[00:22:54] (03CR) 10Parent5446: "(1 comment)" [core] - 10https://gerrit.wikimedia.org/r/22167 (owner: 10Parent5446) [00:22:55] (03PS19) 10Parent5446: Added Uri class and Uri Query classes. [core] - 10https://gerrit.wikimedia.org/r/22167 [00:22:56] (03CR) 10jenkins-bot: [V: 04-1] Added Uri class and Uri Query classes. [core] - 10https://gerrit.wikimedia.org/r/22167 (owner: 10Parent5446) [00:23:30] (03PS9) 10Parent5446: Add referer and origin check to edit token checking [core] - 10https://gerrit.wikimedia.org/r/65418 [00:24:02] (03PS10) 10Parent5446: Add referer and origin check to edit token checking [core] - 10https://gerrit.wikimedia.org/r/65418 [00:24:31] (03CR) 10jenkins-bot: [V: 04-1] Add referer and origin check to edit token checking [core] - 10https://gerrit.wikimedia.org/r/65418 (owner: 10Parent5446) [00:25:06] (03CR) 10jenkins-bot: [V: 04-1] Add referer and origin check to edit token checking [core] - 10https://gerrit.wikimedia.org/r/65418 (owner: 10Parent5446) [00:28:07] (03PS20) 10Parent5446: Added Uri class and Uri Query classes. [core] - 10https://gerrit.wikimedia.org/r/22167 [00:29:09] (03CR) 10jenkins-bot: [V: 04-1] Added Uri class and Uri Query classes. [core] - 10https://gerrit.wikimedia.org/r/22167 (owner: 10Parent5446) [00:32:17] (03CR) 10Parent5446: ">.<" [core] - 10https://gerrit.wikimedia.org/r/22167 (owner: 10Parent5446) [00:34:21] (03PS21) 10Parent5446: Added Uri class and Uri Query classes. [core] - 10https://gerrit.wikimedia.org/r/22167 [00:35:38] (03PS1) 10Matmarex: mediawiki.action.edit.preview: Assorted fixes [core] - 10https://gerrit.wikimedia.org/r/76457 [00:37:39] (03PS11) 10Parent5446: Add referer and origin check to edit token checking [core] - 10https://gerrit.wikimedia.org/r/65418 [00:40:14] (03PS2) 10Matmarex: mediawiki.action.edit.preview: Assorted fixes [core] - 10https://gerrit.wikimedia.org/r/76457 [00:47:55] (03PS1) 10Parent5446: Changed logged out tokens to use session if already there [core] - 10https://gerrit.wikimedia.org/r/76458 [00:48:24] (03CR) 10Parent5446: "@Daniel - See the change dependent on this change for the implementation of tokens for anonymous users with sessions." [core] - 10https://gerrit.wikimedia.org/r/65418 (owner: 10Parent5446) [00:50:55] legoktm: \o/ [00:51:07] :D [00:55:17] (03PS3) 10Legoktm: Parse the input list using a parser function [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 [00:56:40] Elsie: I updated it a bit. We do lose some duplication checking though… ^ [00:56:48] Why's that? [00:58:45] Hm. Well we never really had it for global deliveries to begin with. I'll need to add an extra function to do it locally though [00:59:16] There were two de-duping mechanisms. [00:59:21] Well, three. [00:59:23] SELECT DISTINCT. [00:59:28] A Python set(). [00:59:30] And the key check. [01:00:02] er, which one was the key check again? [01:00:31] [01:00:39] (03PS1) 10Shirayuki: Replace "themselves" by "oneself" [core] - 10https://gerrit.wikimedia.org/r/76459 [01:00:42] Oh, right. Well we're not really supporting that.... [01:02:39] Right. [01:02:45] I'm just saying that de-duping is important. [01:03:41] Yeah, I'll work it back in [01:11:04] (03PS1) 10TTO: Format protection_expiry timestamp in ISO format for API prop=flagged [extensions/FlaggedRevs] - 10https://gerrit.wikimedia.org/r/76460 [01:47:14] Does "This, that and the other" use IRC? [01:47:17] Elsie do you know? [01:47:38] Krenair: I don't think so. [01:47:41] (03PS2) 10Alex Monk: Format protection_expiry timestamp in ISO format for API prop=flagged [extensions/FlaggedRevs] - 10https://gerrit.wikimedia.org/r/76460 (owner: 10TTO) [01:47:42] I don't know for sure. [01:50:49] (03PS4) 10Parent5446: Fix double-parsing of account creation messages. [core] - 10https://gerrit.wikimedia.org/r/47821 [01:52:26] (03PS5) 10Parent5446: Fix double-parsing of account creation messages. [core] - 10https://gerrit.wikimedia.org/r/47821 [02:23:41] Krenair: He's on rarely [02:23:51] [07:23:44 PM] -NickServ- Last seen : Jul 28 10:11:41 2013 (16 hours, 12 minutes, 2 seconds ago) [02:43:11] (03CR) 10KartikMistry: "URL seems 404. Can you add correct URL to download or appropriate note in case of it is not available?" [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76349 (owner: 10Ebrahim) [02:54:21] legoktm: Hi. [02:55:09] grrrit-wm seems to be dead. [02:55:20] (03CR) 10MZMcBride: "(12 comments)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 (owner: 10Legoktm) [02:55:20] Oh, there it goes. [02:56:41] (03CR) 10MZMcBride: "Thank you!" [core] - 10https://gerrit.wikimedia.org/r/76222 (owner: 10Ori.livneh) [02:57:02] Elsie: how slow was that? [02:58:36] Two minutes. [02:58:45] For the (12 comments) CR. [02:58:56] The "thank you!" CR came through quickly. [03:04:42] (03CR) 10KartikMistry: "LGTM!" [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76236 (owner: 10Santhosh) [03:06:57] YuviPanda: Does grrrit-wm have any call/response? [03:06:59] Like... [03:07:02] * Elsie hits grrrit-wm. [03:07:05] nope [03:07:07] i should add some [03:07:09] at some point [03:07:13] Just something silly, yeah. [03:07:17] reasonably trivial [03:07:21] you can add them too! :) [03:07:21] A hug response or something. [03:07:36] well, responding with "Working..." would be appropriate, I think :) [03:07:42] Hah! [03:08:16] :D [03:08:25] Elsie: want to try adding that? :) [03:08:33] Which repo is it again? [03:08:48] I have a working Gerrit account now, allegedly. [03:09:11] Elsie: labs/tools/grrrit [03:09:21] Elsie: you can also send a pull request to it on github, if you want. [03:10:30] (03CR) 10Parent5446: [C: 031] "Haven't tested, but looks good to me." [extensions/Annotator] - 10https://gerrit.wikimedia.org/r/75106 (owner: 10Rjain) [03:15:53] (03PS1) 10Mollywhite: Bug 50762: Add variables to show/hide items in the navbar. [extensions/BookManagerv2] - 10https://gerrit.wikimedia.org/r/76463 [03:26:24] (03CR) 10Parent5446: "(1 comment)" [extensions/Annotator] - 10https://gerrit.wikimedia.org/r/75645 (owner: 10Rjain) [03:28:44] (03PS3) 10Santhosh: Add Nazli font for Persian Script [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76349 (owner: 10Ebrahim) [03:31:19] (03CR) 10Santhosh: [C: 04-1] "compressed the font more using sfntly." [extensions/cldr] - 10https://gerrit.wikimedia.org/r/45499 (owner: 10Kaldari) [03:36:43] (03CR) 10Santhosh: [C: 04-1] "compressed the font more using sfntly." [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76349 (owner: 10Ebrahim) [03:37:27] (03CR) 10Santhosh: "Sorry about previous comment, wrong browser tab :)" [extensions/cldr] - 10https://gerrit.wikimedia.org/r/45499 (owner: 10Kaldari) [03:41:40] (03PS7) 10Yuvipanda: Clean-up UploadWizardDetails [extensions/UploadWizard] - 10https://gerrit.wikimedia.org/r/75595 (owner: 10Rillke) [03:42:30] (03CR) 10Legoktm: "(7 comments)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 (owner: 10Legoktm) [03:43:03] Elsie: ^ [03:48:28] (03CR) 10MZMcBride: "(4 comments)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 (owner: 10Legoktm) [03:57:20] [23:57] ... James_F|Away!~James@wikimedia/JamesF [03:57:20] [23:57] ... ircname "James [03:57:25] I wonder if that's my client or his. [03:58:06] as James always says, "it's [03:58:45] So apparently over 400 German Wikipedians voted to reverse VE's deployment. [04:02:23] Aaron|home: Hey, small job queue question for you. Is there a way I can get a count of how many jobs that are of the "MassMessageJob" class are queued? [04:02:59] hey ori-l. did you see hashar's messages about EventLogging? [04:03:03] they... don't fully make sense to me :) [04:03:10] no. where? [04:03:19] interestingly, on labs-l [04:03:33] legoktm: no way from API, it's possible with CLI access (showJobs.php --group) [04:03:38] oh [04:03:39] actually no [04:03:42] oh, i'm not on that list. i'll look. [04:04:06] Aaron|home: This is for a special page so that will work. I'll take a look at how that script works, thanks [04:04:08] ori-l: err, sorry [04:04:11] ori-l: that was wikitech [04:04:17] ori-l: 'bot dispatcher' thread [04:04:32] ori-l: http://lists.wikimedia.org/pipermail/wikitech-l/2013-July/070864.html [04:04:44] he also mentioned using EventLogging to... relay Gerrit messages to me before :) [04:06:11] what's confusing about that? [04:06:37] ori-l: using EventLogging to relay RecentChanges messages? [04:06:37] to labs? [04:07:12] I guess we can use the same concepts (0mq, etc), but I guess EL by itself can't do it? [04:07:14] I think hashar is being a bit generous with praise for EventLogging; the aspects of it that are desirable and applicable to this problem are mostly attributable to ZeroMQ [04:07:35] well, it can [04:07:50] yeah, but then it will no longer be EventLogging, but something bigger [04:07:54] you have: [04:08:31] a) a wrapper around wfDebugLog that encodes objects to JSON and validates them for conformity with a spec (the latter part is only useful if we expect to need some flexibility / adaptability) [04:09:24] so an interface for emitting JSON via UDP from MediaWiki [04:09:45] "massmessageJob: 2 queued; 0 claimed (0 active, 0 abandoned)" :D [04:09:46] EL also uses UDP? I thought it was 0mq from the apaches... [04:09:48] Elsie: ^ [04:10:04] and then a set of puppet classes that are well-documented and easy to deploy for provisioning a UDP listener that transforms the stream to a ZeroMQ publisher [04:10:21] legoktm: :D [04:10:40] and then a puppetized plug-in architecture for adding subscribers to that stream [04:10:43] Okay, let's say I want to upload a new patchset for a particular Gerrit changeset. [04:10:46] How do I do that? [04:11:11] Elsie: First get it via git review -d #### [04:11:22] Then you make your changes, git commit --amend [04:11:24] git review -R [04:11:28] Do I need to clone the repo first? [04:11:36] Yes [04:11:41] Okay, let me do that first. [04:11:48] ori-l: hmm, true. It'll be easier to extract that out than do something from scratch [04:11:53] I think the primary benefit of using EventLogging stuff is not in any sort of wowee-zoee feature but rather the fact that I bashed my head against all sorts of MediaWiki / Wikimedia production environment idiosyncracies and found workable solutions for them [04:12:01] hehe :) [04:12:10] true! [04:12:10] legoktm: I'll likely need a bit of hand-holding, if you're available for the next few minutes. :-) [04:12:11] or had people smarter than me find workable solutions for them :) [04:12:11] sure :P [04:12:28] ori-l: I think this was confusing to me because... I think petr had put it on labs-l, since this is a toollabs specific thing that is going to consume the IRC feeds and rebroadcast them as redis after some filtering. [04:13:04] it now makes sense to me, since it was on wikitech-l. Initially I Thought it was labs-l [04:13:07] so nevermind. [04:13:30] mzmcbride@gonzo:grrrit$ git clone gerrit:operations/mediawiki-config.git [04:13:33] ori-l: btw, I don't think that IRC RC Feeds thing will be replaced anytime soon, though :( Nobody cares, mostly. [04:13:33] I'm running that now. [04:13:40] I have gerrit: as an SSH alias or something. [04:13:54] Yeah, ori-l's tip about that was awesome :D [04:14:11] I did it in .ssh/config, not .gitconfig. [04:14:12] I think. [04:14:16] I just set a port and some other shit. [04:14:20] Oh, neat. [04:14:25] I read it in the docs. [04:14:30] I didn't pounce on petr with EventLogging stuff for exactly that reason; it's a project that needs someone who is passionate about it so if he wants to reinvent the wheel and experiment with some interesting pieces of software in the process he should go ahead. [04:14:33] Okay, cloned. [04:14:43] Now I want to... [04:14:59] wowee-zowee [04:15:06] I think the W is required. [04:15:10] there's nothing more fatal to a young developer's enthusiasm than to be told that the project they want to embark on is mostly done and "only" requires some work adapting old code. [04:16:09] ori-l: :) [04:16:23] legoktm: So I tried "git review -d 76199". [04:16:25] It blew up. [04:16:44] Do I need to provide a hash instead? [04:16:44] Elsie: What did it say? [04:16:49] No, that should work. [04:17:05] Sec, I'll paste. [04:17:11] It went a bit ape-shit. [04:17:39] https://dpaste.de/GsXRd/raw/ [04:17:56] Oh [04:18:01] YuviPanda: the recently-rewritten eventlogging module in the puppet repo is fairly well documented and does a reasonably good job at explaining how the different components could be composed to facilitate different data flows, if you (or petr) are curious. I'd be happy to help if so, but I'll also endorse some amount of wheel-reinvention.. [04:18:01] Elsie: Try running git review -s first [04:18:11] That command should be a one time run per repo. [04:18:19] Similar error. [04:18:23] bbl [04:18:48] Hmmm [04:19:13] https://dpaste.de/qUwxr/raw/ [04:19:27] * Elsie googles. [04:19:43] Elsie: What does "git remote show gerrit" result in? [04:20:15] It should error [04:20:20] $ git remote show gerrit [04:20:20] fatal: 'gerrit' does not appear to be a git repository [04:20:20] fatal: Could not read from remote repository. [04:21:03] Try... [04:21:23] git remote add gerrit ssh://mzmcbride@gerrit.wikimedia.org:29418/operations/mediawiki-config.git [04:22:11] Okay, new error. [04:22:40] pastebin please [04:22:41] https://dpaste.de/KAyut/raw/ [04:22:53] I think it's using the wrong private key. [04:23:08] Well. There's your answer :P [04:23:08] How do I specify a key? [04:23:25] I'm not sure, that's a git thing [04:23:39] http://stackoverflow.com/questions/4565700/specify-private-ssh-key-to-use-when-executing-shell-command-with-or-without-ruby [04:24:36] ori-l: :D ty, will do at some point [04:25:27] legoktm: Hmmm, not really helpful. [04:25:45] Yeah, not really. [04:26:41] Never mind, got it. [04:26:47] I had to adjust my .ssh/config. [04:27:10] I had... [04:27:12] Host gerrit [04:27:15] I needed... [04:27:16] Host gerrit gerrit.wikimedia.org [04:27:28] Then I have Hostname gerrit.wikimedia.org \n Port ... [04:27:30] Okay. [04:27:52] $ git review -d 76199 [04:27:52] Downloading refs/changes/99/76199/1 from gerrit [04:27:52] Switched to branch "review/jforrester/bug/49998" [04:27:55] :D [04:28:47] So all I need to now is update the file, git commit --amend, and then git review something? [04:29:18] yup! git review -R [04:29:25] Let's see. [04:29:33] Is -R short for --no-rebase ? [04:30:08] Not sure. Lemme check [04:30:23] Does git review have docs? [04:30:40] `man git` is useless for this third-party shit. [04:30:42] > The -R is important here. It tells git-review to not rebase your change against master, which clutters diffs between patch set 1 and 2. [04:31:08] https://www.mediawiki.org/wiki/Git/Tutorial and the See also have links to docs [04:31:12] has* [04:31:12] So... yes. [04:31:25] Yeah [04:36:06] I think I need --all. [04:36:55] Okay, fingers crossed. [04:37:09] And it failed horribly. [04:37:45] Heh. [04:37:58] E-mail address inconsistency. [04:38:12] I'm wondering what e-mail address I actually want to use. [04:38:16] Does it matter? [04:38:31] If you use the same email as you have in github its easier to link those commits [04:38:39] No outing. [04:38:48] Also that email will be saved FOREVER. So maybe. [04:38:53] FOREVER. [04:38:57] I use g@ for GitHub. [04:39:07] g for git, g for gerrit! [04:39:10] Though w@ for Wikipedia. [04:39:20] So much internal conflict. [04:39:26] m for MediaWiki. [04:39:30] Anyway, whatever. Life is short. [04:39:41] It's done. [04:39:53] (PS2) MZMcBride: Enable anonymous use of VisualEditor on es/fr/he/it/pl/ru/sv [operations/mediawiki-config] - https://gerrit.wikimedia.org/r/76199 (owner: Jforrester) [04:40:54] All right. [04:42:15] Congrats :P [04:42:48] Okay, so if I want to submit a new changeset. [04:43:00] Do I need to un--d myself? [04:43:14] I'm currently on branch review/jforrester/bug/49998. [04:43:22] Should I switch back to the master branch? [04:43:27] Or should I just stay in this branch? [04:43:41] Is this a brand new change? [04:43:51] Hmmmm. [04:44:15] Yes, I guess so. [04:44:18] I'm reverting https://gerrit.wikimedia.org/r/#/c/75543/4/wmf-config/InitialiseSettings.php,unified [04:44:24] Partially. [04:45:00] Well there is a git revert command thing. But that is for full commits, not really partial ones. [04:45:06] Okay. [04:45:09] git checkout master [04:45:10] How do you name branches? [04:45:12] I did that. [04:45:14] Then I ran git pull. [04:45:17] git checkout -b topicname [04:45:18] Now I'm about to name the branch. [04:45:26] I mean how do _you_ name branches? [04:45:29] checkout switches branches, -b is to create a new branch [04:45:36] Right. [04:45:36] Oh [04:45:41] Um, usually bug/#### [04:45:55] Lemme see if there is a guideline.... [04:46:20] https://www.mediawiki.org/wiki/Git/Tutorial#Create_a_branch [04:46:22] Not really. [04:46:26] > Use a descriptive branch name. If you are working on a bug, include its id, for example "2012/bug12345". [04:47:09] 2013/ seems superfluous. [04:47:12] The commit has a timestamp. [04:47:24] Anyway, I named it "bug/52202". [04:49:09] legoktm: What does your typical author line show? [04:49:11] Can you paste it? [04:50:50] Okay, so I ran "git commit --all". [04:50:59] And then I ran "git show HEAD". [04:51:03] And I want to edit my commit message. [04:51:08] Should I just go home? [04:51:57] Meh. [04:52:00] I'll edit it in a second. [05:12:18] God help me, I'm committing to Gerrit. [05:12:42] * YuviPanda claps [05:13:16] Elsie: Oh sorry, I went afk a bit. [05:13:22] I forgive you. [05:13:36] I changed my git user.name. [05:13:40] My author line is "Kunal Mehta " [05:13:40] Which may piss off GitHub. [05:13:42] We'll see. [05:13:48] github uses email [05:15:06] All right. [05:16:52] (03CR) 10Tim Starling: [C: 04-1] "(1 comment)" [core] - 10https://gerrit.wikimedia.org/r/17069 (owner: 10Hashar) [05:17:54] Also, I fixed my text editor to automatically add newlines at the end of the file. [05:20:49] I fixed mine to not strip trailing whitespace. [05:29:33] Krinkle: can you comment on https://gerrit.wikimedia.org/r/#/c/65641/ please? [05:35:09] TimStarling: checking, but I might have to defer it to my mental "tomorrow". It's 7am and on weekdays I usually wake up at noon and work until midnight, yesterday I got hooked in a few community projects that I've been working on all weekend, so this is still my sunday. I'll be back after I've slept and enter my regular monday :) [05:37:36] ok, see you tomorrow [05:37:42] (03PS1) 10Santhosh: Load jquery.ime only when user focus on editable fields [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76469 [05:41:42] (03CR) 10Krinkle: "(1 comment)" [core] - 10https://gerrit.wikimedia.org/r/65641 (owner: 10Alex Monk) [05:42:35] "krinkle: the mental tomorrow, today." [05:44:27] "...a really free project where little big games of power, sex of the angels discussions and alikes were secondary." [05:44:34] https://bugzilla.wikimedia.org/show_bug.cgi?id=49860#c35 [05:44:44] (03CR) 10Krinkle: [C: 04-1] "(3 comments)" [core] - 10https://gerrit.wikimedia.org/r/65641 (owner: 10Alex Monk) [05:45:18] people discuss the sex of angels in portuguese? [05:45:56] is that meant to be like the traditional question of their minimum footprint area? [05:46:19] https://en.wikipedia.org/wiki/Sex_of_Angels [05:47:28] Maybe it's a particularly popular movie to discuss in Portuguese? [05:47:46] hard to believe it would be used to make a point on bugzilla in english [05:48:10] (03CR) 10Legoktm: "(9 comments)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 (owner: 10Legoktm) [05:48:20] (03PS4) 10Legoktm: Parse the input list using a parser function [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 [05:48:38] "Many Christians regard angels as asexual and not belonging to either gender as they interpret Matthew 22:30 in this way. Angels are on the other hand usually depicted in painting and sculpture as looking like male human beings. Their names are also masculine. " [05:48:55] curious [05:48:59] https://en.wikipedia.org/wiki/Angel [05:49:58] Elsie: There, I think I addressed everything. [05:52:54] Cool. [05:53:50] Should the current status thing be its own special page? [05:54:19] Probably. [05:54:24] Special:MassMessage/Status or something. [05:54:44] Special:MassMessageStatus is easier [05:54:55] All right. [05:55:04] You'll need to also add a link to it. [05:55:07] Probably below the header. [05:56:20] I was thinking of adding a link to the your message has been queued message [05:56:27] That's fine. [05:56:37] But you'll also need some inter-Special page navigation. [06:00:48] ok [06:04:42] http://what-if.xkcd.com/55/ [06:16:44] (03CR) 10MZMcBride: "(4 comments)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 (owner: 10Legoktm) [06:17:04] legoktm: Have you had a chance to look at tests for input lists? [06:18:04] Not yet, I just glanced at the example which wasn't really helpful [06:18:17] So fix it. ;-) [06:30:08] (03CR) 10Krinkle: [C: 04-1] "See bugzilla, two blocking bugs." [core] - 10https://gerrit.wikimedia.org/r/70036 (owner: 10Krinkle) [06:31:30] (03CR) 10Krinkle: "@Tim: Both for the and protocol relative urls, what kind of non-browser clients are you referring to (I can think of some, but like" [core] - 10https://gerrit.wikimedia.org/r/70036 (owner: 10Krinkle) [06:50:13] ori-l: Do you still control the test instance of Bugzilla? [06:50:16] On Labs. [06:50:55] no, i gave it to andre [06:51:15] Hmm. I was hoping to change the skin to be more Vector-like. [06:51:20] The search bar is misplaced. [06:51:23] As are the login links. [06:51:26] And there are no tabs. [06:51:50] I mean, if we're gonna get all cutesy, we may as well go overboard. [06:51:55] "Add comment" tab and such. [06:52:30] Anyway, I'll bother Andre about getting access. Thanks. [06:52:54] http://boogs.wmflabs.org/ has an outdated home page. [06:53:07] Probably just a matter of `git pull`. [06:53:45] I added you and legoktm as admins [06:53:59] just let andre know if you plan on doing anything potentially destructive [06:54:24] Oh, cool. Thanks. :-) [06:54:31] Is the account called "boogs"? [06:54:42] project-bugzilla [06:54:46] the bugzilla account, you mean? [06:54:50] the instance is boogs, yes [06:55:22] I think I might have the bugzilla admin account credentials somewhere [06:55:24] I'm trying to figure out where its home dir is. [06:56:10] probably under /srv/bugzilla [06:56:22] Okay. [06:56:34] I think I have to get an SSH-agent thing set up before I can access it. [06:56:46] I can only get into tools-login and bastion right now. [06:57:14] OK, I'll let legoktm help with that [06:57:20] Heh [06:57:22] Ok :P [06:57:42] > Ori.livneh added you to project Nova Resource:Bugzilla [06:57:52] Hot-damn, Echo notification and everything. [06:58:01] https://wikitech.wikimedia.org/wiki/Help:Access#Using_ProxyCommand_ssh_option [06:58:17] (03CR) 10Robert Vogel: [C: 032 V: 032] "Is the initial commit." [extensions/WebPlatformAuth] - 10https://gerrit.wikimedia.org/r/76102 (owner: 10Robert Vogel) [06:58:33] legoktm: Thanks, taking a look now. [06:59:04] (03CR) 10Robert Vogel: [C: 032 V: 032] "Is inital commit" [extensions/WebPlatformSearchAutocomplete] - 10https://gerrit.wikimedia.org/r/76101 (owner: 10Robert Vogel) [07:01:23] Hmmmm. [07:04:55] legoktm: I decided to use ssh-add. [07:05:04] Because I want to be able to ssh from bastion to other hosts. [07:05:21] And the .ssh/config didn't seem to allow that. [07:06:02] No, I think it does [07:06:08] I can ssh from bastion into other things [07:06:26] I can also do stuff like ssh legoktm@legoktm.pmtpa.wmflabs [07:06:33] (from my laptop) [07:06:34] Hmmmmm. [07:13:17] I think I may have broken my SSH key... [07:14:20] No, we're good. [07:22:44] one needs ssh-agent AND ForwardAgent=yes [07:25:33] legoktm: https://wikitech.wikimedia.org/wiki/Managing_Multiple_SSH_Agents [07:25:37] legoktm: What do you do about that? [07:26:48] Seems….complicated. I'm just fine with using one ssh key everywhere [07:32:12] (03PS1) 10Santhosh: Change upstream URL for Shapour font for Pahlavi script [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76470 [07:32:23] legoktm: I don't think you're using only .ssh/config. [07:32:26] I think you're using ssh-add. [07:34:19] (03PS1) 10Spage: Try to use new Agora checkmark style in Login form [core] - 10https://gerrit.wikimedia.org/r/76471 [07:42:33] (03CR) 10Spage: [C: 04-1] "(3 comments)" [core] - 10https://gerrit.wikimedia.org/r/74662 (owner: 10MarkTraceur) [07:43:49] (03CR) 10KartikMistry: [C: 031] "LGTM!" [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76470 (owner: 10Santhosh) [07:48:52] (03PS1) 10Legoktm: Actually enable the MassMessage.alias.php file [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76473 [07:49:25] Elsie: If you want to use your new +2 powers ^ [07:50:20] I finally submitted a few patch sets. [07:50:23] Let's not get crazy. ;-) [07:51:11] Elsie: welllll done [07:51:21] hashar: Thanks. :-) [07:51:40] I was almost going to fill in a bunch of bugs related to your lack of patches :-D [07:51:54] https://gerrit.wikimedia.org/r/#/c/76468/ [07:52:11] https://gerrit.wikimedia.org/r/#/c/69982/ [07:52:24] https://gerrit.wikimedia.org/r/#/c/76199/ [07:52:29] Those are the three I did, I think. [07:53:25] you even managed to amend a patch \O/ [07:53:57] I cheated on 69982. [07:54:02] (03PS1) 10Legoktm: Add a new Special:MassMessageStatus page, which shows how many jobs are queued [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76474 [07:54:47] (03CR) 10Daniel Kinzler: [C: 031] "restoring CR+1 from PS 6: Seems fine, but I havn't tested it, and we have no unit tests for this either. So, before merging, please test :" [core] - 10https://gerrit.wikimedia.org/r/76082 (owner: 10Hoo man) [07:54:58] Hrmph, whitespace. [07:58:06] (03CR) 10MZMcBride: "(3 comments)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76474 (owner: 10Legoktm) [07:59:47] (03CR) 10Legoktm: "(2 comments)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76474 (owner: 10Legoktm) [08:00:35] (03PS2) 10Legoktm: Add a new Special:MassMessageStatus page, which shows how many jobs are queued [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76474 [08:00:46] Oh, I forgot to fix the whitespace. [08:01:14] legoktm: Hold on a sec. [08:01:19] (03PS3) 10Legoktm: Add a new Special:MassMessageStatus page, which shows how many jobs are queued [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76474 [08:01:24] * legoktm holds. [08:02:24] Oh, you already submitted a new patchset. [08:02:53] (03CR) 10MZMcBride: "(1 comment)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76474 (owner: 10Legoktm) [08:04:39] (03CR) 10Legoktm: "(1 comment)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76474 (owner: 10Legoktm) [08:04:51] (03PS4) 10Legoktm: Add a new Special:MassMessageStatus page, which shows how many jobs are queued [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76474 [08:12:28] (03CR) 10Tpt: [C: 04-1] "(2 comments)" [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/72933 (owner: 10Rtdwivedi) [08:16:59] (03PS4) 10Physikerwelt: Expose LaTeXML Settings [extensions/Math] - 10https://gerrit.wikimedia.org/r/75828 [08:24:14] (03CR) 10Nikerabbit: [C: 032] Improve documentation for $wgRecentChangesFlags [core] - 10https://gerrit.wikimedia.org/r/74659 (owner: 10Umherirrender) [08:26:57] (03Merged) 10jenkins-bot: Improve documentation for $wgRecentChangesFlags [core] - 10https://gerrit.wikimedia.org/r/74659 (owner: 10Umherirrender) [08:28:04] (03CR) 10Nikerabbit: "(1 comment)" [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/73448 (owner: 10Umherirrender) [08:28:31] (03PS3) 10Nikerabbit: TypeError: window.navigator.userLanguage is undefined [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/73448 (owner: 10Umherirrender) [08:32:41] hashar: 1) morning. 2) http://ascii.io/a/4428 [08:34:02] ori-l: oh my god [08:34:11] ori-l: does it record your terminal ? [08:34:14] much like "script" ? [08:34:30] yeah, you've never seen ascii.io? it's pretty handy [08:34:35] but check out the video! :P [08:34:38] a while back I wanted to replay "script" command in the browser, eventually gave up reimplementing a terminal in javascript [08:38:19] I've never seen ascii.io. [08:39:40] ori-l: so you install extensions via puppet ? :D [08:39:51] all of that is nice really [08:41:05] hashar: https://git.wikimedia.org/blob/mediawiki%2Fvagrant.git/8842cdc0f6076be657d3b80e0bd0ab37656d5468/puppet%2Fmanifests%2Froles.pp#L201 [08:41:32] yeah [08:41:38] I don't think it is going to scale up very well though [08:41:47] since you got to define extension dependencies in the puppet manifest [08:42:00] I really need to get someone to work on composer [08:42:01] :-D [08:43:28] that will be for after my vacations [08:43:46] ori-l: petan send a mail on wikitech-l about having bots pulling events from RecentChanges [08:43:58] I was wondering whether ZeroMQ could be used as a queue [08:44:02] :o [08:44:06] yes I think it coul dbe [08:44:13] where people can subscribe there and get some kind of history they haven't processed [08:44:15] but there is no zeroMQ server on labs [08:44:23] petan: that can be solved :-] [08:44:25] (03CR) 10TheDJ: "Well, do we really want to add a special class for that ? I guess we could (might also be useful for userscript/wiki customization), but I" [core] - 10https://gerrit.wikimedia.org/r/74391 (owner: 10TheDJ) [08:44:32] yes, definitely [08:44:36] in redis it works the way that as you pull stuff from queue, it get removed from it [08:44:44] so once you process edit, it disappear from it [08:44:56] though adding some buffering in ZMQ will probably make it less stable and less likely to scale [08:45:23] not necessarily [08:45:51] ZMQ provides different socket "types" that each represent a different message-oriented architecture [08:46:01] like dealer, publisher-subscriber, request-response, round robin, etc [08:46:12] (03CR) 10Legoktm: [C: 032] Actually enable the MassMessage.alias.php file [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76473 (owner: 10Legoktm) [08:46:19] petan: the whole point of my message was to make sure you don't rely on the feeds provided on irc.wikimedia.org cause we want to remove that service at one point. :) [08:46:25] (03Merged) 10jenkins-bot: Actually enable the MassMessage.alias.php file [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76473 (owner: 10Legoktm) [08:46:56] some of those buffer by default, all can be configured to buffer [08:46:58] hashar: I don't imagine irc.wikimedia.org will be removed in the next five years... [08:47:12] Or ever, really. But at least. [08:47:22] hashar: yes that's another advantage of my service, you will not remove it, you will replace it with something else I guess. And once that happen I simply update the system which parses the changes to new one and users will not need to change anything [08:48:11] you should separate data encoding from transport [08:48:13] petan: na let me say it otherwise: do not use irc.wikimedia.org has a RC provider :-] [08:48:30] those are two separate problems, even if some platforms bundle them [08:48:43] hashar: ok so what is a better provider? ;) [08:48:48] Everyone else uses irc.wikimedia.org as an RC provider. [08:48:51] I can't hook to UDP messages [08:48:59] petan: the RC2udp code in MediaWiki is horrible [08:49:13] ori-l: can we send to zmq using udp packets ? [08:49:15] whatever, it's still the only solution we have [08:49:19] don't be like Ryan :P [08:49:19] There's a changeset somewhere which makes it slightly less horrible [08:49:40] he told me: you must not ask people for passwords in your application, you need to use some other service, like OAuth [08:49:52] just a minor issue: there is no such thing on whole wikimedia project [08:49:56] petan: I have concerned in having to maintain rc2udp, it broke a while back and it has been an horrible time for the platform team with various bots people constantly harassing us ;) [08:50:11] irc.wikimedia.org is supported and will remain supported. [08:50:28] hashar: https://git.wikimedia.org/blob/operations%2Fpuppet.git/ae908d3f6c6e7486b1b3d184a1edc99ff0ee0232/modules%2Feventlogging%2Fmanifests%2Fservice%2Fforwarder.pp [08:50:30] ok, it may not be best service, hashar we all know, but you can't shut down something without providing alternative [08:50:31] https://gerrit.wikimedia.org/r/#/c/52922/ [08:50:43] Elsie: na we want to phase it out and have bots migrate to something else. The question is what :-] [08:50:55] 1) yes. 2) it's puppetized. 3) for wikimedia. 4) with monitoring. :) [08:50:59] Elsie: but I agree with you, for now irc.wm.o is supported indeed. [08:51:04] the hell are those 'LOST' jenkins-bot failures? [08:51:08] hashar: Even if you had the perfect solution today, the deprecation period would be a year or two, at least. [08:51:32] most important: there is nothing else than irc so irc.wikimedia.org is best solution for now [08:51:58] no matter if you like it or not, until something better arrives we can't fix it [08:52:23] hashar: People have been talking about switching to XMPP for... five years or so, maybe. [08:52:26] Probably longer. [08:52:34] Daniel K. did some work on it at some point, as I recall. [08:52:59] ori-l: so potentially we could have the RC feed sent over UDP to ZMQ, and then we could get redis subscribers that would write the RC feed in redis queues for bots to consume? [08:53:13] hashar: yes, though 0mq can queue [08:53:22] but redis works, yeah [08:53:35] ori-l: the concern is that each bot will need their own RC queue. So I am not 0MQ could scale that many queues [08:54:10] hashar: you might be surprised, but regardless redis works well for this too, i agree [08:54:29] yeah I am sure redis is the proper tool [08:54:36] I am more concerned about reusing irc.wm.o :-] [08:55:22] petan: how would you handle several bots consuming the same RC feed? Would you duplicate data in redis ? [08:55:50] of course [08:55:59] not necessarily [08:55:59] that's the only way to do it securely [08:56:12] you could use redis pub/sub and decide that bots that aren't around to get a message don't get it [08:56:18] I would put same data to 2 separate queues [08:56:34] because the queues are supposed to be secret [08:56:44] so that nobody can break queue of someone else [08:57:13] that's the security model YuviPanda proposed, the redis keys would have secret names (containing a hash) so that other wouldn't be able to read them [08:57:27] but given the production RC feed is huge, that means the queue will be very huge [08:57:36] depend [08:57:37] * s [08:57:39] might be a trouble to duplicate all of that as the number of bots grow [08:57:42] queue contains filtered edits [08:57:52] that means, only edits that people subscribe for [08:58:01] also queue has a max size [08:58:04] hashar: the standard approach is to implement something like a bounded queue by calling LTRIM on every insert [08:58:16] if max size is reached, the oldest records are dropped [08:58:29] so you could limit each bot queue to say 100k events [08:58:41] so 1 bot can take up to N mb of redis [08:58:47] of course [08:58:53] I think [08:58:57] YuviPanda is redis expert not me [08:59:01] so at least you can limit it :] [08:59:17] as I said, this dispatcher thing is going to be very scalable, it can push data to other things than redis [08:59:26] now to feed the redis queue, I would really prefer you guys using a zmq subscriber instead of the irc.wm.o [08:59:35] it could eventually write them to a file, or send them using tcp / udp somewhere [09:00:00] BUT hashar, there is no zmq subscriber [09:00:00] we can get an extension that hook in RC and generate the UDP packets to zmq. That would be a nice replacement to rc2udp [09:00:19] that's what we all talk about, reason why I use irc.wm is that it's the /only/ provider out there [09:00:41] but you still have to write all the code to reverse engineer the ugly IRC formatted messages [09:00:50] that's already written [09:00:55] I stole it from wm-bot [09:00:58] sigh. [09:01:01] :( [09:01:03] ok. [09:01:08] :0 [09:01:10] why sigh [09:01:31] wm-bot's rc feed kind of works [09:01:35] not that it couldn't be improved [09:01:43] understatement of the century [09:01:44] I'm not sure why everyone is hating on irc.wikimedia.org. [09:01:59] neither am I [09:01:59] great, so there is no problem [09:02:01] gl;hf. [09:02:04] Elsie: it is basically a huge hack in core :-] [09:02:07] It's one of the most used and most important services that Wikimedia runs. [09:02:19] Elsie: what's your point? [09:02:37] That the hate seems unjustified. [09:02:48] jesus christ. [09:02:59] Nobody likes parsing the damn thing. [09:03:04] It's awful. [09:03:06] I agree. [09:03:09] we all know how irc.wm.o is important to bots users, but that is really a ugly way to provide a feed of metadata. There are irc color codes as separator, one has to parse it often with their own custom codes [09:03:26] Yup. [09:03:30] and everytime we change something in core that has the potential to make all bots to not recognize the format anymore [09:03:35] thus making people angery [09:03:36] hashar: that's why I am writing this dispatcher, it will provide the data in xml or json or whatever else I can find libraries for [09:03:39] hashar: Yup. [09:03:54] hashar: And yet it's been that way for like a decade. [09:03:59] Elsie: the problem is not with irc.wm.o existing. [09:04:00] providing a nice formatted feed would let people easily read the feed and the whole thing will be way easier to maintain. [09:04:03] it should continue to exist. [09:04:12] but it should not be the exclusive broker of data. [09:04:17] they will just say: hey dispatcher, give me a queue of all edits to Wikipedia:.* on english wikipedia in xml, tyvm [09:04:18] it should consume a well-formatted feed [09:04:19] I agree. [09:04:23] What happened with XMPP? [09:04:26] that is made available to other applications. [09:04:45] Elsie: people on IRC asked too many annoying questions and developer interest vanished. [09:04:47] just a guess. [09:05:02] Elsie: forget about XMPP, as far as I know nobody among platform engineering wants to use that. [09:05:34] funny they didn't ask them via XMPP, though. [09:05:36] Well, before we dismiss it, it might be good to know why it's unsuitable. [09:05:38] (03PS1) 10Jack Phoenix: VoteNY: manually work around bug #46029 for now. [extensions/VoteNY] - 10https://gerrit.wikimedia.org/r/76476 [09:05:50] Elsie: https://www.mediawiki.org/w/index.php?title=Talk%3ARequests_for_comment%2FStructured_data_push_notification_support_for_recent_changes&diff=737350&oldid=525600 btw [09:06:02] (03CR) 10Jack Phoenix: [C: 032 V: 032] VoteNY: manually work around bug #46029 for now. [extensions/VoteNY] - 10https://gerrit.wikimedia.org/r/76476 (owner: 10Jack Phoenix) [09:06:43] petan: I'm sorry if I'm being cross, but I just want to explain. [09:07:03] What you're proposing to do is to fix the problem in a thousand places. [09:07:08] Instead of fixing it in the one place where it should be fixed, which would also fix those thousand other places. [09:07:09] (03PS1) 10Umherirrender: Clean up comment style in DefaultSettings.php [core] - 10https://gerrit.wikimedia.org/r/76477 [09:07:24] (03CR) 10Umherirrender: "changes from Patch Set 1 to 3 looks good." [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/73448 (owner: 10Umherirrender) [09:07:37] You'll get far parsing the IRC feed, but maintaining it is a full-time job, and you'll eventually slack and it'll be another piece of infrastructure rot. [09:07:40] ori-l: I can't fix that one place which needs to be fixed, I am not employed by WMF as ops :P [09:07:55] but I can fix the other thousands places, which IMHO also need a fix [09:08:10] so lets fix the one place whatever it is :-] [09:08:15] will save everyone ton of time [09:08:15] which I can't [09:08:39] by the one piece, I guess you are referring to how MediaWiki sends IRC messages correct? [09:08:42] ori-l: this thing is open source and it run on too labs, I can't do more to make it more open and possible for anyone else to maintain [09:08:42] aka rc2udp [09:08:59] ori-l: or overtake if I didn't [09:09:11] petan: have you tried asking for support in settings this up from wikimedia folks? [09:09:21] petan: or even listened long enough to have this help be offered to you? [09:09:22] Heh. [09:09:33] I tried asking Ryan to set up OAuth 2 years ago [09:09:42] (03CR) 10Ebrahim: "Thanks. farsiweb.ir is down but the link was not wrong." [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76349 (owner: 10Ebrahim) [09:09:47] nothing happened since then and that kind of killed my trust :P [09:09:56] OAuth will happen in the next year. [09:10:08] I heard this years ago [09:10:14] So if you ask about a better feed now, you'll have it by 2016. ;-) [09:10:15] chris steipp is working on OAuth :-) [09:10:19] lol [09:10:23] 2016 [09:10:25] petan: OAuth fails under plattform engineering [09:10:28] I'm not really kidding. [09:10:34] what we do know is that we have the team to focus on a project per quarter [09:10:37] like having LUA [09:10:37] hashar: falls, not fails. :-) [09:10:42] Lua, not LUA. [09:10:43] Elsie: :-]]]]]]] [09:10:54] Elsie: I don't really care, if that thing will be out in 2016 I will change switch few lines of code and dispatcher will still works as it is [09:10:57] last quarter was SUL2 iirc [09:11:03] I am not sure what is going to be next quarter [09:11:05] petan: I agree with you, BTW. [09:11:19] I think relying on vaporware is bad practice and the abstraction layer you're describing seems sane to me. [09:11:36] and we will probably stick Facebook Hiphop soon, maybe to have it land by the end of this winter (aka march 2014 maybe) [09:11:41] Everyone can continue to hate on irc.wm.o, but it's stable and supported. [09:11:49] So the downsides here seem pretty limited. [09:11:58] Other than "it should be better." [09:12:04] and what's more important, /it works/ [09:12:05] Which is how I'd describe most software. [09:12:16] Elsie: or put otherwise, irc.wm.o is so badly designed that nobody want to touch it being afraid to attack bots owner angriness :-] [09:12:22] * ori-l facepalms. [09:12:34] Yeah, you can't change the feed. [09:12:35] It's an API. [09:12:35] A very bad API. [09:12:37] Elsie: I received threat emails by bots owner when I worked on fixing up the rc2udp last year [09:12:41] That Wikimedia created and that Wikimedia will support. :-) [09:12:42] hashar: but why you need to replace the thing? just provide alternative and wait for people to switch [09:12:54] Yeah, you're not allowed to break it. [09:12:57] It's an API. [09:12:58] there is nothing else than irc feed and yet people are blaming it [09:14:19] so [09:14:25] I'm still not sure why XMPP is off the table. [09:14:33] so start something else and let people decide if it's really better or not [09:14:42] instead of blaming only working solution we have :P [09:14:44] let me says it once more: rc2udp + IRC formatting to provide data = HUGE LAME UNMAINTABLE PIECE OF CRAP [09:15:00] It's been maintained for like ten years. [09:15:05] no argument agains that. but it's the only thing we have [09:15:09] I think we can safely dispel the myth that it's unmaintainable. [09:15:09] na it is not maintained [09:15:12] we avoid breaking it [09:15:15] that is slightly different [09:15:15] so until someone (from ops / wmf) fix that we have to use it [09:15:23] much like some abandon ware software everyone is afraid to touch [09:15:29] that is very common in IT. [09:15:35] (we == community people with little powers to fix these things) [09:16:26] Elsie: you're playing a very unsophisticated game. A structured data stream could have a client that transforms it to the input format that the IRC feed expects, and IRC service would continue without interruption. [09:16:58] ori-l: I agree. Will we see that this decade? [09:16:59] I don't think there's a sophisticated enough developer for it, though, since petan is apparently committed to following through the first exciting idea that popped in his head without applying any scrutiny to it. [09:17:04] That's petan's point. [09:17:50] (03PS16) 10Rtdwivedi: Addition of code that will be used for setting up of editing system in PHP instead of JS. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/72933 [09:17:51] (03PS4) 10Rtdwivedi: Toggle header-footer visibility. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/76307 [09:17:52] (03PS12) 10Rtdwivedi: Moved the whole display of fields to PHP. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/74850 [09:17:53] A structured data stream sounds great. As soon as we have one, I'm sure it'll get a lot of use. Until then... [09:17:56] (03CR) 10Santhosh: "The most useful URL here is the version control repository of the font if any." [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76349 (owner: 10Ebrahim) [09:18:51] so my point is that feeding structured data is not necessarily hard [09:19:11] one need an extension that hook in RC feed, format the metadata and send them to zero mq [09:19:22] from then a subscriber will feed the data in redis [09:19:22] done [09:19:25] Will Chris Steipp be working on that? :-) [09:19:37] (03CR) 10Legoktm: "(4 comments)" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 (owner: 10Legoktm) [09:19:41] that will be a nice piece of infra which is would reuse tools wmf engineers are able to omaintain [09:19:46] (03PS5) 10Legoktm: Parse the input list using a parser function [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 [09:19:47] (03CR) 10jenkins-bot: [V: 04-1] Parse the input list using a parser function [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 (owner: 10Legoktm) [09:19:57] Elsie: it is unrelated, Chris is working on security / auth :-] [09:20:08] hashar: Who's going to build the structured data stream? [09:20:20] nobody, now. [09:20:33] Because we've discussed this on IRC and IRC RUINS EVERYTHING. [09:20:44] the idea is to move the burden of structuring data from bots people to mediawiki/core [09:20:45] since apparently it's not an area you can engage in without having your nose rubbed in every unfulfilled WMF project. [09:20:48] so it will benefit to everyone [09:20:54] btw here is example interface for dispatcher if you want to change anything tell me when it's a good time to design it: https://wikitech.wikimedia.org/wiki/Bot_Dispatcher#Example_usage [09:21:38] it's written in python despite I loathe it :D [09:21:39] hashar: You didn't answer the question. [09:21:50] Elsie: ori did :-) [09:22:12] wikibugs is going to resolve this bug. [09:22:13] Wonderful. [09:23:07] (03PS6) 10Legoktm: Parse the input list using a parser function [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 [09:23:12] Elsie: patch converting a UDP log stream that is formatted in an ad-hoc fashion to JSON: https://gerrit.wikimedia.org/r/#/c/75560/ [09:23:32] Elsie: here is the metadata formatter right now https://github.com/wikimedia/mediawiki-core/blob/master/includes/RecentChange.php#L736 RecentChange::getIRCLine(). [09:23:38] Elsie: puppet class for transforming UDP feed to 0MQ publisher: https://git.wikimedia.org/blob/operations%2Fpuppet.git/ae908d3f6c6e7486b1b3d184a1edc99ff0ee0232/modules%2Feventlogging%2Fmanifests%2Fservice%2Fforwarder.pp [09:23:43] haha I would happily get hired by wmf for a contract to implement this structured data stream instead of irc... but I guess people in wmf wouldn't like idea of it being written in c# (or in worst case in c++) because they all love python [09:23:51] (03PS2) 10Umherirrender: Clean up comment style in DefaultSettings.php [core] - 10https://gerrit.wikimedia.org/r/76477 [09:23:57] Elsie: the title is represented with "\00314[[\00307$title\00314]]". That is really not idea [09:24:08] petan: i'm glad we've finally come to the heart of the issue [09:24:15] Elsie: the piece of crap could be replaced to some format accepted by zmq. I guess it is json. [09:24:23] hashar: You realize that I designed and run an IRC bot that uses the feed? [09:24:27] Elsie: so instead you will send something like; { "title": $title } [09:24:29] I'm familiar with the insanity of it. [09:24:31] Elsie: The parser function patch is ready for merging now I think. [09:24:52] Elsie: would you at least agree that consuming a json feed would be simpler for bots owner than parsing the horrible IRC feed? :° [09:25:02] hashar: they can have a choice [09:25:07] not all of the bots have active maintainers [09:25:12] yeah I agree [09:25:17] we can still maintain the IRC feed [09:25:21] hashar: Sure. Where can I find the JSON feed? [09:25:23] and provide a json feed in parralel [09:25:28] hashar: would you at least agree that using bot dispatcher solves this issue? :P because it provides json [09:25:31] then bots owner can migrate from a feed to another one [09:25:54] petan: yeah that solve the issue, but it still rely on an horrible piece of infrastructure [09:26:01] you can always subsribe to --pagerx .* [09:26:07] petan: that needs to be solved one level higher, aka in mediawiki itself [09:26:10] which will basically convert whole irc feed to json [09:26:24] hashar: that's level I am not allowed to touch ;) [09:26:25] so have PHP feed json directly? [09:26:30] wait a moment [09:26:37] (03CR) 10Umherirrender: "Break more long lines" [core] - 10https://gerrit.wikimedia.org/r/76477 (owner: 10Umherirrender) [09:26:39] I don't think we need to solve this on php level [09:26:53] the current UDP thing is good (I like it at least) [09:27:06] it's very fast and can be easily redistributed in different format [09:27:09] using external daemons [09:27:23] the formatting chain would be roughly: PHP object --> IRC message --> json --> whatever bot use [09:27:41] the IRC message could be stripped and the chain would be a bit simpler: PHP object -> json -> whatever bot use [09:28:06] what about PHP ---- udp ----> daemon ----- json -----> queue server (mq-whatever) [09:28:22] so just have the PHP code send directly to the queue server ? [09:28:51] instead of writing the daemon write an extension that hook in RC, craft the json using the same specification as RecentChange::getIRCLine() [09:28:54] idk if that is really possible, but if yes that would be probably best [09:29:02] have it send to Zmq using the event logging extension. Done. [09:29:33] though that would need you to adhere to EventLogging formatting, discovering how to use json schema and the sort [09:29:36] but that is supported [09:29:39] Elsie: Oh wait its actually not ready for merging. I'll fix that tomorrow though [09:29:42] in the end you will have a much more robust systema [09:29:57] and all bots authors would have a very easy to parse feed [09:29:58] I wait with baited breath. [09:30:03] :P [09:30:14] hashar: sounds great, when we can expect that? [09:30:31] invest sometime on eventlogging? : -] [09:30:33] (03CR) 10Legoktm: [C: 04-1] "Missing qqx i18n messages" [extensions/MassMessage] - 10https://gerrit.wikimedia.org/r/76333 (owner: 10Legoktm) [09:30:57] RecentChange_save is the hook you want to use [09:32:47] I willr eply on list [09:36:51] (03PS5) 10Rtdwivedi: Toggle header-footer visibility. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/76307 [09:36:52] (03PS13) 10Rtdwivedi: Moved the whole display of fields to PHP. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/74850 [09:37:21] (03CR) 10Ebrahim: "There isn't such thing for this great font unfortunately. What can I do? Removing font link?" [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76349 (owner: 10Ebrahim) [09:38:18] hashar: I think you should start a new thread for that [09:39:27] I am wondering how easy it is to create a client for that MQ server you talk about [09:39:37] because creation of client for irc server is matter of few minutes [09:39:41] irc protocol is very simple [09:39:42] petan: it is all provided by EventLogging already [09:39:57] hashar: I am not talking about developers of mediawiki [09:39:57] and let me restate it once more: the irc server need to disappear one day [09:39:59] :] [09:40:00] but developers of bots [09:40:17] (03CR) 10Tpt: [C: 04-1] "(4 comments)" [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/72933 (owner: 10Rtdwivedi) [09:40:32] if I was a developer of a bot using a language that doesn't have library for that MQ server, how would I write one, how complicated the protocol is? [09:40:55] ZMQ is basically the dispatcher you want to write [09:41:10] (03PS6) 10Rtdwivedi: Toggle header-footer visibility. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/76307 [09:41:11] (03PS14) 10Rtdwivedi: Moved the whole display of fields to PHP. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/74850 [09:41:15] the events are send in ZMQ, then there is a subscriber consuming the message and relaying there to some place, for example to redis. [09:41:19] I don't think so, unless ZMQ support regex filtering and mediawiki RC flags filtering :P [09:41:21] the bots would read from redis. [09:41:37] we could even have a ZMQ subscriber that write to a mysql database or to some flat files. [09:42:00] yes that's almost exactly what I am writing, I just doubt it would be so flexible [09:42:30] like you can hardly tell ZMQ: provide me a queue of all bot edits on english wikipedia on pages that start with Wikipedia:Blah and are done by user Bob [09:42:52] (you can do that with dispatcher) [09:44:11] you can do that with the ZMQ subscriber :-] [09:44:17] how [09:44:38] just like you want to do it in your dispatcher [09:44:49] it is just that your code will consume messages from ZMQ instead of from IRC. [09:44:57] and will get a json stream instead of the horribly formatted messages [09:45:08] of course I can do that, I was thinking that ZMQ can do the whole job of dispatcher [09:45:30] if it's just a replacement for irc feed, then it's fine [09:46:00] in your use case, think of ZMQ as a replacement for the irc.wm.o service [09:46:14] you still need a client (in ZMQ it is a subscriber) [09:47:10] (03PS8) 10Alex Monk: Provide a JSON recent changes feed. [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [09:47:41] Krenair: touché [09:48:08] added self to reviewers [09:48:42] the rebae was just because jenkins screwed up and 'LOST' stuff :/ [09:49:41] (03CR) 10jenkins-bot: [V: 04-1] Provide a JSON recent changes feed. [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [09:50:08] let's go with that patch [09:50:23] i haven't read it yet but it has a 'looks great' from roan on PS2 [09:50:38] which puts it firmly in the deployable range [09:51:04] Can we have Roan back, please? [09:51:14] The VisualEditor team stole him. [09:51:20] And I'd like him returned. [09:52:02] I think victor sort of forgot about that change [09:52:18] So I picked it up and tried to move everything around into how I think Krinkle wanted it [09:54:32] Krenair: 'By default, this contains a 'default' entry pointing to old $wgRC2UDP paramters', but array underneath is empty. Confusing. [09:54:33] oh my god [09:54:42] people already implemented a JSON RC feed [09:54:43] http://gerrit.wikimedia.org/r/#/c/52922/ [09:54:44] crazy [09:55:16] I was not happy until that link loaded. I thought it was something separate completely by someone else -.- [09:55:59] Anyway that thing will still try to send over IRC (unless you give it code and config to send elsewhere) [09:56:14] can we send to both ? [09:56:24] yes [09:56:28] yep [09:56:29] Unfortunately this means that some things get broken over newlines [09:56:30] ah $wgRCLiveFeeds seems to accept different settings [09:56:31] it's a very good patch [09:56:38] And I don't have much of a solution to it [09:56:39] Looks great otherwise. [09:56:42] so we could get the old UDP and in parallel the stuff to send to ZMQ ? [09:56:49] yes [09:56:55] I hate the getLine() :-] [09:57:17] the idea or the implementation in UDPRCFeedEngine? [09:57:25] and we need to migrate tests probably [09:57:31] Krenair: just the method name [09:57:34] tests need some updating. I'm doing that now [09:57:46] getLine is inherited from the current IRC formatter which indeed craft a line [09:57:55] whereas with a JSON RC based engine, line makes little sense [09:58:06] but that is nitpicking really [09:59:52] I'm open to suggestions if you have any [09:59:53] (03PS9) 10Alex Monk: Provide a JSON recent changes feed. [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [10:00:42] Krenair: I think I am the one that wrote the tests to ensure we had no regression when changing RC [10:00:46] to avoid breaking bots [10:01:07] Krenair: I have a couple of quick suggestions, would you like them here or on the patch? [10:01:31] on the patch please [10:01:36] petan: so seems we are going to have RC events formatted as JSON :-} [10:02:07] hashar, this patch can't go in until the multiple-line IRC problem is solved somehow [10:02:30] and we will have to notify the change to bots people [10:02:56] seems the tests are failing because of the deprecation [10:03:07] probably easy to handle [10:03:36] jenkins passed PS9 [10:04:17] :-] [10:05:21] Krenair: 'By default, this contains a 'default' entry pointing to old $wgRC2UDP paramters', but array underneath is empty. Confusing. [10:05:33] This is because that is done in includes/Setup.php now [10:05:57] i saw. still worth mentioning that or rephrasing. [10:06:16] I'm thinking of separating out the formatter stuff from the sender stuff [10:06:34] but then we end up with classes with only one useful method [10:07:39] maybe we should have some config thing which simply points to the right callable for each format and sending engine [10:08:02] instead of classes extending RCFeedEngine [10:10:03] ah I know what I dislike [10:10:09] it mix IRC and JSON in the same class [10:10:25] yes, that's what I was talking about [10:11:26] sorry was not paying attention [10:13:58] (this idea was by Krinkle in https://gerrit.wikimedia.org/r/#/c/52922/4/includes/RecentChange.php ) [10:15:24] (03CR) 10Ori.livneh: [C: 04-1] "The logic for emitting log records via UDP is implemented more carefully in wfErrorLog() in GlobalFunctions.php -- there's support for IPv" [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [10:15:51] Krenair: ^ [10:16:54] I had no idea about that. [10:17:48] it's like the one area of obscure mediawiki lore i'm halfway competent in, so let me have my moment here :P [10:18:41] Feels like something that was made for the WMF cluster and touched by no one else [10:18:55] ......no comments [10:19:04] (03PS2) 10Daniel Kinzler: (bug 51621) Make SBL aware of ContentHandler. [extensions/SpamBlacklist] - 10https://gerrit.wikimedia.org/r/75867 [10:19:14] (03CR) 10Hashar: [C: 04-1] "Thank you everyone for working on this. That will eventually let us phase out the horrible IRC formatting one day =)" [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [10:19:50] Krenair: that is definitely wmf specific. Tim would know about the history [10:20:10] possibly because we had some bots parsing Special:RecentChanges and relaying to the #wikipedia channels here in ffreenode [10:20:25] I guess the bots get k-lined because they spammed the servers and we went with our own IRC server [10:20:32] then bots user started parsing the feed [10:21:36] json all the things! [10:21:48] https://en.wikipedia.org/wiki/File:NSApolygraphvideo.webm [10:21:53] ori-l: ಠ_ಠ [10:22:09] json some of the things! [10:22:21] hashar, actually we were discussing wfErrorLog's udp support [10:22:56] (03PS3) 10Daniel Kinzler: Detect leakage of error_reporting state in tests. [core] - 10https://gerrit.wikimedia.org/r/74170 [10:23:19] (03CR) 10jenkins-bot: [V: 04-1] Detect leakage of error_reporting state in tests. [core] - 10https://gerrit.wikimedia.org/r/74170 (owner: 10Daniel Kinzler) [10:25:40] Krenair: yup I changed my reply when I have seen ori comment [10:25:57] I feel horrible to ask you guys to split the classes a bit more though [10:26:28] It makes sense [10:26:32] Krenair: splitting formatting would let one write nice unit tests though [10:26:49] (03CR) 10Ori.livneh: "You might actually get away with using wfErrorLog without splitting it up if you modify the $wgRCLiveFeeds configuration array to use URIs" [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [10:26:53] But we end up with one-method classes [10:27:35] RC_Formatter_Json::format() RC_Formatter_IRC::format() :] [10:28:07] but then it is trivial to add a new formatter and the code is well isolated and thus easy to unit test [10:31:11] ori-l, I feel like actually using wfErrorLog would be a bad idea [10:31:40] how come? [10:31:44] Because someone might come along, change something in the function and suddenly all RC messages are sent to some error log somewhere [10:32:03] Honestly I wouldn't think to check all calls to such a function to make sure they're not doing anything inappropriate [10:33:11] I don't think it's a sensible concern, tbh. You specify the UDP connection URI as a parameter to wfErrorLog; it's not like the destination is a private implementation detail [10:33:37] Hello the page download of my extension is empty. are there a problem ? http://www.mediawiki.org/wiki/Special:ExtensionDistributor/LinkedWiki [10:33:41] it's also used extensively on the cluster [10:34:16] I might add a comment to the doc for wfErrorLog [10:34:20] But it still seems wrong [10:34:27] Krenair: look at $wgDebugLogGroups in InitialiseSettings [10:34:34] well, 'wgDebugLogGroups'. [10:34:42] ~line 4137 [10:35:10] Just like RecentChange::notifyRC2UDP became RecentChange::notifyLiveFeeds with the old one deprecated, if we're going to use wfErrorLog for purposes unrelated to error logging, we should rename it and deprecate the old name [10:35:23] yeah, I was going to say that myself [10:35:53] but it's used in extensions too, so pick your battles [10:36:06] Ugh. Which? [10:36:20] sorry now this ok... strange... [10:36:26] if i said eventlogging [10:37:05] >_> [10:37:07] ...nevermind. anyways, renaming it might be worthwhile. [10:38:03] and as you can see from the list in InitialiseSettings, it's already use as a generic bus to ferry different kinds of content [10:38:34] and there's quite a lot of infrastructure and software for working with such streams [10:38:44] deployed on the cluster, I mean. [10:39:20] I think it should be deprecated in favor of something like 0MQ, but you're not complicating the migration by using the same centralized dispatch point that everything else is using [10:40:18] in fact you're arguably facilitating it [10:40:35] got it :) [10:43:09] ok, it's nearly 4 AM here [10:43:50] * ori-l waves [10:44:41] waves [10:44:42] :) [10:45:33] Now I'm going to have a class like this: [10:45:47] class UDPRCFeedEngine implements RCFeedEngine { public function send( $line, $feed, $prefix ) { wfErrorLog( $prefix . $line, $feed['uri'] ); } } [10:45:48] -.- [10:47:12] (03PS1) 10Aude: Add DataTypes extension, a dummy version with i18n only for the branch [core] (wmf/1.22wmf12) - 10https://gerrit.wikimedia.org/r/76480 [10:55:16] (03PS1) 10Rillke: Make jquery.autosize aware of style changes [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76482 [10:58:42] (03CR) 10Rillke: "Please note there a slightly different approach at 45cc4a57ee44f1490ec6512d102cc11f7974358f" [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76298 (owner: 10Rillke) [10:59:55] (03PS1) 10Aude: Add DataTypes extension, a dummy version with i18n only for the branch [core] (wmf/1.22wmf11) - 10https://gerrit.wikimedia.org/r/76483 [11:03:34] (03PS1) 10Aude: add DataTypes to make-wmf-branch [tools/release] - 10https://gerrit.wikimedia.org/r/76485 [11:05:35] (03PS10) 10Alex Monk: Provide a JSON recent changes feed. [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [11:05:43] well there you go ori-l-away & hashar [11:06:01] Reedy: aude: is there a way to find out how often a specific API module is used or if it's used at all? [11:06:34] I'm wondering if the wikibase module wbcreateclaim is ever used.. [11:07:14] Tobi_WMDE: it is! [11:07:21] bots use it [11:07:33] how often, not sure [11:07:42] what's the issue? [11:07:45] aude: I doubt, 'cause it's broken [11:07:48] completely [11:07:50] why? [11:07:55] and probably always was [11:07:59] how? [11:08:20] aude: let's switch to wikidata channel [11:08:24] k [11:09:13] {"id":1828,"type":"edit","namespace":0,"title":"Main_Page","full_title":"Main [11:09:14] Page","comment":"","timestamp":"2013-07-29T11:04:22Z","user":"Krenair","bot":true,"minor":0,"patrolled":1,"old_len":368,"new_len":371,"old_revision":1709,"new_revision":1817,"url_server":"http://localhost","url_script_path":"/MediaWiki/Git/core/index.php","url_article_path":"/MediaWiki/Git/core/index.php/$1"} [11:09:24] Krenair: that was fast [11:09:24] Tobi_WMDE: we have api logs [11:09:33] haven't figured out how to fix that newline yet hashar [11:09:41] Tobi_WMDE: so potentially we could parse the api query log and find out what is being used. [11:10:16] Krenair: space replaced by "\n" ? scary [11:10:27] Krenair: you can write a small test for it :-] [11:10:56] I think it just happened to break there [11:11:21] and it was caused by my software that received the UDP packet and sent it on to IRC, rather than MediaWiki [11:12:14] Krenair: you can also use netcat to setup a UDP listener on your work machine. [11:12:51] something like: netcat -l -p 1666 127.0.0.1 [11:13:17] Tobi_WMDE: 6800 occurrences of webcreateclaim in our api.log since it rotated (a few hours ago probably) [11:13:41] hashar: thx! that helps.. [11:14:10] Tobi_WMDE: log got rotated at 6:25am or 5 hours ago roughly [11:15:37] Tobi_WMDE: and it is used by someone named AudeBot. No idea who it can be [11:15:39] *smiles* [11:15:50] hah [11:15:51] it's not broken [11:16:08] Krenair: could you comment on https://gerrit.wikimedia.org/r/#/c/52922/ ps 10 about what you changed ? :) [11:16:18] i have some more stuff to fix first [11:16:19] it works fine, although the way the modules behave might be inconsistent [11:16:22] Krenair: just as an introduction :-] [11:16:49] that would be ~50,000 edits so far :) [11:16:54] with create claim [11:20:54] (03CR) 10Nikerabbit: [C: 032] Use one call to .attr instead of two and linebreak consistency [core] - 10https://gerrit.wikimedia.org/r/76345 (owner: 10TheDJ) [11:23:31] (03Merged) 10jenkins-bot: Use one call to .attr instead of two and linebreak consistency [core] - 10https://gerrit.wikimedia.org/r/76345 (owner: 10TheDJ) [11:23:37] (03PS11) 10Alex Monk: Provide a JSON recent changes feed. [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [11:25:13] (03CR) 10Siebrand: "This library originates from https://github.com/jackmoore/autosize. We are using a different version from the current version v1.17.2 - 20" [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76482 (owner: 10Rillke) [11:27:08] (03PS1) 10Robert Vogel: Initial commit [skins/BlueSpiceSkin] - 10https://gerrit.wikimedia.org/r/76489 [11:28:13] (03CR) 10Robert Vogel: [C: 032 V: 032] "This is sitll an early stage but the code is okay" [skins/BlueSpiceSkin] - 10https://gerrit.wikimedia.org/r/76489 (owner: 10Robert Vogel) [11:36:00] (03CR) 10Rillke: "That's a fine idea. Update follows." [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76482 (owner: 10Rillke) [11:36:20] AzaToth: I switched to a theme that does colors. Looks very nice now :) [11:36:42] nice [11:43:24] (03PS4) 10Matthias Mullie: Send Echo notifications for AFTv5 [extensions/ArticleFeedbackv5] - 10https://gerrit.wikimedia.org/r/74166 [11:44:26] YuviPanda, I like grrrit-wm now. nice work with that [11:44:32] Krenair: :) [11:45:03] (03CR) 10Matthias Mullie: [C: 031] "I assume your rebase means you're ok with the change?" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/74617 (owner: 10Matthias Mullie) [11:45:09] Krenair: should replace wikibugs at some point, I guess [11:45:15] (03CR) 10Matthias Mullie: "Depends on https://gerrit.wikimedia.org/r/#/c/74617/" [extensions/ArticleFeedbackv5] - 10https://gerrit.wikimedia.org/r/74166 (owner: 10Matthias Mullie) [11:45:29] would be good [11:45:42] YuviPanda, wikibugs apparently parses wikibugs-l emails [11:45:56] Krenair: indeed. [11:46:21] hashar, have you seen https://gerrit.wikimedia.org/r/#/c/71932/ ? [11:56:37] (03PS2) 10Rillke: Update jquery.autosize [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76482 [11:59:51] (03CR) 10Rillke: "Voilà !" [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76482 (owner: 10Rillke) [12:06:25] (03CR) 10Matmarex: "Krinkle, this is an annoying and visible problem, so I'd appreciate if you either removed your -1 or provided some constructive feedback." [core] - 10https://gerrit.wikimedia.org/r/75671 (owner: 10Matmarex) [12:07:25] i need a javascript person to comment on https://gerrit.wikimedia.org/r/#/c/36391/6 . anyone? please? [12:08:05] Krenair: I am barely reviewing code nowadays [12:13:56] Krenair: replied :] I did some similar change earlier. [12:20:36] (03CR) 10Amire80: [C: 04-1] "(1 comment)" [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76469 (owner: 10Santhosh) [12:21:33] Krenair: and we talked about $wgSquidServersNoPurge in #wikimedia-operations :] [12:25:32] hashar: hallo. [12:25:37] hi [12:26:08] I need a reminder: bastion1 is the host that runs http://en.wikipedia.beta.wmflabs.org/wiki/Boleyn_family , right? [12:26:38] no idea what bastion1 is :-] [12:26:54] the URL above is being served by the beta cluster (labs project is named "deployment-prep" ) [12:27:13] which is served by some instances caches and mediawiki instances that reproduced production [12:27:20] mmmm... to ssh to it, do I do `ssh bastion.pmtpa.wmflabs' ? [12:27:26] or something else? [12:27:40] the work machine for beta is deployment-bastion.pmtpa.wmflabs :) [12:27:54] that is the equivalent of tin.eqiad.wmnet in production [12:28:10] if you get an access to prod, you should feel at home [12:28:23] aharoni: what do you want to do ? [12:29:24] mlitn: Hi. Can you give me a partly dump of the dewiki-aftv5-db? Doesn't want to error-prone parse the html-output of api [12:29:30] hashar: I'll just tell the whole story: there's a very weird bug on that page - it looks like a message isn't loaded. It is loaded correctly on my local machine, which runs the same Git revision and on the English Wikipedia as well. [12:29:38] So I need some way to debug it. [12:30:35] That's because the cucumber tests run there. [12:30:56] aharoni: what message? [12:31:00] a message? [12:31:05] * aude does not see anything wrong with the page [12:31:11] is it provided by a json structure served out of bits ? [12:31:43] uls looks fine [12:32:16] aude, hashar : you can reproduce it like this: click the ULS cog icon, click "Input" on the sidebar, bring your mouse pointer to the "How to use" link near the "Internation phonetic alphabet" button. [12:32:30] check your browser's status bar to see where the link points [12:32:37] you'll see a message that wasn't properly loaded. [12:33:57] aharoni: confirm [12:34:54] hashar: so, to debug it on my local machine, I'd do the simple and stupid thing and add some log messages to the source where that message is supposed to be loaded, though I'm not sure that it's OK to do it on that host... is it? [12:34:54] sends me to http://en.wikipedia.beta.wmflabs.org/wiki/%3Culs-ime-helppage%3E :D [12:35:03] so there is some l10n message not properly updated [12:35:09] aharoni: i think it' sokay [12:35:18] it is okay, but defer to hashar [12:36:41] aharoni: although i think it's a js issue, so console would work [12:36:48] cp: `/data/project/apache/common-local/wmf-config/ExtensionMessages-master.php' and `/usr/local/apache/common-local/wmf-config/ExtensionMessages-master.php' are the same file [12:36:48] Update of MediaWiki localisation messages failed [12:36:50] .... [12:36:54] I HATE OUR L10N INFRA [12:37:19] Ah! So maybe you found it. [12:37:29] yeah [12:37:35] l10n is broken on [12:37:37] beta [12:37:45] l10n update? [12:38:09] oh, it might be our fault :/ [12:38:32] we removed the offending wikibase code from master, if it's what i think [12:38:54] let me know if it works now? [12:39:28] like, now-now? [12:39:30] not yet. [12:39:36] did you update anything? [12:39:39] caused by some change that made cp to die out the mw-update-l10n script [12:39:39] hehe [12:39:43] * aharoni waves at arrbee [12:40:02] hello aharoni [12:40:36] aharoni: no, does the l10n update run now? [12:41:18] * aharoni never properly learned about the internals of l10n-update [12:41:26] ok [12:41:36] i've not run in on beta and believe it is automated [12:41:57] * aude ran it on my wikifarm and thinks wikibase broke it for master [12:46:45] so, hashar aude, any idea how to get it fixed? [12:46:57] yeah looking at it [12:47:03] mw-update-l10n needs to be tweaked [12:47:05] i'll let hashar look [12:47:09] +1 [12:47:25] beta only has one source directory [12:47:31] * aude just curious if it still fails or if we fixed it [12:47:40] and cp fails when trying to overwrite the file with itself [12:47:49] ok [12:47:55] that might be a different problem then [12:48:05] yup [12:48:10] been the case for a few weeks [12:48:13] ok [12:48:37] yes , it would be great to be able to test l10n update on beta [12:50:19] back in a bit.... [12:51:01] yeah I need to have it monitored [12:51:05] or replaced by a jenkins job [12:53:15] (03PS1) 10Umherirrender: Add more parser tests for variables [core] - 10https://gerrit.wikimedia.org/r/76493 [13:02:05] (03PS7) 10Matmarex: Provide month names via mediawiki.language.months RL module [core] - 10https://gerrit.wikimedia.org/r/36391 (owner: 10Wizardist) [13:04:29] (03CR) 10Matmarex: "So, could somebody look at this?" [core] - 10https://gerrit.wikimedia.org/r/36391 (owner: 10Wizardist) [13:04:33] (03PS8) 10Rjain: Added the permissions plugin [extensions/Annotator] - 10https://gerrit.wikimedia.org/r/75645 [13:06:19] (03CR) 10Rjain: "(1 comment)" [extensions/Annotator] - 10https://gerrit.wikimedia.org/r/75645 (owner: 10Rjain) [13:06:21] (03PS1) 10Matmarex: mw.config: Deprecate wgMonthNames and wgMonthNamesShort [core] - 10https://gerrit.wikimedia.org/r/76497 [13:12:32] (03PS1) 10Umherirrender: Added CoreVariables class and Parser::setVariableHook [core] - 10https://gerrit.wikimedia.org/r/76498 [13:16:58] ok got a patch [13:17:28] anomie: are you around by any chance ? : -] [13:17:38] hashar: Yes [13:17:38] aude: I got some shell review for you if you are in the mood for it [13:17:39] https://gerrit.wikimedia.org/r/76500 [13:17:41] anomie: https://gerrit.wikimedia.org/r/76500 [13:17:42] :-] [13:17:47] * anomie looks [13:18:05] it is an oddity on labs where the php extension messages ends up being the same file [13:18:12] we have to copy it locally on tin to refresh the message [13:18:19] but on labs that is always the NFS share /data/project [13:18:46] also I am not sure how scary it is to run mw-update-l10n in production [13:20:58] (03CR) 10Matthias Mullie: [C: 032] link-from-page should be linked in html email outputformat [extensions/Echo] - 10https://gerrit.wikimedia.org/r/75781 (owner: 10Bsitu) [13:21:07] (03Merged) 10jenkins-bot: link-from-page should be linked in html email outputformat [extensions/Echo] - 10https://gerrit.wikimedia.org/r/75781 (owner: 10Bsitu) [13:23:59] (03PS1) 10Matmarex: updater: Move rev_sha1 addition before convertUserOptions [core] - 10https://gerrit.wikimedia.org/r/76502 [13:25:40] (03PS2) 10Matmarex: updater: Move rev_sha1 addition before convertUserOptions [core] - 10https://gerrit.wikimedia.org/r/76502 [13:25:53] anomie: thank you :-] [13:32:24] hashar: looking [13:32:48] aude: anomie reviewed it as well [13:32:49] :) [13:33:04] k [13:33:26] 1 + 1 = +2 :) [13:33:32] if it only worked like that.... [13:34:13] aude: If we switch from gerrit into $calculatorApplication ... [13:34:20] :) [13:35:03] hi hoo. i need you [13:35:19] https://gerrit.wikimedia.org/r/#/c/36391/ needs a comment. :) [13:35:19] :) [13:35:38] not that again :P [13:35:39] I still don't really have a strong opinion [13:35:48] you didn't comment the least time :P [13:36:30] well, i want this merged in one form or another, and apparelty all of WMF's javascript guys are busy writing visualeditor [13:36:36] Because I didn't have a strong opinion then either [13:37:12] MatmaRex: have you tried pinging Santhosh? [13:37:30] sumanah: he's a reviewer on the patch [13:37:50] he complained that the previous version of that patchsucked, and i rewrote it since [13:38:21] MatmaRex: yes, I mean, have you tried reaching out to him personally, as you just reached out to hoo :) [13:38:25] Sometimes that helps. [13:38:38] he's rarely on irc [13:38:45] He responds to email. [13:38:49] Just a suggestion. Trying to help. [13:39:51] Those are Gerrit. Personal email gets a better response. [13:39:56] As marketing companies have known forever. [13:39:59] Hi santhosh! [13:40:17] if i had to write a few personal emails for each change i submit, i'd bedoing nothing allday but writing mails [13:40:29] (sorry, i need to fix my spacebar) [13:40:29] MatmaRex: Most people respond more to personal email than to automatic notifications. It's just a fact of life. When you really really want something you spend a little extra time. Most of the time you don't. [13:40:58] Again, I am simply making a suggestion to help you get what you want. [13:41:01] sadly it looks like my throughput is too much for wmf's engineers to catch up to [13:41:03] Please take it in that spirit. [13:41:07] and myself i'm merging more hanges than l10n-bot [13:41:09] changes* [13:41:15] * sumanah leaves conversation [13:41:50] (03CR) 10Hoo man: "While I don't really have a strong opinion on what we do here it IMO has to be as consistent with other code as possible. That includes bo" [core] - 10https://gerrit.wikimedia.org/r/36391 (owner: 10Wizardist) [13:44:56] thanks hoo [13:47:02] (03PS1) 10Rillke: Make Special:AggregateGroups using jquery.chosen [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76503 [13:47:08] (03CR) 10jenkins-bot: [V: 04-1] Make Special:AggregateGroups using jquery.chosen [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76503 (owner: 10Rillke) [13:48:51] (03CR) 10Rillke: "This building system does not honer when depending on external scripts. It should not vote with -1 for reasons like "Identifier 'search_co" [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76503 (owner: 10Rillke) [13:54:39] (03PS2) 10Rillke: Make Special:AggregateGroups using jquery.chosen [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76503 [13:56:39] (03PS3) 10Rillke: Make Special:AggregateGroups using jquery.chosen [extensions/Translate] - 10https://gerrit.wikimedia.org/r/76503 [13:59:16] (03PS9) 10Rjain: Added the permissions plugin [extensions/Annotator] - 10https://gerrit.wikimedia.org/r/75645 [14:01:17] (03CR) 10Rjain: "Same comment as patchset 8" [extensions/Annotator] - 10https://gerrit.wikimedia.org/r/75645 (owner: 10Rjain) [14:11:54] (03PS4) 10Manybubbles: Add tests for CirrusSearch. [qa/browsertests] - 10https://gerrit.wikimedia.org/r/75793 [14:14:06] hashar: I'm not sure what still has to be done to get CirrusSearch into beta but I'd really like to get it done today if possible. It'd make me feel better. [14:15:37] hashar: will you help walk me through it? I think I need configure those machines with puppet. and my irc client is broken.... [14:18:53] manybubbles: the extension needs to be registered in mediawiki/extensions.git [14:19:13] you need a bunch of configuration in operations/mediawiki-config.git whenever it is merged it is deployed on beta automatically [14:19:50] and the instances should be configured using puppet class which are applied in wikitech console [14:20:00] and I guess you are already fulfilling all the above [14:20:46] hashar: I don't know about mediawiki/extensions.git [14:20:57] manybubbles: luckily that path is covered already [14:21:02] cool [14:21:18] manybubbles: that is a "super project", its submodules are some extensions and they are updated by Gerrit automatically [14:21:35] I wonder if ^demon did the instances - which project are they in in wikitech? [14:21:38] so whenever an extension receive a new change in their master branch, Gerrit update the corresponding submodule in mediawiki/extensions.git [14:21:45] (03PS1) 10Matmarex: Expand the info in 'resettokens-watchlist-token' [core] - 10https://gerrit.wikimedia.org/r/76506 [14:21:54] then on beta we just fetch that repo from time to time and thus we have the extensions code autoupdated :-] [14:22:06] manybubbles: the project for beta is named deployment-prep [14:22:28] hashar: can you add me to the project? I think it is hidden from me. [14:22:55] manybubbles: labs account being manybubbles ? [14:22:59] yeah [14:23:17] manybubbles: you are in :)) [14:23:27] manybubbles: the main work machine is deployment-bastion.pmtpa.wmflabs [14:23:43] the mediawiki core / extensions and config are updated automatically from git master branches [14:23:46] as user mwdeploy [14:23:53] usually you don't need to touch that on the instance [14:24:06] cool [14:24:17] let me make sure the elasticsearch machines actually have elasticsearch installed [14:24:34] (03CR) 10MarkAHershberger: [C: 032 V: 032] updater: Move rev_sha1 addition before convertUserOptions [core] - 10https://gerrit.wikimedia.org/r/76502 (owner: 10Matmarex) [14:30:38] (03CR) 10Mdale: [C: 031 V: 031] "Looks fine, but ops should approve." [extensions/TimedMediaHandler] - 10https://gerrit.wikimedia.org/r/63418 (owner: 10J) [14:32:14] (03PS1) 10Hashar: beta-code-update replaces upstart service [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/76507 [14:32:23] (03CR) 10Hashar: [C: 032] beta-code-update replaces upstart service [integration/jenkins-job-builder-config] - 10https://gerrit.wikimedia.org/r/76507 (owner: 10Hashar) [14:43:08] hashar: so once those machines run puppet they should have elasticsearch and me as a sudoer [14:43:53] (03PS1) 10Hashar: fss.ini used '#' for comments, which is deprecated [php/FastStringSearch] - 10https://gerrit.wikimedia.org/r/76511 [14:44:05] hashar: so once we decide to merge ^demon and my config changes one of us will have to shell into one of the mediawikis and run some maintenance scripts [14:45:51] manybubbles: you get root sudo on all instances [14:46:28] manybubbles: you will have to apply the puppet classes on your instance, that is done in wikitech web console [14:46:41] hashar: already done the puppet [14:46:49] manybubbles: the mediawiki-config stuff is automatically deployed on beta whenever it is merged in master branch of operations/mediawiki-config.git [14:47:07] hashar: is there a machine I should use for maintenance scripts? [14:47:30] deployment-bastion.pmtpa.wmflabs [14:47:48] I usually run them as the mwdeploy user, so I have an alias: alias mwdeploy='sudo su --login --shell /bin/bash mwdeploy' [14:48:01] though maybe you can just do something like: sudo -u mwdeploy mwscript foobar.php [14:48:42] (03CR) 10Siebrand: "(1 comment)" [core] - 10https://gerrit.wikimedia.org/r/76506 (owner: 10Matmarex) [14:49:33] cool - question about the mediawiki-config - when does that get pushed to production? I don't want to accidentally do that at all [14:50:09] so when you merge in [14:50:16] you have to deploy in production right away [14:50:19] thought that is done manually [14:50:32] but we have a way to only change beta [14:50:50] getRealmSpecificFilename( 'foobar.php' ); is a way to load different PHP files in prod and labs [14:51:22] hashar: I thought something like that. [14:51:27] so possibly we could have a search-prod.php that would hold the settings for production [14:51:32] and search-labs.php for beta :-] [14:51:54] hashar: so I think we've done this right then: https://gerrit.wikimedia.org/r/75507 [14:51:54] there is a few example lying under wmf-config/ you can have a look at the lucene* files [14:52:19] it doesn't contain the prod config but we'll get to that once we know how it'll be deployed to prod [14:52:34] we require lucene-common.php which in turns does the nasty: require( getRealmSpecificFilename( "$wmfConfigDir/lucene.php" ) ); [14:52:43] that would load either lucene-prod.php or lucene-labs.php [14:58:22] (03PS1) 10Hashar: build status for beta-code-update job [integration/docroot] - 10https://gerrit.wikimedia.org/r/76512 [14:58:37] (03CR) 10Matmarex: "(1 comment)" [core] - 10https://gerrit.wikimedia.org/r/76506 (owner: 10Matmarex) [14:58:49] (03CR) 10Hashar: [C: 032] build status for beta-code-update job [integration/docroot] - 10https://gerrit.wikimedia.org/r/76512 (owner: 10Hashar) [14:58:49] (03Merged) 10jenkins-bot: build status for beta-code-update job [integration/docroot] - 10https://gerrit.wikimedia.org/r/76512 (owner: 10Hashar) [14:59:00] (03CR) 10Siebrand: [C: 04-1] "(1 comment)" [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/72933 (owner: 10Rtdwivedi) [14:59:48] (03PS2) 10Matmarex: Expand the info in 'resettokens-watchlist-token' [core] - 10https://gerrit.wikimedia.org/r/76506 [15:00:44] hashar: I just remembered - installing this extension currently requires symlinking some scripts into $IP/maintenance - it looked like something other extensions did - how do I set that up? [15:01:00] that is horrrible [15:01:02] :-D [15:01:20] sorry - is there something better? [15:01:24] It is easy for me to change [15:01:41] ahh I think I know why [15:01:50] mwscript myjob.php [15:02:10] mwscript would not find it so one has to do something like: mwscript extensions/CirrusSearch/maintenance/myjob.php [15:02:15] probably why people create symlinks [15:02:48] there are none in beta, the MediaWiki code is an exact copy of master without any local hack [15:02:57] it actually get rid of any local modification if there are any [15:04:05] (03PS2) 10Santhosh: Load jquery.ime only when user focus on editable fields [extensions/UniversalLanguageSelector] - 10https://gerrit.wikimedia.org/r/76469 [15:04:57] well once I'm installed I can see if that works [15:07:51] manybubbles: commented on https://gerrit.wikimedia.org/r/#/c/75507/ (cirrus search mw conf) [15:08:16] hashar: Thanks! I'll read it soon! [15:14:17] hey aharoni. [15:14:24] do you want rights on hewiki to be able to do imports? [15:14:32] I can grant you those [15:14:36] that was fast :) [15:14:42] well, now that you have access *you* can grant you those ;) [15:14:42] the labs hewiki , you mean? [15:14:45] yeah [15:15:00] well, that would probably be ok. [15:15:04] let me do that then :) [15:15:13] how much to import, i wonder [15:15:15] (03PS1) 10Manybubbles: Get maintenance scripts to respect MW_INSTALL_PATH [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/76513 [15:15:30] what's your username there? [15:16:47] (03CR) 10Manybubbles: "If you want to run these in the 'normal' labs setup where you clone cirrussearch to /srv/CirrusSearch and symlink it into /srv/mediawiki/e" [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/76513 (owner: 10Manybubbles) [15:17:22] aharoni: what is your username? [15:18:27] * YuviPanda pokes aharoni some more [15:22:41] * YuviPanda pokes aharoni again [15:23:41] (03CR) 10Nikerabbit: [C: 04-1] "(2 comments)" [extensions/cldr] - 10https://gerrit.wikimedia.org/r/45499 (owner: 10Kaldari) [15:29:19] YuviPanda: probably Amire80 as usual [15:29:31] hi YuviPanda [15:30:49] hey Nikerabbit [15:30:50] Nemo_bis: that doesn't seem to exist? [15:31:33] manybubbles: I am off will be back later tonight [15:32:39] manybubbles: make sure to warn Mobile and QA whenever you switch beta to CirrusSearch, that might break their Selenium tests :-] [15:32:50] hashar: I'm sure it will:) [15:34:20] YuviPanda: hmpf, only a test account on the en wiki too [15:34:42] yeah i'll just wait for him i guess [15:37:44] off be back tonight [15:39:13] YuviPanda: i'm back, was cooking :0 [15:39:40] (03CR) 10Nikerabbit: "Missing @since." [core] - 10https://gerrit.wikimedia.org/r/74320 (owner: 10Aaron Schulz) [15:44:46] (03CR) 10Demon: [C: 032] Get maintenance scripts to respect MW_INSTALL_PATH [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/76513 (owner: 10Manybubbles) [15:44:48] (03Merged) 10jenkins-bot: Get maintenance scripts to respect MW_INSTALL_PATH [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/76513 (owner: 10Manybubbles) [15:45:55] YuviPanda: sorry - created Amire80 at http://he.wikipedia.beta.wmflabs.org/ [15:46:11] aharoni: done [15:46:12] :) [15:46:46] (03CR) 10Demon: "Functionally ok, but I'm kinda sad that we can't do this with the bulk API better. Since we're already storing it in a static variable, is" [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/75151 (owner: 10Manybubbles) [15:46:48] aharoni: can you verify? [15:49:07] (03CR) 10Manybubbles: "I suppose I'm not clear enough on the process model in php to do that." [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/75151 (owner: 10Manybubbles) [15:51:06] (03CR) 10Nilesh: [V: 032] Removed rescorer, added maven dependencies for testing [extensions/WikidataEntitySuggester] - 10https://gerrit.wikimedia.org/r/76033 (owner: 10Nilesh) [15:54:32] (03CR) 10Nilesh: [V: 032] Added unit tests for AbstractEntitySuggesterServlet and WebClientRecommender [extensions/WikidataEntitySuggester] - 10https://gerrit.wikimedia.org/r/76013 (owner: 10Nilesh) [15:56:50] (03CR) 10MarkAHershberger: "Saw this in 1.19 after installing the Facebook extension." [core] - 10https://gerrit.wikimedia.org/r/18073 (owner: 10IAlex) [15:56:51] (03CR) 10MarkTraceur: "(2 comments)" [core] - 10https://gerrit.wikimedia.org/r/74662 (owner: 10MarkTraceur) [15:59:41] chrismcmahon: ping? [16:08:02] hi YuviPanda [16:08:07] hey chrismcmahon [16:08:16] uploadwizard tests seem to be failing on master for me [16:08:28] YuviPanda: hrrm, ok [16:08:39] YuviPanda: looking now... [16:08:40] so is it just me? [16:08:43] or are they actually failing? [16:08:57] YuviPanda: not sure, they were OK Friday I think, let me check [16:09:03] ty [16:09:08] andre__, greg-g: Just for reference (re ops v. shell), nearly anything database-content related is shell. ops only relate to databases for swapping databases out. [16:09:23] And even database content can sometimes not require shell, if the table is replicated. [16:09:56] Elsie, thanks, that's helpful [16:11:09] No problem. [16:14:05] YuviPanda: UW test on beta commons is failing for me because login looks broken on the beta cluster. Works OK on test2wiki though. [16:14:17] chrismcmahon: mhm, so just vagrant acting up again, then [16:14:46] YuviPanda: people have been futzing with CentralAuth on the beta cluster recently, that might be the root cause [16:14:48] (03CR) 10Anomie: [C: 04-2] "Yes, Iaa5a71ec also fixes that issue, as well as other issues with these functions which your patch doesn't touch. Better to try to get th" [core] - 10https://gerrit.wikimedia.org/r/74977 (owner: 10Liangent) [16:15:22] chrismcmahon: heh, okay. But I'll now need to dig into vagrant, I thin [16:17:09] Elsie: is this re the translation thing? There was an email from niklas about how ops needed to help somewhere (I didn't look at details, though) [16:17:23] (email to me and language team) [16:17:25] greg-g: Yes. [16:17:38] Niklas likely meant "ops" as a very broad term. [16:17:49] * greg-g nods [16:17:59] possibly [16:18:01] * greg-g shrugs [16:18:11] but thanks for the heuristic [16:18:34] If it's switching a master or doing something else that's crazy (pooling new DB servers), it's ops. Otherwise, it's shell. [16:18:48] For that bug, I think it just needs someone to log in and check the tables. [16:19:01] And depending on what they find, figure out what happened. [16:19:08] (03PS1) 10Catrope: Update VisualEditor to master [core] (wmf/1.22wmf12) - 10https://gerrit.wikimedia.org/r/76523 [16:19:12] Hopefully data hasn't simply gone missing. ;-) [16:19:19] yea :/ [16:19:20] +h [16:19:45] (03CR) 10Catrope: [C: 032 V: 032] Update VisualEditor to master [core] (wmf/1.22wmf12) - 10https://gerrit.wikimedia.org/r/76523 (owner: 10Catrope) [16:20:03] (03PS1) 10Catrope: Update VisualEditor to master [core] (wmf/1.22wmf11) - 10https://gerrit.wikimedia.org/r/76524 [16:20:09] (03CR) 10Catrope: [C: 032 V: 032] Update VisualEditor to master [core] (wmf/1.22wmf11) - 10https://gerrit.wikimedia.org/r/76524 (owner: 10Catrope) [16:21:45] thanks YuviPanda. now what exactly do I import - a big dump? a bunch of articles from Special:Export? [16:21:58] yeah, aharoni [16:21:58] that's the easiest way to do it [16:22:22] kthx [16:24:01] Reedy: Care to explain yourself re https://gerrit.wikimedia.org/r/#/c/76108/ ;) [16:24:16] I just pulled that down onto fenari, it was merged but never deployed [16:24:55] aude: ^ [16:25:43] (03CR) 10Parent5446: [C: 04-1] "(12 comments)" [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [16:30:08] Reedy, aude: Given that you guys didn't answer and the WMF branches are supposed to reflect the deployed state, I'm reverting that change [16:30:31] RoanKattouw: mh [16:30:34] (03PS1) 10Catrope: Revert "Update Wikibase" [core] (wmf/1.22wmf12) - 10https://gerrit.wikimedia.org/r/76528 [16:30:43] (03CR) 10Catrope: [C: 032 V: 032] Revert "Update Wikibase" [core] (wmf/1.22wmf12) - 10https://gerrit.wikimedia.org/r/76528 (owner: 10Catrope) [16:30:46] If you give me a minute I can check that change [16:30:53] or no :P [16:31:09] hoo: I'm not comfortable deploying Wikibase changes on my own without Reedy or aude present [16:32:00] (03CR) 10Nilesh: [C: 032 V: 032] Added unit tests for AbstractEntitySuggesterServlet and WebClientRecommender [extensions/WikidataEntitySuggester] - 10https://gerrit.wikimedia.org/r/76013 (owner: 10Nilesh) [16:32:14] (03CR) 10Nilesh: [V: 032] Added unit tests for AbstractEntitySuggesterServlet and WebClientRecommender [extensions/WikidataEntitySuggester] - 10https://gerrit.wikimedia.org/r/76013 (owner: 10Nilesh) [16:32:25] RoanKattouw: Probably wasn't this important anyway :P [16:32:37] If it was, you'd think they would've actually deployed it :P [16:34:35] https://github.com/wikimedia/mediawiki-extensions-Wikibase/commit/e91cc44c71ea7704f818ee45626c1ff9525ea248 that one :P [16:34:41] https://github.com/wikimedia/mediawiki-extensions-Wikibase/commit/8f3d8c16c2af787d5fccb6dc918039ec2eb6d349 [16:34:51] that's minor stuff [16:35:18] but needs to be there when wikdata.org goes wmf12 [16:36:32] which is tonight [16:38:08] Seems like the new gerrit bot is ignoring me... [16:38:17] hoo: what happened? [16:38:25] (03PS1) 10Hoo man: Revert "Revert "Update Wikibase"" [core] (wmf/1.22wmf12) - 10https://gerrit.wikimedia.org/r/76530 [16:38:43] hoo: it's as slow as gerrit is, and no slower :) [16:38:44] YuviPanda: It took 3 minutes to report the above change... the old bot was faster... [16:38:48] hoo: true [16:38:51] Really? [16:38:56] gerrit stream events is going to be slower [16:39:02] the previous one operated on a file system level [16:39:04] so was faster [16:39:12] I see [16:39:13] it was a direct hook [16:39:17] inside the gerrit repos [16:39:21] this runs on toollabs [16:40:05] mh [16:42:06] <^demon> The major benefit of it being each gerrit upgrade won't break the bot so badly :p [16:42:08] <^demon> (And it means I don [16:42:15] <^demon> 't have to maintain this junk in puppet anymore) [16:43:58] :D [16:45:31] ^demon: Instead, there's an upcoming cron job/query to maintain. :D [16:45:52] <^demon> I'll take one cron over like 5-6 crappy python scripts :) [16:47:58] (03PS1) 10Nikerabbit: Fix test cases in SemanticMaps as well ala I95c0e0c28 [extensions/SemanticMaps] - 10https://gerrit.wikimedia.org/r/76532 [16:50:22] (03PS1) 10Demon: Add CirrusSearch to future wmf branches [tools/release] - 10https://gerrit.wikimedia.org/r/76533 [16:50:53] (03PS1) 10Umherirrender: Add expensive parser functions {{REVISION*:}} [core] - 10https://gerrit.wikimedia.org/r/76534 [16:51:54] (03PS3) 10CSteipp: Add methods to get raw request in WebRequest [core] - 10https://gerrit.wikimedia.org/r/70747 [16:52:19] (03PS4) 10CSteipp: Add methods to get raw request in WebRequest [core] - 10https://gerrit.wikimedia.org/r/70747 [16:54:50] wtf [16:54:51] string(63) "Table 'sandwiki.unittest_objectcache' doesn't exist (localhost)" [17:00:44] <^demon> Nikerabbit: Language question in MW. If I've got the 2 or 3 letter language code, how do I get the english name of that language? [17:00:57] <^demon> Don't care about non-english for this particular case (it's an internal thingie) [17:01:30] ^demon: grep language/Names.php for it. i don't think it's stored anywhere but comments in there [17:02:14] There's also {{#language:}}. [17:02:14] <^demon> Ew. Thanks :) [17:02:18] If you're doing it on-wiki. [17:02:24] I think {{#language:en}} --> English. [17:02:25] ^demon: there is Language::fetchLanguageName( $x, 'en' ); [17:02:25] Elsie: but that returns the name in the language [17:02:43] there is also {{#language:en|en}} I think [17:02:46] <^demon> Elsie: This is in PHP, not on-wiki :) [17:02:47] or perhaps it was ¤languagename [17:02:56] hashar broke my unit tests [17:03:11] The unit tests are le broken. Le sigh. [17:04:54] Unit tests, broken! say it aint so [17:04:59] (03PS2) 10Umherirrender: Add expensive parser functions {{REVISION*:}} [core] - 10https://gerrit.wikimedia.org/r/76534 [17:06:21] (03CR) 10Demon: [C: 04-1] "(2 comments)" [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/75884 (owner: 10Manybubbles) [17:10:59] (03CR) 10Umherirrender: "Patch Set 2: Removed trailing space" [core] - 10https://gerrit.wikimedia.org/r/76534 (owner: 10Umherirrender) [17:13:32] MatmaRex: are your test patches still unmerged? [17:14:43] (03PS3) 10Manybubbles: Fix suggestion queries containing a space. [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/75884 [17:15:13] (03CR) 10Manybubbles: "(1 comment)" [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/75884 (owner: 10Manybubbles) [17:15:32] (03CR) 10Nikerabbit: [C: 04-1] "(3 comments)" [core] - 10https://gerrit.wikimedia.org/r/73991 (owner: 10Cscott) [17:19:03] Nikerabbit: thanks for the review, that patch has been unloved for quite a while. [17:19:33] (03PS1) 10Cmcmahon: don't run VE test on any IE version [qa/browsertests] - 10https://gerrit.wikimedia.org/r/76536 [17:20:02] (03CR) 10Cmcmahon: [C: 032] don't run VE test on any IE version [qa/browsertests] - 10https://gerrit.wikimedia.org/r/76536 (owner: 10Cmcmahon) [17:20:14] (03CR) 10Demon: [C: 032] Fix suggestion queries containing a space. [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/75884 (owner: 10Manybubbles) [17:20:16] (03Merged) 10jenkins-bot: Fix suggestion queries containing a space. [extensions/CirrusSearch] - 10https://gerrit.wikimedia.org/r/75884 (owner: 10Manybubbles) [17:24:42] (03CR) 10MarkTraceur: "OK, I think I see where the confusion is:" [core] - 10https://gerrit.wikimedia.org/r/74662 (owner: 10MarkTraceur) [17:25:19] ori-l-away, hashar: Ugh, someone came along and wrote comments for functions that I only renamed, but otherwise left untouched [17:25:24] (03CR) 10Anomie: "(1 comment)" [core] - 10https://gerrit.wikimedia.org/r/76082 (owner: 10Hoo man) [17:25:32] (03CR) 10Anomie: [C: 04-1] Make APIEditBeforeSave give the whole revision [core] - 10https://gerrit.wikimedia.org/r/76082 (owner: 10Hoo man) [17:26:51] :) Good catch [17:27:03] (03CR) 10Alex Monk: "(5 comments)" [core] - 10https://gerrit.wikimedia.org/r/52922 (owner: 10Victor Vasiliev) [17:27:31] (03PS17) 10Rtdwivedi: Addition of code that will be used for setting up of editing system in PHP instead of JS. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/72933 [17:27:32] (03PS7) 10Rtdwivedi: Toggle header-footer visibility. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/76307 [17:27:33] (03PS15) 10Rtdwivedi: Moved the whole display of fields to PHP. [extensions/ProofreadPage] (pagePagesRefactoring) - 10https://gerrit.wikimedia.org/r/74850 [17:28:08] (03CR) 10MarkTraceur: "Yeah,