[07:41:05] good morning! [08:11:23] phedenskog: yo! [08:11:37] hey! [08:18:54] morning [08:28:11] hepp! [08:32:32] phedenskog: issue fixed! i can run profiles now https://github.com/axemclion/browser-perf/issues/62#issuecomment-221506879 [08:33:17] yeah I saw it the same second on Github. I love Open Source and that he could help us so fast :) [08:33:45] what kind of number did you get for images? and did you feel that it still was slow? [08:34:41] I wonder if we should try to hack it to count the total time spent on parsing images, to have a number that we can show. [08:39:06] joakino: I checked the output you posted, nothing strange there right, so maybe it don't catch it? I'll run on my device and check [08:42:37] phedenskog: that was on wifi [08:42:47] gonna run it on 2G and see what happens [08:43:01] ok! [08:48:22] phedenskog: nothing too weird, gonna rerun [08:48:39] on manual run the scrolling is not blocked when loading like on the YT video [08:48:54] i'll gist the numbers in a bit [08:49:25] I wonder why is that [08:49:36] could the problem be with your fingers? :D [08:50:50] maybe it's because it's a clean session, hmm [08:51:31] phedenskog: i have the theory that accesibility could have slowed down chrome a lot [08:51:39] i recently disabled it (and lastpass) [08:51:44] phuedx: https://gist.github.com/joakin/c900fae8eddc57d95ceedeb1892aac80#file-edge-txt [08:51:49] oops, i meant phedenskog [08:51:56] damn ph* [08:52:01] anyways, seems normal [08:52:32] if you enable it? [08:52:46] I love that loadTime :) [08:53:38] gonna try enabling it, but without lastpass enabled listening to it it may do nothing [08:53:39] but still even though this is only one run, it would intreresting to compare with/without lazy loading. [08:53:40] lets see [08:53:50] phedenskog: mobileaction=beta? [08:54:00] yes don't you think? [08:54:05] yep [08:54:09] to get some kind of feel for the metrics [08:54:31] everything that can help us quantify the change is good I think [08:55:14] i'll try [08:55:26] keeps changing between G and E, which is not good [08:57:50] freaking CN banner is destroying load time on gprs [08:58:18] that load time is /beautiful/ [08:58:19] 2g? [08:58:23] Edge [08:58:29] whatever that means [08:59:10] joakino: ~100kbps down is typically what that means [08:59:33] phedenskog: beta metrics are insane, not sure if it is my network conditions (changing between E and G) or that it is going to the US to the php servers instead of Ams to the varnish ones [08:59:42] phuedx: 👍 [08:59:53] keeps switching to Gprs though [09:00:52] woah [09:01:00] visualeditor translations on mw.org are borked [09:03:47] phedenskog: i can't get meaningful metrics to compare, beta is loading a CentralNotice banner for me, while stable isnt, so it takes longer [09:05:15] joakino: i added a note about proxies and centralauth to the staging server notes: https://www.mediawiki.org/w/index.php?title=Reading%2FWeb%2FSetting_up_a_staging_environment&type=revision&diff=2147065&oldid=2145798 [09:05:27] i still doubt that folk will bother setting up their own staging server :/ [09:06:05] phedenskog: i've been considering writing a few benchmarks for certain parts of the mobileformatter [09:06:14] with a php benchmarking library [09:06:36] they'd be part of the codebase and runnable, but not run during integration or anything like that [09:06:38] phuedx: i'll do it, i've added to the todo list,w [09:06:57] i'm just curious about the recent additions to the formatter [09:06:58] lol vim keys ,wq [09:07:01] ^ lol [09:07:24] joakino: ah I see. I'll try see how it works for me but I need phuedx pi, don't have that yet, waiting for the paycheck . [09:07:39] phuedx: cool [09:37:42] phuedx: how do you diagnose the false positives? [09:37:44] manually? [09:37:50] manually [09:38:07] the first five lines of the stacktrace in jenkins is usually a good start [09:38:31] the API request to create a page, for example, will fail with "this wiki is in read-only mode…" [09:38:39] which is usually a good indicator that something's up [09:38:48] i was thinking on opening a task for them to think about not sending alerts when bc is not normal or showing the beta cluster status on the alert [09:39:09] phuedx: makes sense, do you think ^ would be useful? [09:44:30] joakino: i think skipping the scenarios if the environment is in read-only mode would be a positive step [09:44:39] i'd be up for co-authoring a task :) [09:45:25] phuedx: feel free to edit https://phabricator.wikimedia.org/T136172 [09:46:07] hopefully we can get somewhere that gives us less false positives [09:49:37] joakino: left a comment, rather than editing the task [09:50:11] joakino: you'll note that some scenarios do pass in read-only mode [09:51:09] phuedx: tagging tests with read write permissions sounds very error prone to me, i'd rather not run the tests against that environment [09:51:34] why? again, it simply means that you can't make edits, but you can read information and use the ui [09:51:56] i'd rather know that users can use parts of the ui in read-only mode [09:52:01] that we support read-only mode [09:52:05] rather than simply bomb out [09:57:49] or maybe we should have tests for read-only mode but de-prioritize test reports from flaky environments instead [09:57:54] ^ joakino [09:59:54] joakino: my reasoning was that requiring write access during our scenarios is actually fairly exceptional [10:00:00] but maybe you're right [10:06:44] phuedx: it makes sense if we want to properly support read-only mode [10:06:54] phuedx: but a lot of tests setup involves writing to test read features [10:06:57] which is the problem I sees [10:06:58] see [10:07:51] that's why i'm not sure how many read-only features you can browser test without writing anything in the test setup [10:08:03] there are a few, like UI stuff definitely [10:08:44] so maybe i should flip the tag suggestion to "@read-only" [10:08:51] thinking about it, for those it may be interesting to tag @readonly [10:08:56] ^^ [10:08:57] boom [10:09:10] mind if i copypasta this discussion to the ticket and change my suggestion? [10:09:16] sure [10:09:19] go for it [12:29:04] how can I get ConfirmEdit/QuestyCaptcha working in MobileFrontend? [12:29:15] it will not show up, yet still block the registration [12:32:36] okay, I just figured out there's a extrafields that could be used inside the code. I'll investigate further.. [12:33:54] okay it's just not shown. thanks anyway [13:10:37] some love for ref popups on mobile web ;) https://twitter.com/libbybrittain/status/733771681567637504 (dr0ptp4kt , shall we RT?) [13:11:17] HaeB: yeah, looks like a good one [13:11:29] HaeB: i'm going for a jog in a few minutes. ttyl! [13:12:56] (done, have a good run ;) [15:23:59] bearND /cc dbrant|brb mdholloway: can we have a short discussion about the future of JsonPageLoadStrategy sometime? at one point, i think we had multiple strategies but now it seems we just have JPLS. i was hoping to discuss how we could improve this class while still keeping support for MW-only (no Content Service) wikis working [15:26:43] I can meet in the batcave in 4 minutes, 30 after the hour [15:29:17] bearND: that'd be cool. i'll see you there momentarily /cc dbrant mdholloway [15:29:30] niedzielski: bearND: be there in 1-2 mins [15:32:52] niedzielski: bearND: boo, left my headphones at home (luckily only about 5 mins away) -- you guys can fill me in at standup if you don't mind [15:43:04] mdholloway: we're gonna reschedule [15:53:12] joakino jdlrobson leads meeting [16:07:52] 13 crashes so far in the new beta. here's one: "OutOfMemoryError: OutOfMemoryError thrown while trying to throw OutOfMemoryError" [16:20:50] haha [16:51:21] niedzielski: that error is hot [16:54:37] phuedx: it threw me [16:57:10] niedzielski: that's an awesome error [16:57:11] so meta [16:57:24] i want to retweet it [16:58:12] no one has tweeted it yet, so you just want to tweet it joakino haha [16:59:10] it's exceptional [17:00:02] i want to retweet niedzielski [17:01:47] 👍 a tweet in the hand is worth two in the bush [17:57:26] phuedx: is reading-web-staging-3 functional? what's the url? [18:24:42] bmansurov: normal url, reading-web-staging.wmflabs.org [18:24:54] the proxy was redirected to -3 [18:24:57] gonna head out, bb! [18:25:08] joakino: cool thanks [18:25:11] later [18:47:24] jdlrobson: i don't see timo's changes on staging [18:47:34] still shows the spinner and border [18:56:36] bmansurov: hey you there? [18:56:40] im having issues with reading web staging [18:56:58] I switched branch on cd /srv/vagrant/mediawiki/extensions/MobileFrontend/ [18:57:09] but it's not being reflected in Special:Version [19:02:56] phuedx: still around? did you setup a new repo? [19:05:49] ahhh now it is ssh reading-web-staging-3 not 2 [19:06:40] jdlrobson: did you use the new staging env? [19:07:07] jdlrobson: yes, that's right [19:07:36] still doesnt seem to be reflected in the code though hmmm. [19:08:00] I've checked out 6725fa037ae529e6052c5d64ad75ac889fc1bccb [19:08:13] * jdlrobson but it's telling me head is something else on http://reading-web-staging.wmflabs.org/wiki/Special:Version [19:21:31] jdlrobson: ok let me take a look [19:32:22] jdlrobson: having looked at it, I'm confused too. I can't even figure out how to restart the webserver. apache2 isn't installed. [19:32:56] neither is nginx [19:43:56] Hi, if anyone has read T135571, I'm wondering if the server side change has been deployed and I am ready to perform QA on it since it is still under QA signoff. [19:43:56] T135571: [BUG] [Content Service] Tapping random causes an unknown error sometimes - https://phabricator.wikimedia.org/T135571 [19:46:40] bmansurov: which browser you using? [19:46:49] dr0ptp4kt: chrome [19:46:58] bmansurov: for the beta hovercards on / user used modal to suppress [19:47:03] bmansurov: hmmmmmmmmmmmmmmmmm [19:47:07] bmansurov: one moment [19:47:20] bmansurov: i was testing in firefox [19:47:28] bmansurov: you want to check in ff and i'll check on my side in chrome? [19:47:30] dr0ptp4kt: ok let me test it on ff [19:48:30] dr0ptp4kt: working on ff too [19:48:46] bmansurov: you using network inspector and live http headers or something else? [19:48:55] dr0ptp4kt: just logging the info [19:49:06] bmansurov: oh, console? [19:49:07] dr0ptp4kt: but let me check the network [19:49:09] yes [19:49:22] bmansurov: hmm, curioser and curiouser [19:49:48] dr0ptp4kt: sendBeacon in action [19:49:50] :) [19:50:03] bmansurov: yeah. i had to remind myself to disable dnt [19:50:13] bmansurov: i guess that's a separate matter [19:52:01] bmansurov: ARGH [19:52:17] bmansurov: Network in dev tools in chrome isn't listing URLs accessed. HUH? [19:52:32] that's weird [19:53:01] bmansurov: that ever happen to you? [19:53:12] bmansurov: ugh, i guess i'll have to restart chrome, too [19:53:19] dr0ptp4kt: it's happening in ff [19:53:31] bmansurov: lol [19:53:34] dr0ptp4kt: dnt maybe? [19:53:52] bmansurov: no, i mean like it isn't showing *anything* [19:53:54] bmansurov: and dnt is turned off [19:54:02] dr0ptp4kt: make sure you select all requests [19:54:15] at the top of the network tab [19:54:25] bmansurov: d'oh [19:54:30] click on the filter if it doesn't show [19:54:36] bmansurov: yay [19:54:48] bmansurov: ok, now back to me checking in chrome [19:55:03] dr0ptp4kt: ok, i'm done checkin in FF. it's working [19:55:08] i see it in the network tab too [19:55:13] bmansurov: so you see network requests in ff? [19:55:16] yes [19:55:17] bmansurov: hang tight [19:56:26] nicholas-tsg: these changes aren't out yet. can you try in a couple days? [19:57:20] bmansurov: ok, i'm seeing network requests in chrome. lemme try ff after a restart [19:59:36] bmansurov: ok, now it's sending click events in ff, at least when i'm *not* logged in (i have all stuff deleted upon browser restart) [19:59:53] bmansurov: lemme do a fresh ff browser instance and log in to mw in ff [20:00:22] bmansurov: btw, i had 'all' network events on in ff [20:00:30] ok [20:00:31] both when i added the comment and now [20:02:21] bmansurov: and it's working for logged in on ff, too [20:02:36] bmansurov: ok, so it's just the dwellButAbandoned piece. i have no idea why it wasn't sending on ff [20:02:39] that's strange [20:02:43] yay, i'm about to finish the remaining part [20:05:24] ^ bearND not sure if you saw nicholas-tsg question [20:06:15] ok @niedzielski I'll check again in 2 days [20:06:29] nicholas-tsg: thanks :) [20:50:15] jdlrobson: thanks [22:13:39] bmansurov: sounds like https://gerrit.wikimedia.org/r/#/c/290749/1/resources/ext.popups.targets/desktopTarget.js is only adding a mouseleave blur event ? [22:13:49] it's deceptively big diff due to the chaining youve introduced [22:14:00] i think the commit message could be clearer. If you can do that i'll happily merge [22:18:36] jdlrobson: do you mean I should clarify that chaining was introduced? [22:18:46] bmansurov: i updated the commit message [22:18:50] and commented [22:18:51] let me see [22:19:04] disabled popups means a different thing to me [22:19:19] so the patch didnt meet the expectations i had when reading commit message [22:19:23] looks good [22:19:24] thanks [22:19:40] wait [22:19:45] body needs changing too [22:19:51] i'll do so [22:20:40] jdlrobson: done [22:52:10] reets nzr: so this zeplin stuff is pretty cool. i can get the exact specs! just a heads up that the width of this project appears to be half that intended https://app.zeplin.io/project.html#pid=56b3ef430f64a4eb0f1b9d27&sid=571529f2c7312f087aab1b4d [22:52:27] reets nzr: i think it should 480dp wide