[07:41:05] morning! [12:19:30] hey joakino [12:22:12] hi phuedx [12:22:49] how're you? [12:23:18] fine! phabricating around [12:23:23] i see [12:23:34] *so i see [12:28:25] phuedx: how are you doing [12:28:38] phuedx: my puppy is causing massive destruction today [12:28:47] classic puppy [12:28:59] yeah -- tired but doing a little better [12:29:12] hopefully going for a boulder tonight, which'll be the first time in a while [12:29:25] she grabbed a kitchen paper roll and played with it all through the apartment [12:29:51] brb [12:30:00] 👍 [12:43:00] just taken delivery of 1kg of coffee [12:43:03] i am happy [12:43:06] also, backsies [12:50:05] awesome phuedx ☕️😁 [12:50:09] i'm out for lunch! [12:50:11] bb! [14:26:38] joakino: hi :) Do you mean that I should write an e-mail to the mailing list or directly to you? -> https://phabricator.wikimedia.org/T91752#1523691 ?? :) [15:48:35] joakino: you around? [15:56:53] rmoen: he went for lunch ~3 hours ago [15:56:59] must've been a /big/ lunch [15:57:08] phuedx: hah must be [16:01:42] missed joakino for meeting, do i need to be in prioritization. [16:02:18] there's a prioritisation meeting? [16:02:27] oh [16:02:29] yes, there is [16:02:36] it's tech lead/product for defs [16:02:37] rmoen: hey [16:02:43] ok [16:02:45] other engineers can attend if they so wish [16:02:47] sorry dude, the clock on my phone is wrong [16:02:55] ^ read: had a nap [16:02:59] after a /big/ lunch [16:03:07] I'm gonna coffee up. See you at standup [16:04:13] joakino: we can chat later ? [16:08:11] rmoen: we can talk now, free till standup [16:10:27] is gerrit functioning for y'all? [16:10:30] Making coffee. 5 minutes ? [16:11:42] rmoen: 👍 [16:12:18] FlorianSW: I meant that /you/ should send an email (since you worked on the feature 😁) after it's public on enwiki [16:12:28] joakino: can you access gerrit? [16:12:29] FlorianSW: sorry if I confused you [16:13:01] joakino: ah, ok :) [16:13:03] phuedx: yerp, seems to work fine [16:13:30] derp [16:14:39] FlorianSW: how can i see the editor change? (force it) [16:15:56] joakino: on beta labs? Unfortunately I haven't thought about that :P So you need to have an account with a user id, that must not be divisible by 2 and you need to be in the beta mode. [16:16:18] joakino: locally you can simply change the check in mobile.editor/init.js to be always true [16:16:21] FlorianSW: only logged in users? [16:16:43] joakino: yes, anonymous users has null as a user id :) [16:18:05] FlorianSW: we don't want the change for anon? seems like it could be helpful [16:18:13] ok found a user that shows it :D [16:19:26] joakino: great :D It would be more difficult to bucket anonymous users. I think it would be helpful, but the idea was to first look how the user use it and maybe expand the test or remove it :) [16:20:27] FlorianSW: 👍 [16:20:58] FlorianSW: we have ways of bucketing anons via localstorage tokens, we've done it before. ping the team if you want to expand it or need help with anything [16:21:59] joakino: I would need help by analyzing the data, since I haven't access to the eventlogging data :) The new toolbar editor is logged with a different editor type in eventlogging :) [16:42:38] FlorianSW: oki! lets keep in touch and we'll support you and help you analyze it [16:43:09] joakino: great, thanks :) I think it will take some weeks after we can check, the results :) [17:17:09] ye! [17:17:58] mbinder: mhurd_afk sorry got logged out at the worst time.. [17:18:59] back in a bit y'all [17:19:30] so pretty seriously i'm not seeing any js run on mobile in production [17:19:39] can anyone else replicate this on https://en.m.wikipedia.org/wiki/Pond in an incognito window.. [17:20:55] or https://en.m.wikipedia.org/wiki/Dengue ? [17:24:04] ^ bmansurov rmoen joakino phuedx|afk any of you able to replicate [17:24:06] https://en.m.wikipedia.org/wiki/Platelets also [17:25:31] all 3 urls work fine [17:29:23] mm wtf [17:29:40] bmansurov are you anon, in an incognito window with cache disabled [17:30:27] jdlrobson: yes to all 3 questions [17:30:51] rmoen: can you check? [17:30:58] maybe it's a usa thing [17:32:54] jdlrobson: i'm in the usa [17:38:13] i dunno i'm clutching at straws. Roan can't replicate it either [17:38:17] or Ori [17:38:25] bmansurov jdlrobson afk for a sec, almost burned the house down [17:38:30] but i see tweets on twitter suggesting i'm not alone [17:39:20] rmoen: oh man, be careful [17:42:22] bmansurov: and to be clear you don't see a blue button on https://en.m.wikipedia.org/wiki/Platelets saying Search ? [17:43:05] jdlrobson: no i don't [17:43:31] joakino: around? ^ [17:43:33] jdlrobson: https://en.m.wikipedia.org/wiki/Platelets in a Firefox private window seems to work, do you want me to log in? [17:44:07] jdlrobson: i can't replicate either, watchstar opens the pane fine and the searchbar is js enabled on all those pages :S [17:44:15] gonna try the phone [17:46:04] jdlrobson: chrome android incognito seems to work fine too [17:46:20] jdlrobson: confirmed, no js on https://en.m.wikipedia.org/wiki/Pond in a new incog window [17:46:22] S Page is not having issues either and he's sitting next to me [17:46:32] but i can consistently replicate it on Firefox and chrome [17:46:38] damn [17:46:39] wtf [17:46:40] i can replicate [17:47:14] https://en.m.wikipedia.org/wiki/Platelet is fine though [17:47:18] cache ? [17:47:32] firefox works fine for me too [17:47:54] jdlrobson: what's going on in the network tab? 404s? [17:47:58] i shouldn't be breathing these fumes, taking laptop outside [17:48:21] joakino: so the problem seems to be old HTML being served with async js [17:48:43] basically the bug i investigated last week [17:48:52] but i was told it would be purged before going live [17:49:53] jdlrobson what does mw.loader.getState( 'mobile.search' ) return for you? I assume it's not "ready" [17:50:14] jdlrobson: why is it happening now [17:50:24] or has it been happening since thursday [17:51:21] i'm trying to work out in wikimedia-perf [17:51:36] spagewmf: it's registered - i know the route cause [18:10:50] so i'm assuming this was a solved problem on mobile [18:10:53] *desktop [18:11:06] and we can do whatever they did.. but right now having trouble getting the right people [18:11:37] joakino: got you a minimum test case for https://phabricator.wikimedia.org/T108368 [18:11:43] so feel free to triage as appropriate [18:12:56] thx jdlrobson, i'll have a look tomorrow [18:29:07] vibha: kaity either of you have time to talk about iOS navigation in 5.0? [18:39:48] hmmm anyone else have a script error complaining about strict MIME type checking? [18:39:52] bgerstle: sure [18:40:00] kaity what time works for you? [18:40:00] bgerstle: 2 min [18:40:06] ok [18:40:52] nvm, its just me [18:46:56] bgerstle: sorry ready anytime now [18:47:05] k [18:49:15] kaity: give me one second to get into a somewhat demo-able state [18:55:04] kaity: ok, ready if you are. do you want to grab a room or should i call you? [18:55:32] jdlrobson: hey are you guys aware of a bug in stable where everything seems shifted to the right? [18:55:37] jdlrobson: https://www.dropbox.com/s/yf0lany4557mbsi/IMG_9394.jpg?dl=0 [18:56:05] jdlrobson: the toast is shifted to the right in that one [18:56:08] jdlrobson: https://www.dropbox.com/s/ibao4x2kyykjq49/IMG_9395.jpg?dl=0 [18:56:21] edit and watch are cut off on right side [18:57:10] this is in safari on ios 8.1 [19:36:13] gtg -- g'night y'all [19:36:21] review incoming tomorrow morning [19:37:02] jdlrobson: based on your comments https://gerrit.wikimedia.org/r/#/c/229859/, I'm assuming you are implying that this is the wrong approach. It seems there will be a delay in rendering as the mobile.head init is async.. so not sure if my change introduces that mainly exposes the delay as the modification to the bar is now happening. [19:59:24] niedzielski: i actually didn't realize that SimpleDateFormat wasn't thread-safe! [19:59:47] niedzielski: what do you mean by skipping the member variable? [20:01:02] dbrant: er, yes i could have been cleaer [20:01:05] clearer, even [20:01:23] dbrant: i was thinking if we just return a new instance each time that would be one way around it [20:02:38] dbrant: i used joda time in the past. i think there are a couple caveats for android use but i think those are solved in this fork https://github.com/dlew/joda-time-android [20:02:55] and corresponding blog post http://daniel-codes.blogspot.com/2013/08/joda-times-memory-issue-in-android.html [20:04:24] niedzielski: ah yes, thanks. I think I'd probably prefer that for now. (returning a new instance) [20:04:47] dbrant: ok cool. other than that, the patch looked great to me [20:07:04] rmoen|lunch: the approach is the one Timo mentioned - about not hardcoding if ( mw && mw.mobileFrontend ) [20:07:07] these will not work any more [20:07:33] there are 2 issues here - fixing cached pages and future proofing it - i would suggest focusing on latter to start with [20:07:51] sorry i can't explain more as i'm tackling similar related problems [20:13:46] rmoen|lunch: https://gist.githubusercontent.com/Krinkle/b12a50a3e34edf25d905/raw [20:39:12] kaity: nope raise a bug [20:40:52] jdlrobson: thanks! couldve raised it but didnt know if it was already caught [21:02:37] jdlrobson, Krinkle. I see the point about using ResourceLoader::makeInlineScript. However, this does not resolve the issue as the emit is happening before the module listening is present. So I'm not sure how this is the wrong approach. [21:03:06] either way, I've made the suggested change https://gerrit.wikimedia.org/r/#/c/230653/ [21:03:20] rmoen: The condition is for 'window.mw' which doesn't exist because its loaded asynchronously. So the code is ignored as if(falsey) [21:04:35] dbrant: sorry we are trying to figure out a room [21:04:39] I understand. however it seems the emit is still not firing ;/ [21:04:42] kaity: np! [21:05:17] rmoen: That's because the script tag isn't being echoed [21:05:24] ;-) [21:05:55] Krinkle: oh $wgOut->addScript( ResourceLoader::makeInlineScript( $script ) ); would work ? [21:06:18] No that would result in a syntax error for ' you want to either 'echo' or out->addHtml() [21:06:51] addHtml is recommended, but probably won't work properly since most of mobile uses [21:06:53] echo still [21:09:38] Thanks Krinkle. Though now throwing error Uncaught TypeError: Cannot read property 'emit' of undefined [21:10:16] rmoen: That suggests it was already broken, except hitting the 'mw.mobileFrontend' case so it was ignored silently before [21:10:27] You'll need to ensure somehow that that method exists or is loaded first, or add a dependency [21:10:53] the module that provides that method should be light-weight, in the top module, and required using mw.loader.using() probably [21:11:04] Deferring to Jon for that [21:12:25] So instead of if(mw.mobileFrontend) use mw.loader.using('mf.emit', function () { mw.mf.emit( 'x' ); }) or something like that [21:12:35] Gotcha [21:27:36] dbrant: Did your production release go off without any issues? [21:28:41] Deskana: it's in a staged rollout as we speak (20% right now, 100% by the end of today). No issues so far! [21:29:43] dbrant: Excellent. Yesterday saw the biggest decrease of zero results rate that we've seen since we started monitoring it, although it is admittedly not a statistically significant variance from the normal decrease at this stage. [21:29:58] We'll see what happens in the next few days! [21:30:30] Deskana: yep, the next few days will tell. [22:06:26] dbrant: https://www.dropbox.com/s/qjs39s6uis0yco4/link-preview-13.png?dl=0 [22:16:44] bmansurov: can you take a look at https://gerrit.wikimedia.org/r/230680 [22:16:53] i'm gonna merge Sam's patch but it introduces a new regression ^ [22:16:59] k [23:32:52] kaity: i sent you an notification from HockeyApp [23:33:04] i uploaded a build w/ what we talked about: conventional iOS navigation transitions [23:33:09] bgerstle: cool ill download [23:33:51] kaity: one bug you'll need to work around for now is that the webview doesn't load correctly, so you'll need to pull to refresh to fill it w/ content [23:34:16] plan to fix that soon, but at least this way you can play w/ it a bit and i can talk to corey about it tomorrow [23:35:48] bgerstle: the version i'm getting seems to be an old one [23:36:06] kaity: it's labeled as 4.1.7(168) [23:37:01] but should say it was uploaded a few minutes ago [23:37:41] bgerstle: hmm it says aug 4 [23:38:11] oh and it says 5.0 (168) [23:38:35] kaity: try going to the HockeyApp app [23:39:45] bgerstle: i got the right one now [23:39:53] cool [23:39:55] sorry for the mixup [23:40:16] bgerstle: sorry must've been my hockey app not updating [23:41:36] bgerstle: transitions looks good [23:41:47] bgerstle: can't get edge swipe to work [23:41:52] really? works for me [23:42:04] hm [23:42:07] only sometimes, actually [23:42:08] weird [23:42:14] worked after i tapped to show the nav bar [23:42:27] bmansurov: can you revert that change you just merged [23:43:13] kaity: also works OK from the webview [23:43:27] probably some gesture conflicts w/ the current table-based view we have for the "reader"/native view [23:43:33] i can play around w/ it tomorrow [23:43:44] jdlrobson: rob's patch? [23:43:51] bmansurov: yeh i posted a revert [23:44:03] i don't want us rush fixing this. [23:44:10] we might as well not load that code in the head if we are doing this [23:44:51] bgerstle: every other page seems to load toolbar at top then no toolbar [23:45:08] The only reason we are loading JavaScript in the head is to avoid a flash of unstyled content [23:45:15] if we introduce one then there's no reason to load in the head right? [23:45:23] kaity: the toolbar goes away after you swipe/tap [23:45:41] we could also be hiding it [23:45:50] bgerstle: right, maybe i'm just getting bugs [23:45:53] yeah [23:46:01] bgerstle: lots of content not loading in native cards too [23:46:05] still some kinks to work out, this was just the first attempt [23:46:06] ok [23:46:08] i'll have to look at that [23:47:14] bgerstle: its cool, transitions are looking good [23:47:32] good, hope it at least gives you a better idea [23:47:52] i have a feeling we'll talk about it more tomorrow when corey's back [23:48:24] bgerstle: so you don't have the link previews coming up from the bottom? [23:48:34] kaity: nope, just pushing from the right [23:49:33] kaity: it would be possible to do the transition i described where they come up from the bottom as the previous content slides out to the left [23:49:34] am writing up in more detail on the patch [23:49:37] but would require some reworking [23:50:19] bgerstle: i got an animation program working so i can experiment with a lot of stuff [23:50:25] sweet! [23:50:27] yeah, that makes sense [23:50:43] bgerstle: can you view this ok? https://www.dropbox.com/s/l4goi0c71camv4e/link-preview.mp4?dl=0 [23:51:09] bgerstle: (thats more for the android style) [23:51:39] kaity: yeah [23:51:47] that's basically what we had before [23:53:02] oh kaity apparently the edge swipe is only possible while the back button is visible :-/ [23:53:27] bgerstle: oh i got it working now [23:53:33] ? [23:55:49] kaity: the edge swipe? [23:56:01] bgerstle: ya the edge swipe [23:56:18] bgerstle: when i made the back button visible i mean [23:56:20] kaity: cool, it seems to only work when the navigationBar is visible [23:56:21] yeah [23:56:29] that might be by design which i'm not crazy about [23:56:34] never realized it behaved that way [23:58:33] kaity: anyway, i'm out for the day [23:58:36] talk tomorrow o/ [23:58:45] bgerstle: have a good night! [23:59:10] thanks, you too!