[13:40:58] 10ORES, 10Scoring-platform-team, 10Graphite: Add a graph of ORES Celery task queue length - https://phabricator.wikimedia.org/T182914 (10akosiaris) We can probably get away with reusing https://github.com/oliver006/redis_exporter that we already use. It does have a `check-keys` parameter that allow us to cou... [13:43:27] 10ORES, 10Scoring-platform-team, 10Graphite: Add a graph of ORES Celery task queue length - https://phabricator.wikimedia.org/T182914 (10Ladsgroup) That looks awesome! the key is "celery" [13:43:54] 10MediaWiki-extensions-ORES, 10Scoring-platform-team, 10Continuous-Integration-Infrastructure, 10MediaWiki-extensions-Newsletter, and 3 others: Jenkins jobs for npm-test fail on project with deps on node-gyp which requires python2.7 - https://phabricator.wikimedia.org/T213944 (10zeljkofilipin) [13:51:29] 10MediaWiki-extensions-ORES, 10Scoring-platform-team, 10Continuous-Integration-Infrastructure, 10MediaWiki-extensions-Newsletter, and 3 others: Jenkins jobs for npm-test fail on project with deps on node-gyp which requires python2.7 - https://phabricator.wikimedia.org/T213944 (10zeljkofilipin) a:05zeljkof... [14:59:09] o/ [14:59:22] hey Amir1! How are you doing? [14:59:56] halfak: hey, Good. Still slow, had to install things on my new laptop but it works for now [15:04:15] Gotcha. In your absence we have been focusing on Jade as discussed. [15:04:40] I've been working on some wireframes. And I've been thinking about what type of UI development strategy would make sense. [15:06:08] Do you know what JS ui framework we should be using for new development? [15:08:47] Amir1, ^ [15:17:58] 10MediaWiki-extensions-ORES, 10Scoring-platform-team, 10Continuous-Integration-Infrastructure, 10MediaWiki-extensions-Newsletter, and 3 others: Jenkins jobs for npm-test fail on project with deps on node-gyp which requires python2.7 - https://phabricator.wikimedia.org/T213944 (10hashar) So the fix is strai... [15:19:57] halfak: at meeting now, today is wikidata day [15:21:22] Uh. Right. [15:23:06] halfak: draftquality for enwiki seems to only pick up n=167186 (which should probably be close to 200k?) [15:23:25] and almost all of them (but ~10) are "OK" [15:23:35] That sounds right to me. [15:23:56] It's possible that many of the revisions have been suppressed. [15:24:06] But so many? [15:24:27] Is it maybe that mwapi is not getting these revisions because I'm not a local sysop (I can still read them, but not via local rights) [15:24:43] Looks like we had 201261 for the last model training. [15:25:04] Any that are not "OK" should be deleted. [15:25:17] So if you're getting any of those, I would assume that includes the deleted ones. [15:25:49] Are you seeing any clear errors/warnings during the process (or maybe it's still warning about "rvslots" constantly?) [15:26:00] Just 3(!) that are labeled "vandalism" [15:26:26] afk for a bit [15:33:23] Yeah. that seems wrong. [15:33:45] When you get back, see if you can find a good example set of revisions you can't gather data for and we can check them out. [15:41:05] Hi everyone, I'm curious if there is a way of including a value for the threshold in an optimisation query like: https://ores.wikimedia.org/v3/scores/enwiki/?models=damaging&model_info=statistics.thresholds.true.%27maximum%20!precision%20@%20f1%20%3E=%200.1%27 [15:42:12] ThresholdT, you want to include the threshold value in the query? [15:42:25] Or in the output? [15:42:49] In the query. [15:43:38] Hmm. We don [15:43:42] t have anything like that [15:43:56] But you can ask ORES to output the whole table and work from there [15:44:11] https://ores.wikimedia.org/v3/scores/enwiki/?models=damaging&model_info=statistics.thresholds.true [15:44:15] The table is in order [15:45:13] Alright. Yes that's fine too. Just wanted to make sure I'm not missing anything. [15:45:41] Thanks. [15:46:57] I never thought of including threshold as something you might want to query for because it has no clear meaning. [15:47:07] No operationalizable meaning anyway. [15:47:16] What are you using it for? [15:47:36] Maybe we could add the ability to query by threshold [15:49:30] 10Scoring-platform-team (Current), 10editquality-modeling, 10Spanish-Sites, 10User-Ladsgroup, 10artificial-intelligence: Train/test damaging & goodfaith models for Spanish Wikiquote (eswikiquote) - https://phabricator.wikimedia.org/T209784 (10Halfak) Just barely caught this after merging. You built thes... [15:50:13] I'm not sure what it could be used for either. I have started building a little UI these past days to generate these query URLs (with drop-down menues and such) and wanted to check if I'm paying attention to everything. [15:50:20] wikimedia/editquality#448 (revert-178-eswikiquote - 6b8d741 : Aaron Halfaker): The build failed. https://travis-ci.org/wikimedia/editquality/builds/482951964 [16:23:10] Understood, ThresholdT :) Let me know if you change your mind. [16:23:20] * halfak goes back to digging through emails. [16:24:09] Will do, thanks again [16:55:41] o/ awight [16:55:50] harej, good morning! [16:55:57] I'm working on Tech Management updates. [16:56:08] Anything you'd like me to report for last week? [16:56:19] I imagine hoo unblocked a lot for awight :) [16:58:46] holla [16:59:17] 10Jade, 10Scoring-platform-team (Current), 10Design: Jade Wireframes: Entity edit mode - https://phabricator.wikimedia.org/T212374 (10Halfak) https://docs.google.com/drawings/d/1urHCaEycUs0n-63bTcWLOJtaRE7D08yWd8nQS6W6Jd0/edit This one covers the creation of endorsements and proposals. This is now ready... [17:05:12] I have nothing particularly interesting to add. [17:07:11] Nothing that you're not already talking about, that is. [17:09:55] Cool. I'll be talking about progress on eswikiquote and Jade wireframes. [18:05:45] halfak: Seems that the only vandalism revision it actually fetched indeed weren't deleted [18:07:30] Aha! I wonder if they *were* deleted but have been re-created. [18:08:00] Maybe our data gathering process needs work. I can't remember how it moves from the deletion log (which contain CSD comments) to gathering rev_ids. [18:08:06] " 19:08, 20 February 2017 Beeblebrox (talk | contribs) restored page Everipedia (98 revisions restored: restoring pending userfication) " [18:08:08] Seems so [18:09:51] ouch… revscoring extract needs --login otherwise it acts as anon [18:10:01] Oh nice. That's a good flag. [18:10:53] God… I wasted so much time on running this all/ getting it to run with the broken dataset :S :D [18:11:27] mwapi.errors.ClientInteractionRequest: …required', 'id': 'TOTPAuthenticationRequest', 'provider': 'Two-factor authentication (OATH).'}]) [18:11:42] halfak: Anyway to make that work easily? :S [18:11:49] hoo, mwapi has a utility for this. Probably an easy fix. [18:11:58] * halfak digs [18:13:03] mwapi.cli.do_login [18:13:06] halfak: ^? [18:13:35] Yes. [18:22:11] 10MediaWiki-extensions-ORES, 10Scoring-platform-team, 10Continuous-Integration-Infrastructure, 10MediaWiki-extensions-Newsletter, and 3 others: Jenkins jobs for npm-test fail on project with deps on node-gyp which requires python2.7 - https://phabricator.wikimedia.org/T213944 (10Krinkle) >>! In T213944#489... [18:24:32] meh, can't push to a revscoring branch [18:24:53] Oh! Let me try to fix that. I think we have a group I can add you to. [18:25:47] hoo, you should have an invite. [18:25:55] That should solve your access to all the repos. [18:26:09] halfak: https://github.com/wikimedia/revscoring/pull/421 [18:27:00] hm draftquality/utilities/fetch_text.py still uses that function [18:27:12] but that doesn't seem to be used/useful anymore [18:27:25] all it does is log in anyway [18:27:32] hoo, added a comment. [18:27:35] Otherwise looks good. [18:29:42] halfak: What about the drafquality thing? Is that a reason to reatin the function? [18:30:21] Hmm. Probably not. The only reason I can imagine is when you're doing quick iterations on features and you want to preserve the text for repeat extraction. [18:30:32] As it stands, we don't really do that. [18:30:44] Once every blue moon, we add a feature and rebuild everything from scratch. [18:30:55] If you want to trim that out, I'll support it. [18:31:31] Maybe in the long term, we can have a flag in "revscoring extract" that just sticks the base datasources in a cache for reuse. [18:31:38] That would be useful beyond just text. [18:31:49] E.g. with item quality & property suggestor [18:33:02] hoo, I think I'm going to head out. I've got a tension headache building from a broken rib from last week. awight, should be around if you need anything. Otherwise, I'll check back in after a few hours to see if I can review anything for you. [18:33:31] Anything more I can help with before I go lay down (very gently) [18:33:35] halfak: That's fine, have a rest :) I can continue from here (got the extraction running) [18:33:35] ? [18:33:50] Awesome! Godspeed :) [18:34:04] harej, ^ [18:38:20] awight and halfak and harej Do you mind if we move out our ORES doc meeting for today until later this week or next week? [18:38:29] We could catch up at all hands [18:38:34] +1 [18:40:31] We can catch up at all hands [18:40:36] I canceled it for the week. Let's regroup when we are all togetherI feel like I need to get 2x as much done to make way for next week... [18:41:17] (luckily 2 x 0 is still 0) :-) [18:44:05] ERR_DIV_BY_ZERO [21:20:12] o/ [21:48:03] Woot, I've got php-ml working to train and successfully predeict if a given IP is going to be a proxy / vpn / webhost :) https://i.imgur.com/s02HHL3.png [22:39:16] (03CR) 10jenkins-bot: Localisation updates from https://translatewiki.net. [extensions/JADE] - 10https://gerrit.wikimedia.org/r/485935 (owner: 10L10n-bot) [23:27:40] huh, today I learned you can do machine learning in PHP