[00:13:01] kaldari, not sure why I Can't find this: http://en.m.wikipedia.beta.wmflabs.org/wiki/Special:TopicTag/American_Female_Pop_Singers [00:28:16] JonKatz: that tag hasn't been enabled. You can see the tags here: https://gerrit.wikimedia.org/r/#/c/209242/1/wmf-config/InitialiseSettings-labs.php [00:29:14] bmansurov thanks! [00:29:18] np [07:58:40] good morning [08:00:20] good morning [08:01:03] bearND: it is late on your side, but do you mind if switch the android app from maven to gradle ? [08:01:17] the job is passing just fine :d [08:01:31] hashar: I'm in Europe, so I just got started [08:01:35] HO [08:03:37] going to be bold and just do it [08:03:55] though might need a confirmation the shell commands are correct [08:04:15] ./gradlew -q clean && ./gradlew checkstyle && ./gradlew -q assembleAlphaDebug [08:04:21] which is at https://gerrit.wikimedia.org/r/#/c/210197/3/jjb/mobile.yaml,unified :D [08:04:38] hashar: You could also combine the command line [08:05:05] ./gradlew -q clean checkstyle assembleAlphaDebug [08:05:17] if you want [08:05:27] I assumed -q to be quiet and you wanted verbose for 'checkstyle' :} [08:05:55] hashar: no, quiet is fine, too [08:06:52] hashar: i would like to have the reports XML file linked in a similar fashion as was done for the maven-checkstyle task [08:08:14] yeah the artifact path does not grab the reports [08:08:22] it is build/outputs/ to get the .apk [08:09:19] morning [08:10:18] hashar: checkstyle report is in wikipedia/build/reports/checkstyle/checkstyle.xml [08:11:42] hey joakino :) [08:11:49] good morning! [08:11:51] hi jdlrobson2 [08:11:55] morning hashar :D [08:12:07] good morning party [08:12:12] jdlrobson2: glad to see you finally relocated to an Europe friendly time zone [08:12:23] hashar: haha.. only temporarily :) [08:12:44] I will steal your passport so you end up stuck in France *grin* [08:17:01] that sounds awful.. ;-) [08:19:02] jdlrobson2: i've called air maroc and everything seems to be fine [08:19:12] sweet [08:23:44] bearND: it is now capturing the files in build/reports/ . I have also tweaked the checkstyle reporter to fail the build on any lint issue (i.e. it fails if there is 1 error of either low/normal/high priority) [08:25:35] hashar: if you mean "fail on any checkstyle issue" then I'm with you. Great. Note that we could also run lint using Gradle: ./gradlew -q lintAlphaDebug (but that should be a separate task for now since we still have many failures) [08:26:15] bearND: the current build mentions -Xlint:deprecation and -Xlint:unchecked [08:26:45] hashar: yes. That's some tech debt we Android devs need to tackle [08:27:03] so gotta slightly refactor the job to be able to pass goals to it [08:28:00] ok [08:41:03] bearND: the lint goal generates a HTML report but the directories are flatten :/ [08:41:04] https://integration.wikimedia.org/ci/job/apps-android-wikipedia-gradlew-lint/1/ [08:41:16] Jenkins would not show up the html anyway [08:47:54] hashar: If I click on the html file it does open correctly, though [08:48:28] I am wondering whether there is a Jenkins plugin to parse the .xml file [08:48:36] [08:48:37] but yeah, an XML based integration would be nice to have [08:48:53] any clue what is the name of the software that generate that? [08:49:03] I guess its from android [08:49:18] probably the android-gradle plugin [08:50:06] ah if only it reported them in checkstyle or junit formats :D [08:52:43] hashar: there's a Jenkins plugin for that: https://wiki.jenkins-ci.org/display/JENKINS/Android+Lint+Plugin, but the link doesn't seem to open right now for me [08:52:59] yeah jenkins-ci is dead somehow [09:01:12] bearND: I am deploying the current state [09:04:03] hashar: sounds good [09:04:16] deployed [09:04:21] and I sent a dummy change at https://gerrit.wikimedia.org/r/210289 [09:08:49] {done} [09:10:16] hashar: great! thank you very much! [09:11:59] bearND: the task really intimidated me since I have no android/java/gradle knowledge at all :D [09:49:53] hey y'all [10:07:43] hi phuedx [11:00:14] hey joakino [11:00:15] how're you? [11:00:33] * phuedx is currently listening to a wicked death metal album [11:00:40] with a very pretty ending track [11:00:42] hey phuedx [11:00:54] i'm doing spanish hiphop :) [11:02:14] sylosis - dormant heart [11:02:16] would recommend it [11:02:23] ooh spanish hiphop? [11:04:04] spanish hip hop hey? [11:06:48] phuedx: did you mean to move https://phabricator.wikimedia.org/T98759 to code review? [11:06:58] just saw that - does that mean you have a patch? [11:07:38] jdlrobson2: gerrit-bot doesn't seem to have tagged it with the patch [11:08:15] hmmf okay, i almost started work on that. close call :) [11:08:42] i'd assigned it to myself but hadn't moved it to doing, sorry [11:08:47] yeh not sure what happened [11:08:52] i was probably looking at an old board :) [11:09:29] * jdlrobson2 needs to get an auto refresher extension [11:11:14] phuedx: Notice: Undefined index: links in /Users/jrobson/git/vagrant/mediawiki/extensions/MobileFrontend/includes/skins/SkinMinerva.php on line 43 [11:11:26] jdlrobson2: ... i know -- i'm looking [11:11:36] phuedx: apart from that looks good to go :) [11:20:11] ok [11:20:15] gonna head back home [11:20:29] and have lunch with george before he heads out for the afternoon [11:20:36] (and lisa and harry, obvs) [11:22:24] jdlrobson2: verified [11:22:30] can you give that a /thorough/ test [11:22:34] tfinc: yt? [11:22:35] BOOM [11:22:38] of course [11:22:40] if so, CONGRATULATIONS!!! [11:22:44] right [11:22:48] AWAY!!1 [11:22:51] phuedx|BIKE: of course he's there - he gets no sleep now he's a dad haha [11:23:01] it's a special state [11:23:04] relish it [12:10:50] yo [13:50:24] dbrant: thanks for the PageProperties hint, that makes a lot more sense [13:51:02] mdholloway: anytime [14:47:44] dbrant: so, correct me if i'm wrong, but since a SearchResults object is based on PageTitles, I'm going to need to have FullSearchArticlesTask request pageprops from the api and create some kind of flags for whether each result is a disambiguation or main page, and modify the PageTitle constructor to accommodate this [14:48:49] dbrant: unless there's some easier way to get from a PageTitle to PageProperties that I'm not seeing [14:50:50] mdholloway: yes, you're thinking along the right lines... [14:51:01] mdholloway: actually, got time for a quick hangout? [14:51:15] dbrant: sure [15:20:36] dbrant: hey there. i saw your comment on appcompat patch. can i bug you in the batcave sometime today? i had some questions on the menu bit [15:21:10] niedzielski: hello! i can do a hangout anytime [15:31:28] jdlrobson, joakino: do you know why the photolistapi class references an image width that doesn't exist? wgMFThumnailSizes.medium, but only tiny and small are defined [15:31:33] \o [15:31:44] it used to exist phuedx.. [15:31:52] i imagine whoever removed it missed one [15:33:16] missed a mention or mistakenly deleted it? [15:34:17] not sure [15:34:19] could be either :) [15:34:36] * jdlrobson looks for evidence [15:34:46] right now thumbnails won't work on photolists :/ [15:35:01] phuedx: i saw :/ [15:35:12] ioh [15:35:16] have you raised a bug? [15:35:20] because i know the fix! [15:35:26] phuedx: yup thought you saw it? [15:35:26] i even have it lined up locally :D [15:35:34] i probably have [15:35:35] https://phabricator.wikimedia.org/T98833 [15:35:49] oh i did see it [15:37:55] dbrant: patch updated! [15:38:09] niedzielski: thx! will look... [15:38:16] dbrant: thanks! [15:47:37] dbrant: ok, the event logging patch is about ready except for one question i had with respect to bearND's comments from yesterday [15:48:04] mdholloway: yep? [15:48:15] mdholloway: what was the question? I'm looking at the patch right now [15:48:22] dbrant: bearND: which had to do with passing the sampling rate into the class constructor -- did i understand that correctly? because that's what i have in the latest patch but it's a departure from what's happening in all of the other funnels [15:48:49] *yesterday's latest, that is, but i haven't changed it yet today either [15:49:34] jdlrobson: https://gerrit.wikimedia.org/r/#/c/193017/ [15:49:40] phuedx: on it :) [15:49:40] according to git bisect [15:49:46] dbrant: i was about to take the rate parameter back out of the constructor but i think i'm getting conflicting directives (or misinterpreting them) [15:52:41] ohh nos [15:52:52] this is why we don't panic remove things ;-) [15:52:58] but good detective work [15:53:29] ok phuedx anything you need from me before i leave straight after the meeting? [15:53:50] jdlrobson: nopers -- y'all have a good evening [15:55:06] ok cool. i went a bit crazy today sorry [15:55:16] (ie. lots of patches) [15:55:17] it's all good stuff though [15:55:24] had no one to distrat me [15:56:57] mdholloway: dbrant: how do you think about having the sample rate in the constructor (even the Funnel constructor)? [15:59:44] bearND: mdholloway: that would be somewhat cleaner... byt my concern is: do we have any funnels that exist for the lifetime of the app? We would want the funnel to pick up changes to the remote variable with a greater frequency than that, so checking it in the log() function may be beneficial... [16:01:32] bearND: dbrant: this one, i think, only exists for any given linkclick-(navigate|cancel) pair, but i'm not sure about others [16:09:12] dbrant: bearND: ConnectionIssueFunnel, for one, looks potentially longer-lived (at least for the lifetime of the fragment) [16:15:53] dbrant: bearND: It looks like SessionFunnel does in fact exist for the lifetime of the app [16:15:57] dbrant: OK, I understand your concern. mdholloway: I think we can still check the RemoteConfig as part of the log calls, and just pass in the default sample rate in the constructor. What do you think? [16:17:05] bearND: sounds fine to me. [16:18:46] bearND: dbrant: so should i make that change for each funnel that has sampling implemented, or just handle LinkPreviewFunnel for now and we'll refactor the others in a separate task? [16:21:42] mdholloway: that depends on how much change is involved. If it's a lot then I'd prefer to move it to a separate patch [16:22:06] dbrant: this is kind of bananas. if i call setHasOptionsMenu(false) in the saved pages fragment i can still get the menu from the history fragment [16:23:24] bearND: to me handling the others seems more appropriate as a separate task. [16:23:30] niedzielski: hrm... could this be because you added the calls to super.onCreateOptionsMenu? [16:23:38] mdholloway: sounds good to me. [16:23:56] mdholloway: one more minor thing for your patch: LinkPreviewVersion.LINK_PREVIEW_SAMPLE_RATE could go into the LinkPreviewFunnel. [16:24:09] bearND: cool. i'll get this merged today or die trying. [16:24:14] bearND: will do [16:24:16] it's not a concern of the Version class [16:24:37] mdholloway: thanks [16:24:48] mdholloway: and thank you for your patience [16:25:07] bearND: Yeah, I wasn't sure where to put that but envisioned the Version class as a kind of catchall for temporary sampling variables [16:25:07] dbrant: i don't think so. i just zapped all the menu methods wholesale from the saved pages fragment and still see the menu from the history fragment [16:25:38] dbrant: a small clue is that the menu fixes itself on orientation change [16:26:04] niedzielski: lol... I'm actually not seeing this behavior at all on my API10 device... [16:26:59] dbrant: nooooooo [16:27:46] dbrant: i'm on a GB galaxy note 1 [16:34:30] dbrant: i repro 100% with the following steps: 1) launch the app. 2) open the left panel. 3) tap history. 4) tap the hardware menu button. see that it says clear history even though the trash can is present in the action bar. 5) open the left panel again. 6) tap saved pages. 7) tap the hardware menu button again. see that it still says clear history. [16:37:09] niedzielski: ah yes, i see that now. [17:15:56] dbrant: hm, so i need to do more testing but i think i found a workaround. if we set the showasaction of menu_clear_all_saved_pages to never, it seems to work perfectly [17:16:56] whoever's making sandwiches at jimmy john's today was a true master of the craft. [17:19:29] niedzielski: http://stackoverflow.com/questions/30197508/onprepareoptionsmenu-in-fragment-not-being-called-as-of-appcompat-v22-api-10 [17:22:45] dbrant: new EL patch version is up and passing [17:23:33] dbrant: apps-android-wikipedia-gradlew-lint doesn't seem to like it, but i don't see any output stating specifically why [17:23:57] kaldari, phuedx|BATHTIME can you guys review https://gerrit.wikimedia.org/r/#/c/209777/ ? thanks [17:24:12] dbrant: cool, i'm going to push the workaround up so we have it but hopefully we'll hear back on the SO post [17:26:50] yuvipanda: can you help me w/ some gerrit ACL stuff? [17:27:42] bgerstle: still commuting. I saw email it is on my radar [17:28:03] ok, can you ping me when you get in? would be nice to get some stuff working w/ the CI account as well [17:33:30] kaity joakino rmoen are you guys comign to retro? [17:33:42] kristenlans: i have another meeting [17:34:56] kristenlans, oops [17:40:21] mdholloway|afk: dbrant niedzielski : I hope you've seen the announcement on mobile-l about the Android build changes. We also added a lint task. It's currently failing. It's a non-voting and new task. [17:41:23] bearND mdholloway|afk dbrant : I did! a great victory for us! i was surprised it happened so fast. seems like we've been talking about it for an age and then boom [17:41:24] bearND: ahh, i see [17:42:52] mdholloway|afk dbrant niedzielski : so, for now you can ignore the lint task failures. When we have time we can attack those and hopefully reduce to 0. [17:43:29] kaity will that be an ongoign clash? it's good to have design in retros. There are like 3 things with your name on them right now (in "worked well") :-) [17:43:43] bearND: cool [17:43:49] kristenlans: I'm done at 11 [17:44:04] kristenlans: ya I have design meeting every tuesday 10-11 PST [17:44:27] kristenlans: sorry! [17:45:49] kristenlans: i'll add to the etherpad [17:54:37] dbrant kristenlans: i see some cards in the to do called assigned to the designers. are these ready for reassignment to devs? [17:56:10] niedzielski: yes, the 5-point card is ready to be picked up. [17:57:22] mdholloway: ok, I think I understand some of the confusion about passing in the sample rate to the constructor. I was originally thinking about removing it from the log calls to the constructor of Funnel (or an overloaded version of it). [17:58:00] mdholloway: I don't think it's necessary to actually pass in the rate to the LinkPreviewFunnel constructor since the class knows about the value itself. [17:58:12] mdholloway: I'm sorry for the misunderstanding [17:58:57] bearND: oh, yes, I don't think I ever quite understood that, thanks for clarifying. let me look now, actually, while we're talking [17:58:58] yuvipanda: still commuting? [17:59:14] dbrant: okey dokey. i'll take that one. thanks! [18:01:16] niedzielski: the engineering groundwork that needs to be done is pretty clear, and then I'd suggest letting Vibha know that you've picked it up, so that she'll keep you posted on design updates. [18:03:42] dbrant: thanks! vibha, if there any design changes on Tweet A Fact (tm) please let me know! [18:06:28] bearND: ok, so i'm taking the rate parameter back out of the LinkPreviewFunnel constructor [18:06:56] bearND: so far, so good [18:10:14] bearND: as for how I have it as an optional parameter in Funnel.log() (with overloading to log all events if absent), does that look ok? [18:13:45] mdholloway: sounds good. [18:14:13] mdholloway: one more quick hangout on this subject? [18:14:24] dbrant: let's do it! [18:17:26] anyone know how to view a queue in OTRS? [18:17:42] i just want to see the mobile-ios queue and see its settings (e.g. its email address) [18:18:09] bearND: hey bernd. was they any write up produced from this T89987, "Investigate onboarding libraries"? [18:23:09] niedzielski: yes, let me find the email for that [18:23:17] bearND: thanks [18:25:52] niedzielski: forwarded you the email [18:26:10] bearND: great! thanks [18:27:05] niedzielski: i was looking for some onboarding libraries, but as you can see what I found doesn't really help with the design design wants [18:27:33] niedzielski: The concept of an OnboardingTracker class is interesting, though [18:28:18] bearND: it looks like the second library was just deprecated :| [18:31:34] bearND: hm, yeah looking at the mocks i'm not sure we'll get as much use out of these libs as hoped [18:32:56] niedzielski: aha, there's this: https://code.google.com/p/android/issues/detail?id=170630 [18:33:46] niedzielski: yes, I agree. I'm thinking just a dialog and ViewPager should do the trick. In a previous project we used ViewPagerIndicator for the three dots at the bottom but that hasn't been updated since 2012, and I'm not sure what the replacement lib would be [18:35:35] gonna take a break and get a cup of tea [18:38:55] dbrant: what a scoop! i wonder if this would fix the other issues we're seeing (item duplication between menu and bar, menu not recreated on new fragment). i don't see the fix in aosp yet [18:40:38] bearND: it sounds like the onboarding UI is fully dynamic for the given article. do we have the "grab a fact" method yet? [18:41:37] mdholloway: what's the current status of full-width page images? is it in master on android? [18:41:43] (or better yet, MFE?) [18:42:00] bgerstle: no and no [18:42:06] waiting on code and design review [18:42:30] ok. which task would be best to mark the iOS ticket as blocked by? [18:42:41] so that we can steal all your hard work once it's finished [18:42:51] 😈 [18:43:04] bgerstle: probably T94646 [18:43:11] ok, just wanted to double check [18:43:12] thanks [18:43:30] mdholloway: ping me when the final changes for https://gerrit.wikimedia.org/r/#/c/209138/28 are in [18:44:17] bearND: will do. made some changes to the dismiss() logic that dbrant made in-hangout, and testing now. [18:44:57] bmansurov: hey, one question regarding sending gerrit patches to a branch like "staging" or "dev" [18:44:58] bearND: i've used viewpagerindicator before as well and it worked fine. maybe i can find a fork of that if need be. one of my colleagues used supertooltips pretty effectively on a past project too so i have mostly good feelings about that if we end up needing it [18:45:31] bmansurov: why would it cause rebase/merge hell? When we would want to send to master, just merge "dev" into master directly and done [18:45:40] and train gets cut from master [18:47:02] yuvipanda: any idea when you'll be free to help w/ CI stuff? [18:47:28] niedzielski: The supertooltips project is also abandoned [18:52:05] time for a restart [19:06:41] joakino: sorry, lunch. but think about specific features. we can't merge to master working patches if one of the patches in dev has problem. so, cherry-pick etc. [19:07:13] joakino: if we con't merge dev into master because of one patch, then other tasks won't get shipped either [19:07:31] joakino: if we delay merging that patch only, then rebase etc. [19:08:03] bmansurov: so there's more coordination involved, yes, but that's not something we can't figure out [19:08:03] bearND: API/services meeting? [19:08:47] is the coordination cost higher than the alternative? [19:20:39] phuedx|MEETING: not sure, but creating a new branch doesn't solve the problem directly. [19:44:21] dbrant: hmm, now my events aren't showing up on the EL server at all, and I don't think I could have regressed that badly just making some changes to the logCancel logic [19:44:50] dbrant: and i don't have permission to view the processing log files described here https://wikitech.wikimedia.org/wiki/EventLogging/Testing/BetaLabs#Raw_stream_of_events_.28including_unvalidated_events.29, it appears (do i ask yuvipanda about that?) [19:45:42] mdholloway: did you modify the code to be the chosen one (re: sampling)? [19:45:53] bearND: yep [19:46:28] i can verify through client-side logging that it's getting to the point of sending data over, and the data look in order [19:49:13] mdholloway: hmm, I'm in a meeting right now. But I can try later as well. [19:49:23] bearND: cool, thanks. [19:58:06] kaldari, so my work laptop is down and I have migrated my workstation... I can't remember my passphrase for my deployment key. Any chance you could stage the deploy today? [19:59:13] kaldari, I'm going to keep trying passphrases but in the event I cannot get in would be nice to have a backup [19:59:44] rmoen: sure [20:00:06] kaldari, awesome! so far I've tried about 20 passwords ;/ [20:00:13] with no luck [20:01:31] bgerstle: re etags / caching, RB currently implements https://phabricator.wikimedia.org/T97393 [20:01:39] re segmenting: https://github.com/wikimedia/mediawiki-services-cxserver/blob/80dfe4d426396704bcf1a4b6fb373878b5b7e33b/segmentation/CXSegmenter.js [20:03:16] yuvipanda: ping [20:03:29] gwicke: what was the service you mentioned just now? [20:03:35] oh ta [20:03:36] already done [20:03:52] no way [20:03:59] it does multi-language sbd? [20:04:10] gwicke: for the etag, does RB also implement If-Modified-Since: ? or whatever you use w/ etags? [20:04:38] phuedx: not sure how good they are, but they have been giving that some thought [20:05:08] bgerstle: with etags you can send if-match [20:05:13] browsers do that by default [20:05:14] ah, right [20:05:27] gwicke: if they had a couple more clients then they may have impetus! [20:05:55] bgerstle: I don't think we set the last-modified date consistently yet, but could easily add that [20:06:17] the etag already contains the modification time in the uuid portion [20:06:45] but generally, if you can support the etag then that'll be more precise [20:07:07] etag is fine [20:07:11] probably preferred [20:07:36] gwicke: so... how can we get this functionality in apps? [20:08:02] phuedx: currently they only use that internally, but they'd probably be happy to find collaborators for segmentation [20:08:27] gwicke: cool, thanks for the link [20:08:28] :) [20:08:38] ok -- gonna head off for the night [20:08:47] night phuedx [20:08:49] bgerstle: will respond laterz [20:08:55] (to your textextracts email) [20:09:01] bgerstle: for etags, we might be able to take care of it in RB [20:09:47] gwicke: so it's not possible for the mobileview action? [20:10:16] bgerstle: for mobileview you'd need to add the header, which shouldn't be too hard either [20:10:41] but I *think* it might already be setting Last-Modified [20:11:23] rmoen: any idea if there are any MF dependencies? [20:12:51] gwicke: i'll take a second look [20:13:54] gwicke: doesn't look like it: https://www.irccloud.com/pastebin/uNWHvwkg [20:14:08] is it the X-WMF-UUID? [20:14:26] kaldari, i dont think so but i will test [20:14:52] rmoen: making the submodule update now [20:15:02] bgerstle: hey! sorry, had to run back home because I locked my laptop in my drawer and left the key at home >_> [20:15:06] * yuvipanda reads backscroll [20:15:42] d'oh [20:15:54] yuvipanda: whenever you have a few minutes, would love some help w/ gerrit [20:15:59] bgerstle: so I have no idea about CI - I see legoktm has already responded on that [20:16:04] bgerstle: let me give you guys gerrit rights [20:16:14] yuvipanda: mind hopping on a hangout? [20:16:21] bgerstle: that uuid is new to me; odd that it's not used as an etag if it's actually changing with the content [20:16:33] gwicke: i have no idea [20:16:47] hi [20:17:32] there's no hit for X-WMF-UUID in core [20:17:39] must be the extension, or Varnish [20:17:41] I hope not :P [20:18:21] rmoen: submodule update is built: https://gerrit.wikimedia.org/r/#/c/210467/ [20:18:24] yuvipanda: TL;DR; the iOS people should have admin rights over our repos, and we need to setup privileges for the ci-apps@wikimedia.org gerrit account we created [20:19:14] bgerstle: 1: I just added all of you guys, and 2: I’ve no idea what that is, I think filing a bug / asking the releng team is going to help with that than me [20:19:30] note that I’m actually part of the labs team - I’m only moonlighting helping out with access requests and what not. usually these go through phabricator :) [20:20:10] yuvipanda: yeah, i opened some stuff oh phabricator. i was able to create the gerrit account myself , and figured i could add it just as i would another teammember [20:20:15] lemme take another look at gerrit [20:20:55] yuvipanda: you added us to what, commons? [20:21:32] bgerstle: as owners, yeah. https://www.mediawiki.org/wiki/Gerrit/Project_ownership has more details [20:21:34] note that this doesn’t give you the right to actually add other people, afaik [20:21:44] ok [20:23:53] yuvipanda: alright, i'm giving up for now. will follow up w/ RelEng later. thanks for your help! [20:24:02] yw [20:24:09] bgerstle: you should have +2 rights on that repo now tho [20:25:17] yuvipanda: i had them before [20:25:29] oh? that’s not what that email sounded like [20:25:32] * yuvipanda goes to read it again [20:26:29] > Something came up today w/ commons (ori updated EL endpoint) and I couldn't merge (but he could..?) [20:26:32] anyway, good if it all works [20:26:59] kaldari, submodule update looks good. :) I haven't tested on wmf5 because my machine is still not entirely setup for that yet. Happy to report I'm having great success / speed with vagrant on a machine with 8 cores and 16gb of ram :) [20:27:22] nice [20:29:46] mhurd: https://lists.wikimedia.org/pipermail/wikitech-l/2013-March/067665.html is the thread + scripts about moving github pull requests to gerrit manually [20:30:21] mdholloway: no idea why EL doesn't work. I've also tried the new test URL I got from yuvipanda's patch (http://deployment.wikimedia.beta.wmflabs.org/beacon/event), but then I still don't see any events appearing [20:31:08] bearND: ah, ok. so it's not just me, anyway. [20:31:27] yuvipanda: should the new EL URL already work? [20:31:43] dbrant: I saw you went over some bugs i opened, any feedback ? [20:31:44] it does on prod - all of desktop / mobile web has been hitting new URL for a week [20:31:52] no idea about beta, ask releng team / analytics :) [20:32:23] yuvipanda: bearND: it was working earlier today, only been acting up past hour or two [20:32:27] oh sorry, you mean commons [20:32:33] matanya: anything specifically? [20:32:56] dbrant: just the whole otrs ==> phab thing [20:39:30] dbrant: if i record a task in phabricator that i feel has low priority, should i mark it as such or leave that to you? [20:40:53] niedzielski: you can assign a priority to it. But keep it in the "needs triage" column. (should be default) [20:41:14] matanya: sorry, not sure what you mean [20:41:22] dbrant: phabulous, will do [20:42:08] dbrant: things like: https://phabricator.wikimedia.org/T98139 and https://phabricator.wikimedia.org/T98138 or https://phabricator.wikimedia.org/T98137 [20:44:11] matanya: ah, those are all on our radar in the Andoid-App workboard. I'll triage them as soon as I can. (and thanks for getting those into phab) [20:44:32] np [21:16:58] mdholloway: fyi, on the analytics channel milimetric told me that /var on the EL server was full. That's why the events did not make it through [21:17:28] bearND: aha, thanks. [21:17:47] kaldari: any idea why jenkins is failing here? https://gerrit.wikimedia.org/r/#/c/210200/ [21:18:00] bearND: i just put that down a little while ago and turned to T94646 and updated the commit message to address your comment [21:18:08] bearND: thanks for checking up on the EL server [21:18:20] bmansurov: one sec, help to finish up deployment with rob [21:18:34] sure [21:20:52] mdholloway: good. You should wrap the lines in the commit message when they get too long, see https://www.mediawiki.org/wiki/Gerrit/Commit_message_guidelines [21:20:59] dbrant: i wanted to follow up on our hangout earlier, too. it seemed like there were two pieces: that there are certain dismissals we want to log and others we don't; and that i was going about logging the dismissals incorrectly. which dismissals don't we want to log? my original assumption was that any navigation away from the dialog without following the link counted as a cancel. [21:21:37] mdholloway: there's nothing we don't want to log. [21:22:07] dbrant: ok. [21:22:57] mdholloway: the OnDismissListener will log a "cancel" event in all cases, except when a "navigate" event occurred. [21:33:16] the problem is that it seems never to be reaching onDismiss. [21:42:27] dbrant|brb: however, if i have the LinkPreviewDialog class implement DialogInterface.OnDismissListener, then the approach you suggest appears to work [21:42:58] dbrant|brb: (except that it then eats the "linkclick" events for reasons I'm about to investigate) [21:43:52] bearND: innodb files are killing /var on beta EL instance. I have to drop all databases and do crazy stuff, I'll let you know if/when I get it working again [21:44:58] milimetric: ok, it's getting late for me here, as I'm in Europe already. But please let mdholloway in mobile channel know when it's fixed [21:45:07] will do [21:45:15] thanks [21:45:23] milimetric: bearND: thanks! [21:47:37] mdholloway: :( sorry, I don't have permission to be root on that sql server [21:47:55] I'd hack it since I have sudo but I don't want to mess something else up. I'll ask in labs [21:49:00] milimetric: cool, thanks again for your help. [21:53:32] bmansurov: Looks like Jenkins is failing because it thinks @iconGutterWidth is undefined in https://gerrit.wikimedia.org/r/#/c/210200/3/less/tablet/common.less [21:53:49] kaldari: I see, thanks! [21:54:14] kaldari: is there a way I can make the core dependency visible to jenkins? [21:56:01] bmansurov: not that I know of :( [21:56:40] kaldari: it'd be great if you could merge the core patch ;)) [22:04:15] bmansurov: I have no idea why, but for some reason that core change gives me a fatal error: Fatal error: Class 'Wikimedia\Assert\Assert' not found in /Users/kaldari/Sites/core/includes/libs/ProcessCacheLRU.php on line 132 [22:16:32] kaldari: does restarting the web server help? [22:23:06] mdholloway: the problem got way nastier, basically there's nothing I can really do and I'm waiting for Andrew to come back from vacation Friday or someone else in ops to help [22:24:06] ack, thanks. [22:24:14] milimetric: fyi bearND ^ [22:25:05] milimetric: thanks for letting us know [22:25:07] bmansurov: doesn't seem to. also tried setting my cache to CACHE_NONE. not sure why it's acting so wonky. [22:25:38] milimetric: i started reading a bit about it on releng. [22:26:14] yeah, those guys are trying to help. It may get resolved, it's just over my poor front-end dev head :) [22:26:19] brb [22:26:28] kaldari: ok thanks, hopefully phuedx|zzZ can take a look tomorrow [22:27:53] mdholloway: dbrant|brb: well, since that will take some time I'd be ok merging the EL changes. We know the event went through, and since then that part has not changed materially, right? [22:30:04] bearND: dbrant|brb: right, my most recent changes shouldn't have affected that section (which is to say that if they're being logged on the client side in Funnel.log (as they are) it's probably safe to assume they're going through [22:30:11] ) [22:30:47] mdholloway: in the meantime, you can probably update https://gerrit.wikimedia.org/r/#/c/205649/15//COMMIT_MSG [22:36:40] bearND: done [22:54:41] mdholloway: ok, I freed up a few hundred MB with a DB cleanup [22:54:47] but I expect it'll fill up again soon [22:55:06] for the real fix we need puppet to run and for that we need puppet experts to tell us why it's broken on that instance [23:07:27] yuvipanda: is https://gerrit.wikimedia.org/r/#/c/208315/6 ready to be merged? [23:07:38]