[08:05:03] (03PS1) 10Ladsgroup: Bump ORES to HEAD [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/459498 [08:05:49] o. [08:05:52] *o/ [08:15:47] 10Scoring-platform-team, 10Bad-Words-Detection-System, 10revscoring, 10artificial-intelligence: Add language support for galician - https://phabricator.wikimedia.org/T201142 (10Ladsgroup) @Elisardojm Hey, Yes, It would be great if review the generated list and split it to two lists, one bad words and one... [08:30:45] (03CR) 10Ladsgroup: [V: 032 C: 032] "Getting it in beta cluster." [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/459498 (owner: 10Ladsgroup) [08:48:11] FYI: SREs will soon be doing a "live" test of the switchover process. It should without any repercussions but still keep it in mind. [10:52:15] coolio [10:52:51] akosiaris: Is it fine if I deploy the poolcounter stuff to prod tonight? [10:53:22] Amir1: em, I 'd rather you wait for next week [10:53:36] there's a deployment freeze essentially this week [10:53:48] and no exceptions unless really justified [10:54:27] akosiaris: okay, the reason I asked was because services still have their window: https://wikitech.wikimedia.org/wiki/Deployments#Week_of_September_10th [10:55:07] even tomorrow, maybe we should send an email making this clear [10:56:38] yeah I 'll remove them [10:56:43] thanks for pointing it out [10:57:17] 10Scoring-platform-team, 10ORES: Silence or address E_WOULDBLOCK warning - https://phabricator.wikimedia.org/T152012 (10Ladsgroup) Getting back to this after we implemented poolcounter support inside ORES. mediawiki and thumbor use PoolCounter basically to do the exact same thing. Preventing duplicated intensi... [11:31:00] 10Scoring-platform-team (Current), 10MediaWiki-extensions-ORES, 10MW-1.32-release-notes (WMF-deploy-2018-08-21 (1.32.0-wmf.18)), 10Patch-For-Review, 10User-Ladsgroup: ORES blocking itself on insert and delete - https://phabricator.wikimedia.org/T201918 (10Ladsgroup) 05Open>03Resolved It's done. [11:49:59] 10Scoring-platform-team: Investigate using archiva.wmo to host our wheels - https://phabricator.wikimedia.org/T180606 (10Ladsgroup) 05Open>03declined I boldly decline this as we are on git LFS for wheels now. Feel free to reopen. [13:43:01] (03PS1) 10Ladsgroup: Expose basic ORES API configs to frontend [extensions/ORES] - 10https://gerrit.wikimedia.org/r/459549 (https://phabricator.wikimedia.org/T201691) [13:59:45] Hey folks [14:00:29] I've got a weird head cold that hasn't been getting better. I'm looking to make sure I'm not blocking anyone today and then get some extra sleep. [14:00:46] Amir1, harej: ^ [14:01:28] I don’t think you’re blocking me but if I think of something I’ll let you know. [14:03:47] 10Scoring-platform-team, 10ORES: Silence or address E_WOULDBLOCK warning - https://phabricator.wikimedia.org/T152012 (10Halfak) I don't see how pool counter would help us sequence our tasks in celery. What's your proposal? [14:04:13] harej, BTW, I took a pass through the JADE blog post last Friday, so I think that is ready to move forward once dhorn responds. [14:04:29] I don't think he'll take issue with how we talk about NPP but it'll be good to check. [14:05:03] I wonder if he could delegate that task. Maybe nettrom can talk to us about that when he gets online since he did the studies. [14:06:28] halfak: get rest, don't worry [14:06:44] I can't make it to the sync meeting today too [14:06:50] Thanks Amir1. :) I will. [14:06:57] I need to leave early [14:46:53] (03PS1) 10Ladsgroup: [WIP] Introduce mw.Ores.Api [extensions/ORES] - 10https://gerrit.wikimedia.org/r/459561 (https://phabricator.wikimedia.org/T201691) [14:47:35] 10Scoring-platform-team (Current), 10Gadgets, 10MediaWiki-extensions-ORES, 10Patch-For-Review, 10User-Ladsgroup: Implement JS ORES client in mw-ORES extension - https://phabricator.wikimedia.org/T201691 (10Ladsgroup) With my patches, this works: ``` oresClient = new mw.Ores.Api(); oresClient.get({revids:... [14:58:41] I'm leaving for the day. See you next later [15:26:10] halfak: my JADE session was accepted for WikiConference North America [15:50:53] awight: not attending wikicon? :( [15:50:55] harej: I think this comment was supposed to read "drafttopic"? https://phabricator.wikimedia.org/T203032#4568393 [15:51:16] harej: Medical thing came up, I can explain momentarily [16:02:42] halfak: Amir1: meeting? [16:04:07] awight: I had to leave early. Wrote it here. Sorry [16:29:24] Amir1: All good, thanks for the note! [16:47:26] notconfusing: I read your note wrong, thought u were talking about Wednesday [16:47:29] WHOA [16:47:37] felicitaciones! [16:47:45] wednesday, what? [16:47:56] "to be [gone] wed" [16:48:19] Bad Lip Reading-level wrong [16:48:37] Oh no, it's not this wednesday, it's on Sep 22, I think the video was cutting out at that point. Pero gracias por las felicitaciones. [16:48:39] heh, talk about a dramatic misinterpretation [16:49:01] notconfusing will be wed, but not on wed. [16:49:58] haha yeah, that is funny i didn't think about that [16:51:12] \o/ have a blast! [17:00:32] harej: LMK once you work out a schedule for exploring how editors might want to describe judgments and deliberate. [17:01:47] I'm reading a bunch of theoretical stuff and it's both exciting and anxiety-producing that we'll have a chance to compare against field notes. [17:14:41] awight: right now i'm working on establishing requirements for researchers and tool developers [17:15:29] Like, defining our software requirements? [17:15:36] Right [17:16:27] Cool! https://www.mediawiki.org/wiki/JADE/Implementations#System_requirements or elsewhere? Those should be updated or removed either way... [17:29:05] I'll think of a good place. [17:29:14] I'm working on an email to the WMF research team asking them for input [17:30:57] (03CR) 10Raimond Spekking: [C: 032] Add special page alias for Serbian [extensions/ORES] - 10https://gerrit.wikimedia.org/r/458173 (owner: 10Acamicamacaraca) [17:39:07] (03Merged) 10jenkins-bot: Add special page alias for Serbian [extensions/ORES] - 10https://gerrit.wikimedia.org/r/458173 (owner: 10Acamicamacaraca) [17:42:21] (03CR) 10jenkins-bot: Add special page alias for Serbian [extensions/ORES] - 10https://gerrit.wikimedia.org/r/458173 (owner: 10Acamicamacaraca) [17:56:04] Feeling like I can't design a worthwhile content schema without real-world feedback... I think I'll focus on the secondary schema for a bit. [17:56:33] Either way, I'm happy with the Judgment:Diff/123 convention, it's just a matter of how structured the page content will be. [17:57:01] i.e., the open questions shouldn't block my secondary schema work. [17:57:42] I don't think there's necessarily a right answer; you just have to pick one and hope it works. [18:01:36] I'm biased towards freeform at the moment cos it can express so much more, but I think the pattern is propositions + discussion, and want to see how editors prefer to structure those. It'll be painful to migrate in either direction later, unfortunately. [18:01:51] & freeform doesn't have an obvious way to make the conclusions machine-edible [18:02:40] Asking to clarify -- are you talking about doing away with JSON altogether? [18:04:23] there are a few alternatives, that's definitely an attractive option. [18:05:02] another would be that we just structure so that the freeform proposition has a place in the schema, as well as summary in conclusion, etc. [18:05:20] you mean the comment field? [18:05:44] I think this would be something a bit different [18:06:10] Let's try to not bake too much structure into this before knowing what structures people will prefer. [18:06:20] [ { proposition1, discussion, consensus }, {prop2, etc.} ] [18:06:21] exactly [18:06:40] I'm not baking yet, just trying to make sure we have the ingredients in the kitchen ;-) [18:08:55] It's a tough nut to crack because AIUI, editors are happy-unhappy about all their other workflows, which offer infinite flexibility at a software level but in exchange are tightly regulated at the human level. [19:37:50] (03Abandoned) 10Awight: [WIP] Move to entity-type APIs and a fluent internal interface [extensions/JADE] - 10https://gerrit.wikimedia.org/r/456429 (owner: 10Awight) [19:55:18] 10Scoring-platform-team, 10Wikilabels: Build Wikilabels interface on top of MediaWiki itself - https://phabricator.wikimedia.org/T199879 (10Harej) [19:59:31] 10Scoring-platform-team, 10Wikilabels: Build Wikilabels interface on top of MediaWiki itself - https://phabricator.wikimedia.org/T199879 (10awight) JADE is a bit mismatched at this point, since we don't want collaborative judgments for wikilabels, instead we want independent annotations that aren't influenced... [20:05:55] awight: that task shouldn't be confused with the other task which is explicitly "use JADE as the back-end for Wikilabels," which I might formally decline (what do you think?) [20:07:02] +1 to declining that for now [20:07:17] That task is https://phabricator.wikimedia.org/T153150 [20:08:05] 10Scoring-platform-team, 10JADE: JADE: UI/API for reviewing/refuting how ORES classifies you and your stuff - https://phabricator.wikimedia.org/T148700 (10Harej) [20:08:07] 10Scoring-platform-team, 10JADE, 10Wikilabels: [Discuss] Could Wiki Labels backend be replaced with JADE? - https://phabricator.wikimedia.org/T153150 (10Harej) 05Open>03declined [20:08:24] For T199879, maybe we just remove the JADE project? [20:08:25] T199879: Build Wikilabels interface on top of MediaWiki itself - https://phabricator.wikimedia.org/T199879 [20:08:49] I agree with the idea, although it might not turn out to be practical. [20:09:02] right, I removed the JADE project there [20:09:12] Writing MediaWiki interfaces is an exercise in futility most of the time [20:09:52] harej: haha sorry, I see that you removed the project 4 minutes before I commented [20:09:57] * awight facepalms lightly [20:10:11] snacking for a second. [20:15:49] 10Scoring-platform-team, 10JADE, 10Wikilabels: [Discuss] Could Wiki Labels backend be replaced with JADE? - https://phabricator.wikimedia.org/T153150 (10Harej) As JADE develops it is moving closer to a model where multiple people can collaborate on a single judgment for a given entity. This goes against the... [20:21:43] awight: I reached out to the research team and Leila asked if the pushback from ops is a concern for development of JADE. What do you think? [20:45:38] heck yeah it's a concern! [20:45:46] Without that, we'd be in production today [20:46:09] harej: ^ [20:47:56] Should I be looking for a CC or task, or maybe this was just an email? [20:48:53] FYI the research team has been alerted to JADE a few times already, mostly in the monthly sync meetings we've been having with them. [21:03:51] To answer in more detail, yes I think there's a good chance we'll never be allowed to deploy. The major risks seem to be: * JADE is attracting random side projects, many of which are already being blocked on their own merits, * some of the opposition to JADE is independent of the actual implementation, * lack of a clear statement on our part that the data could be collected without storing as wiki [21:03:57] pages, although JADE is dead without this collaborative aspect. [21:04:17] For the last point, we need to run a pilot to show that the idea is valid. [21:05:06] Ah, on the flip side, I think Ops is interested in helping us now that it's clear that we won [21:05:09] won [21:05:16] won't break the big wikis. [21:05:25] The obstacle at the moment is actually TechCom acceptance. [21:17:58] I have to run around to various med things now, farewell! [21:21:59] It was just an email that I sent to researcgh [21:22:37] Best wishes to your family!