly [15:54:36] guillom: I tried [15:54:38] guillom: for 30minutes [15:54:41] ahaha [15:54:43] guillom: and gave up. but just emailed mark [15:54:44] and it was ok [15:54:55] YuviPanda: It's easy! Once you know how. [15:55:01] +1 for email and virtual paperwork [15:55:06] Joady sent an email once. [15:55:29] Good morning leila [15:55:36] good morning guillom. :-) [15:55:49] guillom: yeah, I was searching to find it [15:55:51] and didn't [15:55:58] guillom: but, I guess mark was happy with it and so... [15:55:58] hello yuvipanda. [15:56:55] o/ leila [15:56:58] YuviPanda: "To delete the holiday line, try clicking on the box on the left of the line, with the list icon. From there you should have the option to delete the line, and then save it. " [15:57:02] hello halfak. :-) [15:57:07] Did you see nettrom's ping re. the showcase? [15:57:21] (for next time) [15:57:21] guillom: I'll try next time [15:57:40] most of my collegemates went to work for ADP, something that makes me not feel particularly happy using it [15:57:45] leila: I'm back in SF tomorrow! [15:57:54] happy to hear it yuvipanda. [15:58:28] YuviPanda: To be honest, I only figured out how to do it because I thought this way was more accountable than keeping a text file with my floating holidays :) [15:58:39] guillom: :D [15:58:49] this was my first one [16:05:12] good morning leila! :D [16:05:55] hello bearloga. :D [17:13:10] o/ bee2502 [17:13:56] leila, Yesterday I met bee2502 -- who is looking to help out with some Data Science work. [17:14:11] ahoy halfak [17:14:18] I gave an overview of revscoring stuff. I figured you'd like to talk about content coverage stuff :) [17:14:21] o/ DarTar [17:14:25] Good trip? [17:14:26] :) [17:14:33] yeah, awesome conference [17:14:55] will report back later @ staff [17:17:01] Cool! Looking forward to it. [17:19:45] leila : Hi , How are you? :) I looked at this link halfak suggested : https://meta.wikimedia.org/wiki/Research:Increasing_article_coverage . It seems very interesting but very broad.I was hoping you could help me with getting started. [17:22:49] hi bee2502. :-) I'm doing well. nice to meet you here. [17:25:47] leila : Hi. Nice to meet you too. [17:26:13] Re increasing article coverage, for the next month or so we will be mostly focused on having the recommendation instance ready for testing so we will need mostly front-end engineering work. After the instance is ready, we may be able to use your help. I think to keep the conversation moving forward, it's good if you send me an email (leila@wikimedia.org) with what broad/specific areas you are interested to contribute to from the research perspective, [17:28:51] bee2502, ^ [17:29:04] leila : Cool ! [17:52:09] leila : I just mailed you. It would be great if you could just take a look at it. [17:52:24] sure, I'll do this by the end of the day bee2502. thanks! [17:54:57] leila : Great ! Thanks a lot. [18:02:46] FYI: https://meta.wikimedia.org/wiki/Meta:Requests_for_comment/Enable_flow_in_the_Research_talk_(203)_namespace [18:02:51] I've started an RFC [18:03:10] About a controversial MediaWiki extension [18:03:12] RIP me [18:03:52] ^ J-Mo, leila, Ironholds, guillom [18:04:03] If you have a moment to comment (or an add it to your todo for later) [18:05:49] commenting now, halfak. Related: do you think we should start up this discussion again? https://meta.wikimedia.org/wiki/Meta:Babel/Archives/2014-04#Enabling_an_opt-in_VisualEditor_preference_on_Meta [18:12:51] oh, and halfak: perhaps ping some of our research colleagues who aren't staff, but who do (or could) document their work in Research NS? I'm thinking people like Mako, Jodi, or Rosta. [18:13:08] +1 [18:14:12] I'll ping Mako now, before I forget. [18:14:14] And thank you for your quick comment :) [18:14:23] any time [18:15:27] Any we have an oppose [18:15:32] By someone who didn't sign their comment [18:20:56] sigh. [18:21:14] it's MarcoAurelio. He's a very active Metapedia admin. [18:22:13] maybe we should email the research-l list, halfak. Best way to contact the relevant researchers, who probably aren't watching their Meta talkpages. [18:22:27] GOod call. I'll draft something and get it out ASAP [18:24:12] awesome. if we can get half a dozen researchers saying "hey, we're the people who use this namespace, and we want this", it will be harder for opposers to shoot down the idea based solely on the argument "I don't like Flow" [18:26:48] Yeah. There's a current of "I hope flow is never successful" too. I dunno what to do about that. [18:27:00] I wish it was "I'd really life these changes to flow." [18:27:09] "I have these specific requirements that are lacking ..." [18:29:02] uh-huh. the schadenfreude in some of the responses to Danny's recent announcement about Flow development was pretty amazing. [18:30:01] J-Mo, where should I link people to try out flow? [18:30:11] Oh wait. I got it https://www.mediawiki.org/wiki/Talk:Sandbox [18:30:15] sorry [18:31:00] email sent [18:31:09] wooo! [18:34:54] 10Quarry: Time limit on quarry queries - https://phabricator.wikimedia.org/T111779#1617653 (10Jarekt) Ok So you are saying I should stop waiting for my http://quarry.wmflabs.org/query/5045 query results? ;) If they are all killed after 20 min., as queries from other tools like CatScan2, than no need for " easy... [18:40:02] Hi, I'm trying to write a query to find out users who have created 25 or more articles and are hence eligible for the Autopatrol right. What's the best way to find out which pages has a user authored? [18:42:00] Niharika_: if revision.rev_parent_id = 0, I believe that indicates that a revision was the first to the page. [18:44:03] J-Mo: Oh, okay. And rev_user will give me the user then. Thanks! That never struck me. [18:44:28] :D [18:44:34] I approve of this method. [18:44:36] +1 [18:44:43] J-Mo, and you self-deprecate around code. For shame, sir! [18:44:48] wait, opposite of that. For no shame, sir! [18:45:27] well, I like to keep people's expectations of my abilities nice and low. [18:45:44] btw, Niharika_ here's an example: http://quarry.wmflabs.org/query/1543 [18:46:32] psht. You are a smart and well-informed cookie [18:46:39] J-Mo: Helpful. :) [18:46:49] * J-Mo blushes and bats eyes at Ironholds [18:46:56] which I guess would be like a Aachener Printen? [18:47:01] That's the most complicated cookie I can think of [18:48:15] Like a bacon cookie [18:48:16] http://www.cookingchanneltv.com/recipes/nadia-g/milk-chocolate-chip-maple-syrup-glazed-bacon-cookies.html [18:48:32] Just wanted yall to know that is a thing [18:49:36] halfak: Also helpful! :) [18:50:25] Oh Wikipedia. https://en.wikipedia.org/wiki/List_of_cookies [18:50:32] RIP my productivity [18:50:38] List of list of cookies [18:50:47] List of featured cookies [18:51:18] I wanted to believe: https://en.wikipedia.org/wiki/List_of_list_of_cookies [18:51:22] I keep some handy in my browser. [18:51:32] Cookies? [18:51:33] Category:Gluten-free cookies with dark chocolate chips, facing left. [18:51:40] lol [18:52:36] Not as good as cookies, but a treat for you halfak: https://tools.wmflabs.org/archaeo/test.html#revertsOverTime [18:52:49] Now with revert predictions in the same chart as actual reverts. [18:53:06] zoom in to see the details per day [18:53:33] guillom, what do the colors mean? [18:53:48] blue = reverts, orange = predicted [18:53:54] Gotcha. [18:53:58] greenish = overlap [18:54:12] I need to get that vandalism predictor up. [18:54:24] So that you can know what reverts were probably for vandalism. [18:54:29] I've just managed to combine the series, haven't gotten around to adding the key yet! [18:55:06] Overall it seems that there are more predictions than there are reverts [18:55:17] (Just an observation, not Science.) [18:55:17] Yeah. That makes sense. [18:55:22] Where did you set the threshold? [18:55:27] true/false? [18:55:33] yep [18:55:41] was the easiest for starters [18:55:46] Yeah... that's ... regretfully not a useful prediction. [18:56:08] The SVC tries to optimize a stat (I think the f-score) and then picks that threshold. [18:56:24] Now that i have a chart, I can tweak the source data however we want :) [18:56:32] :) [18:56:38] I think that 80% is a good threshold. [18:56:47] It seems to be the threshold of "someone should probably look at this" [18:56:48] Good to know! [19:00:36] Noo! Nemo_bis. [19:00:43] s'ok. We can disagree here. [19:00:46] halfak: more MetaPedian opposition. Unless we get a sizable response from the external research crowd, I don't see much hope for this proposal: https://meta.wikimedia.org/wiki/Meta:Requests_for_comment/Enable_flow_in_the_Research_talk_(203)_namespace [19:00:55] Will you beta test my software still? :D [19:01:05] ah, yes. hi Nemo_bis [19:01:53] J-Mo, it's still early. Most of the people who work in the Research namespace aren't highly active Metapedians [19:02:29] yeah, I know. I'm concerned that they will see the intense opposition and be hesitant to participate. [19:02:43] I also haven't seen a substantial complaint raised. [19:03:04] If I were closing this discussion, I'd be troubled by a lack of reasoned argument from the opposition. [19:03:06] the complaints don't need to be substantial. [19:03:07] This is not a vote. [19:03:49] I mean, they should need to be substantial. But I don't have a lot of faith in process where something as polarizing as Flow is concerned. But we shall see… [19:04:02] * J-Mo whistles "Always look on the bright side of life" [19:04:28] Oh Brian. [19:04:41] ^_^ [19:17:38] halfak: ores is down? [19:17:46] It was. [19:17:48] Is it down again? [19:18:24] halfak: either down or very, very slow. [19:18:32] Yeah... Something is definitely up. [19:18:36] I'm looking into it. [19:19:26] Yup. All the workers are offline. [19:19:29] YuviPanda, ^ [19:20:44] * halfak restarts [19:21:24] * guillom just pictured hundreds of blue-collar workers inside the servers, taking a lunch break before retirning to scoring revisions. [19:21:35] returning* [19:21:39] * halfak pokes the workers with sticks [19:21:53] ragesoss, back online [19:21:57] Sorry for the trouble. [19:22:00] I'll monitor for a bit. [19:22:24] halfak: cool. no problem. intermitten failures aren' [19:22:40] t a significant problem for the dashboard. [19:22:48] they just make the builds fail. [19:22:51] :) [19:24:13] Yeah... Still shouldn't happen. [19:24:24] Anyway, we will get better and are quite serious about it. [19:24:32] Thanks for dealing iwth the downtime for now. [19:25:18] no prob. [19:25:27] I'm looking at https://en.wikipedia.org/wiki/Wikipedia:User_access_levels#Autopatrolled and strangely, 'autopatrol' doesn't seem to be a valid group in user_groups table. Is there another name for it? [19:25:37] we've got our first uses of wp10 in production now, btw [19:26:08] halfak: it's a feed of potential DYK candidates by student editors: http://dashboard.wikiedu.org/recent-activity [19:26:10] Oh. autoreviewer. [19:26:26] ragesoss: Neat! [19:26:29] ragesoss, \o/ [19:27:17] BTW if you ever do need to run without ORES, you can. [19:27:18] http://pythonhosted.org/wikiclass/functions.html#wikclass-score [19:27:37] The example on the main page of the docs is better: http://pythonhosted.org/wikiclass/ [19:28:09] halfak: The headings say 'wikclass'. Is that on purpose? [19:28:28] Oh yes. Wikiclass is utilities for Article quality models. [19:28:43] Editquality is the set of utilities for reverted/damaging/goodfaith models [19:28:52] halfak: yeah. we'll do that if we need to, although I'm happy to rely on ORES and not have to worry about that added piece of infrastructure, at this point. [19:29:02] https://github.com/wiki-ai/editquality [19:29:12] halfak: I meant 'wikclass' instead on 'wikiclass'. [19:29:19] Oh! [19:29:20] I thought it might be a typo. [19:29:20] Woops [19:30:08] Thanks for pointing it out. [19:30:12] I just fixed it in master. [19:30:23] Sure :) [19:30:27] Docs will be updated with the next version (which should be coming soon as I run tests on frwiki) [20:59:54] brb guys. Luna goes out and then I join the meeting