[00:00:17] New patchset: Jdlrobson; "separate jQuery shim tests from application tests" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37158 [00:00:17] Alrighty Brion 2.2, 4.0.4, 2.3.6 and 2.3.4 android [00:00:19] all passed [00:00:28] I say we go for it [00:00:39] awesome [00:00:50] tfinc: i'm pulling the trigger on v1.3.3 in the google play store [00:01:02] but first lemme update https://www.mediawiki.org/wiki/Mobile/Release_history [00:01:16] brion: mail mobile tech to let us know what's coming with it [00:01:21] feel free to ref the wiki page [00:01:39] excellent job brion updating the wiki page. you make me happy when you do that [00:02:38] :D [00:03:46] brion, MaxSem, in SkinMobile.php, why do we have both SkinMobile::prepareTemplate() and SkinMobileTemplate:prepareData()? how are their responsibilities different? [00:04:31] ehm, I invented that distinction (as well as wrote these classes), but I don't remember why [00:05:19] please wait, I'm in a middle of some calibrations [00:06:17] okay, we're live on enwiki: https://en.wikipedia.org/w/api.php?action=query&prop=coordinates&format=json&colimit=10&titles=User%3AMaxSem [00:07:05] I'd rather not populate this data so late today, will babysit through it tomorrow [00:10:40] whee [00:12:31] MaxSem, ok, let me know when you have a moment [00:13:02] jgonera, as the matter of time this moment has come [00:13:12] ok ;) [00:14:13] I see that prepareData() uses the data set by the skin to initialize some implementation-specific stuff [00:15:00] the difference might be superficial, but I wanted to shorten SkinMobile::prepareTemplate() [00:16:16] I totally understand that, but maybe we should just split prepareTemplate() into several methods in SkinMobile? and prepareTemplate would only invoke them [00:16:32] as it is now it is really hard to read it and understand what's where [00:17:44] hehe, have you seen Vector? [00:18:22] nope, I'm only looking at things that I have to ;) [00:19:20] tfinc, [04:06:19] okay, we're live on enwiki: https://en.wikipedia.org/w/api.php?action=query&prop=coordinates&format=json&colimit=10&titles=User%3AMaxSem [00:19:20] [04:07:06] I'd rather not populate this data so late today, will babysit through it tomorrow [00:19:24] so, as the author of those classes, would you mind if I tried to refactor this a bit? :) [00:20:00] hey, I don't own this code!;) [00:20:12] WP:BOLD [00:20:33] MaxSem: :) [00:20:35] but first make sure that you know the code base thoroughly [00:20:57] I just wanted to know if you agree that it got a bit out of control and is difficult to modify as is [00:21:08] sure, I won't touch anything I'm not sure about [00:22:12] it's template initialization. it can't be under control, it's always boring:) [00:22:39] MaxSem: how long will it take to populate 80% of the data ? [00:23:10] depends on what's going on with job queue [00:23:28] but I bet several days [00:24:13] brion: tweet about the Android update on https://twitter.com/WikimediaMobile [00:24:37] ah yes :D [00:25:11] https://play.google.com/store/apps/details?id=org.wikipedia still shows 1.3.2, not sure i want to a nnounce until it's dl'able [00:26:22] MaxSem, one more thing: why is there SkinMobileBase if SkinMobile is the only class that inherits from it? [00:26:43] you forgot about SkinMobileWML [00:26:45] no, I'm wrong I take it back [00:26:47] yeah [00:28:34] brion: it can take upward of an hour to update [00:28:54] better an hour than a week :D [00:30:00] we've really fallen behind in our tweets [00:30:13] Maryana: can you make sure to use https://twitter.com/WikimediaMobile more actively when we push new features ? [00:30:33] yeah, i totally forgot we had a twitter! [00:30:44] heh - thanks for reminding me [00:31:30] i've got twitter open, will push the button in a bit once i see the update live :) [00:41:41] New patchset: awjrichards; "Preliminary work for mingle card #146, allowing for account creation" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37163 [00:42:45] New patchset: awjrichards; "Preliminary work for mingle card #140, allowing for account creation" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37163 [00:43:42] New patchset: awjrichards; "Preliminary work for mingle card #140, allowing for account creation" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37163 [00:47:14] New patchset: awjrichards; "Removes no longer needed doSpecialCases() code" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37166 [00:59:02] Maryana: i have most of the backend bits in place for account creation and am not blocked on design. any chance it'll be ready tomorrow? otherwise monday is the next time i'll be ready to work on it [01:02:06] awjr - confused, if you have the backend and design, then it's already ready to go, right? [01:02:16] or did you mean you *are* blocked on design? [01:02:16] derp [01:02:19] s/not/now [01:02:25] oh, haha [01:02:27] :p [01:03:06] it should be relatively straightforward for munaf to mobilify his desktop account creation work [01:03:27] i'll get in touch with him. any particular cutoff you're working with? tomorrow morning/early afternoon? [01:03:50] tomorrow early afternoon would be great, otherwise monday [01:04:00] gotcha. will poke him [01:04:24] thanks :) [01:04:40] anytime! [01:04:55] im out - good night everyone [01:08:18] do all phpunit tests pass for you on master? [01:33:37] New patchset: Jdlrobson; "remove info box from top of login [story 150]" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37175 [01:33:38] New patchset: Jdlrobson; "add placeholder to login elements [story 150]" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37176 [01:33:38] New patchset: Jdlrobson; "style login screen [story 150]" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37177 [01:33:38] New patchset: Jdlrobson; "style error message [story 150]" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37178 [01:57:45] tfinc: just realized we don't have the mobile edit tag enabled on commons.. [01:58:00] hopefully Maxsem can knock that up? [01:58:13] jdlrobson: maryana said she was tracking photo uploads on stat1 with something else [02:00:23] k [02:00:31] I like spying on it though :) [02:00:35] and i'm a mere mortal [02:04:36] New review: Jdlrobson; "Code looks fine but I'm not sure how to test - not seeing any difference before and after!" [mediawiki/extensions/MobileFrontend] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/36809 [14:28:30] what's michelle's irc nick? [14:33:29] jeremyb: jcmish [14:34:01] aha [14:34:02] danke [14:34:28] yuvipanda: who knows how spelling suggestions work in android? [14:34:43] jeremyb: for wiki? [14:34:50] (from the API not from the keyboard) [14:34:51] jeremyb: let me dig up the api call [14:35:03] i mean within the app [14:35:24] who understands how the response is handled. what happens when you click the link and how the link is made in the UI to begin with [14:35:36] action=query, list=search, srsearch: 'term', srinfo: 'suggestion' [14:35:38] in the app? [14:35:44] well, awjr originally built it at a hackathon [14:35:51] hrmmm [14:35:54] but i can mostly tell you stuff [14:35:59] i've worked on it more (the app) [14:36:00] so, STR: [14:36:00] why? [14:36:40] type a page name that exists but where the all lowercase version does not with 1 wrong letter and all in lowercase [14:36:56] it gets suggested that you want the right spelling but still all lower case [14:37:11] click it and you get a page with all lowercase name and blank body [14:37:18] v1.3.2 from market [14:37:44] ah [14:37:45] if you misspell with the correct capitalization then the link for the suggestion takes you to the right place [14:38:59] also, in case it matters, JB 4.1.2 on nexus s [14:39:15] shoudn't [14:39:20] jeremyb: i'll poke at it! [14:39:28] danke [18:16:05] jdlrobson: do you have anything else that you want to add to subject: "Rethinking the mobile web beta" ? . if not i'm going to move forward with #1 [18:16:16] i'm just replying now tfinc [18:16:19] jdlrobson: k [18:16:49] #1 sounds like a plan to me [18:16:59] MaxSem: i'm not sure what updated invites you got but i'm ready to have our 1:1 at 10:30 [18:17:21] tfinc, me too [18:29:04] tfinc, I see you:) [18:29:45] MaxSem if you come to the prioritization meeting today, you should bring up https://bugzilla.wikimedia.org/show_bug.cgi?id=42786 - if you're not planning on being there, i'm happy to bring it up for you [18:32:35] awjr return of sticky cookies! [18:32:42] NOOOOOOO [18:32:45] hard refresh took care of it again [18:32:51] jdlrobson: can you describe exactly what you did:? [18:32:59] or were doing, when you saw the problem? [18:33:04] I was doing the usual thing - switching to desktop then back to movile [18:33:06] i did this twice [18:33:33] it switched from en.wikipedia.org to en.m.wikipedia.org [18:34:00] out of interest.. why do we need these now? Surely the redirector/use of en.m.wikipedia.org takes care of all this? [18:34:18] is the use case people that want to land on the mobile site from their google search? [18:34:50] awjr: just had a thought [18:35:01] i clicked desktop view - > edited the page -> clicked mobile view and it didn't unstick [18:35:05] so maybe editing is related? [18:35:09] jdlrobson: so did you see the desktop version of the site on en.m or was that after switching back to mobile and then going to en.wikipedia.org you were not redirected to en.m? [18:35:21] desktop version on en.m yes [18:35:30] jdlrobson: that is not a problem with sticky cookies [18:35:42] interesting.. [18:36:01] you know sticky cookies do not work if after switching to the mobile view, then on subsequent requests for en.wikipedia.org (from a mobile device) you are not redirected to en.m [18:36:16] jdlrobson: to me that sounds like the cache was polluted, for some reason [18:36:31] which is not good [18:36:58] :( [18:37:04] got to head up to 6 brb [18:37:25] jdlrobson: ok - next time you see it, can you jot down the article that is showing desktop version on en.m? [18:37:53] in fact, it would be good to have the whole URLs of what you were doing [18:38:39] jdlrobson: i have a hunch that this is somehow related to editing but it needs more srs investigation. we should probably open a bug and prioritize it fairly high [18:38:49] it sometimes takes me a while to realize what happened [18:45:07] MaxSem: http://www.geonames.org/img/charts/ws-creditsld8163-NhBKEPcu6P86DGxC7SYivA.png [18:45:39] MaxSem: http://www.geonames.org/img/charts/ws-credits8163-NhBKEPcu6P86DGxC7SYivA.png [18:47:18] MaxSem: http://stats.wikimedia.org/wikimedia/squids/SquidReportUserAgents.htm [18:58:26] awjr, that feature is not that high priority because while we already have an API to get page coordinates live, it currently gives access to 10% of enwiki's coordinates and none from other wikis [19:02:50] MaxSem: credits wise we've gone through 32,732,385 / 60,000,000 [19:03:05] MaxSem: i can just give you access to to the account if you want access to this [19:03:33] tfinc, I don't thik I need more stats from geonames [19:03:36] k [19:06:25] tfinc, so mobile browsers generate ~7.5 times more page views. if we interpolate this, we'll get 500k geo searches/day = 6 requests/second [19:07:06] this should be handlable by yttrium alone (I hope), but peak loads are much more [19:07:26] lets be able to handle at least 2x that then to give us breathing room [19:07:40] i expect it to rise [19:08:09] so I'd say that 3 servers we'll initially have would be enough, but it will not be very future-proof [19:08:20] ok [19:09:57] grr wikivoyage inline styles! [19:10:13] (the wikivoyage homepage) [19:11:23] for GeoData's particular needs, the only things really important for performance are number and frequency of CPUs - since we have no full-text index, the volume of data is pretty small so I/O performance and RAM size are almost irrelevant [19:12:32] Maryana: http://en.wikipedia.org/w/index.php?title=List_of_syphilis_cases&curid=32124842&diff=526589507&oldid=525631656 < :D [19:12:57] hahah, nice. now revert yourself and do in front of a live audience! [19:13:20] i'm a bit confused - i think Erik said he'd drive since he was familiar with his laptop [19:13:38] hmm.. then you might have to whisper over his shoulder to tell him what to do [19:14:18] it's like being back in India :D [19:14:21] (middle men) [19:17:34] i'm just excited to see leaping jon on the big screen [19:20:33] MaxSem: can we get mobile edit tag working on commons? [19:20:50] http://commons.wikimedia.org/w/index.php?namespace=&tagfilter=mobile+edit&title=Special%3ARecentChanges < sad face :( [19:21:31] jdlrobson, it requires an AbuseFilter rule [19:21:46] MaxSem are they easy to add? [19:21:49] we need a more reliable solution [19:21:54] sure [19:22:02] did we get a story written out for that? I forget.. [19:22:05] should raise a bug at least [19:23:34] love the term "dirty diff" [19:23:48] almost as good as "sticky cookies" [19:24:25] brion: can I get away with http://pastebin.mozilla.org/1979938 [19:25:32] jdlrobson: can you chime in about http://pastebin.mozilla.org/1979938 [19:25:49] preilly: _might_ work, but depending on how the template does its output i'm suspicious [19:26:02] may be better to wrap the rest of the stuff in an if [19:28:45] brion: yeah good point or just close out the html near the exit [19:35:05] awjr: were you watching the metrics presentation ? [19:37:43] jdlrobson: did the image finally make it to your user page ? [19:37:52] no tfinc something odd with Erik's machine.. [19:38:05] most strange [19:38:46] preilly: not sure about that paste bin code... [19:38:56] you are trying to just show the banner and notice? [19:39:03] jdlrobson: yes [19:39:16] jdlrobson: they don't want any additional output past that [19:39:17] also seems wrong to put zero code in MobileFrontend.. shouldn't this be done via a hoolk? [19:39:27] you could imagine overriding the render function with a hook [19:39:30] jdlrobson: yes it should be done that way [19:39:43] jdlrobson: yeah that would be the best way for sure [19:39:53] let's do that then :-) [19:40:51] brion: I just forwarded you the TODO list for Zero [19:40:58] tx [19:41:16] brion: preilly have you guy sync'd up about zero arch ? [19:41:28] tfinc: on the basics yeah [19:41:28] tfinc: we talked this morning [19:41:34] great [19:42:00] brion: are you at the point where you can help run any zero tests outside of the thur window? and more importantly can you teach dan how to do it ? [19:42:19] tfinc: not quite yet but we'll try that soon :) [19:43:33] tfinc: The tests involve operations to turn on and off the varnish mobile cache changes [19:44:14] tfinc: Dan could probably make the changes in gerrit but operations would need to merge them and push them to sock puppet and run puppet on the boxes [19:44:19] preilly: yup, i've seen the ACL changes. their really simple. i want Dan to be able to do them with brion as escalation. [19:45:09] tfinc: puppet/templates/varnish/mobile-frontend.inc.vcl.erb [19:47:41] https://gerrit.wikimedia.org/r/gitweb?p=operations/puppet.git;a=blob;f=templates/varnish/mobile-frontend.inc.vcl.erb;h=38a060762f9f9b37f5c5bbae8b69676744f4e4bb;hb=HEAD [19:53:28] [WikipediaMobile] brion pushed 1 new commit to master: http://git.io/hTkCHg [19:53:28] WikipediaMobile/master 415f08e Brion Vibber: Bump version number for WebWorks/PlayBook [19:53:57] Project WikipediaMobile - Nightly builds build #468: SUCCESS in 14 sec: https://integration.mediawiki.org/ci/job/WikipediaMobile%20-%20Nightly%20builds/468/ [19:53:57] Brion VIBBER: Bump version number for WebWorks/PlayBook [19:54:05] Project WikipediaMobile - Nightly builds build #469: SUCCESS in 7.8 sec: https://integration.mediawiki.org/ci/job/WikipediaMobile%20-%20Nightly%20builds/469/ [20:01:08] * MaxSem notes that RoanKattow_, is seriously mutated [20:02:33] tfinc yep i've been watching the metrics meeting - great job jdlrobson and Maryana :) [20:02:47] awjr: then you saw us revert back to the desktop post edit [20:03:10] erik then tried to switch to .m [20:03:18] which still didn't show the mobile interface [20:03:25] yeah - not very well htough (watching from my phone) [20:28:33] New patchset: Jdlrobson; "add an experimental mode to MobileFrontend" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37288 [20:28:36] ^ awjr [20:29:21] jdlrobson: will take a closer look after i finish watching metrics [20:29:28] sure thing [20:29:39] actually, more realistically after the prioritization meeting [20:29:52] :) [20:31:06] also awjr we should rather urgently get https://gerrit.wikimedia.org/r/#/c/37139/ deployed [20:31:09] it is breaking in production [20:31:14] people are tweeting about it [20:31:29] a hell of a lot of phones cannot read sections [20:31:36] no bueno [20:31:51] I need to talk to Michelle about a better way to test this [20:31:59] it keeps coming back [20:32:28] continuously regressing [20:32:43] the test would be to test the site with a certain useragent [20:32:44] jdlrobson: after this i'll see if we can fit in a deployment today otherwise it will need to wait until next week [20:33:31] also awjr tfinc: https://twitter.com/bryanconnor/status/276550328822931456 [20:33:33] another win [20:33:43] booyakasha [20:34:55] jdlrobson: i need to work that into a mission statement for us :D [20:36:14] jdlrobson: i booked us a deployment slot at 3pm pacific (right after the prioritization mtg). i have to eat something and prepare for the prioritization mtg - can you find someone else to review/test the rev? [20:37:44] awjr: thanks dude [20:37:54] np :) [20:59:16] awjr, where's prioritization taking place? [20:59:38] ack no hangout link [20:59:42] MaxSem: one sec [21:00:36] hangout link for prioritization mtg: https://plus.google.com/hangouts/_/066b0d282a2a40436d9f3d429c3f8d12d8ad5a87?pqs=1&authuser=0&hl=en [21:00:45] MaxSem, brion, jcmish ^ [21:01:32] is it meeting timeeeeeee [21:01:39] brion aye [21:01:53] someones in our mtg room [21:01:58] hm, there appears to be an accounting meeting in our meeting room [21:02:09] brion, want to chase them out? :) [21:02:15] be menacing [21:02:27] show them your mean face [21:02:30] we got moved to chambers [21:02:38] aww [21:03:03] * awjr shakes fist [21:08:40] Hi, is there a page explaining what do you get to test when you join the mobile beta? [21:10:18] all I could find was http://blog.wikimedia.org/2011/10/27/wikimedia-mobile-opt-in-beta/ [21:10:35] but that is frozen in time [21:10:44] e.g. no mention to the new image upload [21:19:04] awjr: im confused [21:19:15] my calendar tells me there is a prioritization meeting but there is no one in the room [21:19:24] (R31) [21:19:35] MaxSem: brion ^ ? [21:19:46] jdlrobson: we moved to chambers r37 [21:19:50] ahh [21:19:57] somebody else had stolen our room for a bit [21:20:08] brion: so kick them out [21:21:44] let me know if I should be there too, if not I'm going to code a bit [21:25:56] jgonera: it's optional for engineers [21:26:08] ok [21:26:57] brion: can you send me the link [21:27:00] i'm lost in mingle again [21:27:07] https://mingle.corp.wikimedia.org/projects/mobile/cards?favorite_id=607&view=Iteration+Planning [21:27:09] thx [21:28:28] jcmish: https://bugzilla.wikimedia.org/show_bug.cgi?id=42749 [21:28:36] not sure if that is on your radar [21:28:58] Nope [21:29:00] want me to add it [21:29:06] there's a fix done [21:29:11] hopefully it will be deployed today [21:30:16] gotcha [21:36:02] brion: joining us in r32 ? [21:36:14] we're still in the meeting in r37 [21:38:29] wrapping that up [21:44:13] jgonera: http://www.mediawiki.org/wiki/Mobile/Release_history [21:44:26] qgil_, ^ [21:46:38] jdlrobson, jgonera ok, there is data up in there. But this is no page we could send Wikipedia beta eader to know what are they are actually testing [21:46:46] readers [21:48:41] https://mingle.corp.wikimedia.org/projects/mobile/cards/283 < Maryana [21:49:12] qgil_: what is the context? Is the suggestion to add a link on the settings page to a page which details current beta featureS? [21:50:19] jdlrobson, yes, the idea came when I landed at http://en.m.wikipedia.org/wiki/Special:MobileOptions/BetaOptIn . A link there would be nice so people know what are they testing and therefore can help with more proactive and focused testing [21:50:21] (such page currently does not exist) [21:50:25] woo, that looks somewhat tough, but cool, jdlrobson! i wish we could also put something like that on desktop [21:50:35] MaxSem: should we kill http://en.m.wikipedia.org/wiki/Special:MobileOptions/BetaOptIn in favor of beta? [21:50:39] (via settings) [21:50:50] i had forgotten about that page qgil_ as we don't surface it [21:50:59] hence why it is styled badly :) [21:51:36] jdlrobson, well, whatever page are you seeing whan you click Settings ---> beta On from a mobile device [21:52:05] jdlrobson, it's there for b/c with links lying around in different places such as blogs and twitter [21:52:05] qgil_ http://en.m.wikipedia.org/wiki/Special:MobileOptions [21:52:18] MaxSem: we can redirect to settings now though? [21:52:28] we should ask tfinc if it's safe to remove them [21:52:35] so bit.ly/wmoptin -> http://en.m.wikipedia.org/wiki/Special:MobileOptions/BetaOptIn -> http://en.m.wikipedia.org/wiki/Special:MobileOptions [21:52:35] hmm ? [21:52:56] jdlrobson, exactly. Imagine link there to a page explaining the options currently in beta [21:53:02] MaxSem: what are we looking to remove ? [21:53:04] qgil_: that sounds like a good idea [21:53:16] qgil_: just need time to sit down and do said page and i'd worry about keeping it up to date [21:53:18] 1 title 1 sentence for each with optional link to the project / task would be enough [21:53:19] tfinc, legacy links and UI like http://en.m.wikipedia.org/wiki/Special:MobileOptions/BetaOptIn [21:53:28] MaxSem: kill it [21:53:31] ( qgil_ although hopefully the community can help with that :)) [21:53:32] whee [21:53:47] MaxSem: you are a code assassin [21:54:05] jdlrobson, t's not that you are releasing new features to beta every hour? :) [21:54:07] * MaxSem wipes his dagger [21:54:08] I could help [21:54:18] qgil_: pretty much ;-) [21:54:28] jdlrobson, et's talk [21:54:29] we are creating an alpha mode now... [21:54:30] :P [21:54:40] tfinc, jdlrobson - removing [21:54:41] are you in the office qgil_? [21:54:44] qgil_: http://www.mediawiki.org/wiki/Mobile/Release_history :) [21:54:45] let's talk, I'm interested. sure [21:55:00] tfinc: shown that - we're talking about a more user friendly version [21:55:02] qgil_: we use that page to track whats in the beta [21:55:09] i.e. this is the feature, this is how you get it, this is what it does [21:55:14] watching now [21:55:16] jdlrobson: lets keep is low cost [21:55:17] here's where you raise bugs etc [21:55:44] It would be nice to give better visibility to end users about what they're getting though [21:55:49] maybe we should we blogging about the features more.. [21:55:50] i'd hate to have features slow down because we don't have enough infrastructure to support a bleeding edge feature [21:56:03] jdlrobson: sure [21:56:07] i'm eager to see what you come up with [21:56:20] jdlrobson: alpha features like this sound more like tweets to me [21:56:38] but we still need http://en.m.wikipedia.org/wiki/Special:MobileOptions/Language right? [21:56:40] very little overhead to tweet about them [21:56:49] tfinc: well how about a link saying follow WikimediaMobile to keep track of beta features? [21:56:57] yes [21:57:02] MaxSem: yes [21:57:09] MaxSem: linked from the languages section [21:57:23] could be moved though.. [21:57:37] could become http://en.m.wikipedia.org/wiki/Special:SiteLanguages [21:57:39] or similar [21:59:10] brion: http://www.mediawiki.org/wiki/Mobile/Release_history needs an update for your beta features [22:06:40] Change abandoned: Jdlrobson; "abandoning in favour of a solution similar to https://bugzilla.wikimedia.org/show_bug.cgi?id=41605" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37042 [22:12:48] brion urgent review needed - https://gerrit.wikimedia.org/r/37139 [22:13:08] Change merged: Brion VIBBER; [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37139 [22:15:16] New patchset: MaxSem; "Remove a few antiquated option screens" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37318 [22:21:12] \o/ [22:21:21] New patchset: Jdlrobson; "disable history api usage on Android 4.x phones (bug 41605)" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37321 [22:21:22] New patchset: Jdlrobson; "disable usage of history api in S60 webkit browsers (bug 41407)" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37322 [22:21:22] thanks brion [22:23:18] MaxSem: do we need to remove qqq codes as well? [22:23:25] context https://gerrit.wikimedia.org/r/#/c/37318/ [22:23:34] no, TWN scripts will do it [22:24:22] awesome - that saves me hassle in future [22:25:16] New review: Jdlrobson; "According to Max we don't need to remove the qqq message - this was news to me" [mediawiki/extensions/MobileFrontend] (master); V: 1 C: 2; - https://gerrit.wikimedia.org/r/37156 [22:25:16] Change merged: Jdlrobson; [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37156 [22:25:53] New patchset: MaxSem; "Remove old JSON API completely" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37323 [22:26:44] New review: MaxSem; "https://www.mediawiki.org/wiki/Localisation#Removing_existing_messages" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/37156 [22:29:26] jcmish: will you be available in about 30 mins to help with some quick testing for a bug fix deployment? sorry this is last minute, but the fix is fairly urgent [22:29:35] MaxSem: I was wondering - do you think we could enable a nearby mode on our new alpha when we have it? [22:29:39] since the usage would be so low? [22:29:54] could be a fun use of our experimentation time.. [22:30:17] we don't have geosearch enabled yet [22:30:23] jcmish: im probably going to cherry-pick the patchset to address the issue rather than deploying from master so a quick regression test + test for the fix is all that's needed [22:31:25] also MaxSem i'm not sure about this - https://gerrit.wikimedia.org/r/#/c/36887/ - i'm tempted for us to merge it just to enable him to do what ever he's doing [22:31:58] i looked at http://www.yeshiva.org.il/wiki/ and couldn't translate it - but it seems he is adding additional sections to pages [22:32:20] .il? ask aharoni! [22:32:25] (It seems wrong to put it in the article.. but hey..) [22:32:42] aharoni: can you make sense of what https://gerrit.wikimedia.org/r/#/c/36887/2 is trying to achieve? [22:32:44] it's a religious website [22:33:17] ha, didn't know they use mw, but I'm not very surprised. [22:35:19] if i understand correctly aharoni are they adding a menu to every page? [22:35:31] no idea, i'm trying to undestand. [22:36:34] also aharoni - whilst your here - i have a feeling that the menu is not rtl friendly - when you click it the icons are on the left but they should probably be on the right yes? [22:36:48] which menu exactly? [22:39:42] the three line menu [22:39:46] which is now known as the hamburger [22:39:49] (top right corner) [22:40:14] e.g. http://he.m.wikipedia.org/wiki/%D7%9E%D7%99%D7%95%D7%97%D7%93:MobileMenu [22:40:48] * awjr slaps comcast [22:41:02] aharoni: brb [22:46:39] jdlrobson: yes, the icons must be on the right. [22:47:35] and while you mention that menu - it pissed me off for a while that I can't use the phone's back button to close it. so here, I finally said it. [22:47:48] is it possible to make the back button close it? [22:49:26] it is possible, as long as the history stuff works correctly in the browser :) [22:49:39] l'll leave such implementation details to jdlrobson though [23:00:46] jdlrobson: about yeshiva.org.il, i do see that they added two sections at the bottom, which actually come from the sidebar. These are the two section above the toolbox. [23:01:23] this goes together well with the commit message. is it really a required software feature? - i don't know, i'm not much of a designer. [23:01:51] it's nice to see that they care about the mobile interface and actually try to implement something with it. [23:02:23] and they have advertising banners, which work quite well from the technical point of view, and that's probably a Good Thing. [23:03:24] jdlrobson: the e3 team's deployment is going over, so we're on hold to get that bugfix out [23:04:25] hiring interview @ 7pm .. ouch … thats going to be fun [23:04:37] 7pm [23:04:39] ouch [23:05:30] jdlrobson: ok they're done [23:07:21] brion: how long did it take for 1.3.3 to finally show up in the market ? [23:09:37] jdlrobson: i feel like the alpha mode should have a badge and that badge should be what the mobile team uses in all of its press :) [23:09:37] sweet awjr [23:09:59] It's a dragon blowing fire on the wikipedia logo tfinc [23:10:13] jdlrobson, jcmish, changes are up on testwiki. once you give me the go ahead i'll sync to the cluster [23:10:25] kayo [23:10:29] jorm would probably design that for us ^^ [23:10:44] awjr: looking now :) [23:11:02] jcmish: to test the serious bug you'll need to pretend you are a nokia [23:11:16] jdlrobson: do you have a UA we can use? [23:11:27] k will do [23:11:38] awjr: just grabbing one [23:12:11] tfinc: about 2 hours [23:12:15] k [23:12:22] Mozilla/5.0 (SymbianOS/9.3; Series60/3.2 NokiaE52-2/091.003; Profile/MIDP-2.1 Configuration/CLDC-1.1 ) AppleWebKit/533.4 (KHTML, like Gecko) NokiaBrowser/7.3.1.34 Mobile Safari/533.4 [23:13:14] jdlrobson: and this problem is in non-beta, correct? [23:13:27] correct [23:13:35] previously toggling was impossible [23:13:57] * tfinc spins up tweetdeck [23:14:25] awjr: test http://en.m.wikipedia.org/wiki/Golden_White-eye to verify it is broken in production with your user agent [23:15:20] ew [23:15:24] ok, seems to work fine on testwiki [23:15:25] http://test.m.wikipedia.org/wiki/Test_Photo_Uploads test it works [23:15:26] works for me [23:15:31] jcmish: follow up with https://twitter.com/pablom4t4/statuses/273989223575584769 to see if 1.3.3 fixed the issue [23:15:33] brion: --^ [23:15:54] tfinc: will do [23:16:07] brion: lots of playbook users through twitter. i can see why we needed the fix [23:18:44] awjr: works for me [23:18:45] push at will [23:18:49] cool [23:18:57] jcmish: you confident to push or still testing? [23:20:13] as i keep saying jdlrobson rmoen https://twitter.com/megabee/statuses/276760467060453376 [23:20:36] I think it looks good [23:20:40] i'm going to have to work our new mission statement into our docs [23:20:47] cool syncing now jdlrobson, jcmish [23:20:47] and schwag [23:20:50] kayo [23:21:00] hmm .. speaking of schwag i need to find some for jgonera [23:21:08] tfinc: true true :) [23:22:05] jcmish, jdlrobson: ok changes should now be live [23:23:08] awjr: works for me - i'll reach out to a twitter user [23:23:29] jdlrobson: you might wait just a few minutes [23:23:35] will do [23:23:40] im still seeing the problem on some articles, probably just waiting for cache to refres [23:23:40] h [23:25:42] MaxSem: i do love your branch name btw [23:25:51] you named it [23:26:08] MaxSem: # @todo: Remove in March 2013 < it is not 2013! -2 ;-) [23:26:41] it will eventually be 2013, young padawan [23:27:12] awjr: do we need to wait? [23:27:18] or can we go ahead? [23:27:41] tfinc: talked to pablo on twitter and he's all set [23:27:51] tfinc: anyone else you want me to reach out to? [23:27:57] jcmish: i pushed already but am still seeing the issue on some pages. it is either related to RL cache (which refreshes after 5 mins) or we'll need to manually purge varnish cache [23:28:06] 5 mins should be up in another minute or so [23:28:10] awjr: kayo [23:29:08] ok i htink we need a varnish purge [23:30:47] what will we do in the future when there will be no separate mobile cache and domains so it will not be possible to purge only our cache? [23:32:01] MaxSem: we will cry [23:32:22] or lower our ttl :p [23:32:59] jdlrobson: you're in the office right/ [23:33:13] jcmish: any one else you find in the twitter backlog. also anyone who you see rating us down in the play market [23:33:19] we have the reply feature for our account [23:33:25] allowing bi directional conversation [23:33:26] i'm starting the draft of the blog post on all our exciting beta work. jdlrobson, can we please use the "leaping jon" image of you for this? :) [23:33:36] oh cool [23:33:41] Maryana: please do [23:33:45] :D [23:33:49] as long as its properly licensed [23:33:56] naturally [23:34:02] jdlrobson: if you're in the office can you harass the ops corner to flush the mobile varnish cache IRL? [23:34:15] but very nicely, of course. bribing with whiskey usually works. [23:34:23] awjr: why do we continue to need to do that ? [23:34:36] tfinc bribe ops with whiskey or flush the mobile varnish cache? [23:34:51] later. i know why i do the former. its gets shit done. [23:34:56] hehehe [23:35:18] well right now we're fixing a bug that is affecting cached html for devices that don't dynamically update content, which is why we need to do it [23:35:37] k [23:36:04] jcmish: do you see the reply option in google play ? [23:36:49] New review: Jdlrobson; "I'm still trying to understand the use case - if the idea is to create additional menu items would i..." [mediawiki/extensions/MobileFrontend] (master); V: 0 C: 0; - https://gerrit.wikimedia.org/r/36887 [23:37:16] awjr: who would be the best person to nag? [23:37:22] asher? [23:37:36] jdlrobson: asher first but i think he might be busy. but he'll punt you to the next best person if he is [23:37:38] brion: https://code.google.com/p/chromium/issues/detail?id=113088 [23:37:54] jdlrobson: theoretically any op can do it (we can point them to documentation if necessary) [23:38:22] tfinc: there's a way to build Chromium as an android wrapper which can be used phonegap-like; there's been a couple threads on cordova dev list about it [23:38:29] apparently we should ask in the ops channel.. [23:38:31] main problem is that it seems to eat up 20-40 megabytes :P [23:38:37] yuck [23:38:41] since you're including your own webkit build in your apk [23:38:53] awjr: did you touch the js file inappropriately? [23:39:01] surely we shouldn't be having caching issues [23:39:04] jdlrobson: is there any other way to touch it? [23:39:10] jdlrobson: it's not js related, it's cached html related [23:39:13] i don't know :) [23:39:15] really? [23:39:25] i don't understand.. the fix only touched js... [23:40:03] (mf-application.js) [23:40:25] tfinc hmm no [23:40:31] is this in googleplay console? [23:40:49] this is the debug=true problem no? [23:41:08] jcmish: are you logged into google play with your wikimedia count ? [23:41:13] account* [23:41:20] yup [23:41:32] i see the errors [23:41:36] on the main page [23:42:37] jcmish: you should be able to look at the app detail and right below each review you'll see 'Reply to this review [23:42:38] ' [23:43:11] awjr: ? [23:43:16] ? [23:43:22] jdlrobson: i don't understand.. the fix only touched js... [23:43:22] [3:40pm] jdlrobson: (mf-application.js) [23:43:27] jdlrobson: this is the debug=true problem no? [23:43:52] we need to touch the mf-application.js and resource loader should take care of the rest [23:44:03] jdlrobson: i've done that [23:44:12] mm this makes no sense [23:44:17] tfinc: got it [23:44:17] we shouldn't have to touch varnish cache [23:44:27] jdlrobson: thinking... [23:44:29] I'll start reaching out on there and twitter [23:44:44] brion: did we file that https was not working on android 2.2.2 with upstream phone gap ? [23:44:44] awjr: what if I edit MediaWiki:Mobile.css - will that not trigger a refresh? [23:45:03] tfinc: not yet, i'll put together a minimal test case if i can [23:45:03] [WikipediaMobile] brion pushed 2 new commits to v1.3.3-PlayBook: http://git.io/iFNAJQ [23:45:03] WikipediaMobile/v1.3.3-PlayBook e0273ef Brion Vibber: Apply overflow-y: auto scrolling by default, only disable it on Android.... [23:45:03] WikipediaMobile/v1.3.3-PlayBook 138f685 Brion Vibber: Bump version numbers for PlayBook build [23:45:12] brion: yeah, something so that they know of it [23:45:21] Project WikipediaMobile - Nightly builds build #470: SUCCESS in 12 sec: https://integration.mediawiki.org/ci/job/WikipediaMobile%20-%20Nightly%20builds/470/ [23:45:22] * Brion VIBBER: Apply overflow-y: auto scrolling by default, only disable it on Android. [23:45:22] * Brion VIBBER: Bump version numbers for PlayBook build [23:45:55] hm is jenkins building every branch? odd [23:47:28] reminds me we should fold the firefox os build back into the main repo sometime [23:47:28] sigh [23:47:58] jdlrobson: i feel like i've seen this problem before where it seemed touching and syncing should solve the problem but we've had to refresh the mobile varnish cache. is it possible that references to stale resources get cached with the page content? [23:48:32] awjr: this is a bug in ResourceLoader that we keep trying to get round with a varnish cache flush - we should probably work out what's happening to stop us being dependent on these flushes [23:49:04] maybe we should bug trevor? [23:49:10] jdlrobson: agreed and we are not the only ones dealing with problems [23:49:22] jdlrobson: go for it, i can reach out to the RL guys too [23:49:28] brb going to see if I can find him [23:50:33] jdlrobson: did you get any response from ops irl? [23:51:15] awjr: i haven't asked as I feel that we should avoid this if possible [23:51:34] jdlrobson: in the immediate time frame, i dont think we can. [23:51:36] i'd rather have a broken experience for an additional 1hr and learn a valuable long time lesson [23:51:45] *long term [23:51:59] jdlrobson: i dont think the RL bug is going to be fixed in the next hour :p [23:52:03] Krinkle: there? [23:52:12] im also not even positive the RL bug is the culprit in this particular case [23:52:12] YEs [23:52:14] awjr: I just want to rule out problems with our deployment [23:52:36] jdlrobson: ok, i'm only around for the next hour so im eager to get this done. [23:52:45] The vector heading thing is not RL related (at least not in the sense of the issue with mobile's touch-ing thing) [23:52:49] So Krinkle we have a ResourceLoader related issue - http://en.m.wikipedia.org/wiki/2012_Formula_One_season is serving old javascript for me when I force the user agent to be Mozilla/5.0 (SymbianOS/9.3; Series60/3.2 NokiaE52-2/091.003; Profile/MIDP-2.1 Configuration/CLDC-1.1 ) AppleWebKit/533.4 (KHTML, like Gecko) NokiaBrowser/7.3.1.34 Mobile Safari/533.4 [23:52:53] jdlrobson: is juliusz avoiding irc ? [23:53:06] if I append debug=true i get the latest resources [23:53:20] we keep running into this issue again and again in deployments [23:53:25] jdlrobson: I can't investigate right now, I'm busy with something and only an hour before I'd like to close up. [23:53:26] awjr touched the javascript file that changed [23:53:29] but it hasn't propagated [23:53:37] any theories though Krinkle ? [23:53:41] and mind you we're loading resources in an atypical way for ResourceLoader [23:53:43] jdlrobson: sounds like the module timestamp isn't being updated [23:53:51] it really is biting us every single deployment to the point we are about to ask for a varnish cache flush for mobile [23:54:06] Krinkle is it possible that the module timestamp get cached with page html? [23:54:08] awjr: yes that is also part of a variable that is attractive for blame. [23:54:08] jdlrobson: is the timestamp=XXXXXX up to date in the HTML? [23:54:23] awjr: the way we're using them they are indeed directly in the html with the timestamp [23:54:24] awjr: that can only happen if you don't use RL the way it should be [23:54:38] yes, so that is why we need to flush the mobile vranish cache [23:54:39] brion: what? [23:54:54] [23:55:04] so the startup module and all, that was for nothing? [23:55:13] just hardcoding it anyway [23:55:13] we don't use that quite yet [23:55:16] ok [23:55:43] well, this was the problem since day #1 last year. After every deployment mobile had to purge html cache (not bits) [23:55:56] *nod* [23:55:57] until the timestamps go out of the html that will be that way, no way around that. [23:56:07] it points to a resource [23:56:09] ok this makes more sense now [23:56:12] so i think the current issue we'er seeing is not related to the touch-file thing, but rather dealing with cached html that needs purging [23:56:17] i think the problem will remain until we switch to fully using RL, letting the JavaScript do the loading from the data in the startup module [23:56:21] yep [23:56:25] correct [23:56:26] ok so varnish cache flush it is :( [23:56:27] but this is not what jdlrobson was talking about, right? [23:56:31] \o/ [23:56:33] jdlrobson: you were talking about the bug in RL-mode [23:56:54] with the module not being refreshed, needing an odd touch-file to fix it, right? [23:57:01] Krinkle: i think jdlrobson was not aware of the cached html issue and was thinking the problem was related to the RL bug [23:57:09] ok [23:57:12] awjr: correct [23:57:16] Krinkle: sorry to bother you [23:57:21] no problem at all. [23:57:23] but thanks - this sheds a lot of light on why there is a problem [23:57:33] yay [23:57:34] we don't see this particular thing very often any more [23:57:41] or at least i havent [23:57:56] jdlrobson: is that satisfying enough for you to bug ops IRL now? [23:58:32] no one's responded to me yet on irc :( [23:58:47] oh man my heart stopped for a second…. ran Wikipedia app master on my Android 2.3 phone and got "offline"! [23:58:51] turns out it's… not on the wifi :D [23:58:57] k requested [23:59:02] awjr: ^ [23:59:04] ty jdlrobson [23:59:09] can you join wikimedia-ops [23:59:58] jdlrobson: you probably want #wikimedia-operations