[00:59:04] (03PS1) 10Petar.petkovic: Fix Contributions ORES preference and form override [extensions/ORES] - 10https://gerrit.wikimedia.org/r/399122 (https://phabricator.wikimedia.org/T182911) [02:11:40] 10Scoring-platform-team, 10User-Zppix: Wiki-ai Travis-CI Image upgrade - https://phabricator.wikimedia.org/T183214#3847034 (10Zppix) [04:23:53] 10Scoring-platform-team, 10Phabricator, 10Repository-Admins: Access request: Phabricator Repository-Admins - https://phabricator.wikimedia.org/T182864#3847085 (10demon) 05Open>03Resolved a:03demon Done! Lemme know if you have any questions. [14:12:38] (03PS4) 10Petar.petkovic: Make Contributions ORES preference sticky [extensions/ORES] - 10https://gerrit.wikimedia.org/r/398841 (https://phabricator.wikimedia.org/T182911) [14:13:53] (03CR) 10jerkins-bot: [V: 04-1] Make Contributions ORES preference sticky [extensions/ORES] - 10https://gerrit.wikimedia.org/r/398841 (https://phabricator.wikimedia.org/T182911) (owner: 10Petar.petkovic) [14:17:22] (03PS5) 10Petar.petkovic: Make Contributions ORES preference sticky [extensions/ORES] - 10https://gerrit.wikimedia.org/r/398841 (https://phabricator.wikimedia.org/T182911) [14:19:01] (03CR) 10jerkins-bot: [V: 04-1] Make Contributions ORES preference sticky [extensions/ORES] - 10https://gerrit.wikimedia.org/r/398841 (https://phabricator.wikimedia.org/T182911) (owner: 10Petar.petkovic) [14:41:18] (03PS6) 10Petar.petkovic: Make Contributions ORES preference sticky [extensions/ORES] - 10https://gerrit.wikimedia.org/r/398841 (https://phabricator.wikimedia.org/T182911) [14:54:15] Halfak T183214 is something i'd like your opinion on when you have a chance, I'd like to get it done over the holiday weeks [14:54:15] T183214: Wiki-ai Travis-CI Image upgrade - https://phabricator.wikimedia.org/T183214 [14:54:47] I'm in favor. What is it going to take ot make the switch? [14:54:53] Is jessie even available for Travis? [14:54:57] Yes [14:55:06] I think stretch is too [14:55:18] But i want atleast jessie [14:55:32] Releng is at jessie already with jenkins [14:55:44] I would like to be atleast somewhat similar [14:57:41] halfak: all it takes is me editing a file [14:57:49] And making sure we have all the packages for jessie [14:57:58] Sounds good. Let's do it. [14:58:03] Ok [14:58:12] We'll actually want to move up to stretch soon. [14:58:14] So take good notes. [14:58:15] Let me search to make sure we can actually do it [14:58:26] Heh ok [14:58:40] 10Scoring-platform-team, 10User-Zppix: Wiki-ai Travis-CI Image upgrade - https://phabricator.wikimedia.org/T183214#3848260 (10Zppix) a:03Zppix [15:04:15] halfak: is beaker called python-beaker? [15:10:22] Awight do you know if the beaker package is called python-beaker? [15:11:10] Zppix: I haven’t had to touch that personally, but: pip search beaker [15:11:14] Maybe `Flask-Beaker` ? [15:14:02] halfak: how're you today? [15:19:45] codezee: fwiw, his calendar says “sick as a dog” :-/ [15:20:55] awight: oh, didn't see the calendar ... :/ [15:21:23] Well I might be exaggerating but only slightly :) [15:21:53] although now that i try to i don't think i have access to it [15:25:35] awight: how's weather btw in Peru? humid or pleasant? [15:25:49] Yeah it’s pretty shocking that we’re using G suite, it’s just to save some money. [15:26:16] The rainy season is getting started, so it’s currently both humid and pleasant :) [15:26:47] awight: i suppose only people with official wmf emails have de-facto access? [15:26:49] It’s looking pretty nice where you are, according to the intertubes :) [15:26:59] ^ that seems to be right [15:27:18] it’s really annoying that calendar invites default to exclude “outsiders”, for example. [15:27:49] awight: intertubes? I didn't get that reference [15:28:06] just a T.V. thing to call the Internet... [15:28:29] Of course, I don’t have a T.V. so I’m exposing myself as a poser! [15:28:39] :P [15:29:28] I was bummed by the classifier results with word2vec, when I manually inspected, it didn't predict anything! [15:30:15] rats [15:30:26] Any idea why the numbers looked so good? [15:30:38] I guess that makes for a good regression test ! [15:32:25] awight: its something to do with the nature of the classification I think, for instance, a sample with 100 True and 10000 False, the classifier can learn to predict everything as False and still get precision high... [15:32:45] accuracy.macro here was good but not all statistics were promising when I looked into it [15:33:14] and yes, it calls for tests for this kind of thing [15:33:39] but rn I have to look for alternative features... [15:34:54] aha I think I’ve seen that same issue with accuracy, even for single class classification, when populations are skewed. [15:48:45] 10Scoring-platform-team, 10ORES, 10Wikidata, 10Beta-Cluster-reproducible, 10User-Ladsgroup: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith] - https://phabricator.wikimedia.org/T183266#3848506 (10Addshore) [15:49:02] 10Scoring-platform-team, 10ORES, 10Wikidata, 10Beta-Cluster-reproducible, 10User-Ladsgroup: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith] - https://phabricator.wikimedia.org/T183266#3848518 (10Addshore) a:05Ladsgroup>03None [15:49:49] awight: ^^ :( [15:49:56] /o\ [15:56:43] :O [15:59:55] addshore: fwiw, I think we can fix this with a configuration change. [15:59:59] working on it... [16:00:03] awight: thanks :D [16:10:47] 10Scoring-platform-team, 10ORES, 10Wikidata, 10Beta-Cluster-reproducible, 10User-Ladsgroup: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith] - https://phabricator.wikimedia.org/T183266#3848506 (10awight) I see no ores_model goodfaith for be... [16:14:06] Awight fyi im moving travis-ci tests to jessie to match releng Operating System [16:14:10] Just fyi [16:16:55] 10Scoring-platform-team, 10ORES, 10Wikidata, 10Beta-Cluster-reproducible, 10User-Ladsgroup: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith] - https://phabricator.wikimedia.org/T183266#3848627 (10awight) Okay something weird is going on. T... [16:18:29] wiki-ai/wikilabels#249 (T183214 - f389390 : Devin/Zppix): The build passed. https://travis-ci.org/wiki-ai/wikilabels/builds/318706653 [16:18:47] Zppix: I tried that, there’s no such thing AFAICT [16:19:15] * awight stares blankly at “build passed" [16:19:48] o/ sorry was in meeting. [16:19:57] awight: there isnt? [16:20:03] awight: i just did it [16:20:25] o/ codez [16:20:28] bah. [16:20:33] code[tab] fail [16:20:50] Zppix: https://docs.travis-ci.com/user/reference/overview what is happening [16:20:50] I'm feeling good enough to put in a day's work today (I'm pretty sure) [16:21:05] awight: it is on jessie [16:21:06] halfak: beta “emergency” fix: https://gerrit.wikimedia.org/r/399211 [16:21:17] Zppix: Can you point me to docs that explain this? [16:21:27] awight, eek. beta is dead? [16:21:29] Zppix: cos if that works, I want a matrix that includes stretch. [16:21:31] awight: i can point you to the build log that says this [16:21:38] halfak: https://phabricator.wikimedia.org/T183266 [16:21:44] freaking mystery [16:22:11] awight, gotcha. [16:22:21] I see that this is related to the problem we saw yesterday. [16:22:56] Umm [16:22:58] Weeeird [16:23:09] +1 for inclusing stretch [16:23:09] It isnt on jessie but it says it is? [16:23:15] wat [16:23:44] It says build dist jessie but then in os info it says ubuntu jessie [16:23:45] 10Scoring-platform-team, 10ORES, 10Wikidata, 10Beta-Cluster-reproducible, and 2 others: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith] - https://phabricator.wikimedia.org/T183266#3848672 (10awight) * Per {T181191}, we need to catch any expe... [16:23:49] Err trusty [16:24:04] Zppix: okay that was my experience as well. thank you for not finding the wormhole into 10 years future [16:24:05] Let me figure this out [16:24:15] Zppix: check out the travis docs link I sent [16:24:39] halfak: This is all ready for a CR+2, https://gerrit.wikimedia.org/r/#/c/399211/ [16:24:48] Im going to and i may send a choice worded email to travis [16:24:57] harrr [16:25:15] They’re doing pretty well for being a popular free service. [16:26:55] Damnit [16:27:04] The only options are precise and trusty [16:27:08] awight, best I can reasonably give you is a +1. But if you want a +2, you got it. :) [16:27:13] :/ [16:27:17] kk [16:27:26] Arrg [16:27:36] I really wanted to do this too :( [16:27:47] Actually, looks like I can only +1 :| [16:27:49] awight, ^ [16:28:15] addshore: Want to haha [16:28:19] oops [16:28:26] -releng copypasta [16:28:26] 04Error: Command “releng” not recognized. Please review and correct what you’ve written. [16:28:32] Zppix, yeah that is rough. I agree. I wanted to switch to Jessie a while ago :\ [16:28:55] -shuttup [16:28:55] 04Error: Command “shuttup” not recognized. Please review and correct what you’ve written. [16:29:01] o/ codezee [16:29:06] AsimovBot: stFu [16:29:06] 04Error: Command “stfu” not recognized. Please review and correct what you’ve written. [16:29:06] sorry was in meeting earlier. [16:29:11] lol [16:29:14] AsimovBot: mute [16:29:14] 04Error: Command “mute” not recognized. Please review and correct what you’ve written. [16:29:21] {silencio} [16:29:32] trolled by a bot. [16:29:48] 10Scoring-platform-team, 10User-Zppix: Wiki-ai Travis-CI Image upgrade - https://phabricator.wikimedia.org/T183214#3848686 (10Zppix) 05Open>03stalled Travis CI currently supports Ubuntu Trusty and Percise :/ Will work on AS SOON as if/when travis adds debian OSes [16:30:04] halfak: o/ [16:30:15] no worries, how're you feeling ? [16:30:59] codezee, feeling good enough to be around today. ^_^ [16:31:01] So better-ish [16:31:03] AsimovBot: status [16:31:03] Datos generales - Iniciado: Thu, 14 Dec 2017 09:24:58 UTC - Memoria usada: 7495864 bytes - Operadores: 9 - Ignorados: 0 - Hiperignorados: 0 - Última sugerencia: 305 (Abián) - [16:31:07] awight: i would of expected a CI service to have atleast some form of debian [16:31:21] That date is wrong [16:31:25] Way wrong [16:31:48] https://es.wikipedia.org/wiki/Usuario:-jem-/Asimov [16:32:13] I wouldnt understand it ^ [16:32:15] jem hasn't been around here for a while. [16:32:41] 10Scoring-platform-team, 10User-Zppix: Wiki-ai Travis-CI Image upgrade - https://phabricator.wikimedia.org/T183214#3848708 (10awight) Here's an interesting bug, https://github.com/travis-ci/travis-ci/issues/5075 BanzaiMan says, > This (and/or a variant) has been asked before, to allow builds to run on differe... [16:32:46] It looks like gitlab has an alternative CI service. [16:32:58] Might provide us with a more reasonable image. [16:33:47] 10Scoring-platform-team, 10User-Zppix: Wiki-ai Travis-CI Image upgrade - https://phabricator.wikimedia.org/T183214#3848712 (10awight) [16:33:54] Hmm [16:33:59] What is sbuild? [16:34:39] halfak: when we move to gitlab travis wont work [16:34:50] halfak so we wont have a choice but to change [16:35:17] 10Scoring-platform-team (Current), 10Wikilabels, 10User-Ladsgroup: Develop a backup strategy for campaigns/tasks/labels - https://phabricator.wikimedia.org/T155116#3848720 (10Halfak) I think this is done when we can confirm that wikilabels-dumps is on different physical hardware than the postgres server. @A... [16:39:08] halfak: I think word2vec averaging isn't providing any signal, n_estimators=500, max_depth=5 looked like a decent hyperparams choice but the classifier isn't predicting anything! [16:40:17] codezee, hmmm... That seems pretty sketchy. [16:40:19] 10Scoring-platform-team, 10ORES, 10Wikidata, 10Beta-Cluster-reproducible, and 2 others: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith] - https://phabricator.wikimedia.org/T183266#3848730 (10awight) ORES UI is disabled on beta wikidatawiki.... [16:40:31] Seems like you should be tuning for a thresholds based metric [16:40:33] :) [16:40:48] I imagine we get very high accuracy for not predicting anything [16:40:51] Same with vandalism. [16:40:59] False all the time gets 97.5% accuracy ;) [16:41:41] awight: i can (if noeone beat me) [16:41:43] just ordering a new screen! [16:42:10] addshore haha I self-merged, feel free to bless posthumously. The immediate incident is resolved by disabling ORES on the UI. [16:42:14] But mysteries remain... [16:43:41] halfak: thresholds based metric as in? roc_auc.macro? [16:43:52] codezee, right [16:44:39] 10Scoring-platform-team, 10ORES, 10Wikidata, 10Beta-Cluster-reproducible, and 2 others: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith] - https://phabricator.wikimedia.org/T183266#3848506 (10Catrope) Maybe the maintenance scripts to set up O... [16:46:20] halfak: I’ll go ahead and deploy the ores logging change to beta, eh? [16:46:22] 10Scoring-platform-team, 10ORES, 10Wikidata, 10Beta-Cluster-reproducible, and 2 others: wikidata.beta.wmflabs.org/wiki/Special:RecentChanges InvalidArgumentException No model available for [goodfaith] - https://phabricator.wikimedia.org/T183266#3848750 (10Catrope) I tried running it, but of course that do... [16:46:36] awight, +1 [16:46:47] halfak: n_estimators: [80, 160, 320], max_depth: [3, 5], max_features: ["log2"] seems reasonable? [16:46:51] awight, were you able to confirm the fix in vagrant? [16:47:14] halfak: yes! I wasn’t able to easily fix the systemd scripts though, for 2 reasons discussed in the task. [16:47:19] codezee, yes for a start. Let's stick to that range for now and try higher n_estimators later. [16:47:31] Luckily, my fix is main- vs run- agnostic [16:47:43] awight: roger [16:47:47] Awight what is sbuild? [16:48:04] i'm having a feeling that word2vec averaging over entire article is diminishing the signal [16:48:21] Zppix: Good question ;) http://lmgtfy.com/?q=sbuild :p [16:48:38] I cant say i didnt see the lmgtfy link coming [16:50:06] Love that shit. [16:50:48] My ircbot has a cmd for lmgtfy lol [16:51:26] Its an unbuntu package [16:51:31] Ubuntu* [16:51:44] Unbuntu ubuntu whats the difference 🤷‍♂️ [16:52:45] The “iceberg” burn ward [16:52:57] awight: if we do sbuild i think well have to rework our entire ci [16:53:08] Little more work then i really wanted [16:53:16] ah well thank you for looking into it [16:53:24] But let me actually see [16:53:33] It may be plausible [16:56:16] awight: ill see if i cant find a way to play with it on toolforge and see how it works [17:22:07] 10Scoring-platform-team, 10JADE, 10Design: Design conceptual prototype of JADE integration with MediaWiki - https://phabricator.wikimedia.org/T182829#3848860 (10Halfak) [17:22:09] 10Scoring-platform-team, 10JADE, 10Design: Design conceptual prototype of JADE integration with MediaWiki - https://phabricator.wikimedia.org/T182830#3848862 (10Halfak) [17:22:49] (03PS1) 10Awight: Bump ores submodule [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/399229 [17:23:30] halfak: CR? https://gerrit.wikimedia.org/r/#/c/399229/1 [17:23:40] two patchsets in that chain [17:32:45] 10Scoring-platform-team, 10JADE, 10Design: Design curation/suppression integration with MediaWiki - https://phabricator.wikimedia.org/T183276#3848884 (10Halfak) [17:33:27] 10Scoring-platform-team, 10JADE, 10Design: Design curation/suppression integration with MediaWiki - https://phabricator.wikimedia.org/T183276#3848884 (10Halfak) [17:33:45] 10Scoring-platform-team, 10JADE, 10Design: Design curation/suppression integration with MediaWiki (for JADE) - https://phabricator.wikimedia.org/T183276#3848884 (10Halfak) [17:35:34] awight: I am going to see if debian will be considered by travis soon if not I *may* switch us over to using sbuild with travis [17:35:50] interesting. Best of luck, that sounds like a fun project. [17:35:55] Holler if I can help! [17:35:55] Heh [17:36:08] Ill send them an email [17:36:27] sbuild seems to build packages, but I don’t see how we could use it as a logged CI environment. [17:38:13] thats where i was confused too [17:38:25] Well i sent the email [17:38:30] Now we wait.... [17:49:38] brb [17:53:20] halfak: Ah I see you’re having your most fun meeting. Just surfacing this easy CR from backscroll, https://gerrit.wikimedia.org/r/#/c/398942/1 https://gerrit.wikimedia.org/r/#/c/399229/1 [17:57:56] (03CR) 10Halfak: [V: 032 C: 032] "What does this bump" [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/399229 (owner: 10Awight) [18:02:20] (03CR) 10Halfak: [V: 032 C: 032] Clean up logging config. [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/398942 (https://phabricator.wikimedia.org/T182614) (owner: 10Awight) [18:02:30] awight, {{done}} [18:02:30] You rule, halfak! [18:08:03] (03CR) 10Awight: "Sorry for the laziness. It's the logging, here's a full log for this bump:" [services/ores/deploy] - 10https://gerrit.wikimedia.org/r/399229 (owner: 10Awight) [18:25:53] 10Scoring-platform-team (Current), 10ORES, 10Operations, 10Patch-For-Review: Investigate why ORES logs are being written to syslog despite explicit logging config. Fix. - https://phabricator.wikimedia.org/T182614#3849065 (10awight) This is deployed on the beta cluster, but isn't working. I think I accide... [19:22:05] back in 10 [20:28:07] halfak: Here’s the question I was asking last week—fine of course if this is an open-and-closed issue, but I was hoping to get input from editors and maybe CE. https://www.mediawiki.org/wiki/Topic:U43kfouz8e23gmw3 [20:29:42] Keegan: ^ If you’re interested, the perspective I’m looking for is whether the B-R-D norm is appropriate for our stakeholders using the JADE workflow. [20:30:53] awight: Roger, I'll have a look [20:30:57] o/ [20:31:26] awight: ive been an enwiki editor for 3 years and that tends to be whats perferred [20:32:35] Zppix: Thanks, that reinforces what I hear. The JADE scenario is slightly different, but I wouldn’t be surprised if editors preferred to edit war :p [20:46:48] awight, I think this having ranks avoids edit wars. [20:47:13] E.g. you might think to rank your judgement "preferred" and mine "normal". I revert you. Repeat. [20:47:46] Sorry—I didn’t mean to prejudice either alternative with hyperbole, my above comment was irresponsible. [20:47:56] Oh! thought you were serious :) [20:48:13] Oh damn, I posted as EpochFail. [20:48:14] BRD explicitly exists to avoid edit wars, yet still make forward progress… [20:48:16] well whatever [20:50:53] halfak: i edited your comment so people knew who you are [20:51:07] Zppix, that's not necessary [20:51:37] I can revert [20:51:50] very meta... [20:51:51] If you like [20:52:00] Zppix, you signed your name *in my comment* :P [20:52:09] After i put edit: [20:52:36] It makes sense if you read it [20:52:48] Lol that works too [20:53:02] (I am not used to flow :P) [20:56:38] awight: also i prob wont get a response from travis til tomorrow they are in CET tz [20:58:54] If travis doesnt work out or sbuild, should i go CI shopping? [20:59:21] Cause once we move to gitlabs we will have to anyway [21:01:44] Zppix: Good point! https://about.gitlab.com/comparison/gitlab-vs-travis-ci.html [21:02:05] So? Whats the verdict? [21:02:06] Zppix: Would you mind filing that observation in a new task? [21:02:27] Zppix: It looks like gitlab has its own CI, so IMO that’s a good place to start. [21:03:09] awight: well heres the thing, if travis cant get debian in a timely manner, then i tried sbuild it wont work afaic so we would need more option so ill make that a subtask of the current task [21:03:21] +1 [21:03:32] Check out Gitlab CI if you like [21:04:51] 10Scoring-platform-team, 10User-Zppix: The future of Wikimedia Scoring Platform's CI for repos - https://phabricator.wikimedia.org/T183294#3849464 (10Zppix) [21:04:57] Will do [21:05:39] awight: should i maybe ask releng for their opinion? [21:05:42] Aswlell [21:05:44] Aswell* [21:06:15] Zppix: your call. They’ll probably have lots of varied experience, but FWIW we would be the only *official* WMF project to try anything as crazy as Gitlab [21:06:32] 10Scoring-platform-team, 10User-Zppix: The future of Wikimedia Scoring Platform's CI for repos - https://phabricator.wikimedia.org/T183294#3849464 (10Zppix) [21:06:55] awight: true but they would know who not to even bother with [21:08:25] moving to gitlab? [21:08:45] Yes [21:08:57] I asked em lets see what i get [21:09:56] While im waiting I am going to do some light research into gitlab [21:17:52] 10Scoring-platform-team, 10User-Zppix: The future of Wikimedia Scoring Platform's CI for repos - https://phabricator.wikimedia.org/T183294#3849517 (10Zppix) [21:18:14] awight: added my chat with releng ^ [21:18:21] * halfak does a bunch of email [21:21:51] awight: im liking gitlabs ci [21:22:02] Only problem is who here has experience with docker? [21:22:20] I mean i could learn the ways of docker if need be [21:22:26] But it wont be painless [21:22:51] We'll need to anyway [21:22:55] Kubernetes is in our future [21:22:58] Yeah [21:23:15] Imma make some test repos and screw with gitlabs ci [21:23:27] Ill add you both if you wanna toy along with me [21:28:22] wrong channel but dev seems a little dead, halfak, do you know much about the MW API, specifically `list=blocks`? [21:29:31] TheresNoTime: i dont think api can do range block lists yet [21:30:17] "bkip: List all blocks, including range blocks, that apply to this IP address. You can also specify a CIDR range here, in which case only blocks applying to the entire range will be listed" :/ [21:31:05] trying `list=blocks&bkip=170.24.138.0/24` at the moment to no avail, but I'll try some more on-topic channels ^^ [21:31:16] TheresNoTime: https://www.mediawiki.org/w/api.php?action=help&modules=query%2Bblocks [21:31:22] I’m sure you saw that... [21:31:52] Yeah, issue is https://en.wikipedia.org/w/api.php?action=query&format=json&list=blocks&bkip=170.24.138.0/24 is showing nothing, but at least one IP in that CIDR *is* blocked [21:32:14] Do you know which IP> [21:32:15] ? [21:32:23] Oh interesting [21:32:36] I don't have much experience querying for blocks from the API. [21:32:43] Let's double-check the DB [21:32:47] 170.24.138.53, 170.24.138.57 [21:33:25] Thank you! :) [21:33:26] gotcha, https://en.wikipedia.org/w/api.php?action=query&format=json&list=blocks&bkip=170.24.138.53 [21:33:38] That looks like a bug? [21:33:55] Also possible that you have to URLencode the "/" [21:34:07] nope, https://en.wikipedia.org/w/api.php?action=query&format=json&list=blocks&bkip=170.24.138.0%2F24 [21:34:08] Oh! Yeah! I like that hypothesis [21:34:12] Damn [21:34:22] I did try 170.24.138.0%2F24 :( [21:34:23] lol [21:34:25] Wait. https://en.wikipedia.org/w/api.php?action=query&format=json&list=blocks&bkip=170.24.138.57 [21:35:01] Sorry I'm confused about the problem. [21:35:12] It seems the block is listed for the target ips. [21:35:27] halfak: it is, but bkip is supposed to support CIDR ranges [21:35:31] that part looks borken [21:35:33] Oh I see. [21:35:42] ^ "You can also specify a CIDR range here, in which case only blocks applying to the entire range will be listed" [21:35:48] ah.. [21:35:53] oooh [21:35:56] nice catch. [21:36:01] "only blocks applying to the entire range" as in, the *entire range*? [21:36:21] and not, any blocks affecting a single IP *in* the range? [21:36:22] TheresNoTime: Where do you see that? It would be helpful to add that to the API docs [21:36:50] in the docs, https://www.mediawiki.org/w/api.php?action=help&modules=query%2Bblocks [21:37:23] lemme try with a range which has been blocked.. [21:37:41] I don’t see the language you quoted [21:38:17] oh, https://www.mediawiki.org/wiki/API:Blocks [21:38:18] TheresNoTime: If you want to dive deep, https://github.com/wikimedia/mediawiki/blob/master/includes/api/ApiQueryBlocks.php [21:39:19] awight: okay i like how gitlab does their ci [21:39:20] dammit, got it - https://en.wikipedia.org/wiki/Special:Contributions/176.204.0.0/17 --> https://en.wikipedia.org/w/api.php?action=query&format=json&list=blocks&bkip=176.204.0.0/17 [21:39:22] works [21:39:27] hi [21:39:27] But i hate gitlabs interface [21:39:40] Zppix: Looking forward to being frustrated by its futurism! [21:39:51] awight: ill link to docs for it in task [21:39:57] it only lists when the CIDR queried is the CIDR blocked, thanks for the prod in the right direction folks! [21:40:10] I invited you to a repo to play in with it awight as well (check email) [21:40:21] TheresNoTime: I agree, that’s how I’m reading https://github.com/wikimedia/mediawiki/blob/master/includes/api/ApiQueryBlocks.php#L142-L147 [21:40:48] awight: I read the docs to mean "list any single IP block IN the CIDR provided", which would be more helpful imho [21:40:51] 10Scoring-platform-team, 10User-Zppix: The future of Wikimedia Scoring Platform's CI for repos - https://phabricator.wikimedia.org/T183294#3849546 (10Zppix) Gitlab CI docs: https://docs.gitlab.com/ee/ci/quick_start/README.html [21:42:36] awight: also I have repo notifications figured out for gitlab so when we switch well be golden! [21:43:45] that’s rad. [21:44:00] TheresNoTime: Might be worth a feature request, if there’s a good use case? [21:44:21] It wouldn’t be a difficult code change or a very expensive query. [21:44:38] awight: just drafting up something for phab now :) unless you can think of a better way of doing "find all single IP blocks in a CIDR"? [21:45:01] as currently you'd need to do a query for each possible IP in a CIDR as far as I can see [21:45:31] +1 [21:45:35] +1 from me too [21:45:42] Or do the query yourself on labs :) [21:45:53] Which defeats the purpose of the API, of course... [21:46:19] Mmm, could do, though I like the idea of that being an API call, if only for external tool use [21:50:58] It seems like a reasonable request! [21:51:13] * awight proceeds to sell peers down the river [21:58:51] T183300 if you're interested awight [21:58:52] T183300: Use list=blocks to return all single IP blocks in a CIDR - https://phabricator.wikimedia.org/T183300 [21:59:11] Nicely documented! [21:59:59] TheresNoTime: I think “only if the given CIDR is blocked” is more accurately “if the queried CIDR falls within an existing blocked range” [22:00:11] * awight squints at the PHP [22:00:39] updated as such, thanks :) [22:01:02] I don’t understand 'ipb_range_start' . $db->buildLike( $prefix, $db->anyString() ), [22:01:16] don’t mind me [22:01:46] please do ramble, its educational :) [22:01:50] 10Scoring-platform-team (Current), 10drafttopic-modeling: Fix response processing logic in drafttopic.fetch_page_wikiprojects - https://phabricator.wikimedia.org/T181522#3792934 (10Halfak) Added a note. Otherwise looks fine. [22:03:03] hehe one more data point demonstrating that PHP drives people to desperation [22:14:39] halfak: It’s growing beastly, https://www.mediawiki.org/wiki/JADE/Implementations#System_requirements [22:14:53] `select * from ipblocks where ipb_address LIKE "170.24.138.%";`'l do it db query wise, if my SQL hasn't failed me (tested on tf, seems to return something reasonable) [22:15:17] "Comments and judgments appear on Special:Watchlist when the judged wiki entity is being watched." [22:15:27] Only makes sense for a very small subset of entities [22:23:47] halfak: Why is that? [22:24:06] Only pages can be watched [22:24:30] TheresNoTime: That makes sense, but only works when the CIDR range is on a 8-bit boundary AFAICT [22:24:39] awight halfak I made a brief reply for now https://www.mediawiki.org/wiki/Topic:U43kfouz8e23gmw3 [22:24:57] Unless there's something compelling I'm missing, KISS I'd say [22:24:58] halfak: Should I fix the language “edit or revision on a watched page”, or is the concept wrong? [22:25:18] Keegan: Thanks for the second opinion! [22:25:21] awight: yas, only for a /24 [22:26:51] halfak: I also summarized a few of our talk threads, please BRD if I misunderstood things :D [22:27:55] Keegan, see my response. [22:28:08] There was a use-case in another thread that I linked to. [22:28:32] Honestly, I'm starting to feel like even the preference bit and endorsements is overkill. [22:28:43] A history of judgements would work for the use-case to. [22:29:00] But endorsements looks more like a straw poll and might make sense here. [22:30:12] brb [22:30:41] I still can’t get myself to like the endorsements…. fwiw. [22:31:03] How do you feel about straw polls? [22:31:11] +1 to drifting towards just judgments + comments [22:31:22] straw polls… in general or on-wiki? [22:31:25] halfak: eww [22:31:26] Drifting towards judgement + history [22:31:45] History would replace endorsements. [22:31:47] on-wiki [22:31:51] halfak: use google forms atleast... strawpoll is *shutters* [22:31:58] (judgment, comment)_1, (judgment, comment_2), … yeah [22:32:52] Only reason I'm considering this is because it owuld be easier to map onto mediawiki [22:32:57] And therefor to set up curation [22:33:09] I think endorsements match behavior better [22:33:17] No one builds on the endorsement of another. [22:33:44] Essentially, we'd be abstracting differences of opinion on a judgement history rather than letting them be first-order things. [22:34:03] *No one builds on the judgement of another [22:34:21] unlike a wiki page where most revisions build on a previous revision. [22:35:47] But… judgments (scores + comment), endorsements (set judgment=preferred, comment), discussion (comment that can refer to things)? [22:40:25] GTG but I’m hoping to get this ironed out tomorrow. [22:40:49] I’ll also make an MVP section in JADE/Implementations to clarify what that phase will look like. [22:41:02] ah wait there’s a task somewhere. [22:41:13] https://phabricator.wikimedia.org/T176333 [22:41:36] That’ll be fine. I’ll see if there’s anything to update per our discussions, and link from the wiki page. [22:44:36] Not following this text protocol [22:45:09] Huh? [22:57:15] Zppix, was referring ti awight's parens and "+"s [23:14:21] Ah [23:17:51] Question: for running the ORES locally, `pip install ores && echo -e '{"rev_id": 456789}\n{"rev_id": 3242342}' | ores score_revisions https://ores.wikimedia.org enwiki damaging` it seems this command is still requesting https://ores.wikimedia.org, right? [23:18:10] Because ` https://ores.wikimedia.org` is a host address [23:18:21] and I'd presume it still has limitation on QPS, right? [23:18:46] xinbenlv, you didn't tell it what revisions to score :) [23:18:49] It reads from stdin [23:18:54] Looking for revisions you want it to score. [23:19:00] or --input [23:23:56] do you refer to `{"rev_id": 3242342}` [23:25:42] yes [23:26:03] You can query a set of rev_id's through quarry and download the json-lines format [23:28:45] xinbenlv, ^