[12:02:14] BIGGAR economics sounds like beggar economics, sad for the biggest research universities of europe http://www.unimi.it/news/92897.htm [12:32:48] Biggar? Biggar was my dad's mom's maiden name [12:32:57] It's a Scottish name! [14:06:50] Good morning, Ironholds [14:06:51] :) [14:07:26] {{cn}} [14:07:29] (good morning halfak ;p) [14:07:34] Any chance I can get you to comment here? https://meta.wikimedia.org/wiki/Meta:Requests_for_comment/Enable_flow_in_the_Research_talk_(203)_namespace#Comments [14:07:40] Perfectly OK to oppose. [14:07:43] No hard feelings. [14:07:51] * halfak wants some actual researchers to comment plz [14:08:10] Then again, it's support (WMF)/ oppose (Community) at the moment [14:08:16] halfak, totally! [14:08:22] halfak, can I ask for a favour in return? ;p [14:08:26] sure [14:08:52] do you know of any way I can kick the permissions on a 1002 file so I can edit it? [14:09:12] mikhail created a file for dataset aggregation and my cron jobs can't save modifications to it [14:09:35] Hmm... No but I'll look at the internet with you :) [14:09:52] fair! [14:10:10] When you say 1002 file, where does that 1002 number come from? [14:11:11] stat1002 [14:11:22] specifically, /a/aggregate-datasets/wdqs//wdqs_aggregates.tsv [14:11:31] Oh! Yeah. I can help with this. [14:11:39] * halfak though 1002 was a weird permission number. [14:12:27] So, bearloga can't modify it? [14:13:22] halfak, he's currently unconscious [14:13:30] Gotcha. [14:13:48] So, you can't modify this, but you can make a copy with new perms and modify that. [14:13:52] Will that work? [14:14:35] totes! [14:14:45] can I remove the old version? [14:15:03] because I'll need to throw a patch up to repoint the connector if not [14:15:11] Negative. [14:15:14] alternately I guess I can just wait until Mikhail wakes up and kick him [14:15:15] argh [14:15:22] * halfak digs more. [14:15:24] okay, I'll probably wait on Mikhail then; I'd need him to +2 changes anyhoo ;p [14:15:29] So you have something pointing to this path? [14:15:51] yeah, a set of data retrieval scripts that are gerrit based because gerrit will make things sooo much easier (engineering LIES) [14:16:01] Ha. OK. [14:16:26] I have a REALLY BAD IDEA(TM) [14:16:30] That will probably work. [14:16:53] It looks like we can rename the "aggregate-datasets" directory. [14:17:29] So we can make a duplicate directory (with symlinks?) and then make a quick switch where we rename the current aggregate-datasets and replace it with our symlink'd one. [14:17:38] This is a BAD IDEA [14:17:39] ... [14:17:43] yeah let's wait for mikhail [14:17:44] But it would probably work :) [14:17:46] k [14:17:49] you'd think we'd have a way around this already, darnit [14:17:54] (I mean, sort of having stats run everything) [14:17:56] *short [14:18:12] and stats running /anything/ annoys me [14:18:43] Yeah... It seems like perms in this directory should be set to group writeable. [14:18:56] *thumbs up* [14:18:59] * halfak looks for more researchers to ping about the RFC. [14:19:21] ottomata, see halfak's suggestion around /aggregate-datasets/ above. Wadda I have to do to make it that way? [14:19:32] No! [14:19:36] (this problem will keep coming up as we build more data retrieval workflows) [14:19:37] Don't show ottomata that! [14:19:43] halfak, the group-writeable thing! [14:19:46] not the symlink! [14:19:49] Oh yeah. That's good [14:19:50] :) [14:25:39] Ironholds, hows the R-gineering work going. Anything new coming together soon? [14:25:51] halfak, like, packages or WMF-centric thingies? [14:26:08] Or just MediaWiki/Wikimedia/Human behavior [14:26:13] heh [14:26:40] pretty good! We launched our second set of dashboards (covering Wikidata Query Service) this week - debugging the data retrieval scripts behind them is today's project, hence the permissions questions [14:26:53] Should be moving on to Maps dashboards after that, in conjunction with a framework for user satisfaction [14:27:18] Cool! You guys are cranking 'em out. [14:27:24] Link to WDQS? [14:27:28] Dashboard that is [14:27:36] I would but it's uh broken [14:27:41] Oh. Woops [14:27:55] it turns out that the combination of an R instance without C-side locale variables set, vagrant, and UTF-8 encoded .R files, is bad. [14:27:59] * halfak shakes fist at sleeping bearloga [14:28:15] I'm provisionally blaming the WMF for this because I've never had an R instance built as terribly as the one on that machine [14:28:31] like, how the heck do our machines not ~~set the encoding~~ [14:28:44] Yeah. I'm learning a lot about maintaining systems in production environments. I've seen python go from a beautiful wonderland to a wasteland of intermittent bullshit. [14:29:05] Ironholds, I have the same problem at altiscale [14:29:13] They don't set the encoding. [14:29:17] ugh [14:29:18] Everything is just bytes [14:29:20] Yeah [14:29:35] Had to update a lot of my code to assume 'utf-8' when no encoding is set [14:29:40] gross. [14:30:01] It should be a param, but then I'd have to add the param to EVERYTHING [14:30:54] * halfak calms down [14:30:58] * Ironholds pat pats [14:31:26] Python is the opposite of Wikipedia [14:31:56] I dunno, you did just call it a wasteland of intermittent bullshit [14:32:05] For the editor (library builder) it's wonderful. Everything works. The community is awesome. The tooling is good. [14:32:27] But when you want to put it all together into a package for a user, it's awful. [14:32:43] Which seems to be true of everything as I hear. [14:32:48] But it was surprising for me. [14:33:20] Tying a bunch of different libraries together in an environment with lots of moving parts == headaches. [14:33:50] Running a single script in a curated python environment or on the one server I set up to do this job: beautiful. [14:35:58] Anyway, was going to try to convince you to pull your MediaWiki utilities into this project: https://github.com/mediawiki-utilities [15:20:32] Ironholds: , halfak sorry was in interview [15:20:38] read above chats but still not following [15:20:41] what's up? [15:21:06] TL;DR: permission issues in stat1002:/a/ [15:21:15] Maybe we should enforce group-writeableness [15:21:22] Can we? [15:21:28] ottomata, https://phabricator.wikimedia.org/T111956 [15:21:56] I'm thinking that it might be nice to enforce group-writeable in public dataset directories. [15:24:43] halfak: i think that would be fine [15:24:51] aggregate-datasets ? [15:25:29] If we do that on stat1002:/a/aggregate-datasets, I think it would also make sense to it on stat1003:/srv/public-datasets/ [15:25:32] yes [15:25:34] agree [15:25:58] halfak, Ironholds, could you file ticket, this will need some puppetization and umasks to do the right thing [15:26:08] halfak, I just did [15:26:10] it's linked above [15:26:10] tag it with analytics-kanban and assign to me [15:26:12] oh! [15:26:14] doh [15:26:20] got it [15:26:24] I'll refactor it to include 1003 and assign [15:26:24] thanks! [15:26:29] ok [15:26:44] <3 you both [15:30:54] ottomata, done! thanks! [17:01:14] halfak, wanna see something cool? [17:01:20] Sure [17:01:32] https://pbs.twimg.com/media/COeelxfUwAAag5n.png the flyer my friend Noam is putting out to advertise his final seminar [17:01:53] "the doktoral student pondereth" [17:02:16] Pretty cool [17:02:21] Docktor [17:02:41] I'm going to call him "pondereth" from now on [17:02:51] "what?" "you're the doktoral student pondereth, it says so right on your flier"