[00:10:56] hey DarTar, are we meeting? [00:11:42] DarTar: message me when you're around, I'm switching to another meeting. [00:23:31] lzia: The DarTar is a slippery creature :) [00:23:37] :D [00:23:44] I've given up for tonight guillom. [00:23:49] I also want to catch him to know where to meet him in 7 minutes. [00:23:51] I have another meeting starting in 7 min. [00:23:57] haha! [00:23:58] :D [00:24:02] Well, he asked me to stay late in the office tonight so I'm not going to give up :p [00:24:38] lzia: He said there was an urgent request from Lila, perhaps that's why he's not meeting with you? [00:24:49] no idea guillom. [00:30:25] * guillom is camping at Dario's desk. [10:41:43] Leila is in good company, I just got an email in French from OVH too ;) [13:58:40] Nemo_bis, what's OVH? [14:04:03] hosting service [14:05:01] * halfak reads scrollback and still doesn't understand :) [14:05:26] * halfak hopes that guillom and lzia got to talk to DarTar [14:05:48] guillom, when you get back, ping me and I'll tell you what I know about the urgent Lila request. [14:06:07] Speaking of which [14:06:11] * halfak checks on his progress [14:11:27] * halfak imports monthly counts. [14:11:56] I should really get this dataset up. It's very useful for answering trend questions that lend themselves to monthly totals. [14:39:02] halfak: Hey. I did. And he gave me super-urgent last-minute work and said I should talk to you about it today. Which I probably will, but after I get some breakfast :) [14:39:32] guillom, I should have some useful things for you. [14:39:37] And I have some data pre-generated. [14:41:09] Awesome [14:41:11] biab [15:03:46] halfak: I'm at your disposal in the next 90 minutes. After that I have the Editing quarterly review. [15:47:14] hello halfak. [15:47:22] o/ leila [15:47:43] so, in SoS, I /may/ briefly mention article recommendation's last week and next week work [15:47:54] I'm wondering if you have any item to be added to the list, halfak [15:48:17] Not sure I grok SoS yet, so I'm not sure what would be relevant. [15:48:52] basically it seems to me it is a filter on our standups [15:49:33] so, if we had standups, we would cherry-pick and report the top level items that we think others should know and/or will have impact on them, halfak [15:50:41] I don't know if you have done VE work last week, if you have, we can report it there. If you have done any other research that should be reported, or you plan to do something next week (yeah, we won't be really around), I can report that [15:50:53] I might note that ores is online and ready for light-medium level use with two models -- one that detects vandalism and one that detects overall article quality. We're in the process of deploying a distributed processing system that will allow us to scale to medium-heavy use. [15:51:17] do you have links for that halfak? [15:51:56] http://ores.wmflabs.org/scores/ & https://meta.wikimedia.org/wiki/Research_talk:Revision_scoring_as_a_service#ORES_performance_improvements [15:53:21] great. thanks halfak [15:58:55] Hey guillom, I just saw that I missed your ping 60 minutes ago. [15:59:05] I just have one thing to show you. [15:59:17] * halfak starts making a gist [15:59:59] hey [16:00:01] ok [16:01:44] guillom, https://gist.github.com/halfak/79962a8486c3aeb9c510 [16:02:02] So, there's this table called 'editor_month' that I have been curating on analytics-store [16:02:25] It gives you a count of monthly edits for all registered users in all wikis. [16:02:37] This gist gives you a good description of the table. [16:03:03] Note that the last line filters for 'enwiki' just so that I can take advantage of an index. [16:03:22] * guillom is looking. [16:03:46] You should be able to join this table with EventLogging to figured out whether a user registered via the mobile interface or not. [16:04:13] I have another script to split out mobile edits, but that's unlikely to finish any time soon. [16:08:23] Thanks halfak [16:08:33] guillom, HTH! [16:08:34] I think I mostly understand what the table contains [16:09:02] Oh! One more gotcha in there. [16:09:18] "revisions" is a count of all revisions saved whether the were in deleted pages or not. [16:09:26] "archived" is the count of revisions to deleted pages [16:09:49] So if you want to get a count of the # of revisions in the revision table at the time I ran the query, you want "revisions" - "archives" [16:10:21] A little confusing, but now that the new definitions are stateless, it's important to consider all revisions. In the end, it doesn't make much of a difference. [16:10:57] halfak: It does help; thanks :) However I've never used EventLogging, or even made joins, so I'm afraid whatever Lila needs is going to be above my current SQL skills. If I had nothing else to do by Sunday, I could probably level up and do this, but with my existing commitments I'm not sure there's going to be much I can do. I'll ask Neil to see if he can help, but otherwise I'll simply tell Lila that it wasn't possible to do this in the [16:10:57] little time she gave us. [16:11:29] I think that is very reasonable [16:11:51] joining tables is very easy [16:12:00] i don't know anything about eventlogging unfortuntely [16:12:09] HareJ: Shush. Don't take away my valid excuses! [16:13:48] halfak: BTW another instance of "halfak's documentation makes guillom's life easier": https://meta.wikimedia.org/wiki/Research:Rolling_active_editor and the other standardized metrics pages is super useful. [16:14:11] :D! [16:14:29] I still feel like those pages need some devoted TLC before they will be useful. [16:14:51] Sadly, the metrics standardization work is a good example of something that got derailed by ad-hoc requests. [16:14:54] And VE [16:14:59] For the moment I'm not going to use the exact standardized metric because Labs doesn't have the archive table, but it's still very useful. [16:15:11] Labs doesn't have archive? [16:15:18] It should [16:15:20] I don't think so; it's redacted, isn't it? [16:15:28] Na. Just the text. [16:15:32] OH. [16:15:35] I'm stupid. [16:15:38] Of course. [16:15:43] Even better! [16:15:55] Woot! [16:16:14] You might find the indexes on archive_userindex lacking though. I had some special ones added to our slave DBs [16:16:22] rb [16:16:27] ok (wrong keyboard) [16:16:40] Thanks! [19:32:00] Any idea why a SQL query would run fine in Quarry, but throw a syntax error in CLI? [19:35:08] The query is this one: http://quarry.wmflabs.org/query/4270 [19:35:39] And the CLI says this: [19:35:41] Error: ER_PARSE_ERROR: You have an error in your SQL syntax; check the manual that corresponds to your MariaDB server version for the right syntax to use near 'SET @u = 30; /* activity unit in days */ [19:35:41] SET @T = "20150608"; [19:35:41] SELECT [19:35:41] user [19:35:42] F' at line 13 [20:07:49] AHAH! [20:08:29] So apparently the CLI didn't like the multiple SETs; variables need to be declared in one go. [20:08:39] #TIL. [20:19:53] No, that's not it. [20:19:58] Argh. [20:43:50] So, the problem was that my mysql client had disabled multiple-statement queries. [20:43:55] weee. [21:50:30] I RETURN [21:50:36] AND I BRING AKVAVIT [21:50:43] Hooray [22:13:47] did anything blow up while I waas away? [22:14:01] I'll note that the dashboards are still alive. I had no computer for 2 weeks and they didn't once need hand-holding. [22:29:47] o/ Ironholds [22:29:50] Have fun? [22:30:02] You didn't bring your computer? [22:30:30] halfak, I lost it! [22:30:37] Noooooo [22:30:39] it's somewhere between Amsterdam airport and my hotel in northern Denmark [22:30:40] That's a bummer [22:30:46] It had the only copy of my talk and slides on it [22:30:47] Oh... The airline lost it :\ [22:30:52] Nooo [22:30:56] so I stood up and for 15 minutes, riffed on the contents of my head [22:31:05] and according to the audience this actually worked out phenomenally well [22:31:13] in the future I think unless I'm doing a demo I just won't do slides [22:31:15] Great! [22:31:16] people have to focus on your voice [22:31:33] Whatever gets them to stop reading their email [22:31:56] exactly [22:32:14] okay, I'mma go unpack and unwind; been travelling since 5am EST [22:32:18] laters! [22:34:18] See ya. Rest good :) [22:34:49] will do :)