[12:22:31] o/ [12:46:36] halfak: you have a weird locale. [12:46:48] $ echo 'Καλημέρα' > goodmorning [12:46:48] $ grep 'Καλη' goodmorning [12:46:48] Καλημέρα [12:47:15] probably LC_ALL=C or similar [12:48:07] and scrap the single quotes, just me being cautious, not really neede [13:11:50] (03CR) 10Awight: Kludge to localize string constants in template (031 comment) [extensions/JADE] - 10https://gerrit.wikimedia.org/r/479578 (https://phabricator.wikimedia.org/T211346) (owner: 10Awight) [14:01:09] 10ORES, 10Scoring-platform-team (Current), 10Patch-For-Review, 10User-Ladsgroup: Change default serializer of celery from pickle to json - https://phabricator.wikimedia.org/T206333 (10Ladsgroup) https://github.com/wikimedia/revscoring/pull/419 [14:04:04] wikimedia/revscoring#1563 (json - c1da24d : Amir Sarabadani): The build passed. https://travis-ci.org/wikimedia/revscoring/builds/470993751 [14:17:52] (03CR) 10Ladsgroup: [C: 03+1] Kludge to localize string constants in template (031 comment) [extensions/JADE] - 10https://gerrit.wikimedia.org/r/479578 (https://phabricator.wikimedia.org/T211346) (owner: 10Awight) [14:34:40] o/ [14:34:57] I'm gonna check out my locale. [14:36:05] LC_ALL= [14:36:08] (blank) [14:36:25] Everything else is set to "en_US.UTF-8" [14:36:40] (This is on stat1007) [14:38:33] Same story on my local machine. [14:38:42] akosiaris, ^ any of that look wrong to you? [16:15:07] 10Jade, 10Scoring-platform-team: Explore alternative names for Jade data - https://phabricator.wikimedia.org/T200365 (10Halfak) If "Ode" is an acronym, we should probably capitalize it. I like keeping "Jade" but dropping the acronym. We've already done a lot of socializing the name as "Jade". I also like th... [17:20:41] Would love some review: https://github.com/wikimedia/revscoring/pull/418 [17:21:17] I'm still a bit confused about what that is doing. It looks like it is creating a giant yaml file. [17:21:34] The goal isn't to make it giant [17:21:44] it's to debug with usage like --limit 10 [17:21:49] I see. But the way it writes out, is a big single document. [17:21:58] Why not stream read/write? [17:21:59] maybe there should be a default limit [17:22:15] It outputs to stdout, IIRC [17:22:22] You could still use yaml. [17:22:31] Oh yeah. I get that. I'm just saying write one row at a time. [17:22:56] ah so we can use "head" and process the whole file... [17:23:05] it will be json, not really yaml [17:23:05] Right. Or less or whatever. [17:23:08] and not readable [17:23:13] but not a bad idea [17:23:17] You could use yaml [17:23:32] I mean--nobody asked for this tool, it was just an obstacle I've run into and Marius was puzzled by [17:23:40] How would yaml be one line? [17:23:48] Oh. Don't make it one line :) [17:23:55] Set your own format. [17:23:57] the yaml streaming format is json is what I mean [17:24:00] This is targeted at human readers, right? [17:24:07] yes [17:24:15] discrete tools is fine, too [17:24:24] but I don't think it would be very elegant [17:24:54] I'm imagining a template like "{rev_id}: {label}(\n\t{feature_name}: {feature_value})+" [17:25:32] The tabbing will allow easier tracking of where one revision's features end and another revision's features begin. [17:25:41] Maybe put an extra \n at the end of a revision too. [17:27:52] Amir1: https://github.com/wikimedia/ores/pull/298 ready for review again [17:28:38] awight, I put a more clear suggestion in a comment on that PR. [17:28:46] Amir1: feel free to pick at https://gerrit.wikimedia.org/r/#/projects/mediawiki/extensions/JADE,dashboards/default some more [17:29:26] You'll want to allow the user to specify what field contains the label because it's not always obvious. (e.g. "damaging", "goodfaith", "reverted_for_damage", "wp10", etc) [17:29:53] awight: sure [17:30:16] gotta drive in circles a while longer o/ [17:30:24] Unrealated, amir1 you were right that there is some work to do on the wikidatawiki features post mwbase update. Not too much. But it's gonna take a bit to do because of a lack of tests. [17:32:01] oh okay [17:32:45] Amir1: Thanks! [17:34:15] ^^ [18:22:10] burble. [18:56:10] 10ORES, 10Scoring-platform-team (Current): Support CIDR range whitelist for ORES throttling - https://phabricator.wikimedia.org/T210103 (10awight) The code patch is merged, now we just need some configuration. @akosiaris @Ladsgroup do we have a private repository for semi-sensitive config like this whitelist? [20:02:20] I go eat something [20:02:31] not sure if I'll be back, will tell [20:32:52] I just finished rebuilding models. I'll have a set of PRs soon. [20:35:59] 10Scoring-platform-team (Current), 10articlequality-modeling, 10draftquality-modeling, 10drafttopic-modeling, and 2 others: Rebuild models for new revscoring (2.3.0) - https://phabricator.wikimedia.org/T212530 (10Halfak) [20:39:41] 10Scoring-platform-team (Current), 10articlequality-modeling, 10draftquality-modeling, 10drafttopic-modeling, and 2 others: Rebuild models for new revscoring (2.3.0) - https://phabricator.wikimedia.org/T212530 (10Halfak) https://github.com/wikimedia/editquality/pull/175 https://github.com/wikimedia/article... [20:39:47] 10Scoring-platform-team (Current), 10articlequality-modeling, 10draftquality-modeling, 10drafttopic-modeling, and 2 others: Rebuild models for new revscoring (2.3.0) - https://phabricator.wikimedia.org/T212530 (10Halfak) a:03Halfak [20:39:52] All PRs listed here: https://phabricator.wikimedia.org/T212530 [20:55:50] If I can get a quick review on ^, I'll continue with the config patches and maybe I can get them on beta today :) [20:56:03] awight, you still around? [20:56:21] halfak: can do [20:59:07] Thanks man :) [21:09:37] It's me job :D [21:16:34] awight: I responded to your notes on editquality [21:16:41] I think you're reading the counts wrong [21:17:18] oh good! More of that to come. [21:39:56] My latest git dark magic: v [21:39:59] https://stackoverflow.com/questions/2155887/git-submodule-head-reference-is-not-a-tree-error [21:40:06] Seriously. WTF. [21:40:10] 10ORES, 10Scoring-platform-team: Templatize articlequality makefile - https://phabricator.wikimedia.org/T212533 (10awight) [21:40:34] I think maybe we accidentally put a custom commit into one of our submodules. I'm going to need to kill that in the next config change. [21:58:00] (03PS1) 10Halfak: (WIP) Updates for revscoring 2.3.0. [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/481210 (https://phabricator.wikimedia.org/T212530) [21:59:18] 10Scoring-platform-team (Current), 10articlequality-modeling, 10draftquality-modeling, 10drafttopic-modeling, and 2 others: Rebuild models for new revscoring (2.3.0) - https://phabricator.wikimedia.org/T212530 (10Halfak) Just posted a work in progress: https://gerrit.wikimedia.org/r/#/c/mediawiki/services/... [21:59:37] awight, I just posted a WIP patch for ores deploy config. It has the 00-main.yaml config updates. [21:59:48] But obviously I'm still waiting on new models to be merged. [22:00:22] I'll be hopping onto a meeting and then heading out for the holiday. I suspect that I'll be around a bit over the next couple of days, so if you can finish your review of the model repos, I'll take advantage of that. [22:00:41] * halfak goes to last meeting. [22:04:26] (03CR) 10Awight: "Not merging because WIP." (034 comments) [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/481210 (https://phabricator.wikimedia.org/T212530) (owner: 10Halfak) [22:05:01] I think I've finished reviewing the 4 model repos, ping me if I missed anything. [22:05:16] Take a vacation, though! [22:09:26] Woo! maybe I'll find a bit of time after my meeting so we can get this on beta. :) [22:21:25] I responded to your notes on editquality [22:21:27] awight, ^ [22:29:52] {{done}} [22:29:52] How efficient, awight! [22:36:13] \o/ [22:59:07] 10ORES, 10Scoring-platform-team (Current), 10Analytics, 10Patch-For-Review: Wire ORES scoring events into Hadoop - https://phabricator.wikimedia.org/T209732 (10awight) @JAllemandou I didn't have time to chase down the responsible code, but wanted to let you know that the user redactions look good empirical... [23:19:41] git is so broken. [23:19:55] Every time I try to do something, it breaks. I even tried starting from scratch. [23:20:00] awight, can you try something for me? [23:20:12] yessir [23:20:14] Just update your prod deploy config to most recent [23:20:17] See if it works. [23:20:58] grr, you have the recursive thing on hand? [23:22:15] halfak: wait, which patch should I have? [23:22:18] nothing new in gerrit. [23:22:26] Did you update all the submodules? [23:22:29] No patches. [23:22:33] This fails for me in master [23:22:38] commit 18d3657f677ca63bb4958b99c0dd87354caa1eb8 (HEAD, gerrit/master, master) [23:22:44] that's Dec 17 [23:22:44] git submodule update --init [23:22:56] Are you talking about this repo? [23:22:56] gerrit ssh://awight@gerrit.wikimedia.org:29418/mediawiki/services/ores/deploy.git (fetch) [23:24:41] halfak: Should I be looking at https://gerrit.wikimedia.org/r/#/c/mediawiki/services/ores/deploy/+/481210/ for this experiment, or old master? [23:24:50] Just the current master [23:24:53] No patch [23:25:00] ah yeah I can no-op update :) [23:25:22] Must be working cos it's macos :p [23:25:23] Does it pull down the LFS for editquality without a problem? [23:25:36] lemme rm the directory [23:26:00] Try going into the submodules/editquality/ and doing "git pull origin master" [23:26:02] mmm I don't know how to clear any LFS cache [23:26:17] yes I have updated simply [23:26:50] I can refresh the submodule after deleting [23:27:27] Can you pull the most recent model files? [23:27:32] With a git pull? [23:27:34] there we go, I can rm .git/modules/submodules/editquality/ too [23:28:03] ah ha [23:28:15] congratulations, you broke editquality :p [23:28:20] I didn't do it. [23:28:24] It was like this when I got here [23:28:32] error: Server does not allow request for unadvertised object fd3301e835d1747064010757e1a77d80e2331206 [23:28:35] Fetched in submodule path 'submodules/editquality', but it did not contain fd3301e835d1747064010757e1a77d80e2331206. Direct fetching of that commit failed. [23:29:07] I'm going to try blowing up the submodule entirely and re-adding it. [23:29:13] O_o [23:29:36] I can pull the new master with all of the LFS in my local editquality repo. Why should the submodule work any differently? [23:29:36] Friday evening not challening enuf? ;-) [23:29:45] * halfak is super pissed at git. [23:30:45] https://gist.github.com/myusuf3/7f645819ded92bda6677 [23:30:50] WHy are there so many steps! [23:31:02] First comment: "Works perfectly" [23:31:03] lol [23:31:55] 10ORES, 10Scoring-platform-team (Current), 10Analytics, 10Patch-For-Review: Wire ORES scoring events into Hadoop - https://phabricator.wikimedia.org/T209732 (10awight) Since ORES scores are expensive to recalculate en masse, we only want to refresh scores when a new ORES model or model_version is released.... [23:34:12] Same error. [23:34:16] I think this is gerrit. [23:34:50] Yup. Try: git clone https://gerrit.wikimedia.org/r/scoring/ores/editquality [23:35:08] halfak: https://github.com/AppImage/AppImageKit/issues/511#issuecomment-341885789 [23:35:35] Did that work for you? [23:35:37] aww, that would have been fun. [23:35:40] nope [23:35:57] git submodule update --depth 50 [23:36:01] haha this is evil [23:36:50] 10Scoring-platform-team, 10Gerrit: Gerrit repo scoring/ores/editquality LFS broken (smudge filter lfs failed) - https://phabricator.wikimedia.org/T212544 (10Halfak) [23:36:59] https://phabricator.wikimedia.org/T212544 [23:37:09] OK I give up. I'm running away now :( [23:37:13] Have a good one awight. [23:38:38] 10Scoring-platform-team, 10Gerrit: Gerrit repo scoring/ores/editquality LFS broken (smudge filter lfs failed) - https://phabricator.wikimedia.org/T212544 (10Halfak) Here's the error log: ` git-lfs/2.6.0 (GitHub; linux amd64; go 1.11.1) git version 2.7.4 $ git-lfs smudge -- models/arwiki.damaging.gradient_boos... [23:39:02] halfak: Have you seen the commit which caused this? [23:39:11] "Update submodules to use the new git lfs version" [23:39:13] baaahahaha [23:40:16] I see that fd3301e835d1747064010757e1a77d80e2331206 is present in the editquality repo, though. [23:41:37] oh--note that this commit isn't on the current master branch, that makes me sad and confused. [23:43:10] that fd3301 commit is on the translatewiki branch, though. erp? [23:46:46] halfak: oh wow--editquality's master branch and the fd3301 commit have no shared ancestor. [23:49:03] And the last time the rev histories shared a patch was Nov 28, "Itwiki damaging/goodfaith (#172)". [23:49:22] It's the current origin/master that looks strange, note the missing line breaks in commit messages. [23:50:01] My guess would be a git-lfs history rewrite glitch, but IIRC we rewrote that repo a long time ago? [23:50:23] of course there's no commit to commemorate the event :-/ [23:51:10] yeah, Aug 22 was the rewrite. [23:51:16] Nice mystery. [23:54:53] okay so the submodule was bumped to fd3301 on the day of the LFS rewrite. It was a valid commit at that point, I can see it deployed to production. [23:55:15] The question is when the master branch was switched over to this new history. [23:57:20] GitHub currently only has commits to the "new" master, I assume that's the history switch that would have caused the replica repos i.e. gerrit to lose the commit. [23:58:37] I'm going to fix GitHub.