[13:55:27] halfak, what is the preferred site for reporting bugs nowadays? Phabricator or the GitHub repos? [13:55:46] phab primarity. Thank you! [13:55:52] *primarily. [13:57:22] k [14:18:55] 10Scoring-platform-team, 10Wikilabels: Labeling gadget has non-functional expansion buttons - https://phabricator.wikimedia.org/T250176 (10He7d3r) [14:19:43] halfak, do you know what was supposed to be displayed when we click on the [+] button on Wikilabels? ^ [14:20:22] Maybe some statistics? [15:04:25] Helder, aha! Nice work. That's supposed to expand and contract the worksets. Maybe we don't really need them to expand and contract after all. We could just remove that control. [15:33:24] halfak, in Wikilabels, when a user skips/abandon a task, does it go to a new workset which will be assigned to other users? [15:33:46] yes. thats right [15:34:53] so I assume we can close this issue: https://github.com/wikimedia/wikilabels/issues/58 [15:45:13] Yes. Thank you. [15:45:27] (sorry to be so absent. All meetings this morning) [15:47:14] no problem [16:33:58] accraze, hangouts is failing to start now. This is weird. [16:45:08] halfak: Just to be sure, looking through https://github.com/wikimedia/revscoring/blob/master/revscoring/features/wikitext/datasources/tokenized.py will give me a break down of how tokenization works on revscoring right? Or is there any other module I should be looking at as well? [17:17:37] haksoat, right. That's where all of the important stuff we do related to tokens happens. [17:18:07] Our tokenizer lives in the "deltas" package though. https://github.com/halfak/deltas/blob/master/deltas/tokenizers/wikitext_split.py [17:18:18] (sorry for the delay -- so many meetings!) [17:57:06] posting our async update notes -- [17:57:23] kevinbazira: [17:57:24] Y: [17:57:26] Succeeded at setting up MW Vagrant with MW 1.35 [17:57:28] I think I have figured out a less "painful" way to setup MW Vagrant that is reproducible. I'll write it's documentation and share it. [17:57:30] T: [17:57:32] Documented an easier way to set-up MW-Vagrant here: https://etherpad.wikimedia.org/p/easier-way-to-set-up-mw-vagrant [17:57:34] Aaron and Andy please let me know whether I should share this on another platform or keep it for myself on the office wiki. Thanks! [17:57:36] Looking at the issue about deleting an endorsement that returns an error: jade-bad-content [17:57:38] https://phabricator.wikimedia.org/T247973 [17:57:42] halfak: [17:57:44] Y: Worked on uwsgi/gunicorn memory dynamics and SIGTERM. No luck there. Found some crazy new weird behaviors for uwsgi but nothing useful. I also did the follow-up work on ptwiki articlequality. I should have a new model ready for review today. I pushed up the new labeling campaign for ptwiki. [17:57:46] T: The day is 100% meetings. I did an interview and some management meetings. I have a 1:1 with Andy and two meetings with grad students who are looking to use our topic models. Finally, I'm meeting with Product Strategy to help onboard one of their new contractors. In between meetings, I hope to get a ptwiki PR together and to keep chtnnh unblocked. [17:58:01] haksoat: [17:58:03] Y: [17:58:05] I took a look at the built-in Elasticsearch analyzers and the custom analyzers used by the search team, hoping to find one which can be used for tokenizing word, sentence and paragraph breaks. The closest I found was a pattern analyzer, which doesn't suit our needs as it only tokenizes based on specified regex separators. [17:58:07] T: [17:58:09] I am currently looking through Trey's articles on normalization, tokenization and stemming. I also plan to see how we do tokenization on revscoring and then the ease of creating a custom tokenizer for Elasticsearch since nothing helps with the word, sentence and paragraph break needs so far. [17:58:11] and me: [17:58:17] Y: Continued working on re-enabling hooks and classes to support new schemas and link table structure. Spent some time investigating a possible bitmask solution to handle label data in the DB, although this is probably not the right approach. [17:58:19] T: More of the same, will be working on cleaning up the Link Summary hooks for Jade (aka onPageContentSaveComplete) and will continue experimenting with some different approaches to storing label data. [17:59:05] * Helder wonders what are these Ts and Ys [17:59:56] Helder, Y === yesterday and T === today [18:00:03] thanks [20:25:33] MEETINGS ARE DONE. [20:25:41] \o/ [21:22:18] 10Scoring-platform-team, 10Discovery-Search, 10drafttopic-modeling: Add drafttopic predictions to ElasticSearch index for the Draft namespace where available - https://phabricator.wikimedia.org/T249341 (10EBernhardson) Do these predict the same set of classes, and should they be found with the same `articlet... [21:42:06] 10Scoring-platform-team, 10Discovery-Search, 10drafttopic-modeling: Add drafttopic predictions to ElasticSearch index for the Draft namespace where available - https://phabricator.wikimedia.org/T249341 (10Halfak) These predict the same set of classes. If adding another keyword is not a big deal, it would be... [22:38:03] 10Scoring-platform-team, 10Discovery-Search, 10drafttopic-modeling: super-detect-noop: Support recognizing and updating subsets within an array - https://phabricator.wikimedia.org/T250237 (10EBernhardson)