[09:02:43] yo [09:40:56] morning [10:07:55] yo [14:24:07] back from dad's group [15:47:48] yo [16:22:11] i was curious about the work being done about integrating trending articles into mobile apps, specifically what is being done on the backend to decide on which articles are trending. Is this just the top1000 list from page views api or is there something more? [16:40:11] ebernhardson: might be best to email the reading-wmf mailing list with that question [16:40:17] some apps folk should be online though [16:40:28] coreyfloyd, dbrant ^ [16:41:24] ebernhardson: I'm out today but bgerstle_afk can fill you in when he is around. [16:42:17] i'm around. ebernhardson we're using the pageviews API mostly out of the box, w/ some add'l filtering based on the work done by stephen laporte on top.hatnote.com [17:03:49] joakino: joakino joakino joakino joakino joakino joakino joakino [17:03:52] joakino: joakino joakino joakino joakino joakino joakino joakino [17:03:52] joakino: joakino joakino joakino joakino joakino joakino joakino [17:03:53] joakino: joakino joakino joakino joakino joakino joakino joakino [17:03:53] joakino: joakino joakino joakino joakino joakino joakino joakino [17:03:53] joakino: joakino joakino joakino joakino joakino joakino joakino [17:03:58] PING MATRIX [17:06:11] lol [17:15:41] phuedx: Hi [17:53:18] nice spam [17:54:50] hey d3r1ck [17:56:39] my phone before a meeting with phuedx https://i.imgur.com/NEG51oP.gif [17:57:54] lol [18:15:31] jdlrobson: let's write a script? https://phabricator.wikimedia.org/T124959#2008221 [18:15:51] kaity: android standup [18:17:43] phuedx: script? [18:18:29] dr0ptp4kt: sprint 67 -- https://phabricator.wikimedia.org/project/sprint/board/1755/ [18:18:32] to generate the spreadsheets? yeh that would be good [18:18:58] mobile beta global traffic is interestingly high https://grafana.wikimedia.org/dashboard/db/mobile-2g?panelId=43&fullscreen [18:20:09] mbinder: Internet trouble, I'm reconnecting [18:21:28] mbinder: no update from me [18:21:47] kaity_: OK, I'll tell the guys to reach out if needed [18:22:25] niedzielski: halp: https://translatewiki.net/wiki/Special:ManageMessageGroups [18:22:33] how do i get past this stage again? [18:23:01] bgerstle: you have to resolve the conflict. i can batcave with you after standup [18:23:04] i remember last time there was some check box i didn't think to click [18:23:14] niedzielski: ok, i have standup at 1:30. ping you after? [18:23:25] sounds good [18:24:38] k thx [18:30:35] phuedx: am catching up with the references / jcrespo discussions [18:35:50] thank you, bd808 [18:38:46] phuedx: was 'ParserAfterTidy' not in the right place for the hook you wanted? [18:40:10] niedzielski: hey, have a sec for TWN now? [18:40:50] bgerstle: sure [18:41:29] k [18:41:34] bd808: yes -- but i got all confused this morning and couldn't remember if we have an mf-specific parser cache [18:41:51] sorry "mobile-specific" :/ [18:42:14] if we don't, then we'd pollute the cache (right?) [18:45:15] phuedx: so it seems with the references stuff it's feature flagged and disabled... so I'm not understanding why we could not merge on the beta cluster and run some tests [18:45:45] jdlrobson: merge on the staging node [18:46:20] "merge on the beta cluster" is an interesting idea [18:46:25] what exactly do you mean? [18:47:12] once you've answered, please ask whether or not you can actually do such a thing in -releng and -ops [18:47:17] and then ask //very// nicely [18:47:23] phuedx: yes i'm on that.. let me clarify [18:48:59] phuedx: so reading the ticket jcrespo seems to naturally be concerned about Q1: bloat to the db but he has suggested verifying on a smaller wiki before rolling out to enwiki. Q2: He's also keen to understand performance impact. Feature flagging the call to setExtensionData would allow us to enable this feature on the beta cluster and get some data to answer [18:48:59] Q2. Enabling on a small wiki would help us get a better answer for Q1. [18:52:37] jdlrobson: i understand what you're saying but that's not "merging on beta" [18:52:41] that's enabling on beta [18:52:58] :D [18:55:35] niedzielski: https://github.com/wikimedia/wikipedia-ios/pull/455 [18:56:58] phuedx: got it. Yup bad wroding [18:57:04] bd808: https://phabricator.wikimedia.org/T126137#2009013 seems you can fix that in inline styles (yay) [18:57:10] change is harmless for desktop [18:57:31] jdlrobson: it's that and more [18:57:46] there is a difference in having something in the codebase and disabled and not in the codebase ;) [18:57:51] but yeah, i see what you mean [18:57:59] yeh i commented [18:58:46] suggesting some steps to move this along. If that's okay, I think we can start the work next sprint around references, otherwise I'll look for plan B (and given MobileFormatter will be used looks like we can do that without worrying about splitting parser cache) [19:15:39] phuedx: looks like we have permission to experiment on beta clsuter https://phabricator.wikimedia.org/T125329#2009043 [19:16:02] unlike the navboxes i want us to be crystal clear about benefits of this approach before going ahead with it [19:18:01] jdlrobson: blessing, not permission [19:18:19] let's ask ops? :D [19:20:45] jdlrobson: hi [19:23:23] jdlrobson: i suggest an email to engineering/wikitech-l [19:27:14] phuedx: what are you hoping such an email would achieve? [19:29:57] jhobs: you forgot to update column on https://phabricator.wikimedia.org/T124317 [19:30:06] and fyi you have some npm issues [19:30:08] jdlrobson: thanks sorry I do that all the time [19:30:11] yeah I'm fixing them now [19:30:23] trying to figure out where the documentation types are documented tho [19:30:31] since they seem to be different from what we use in MF [19:31:06] found it [19:35:56] jhobs: gave you some review too [19:36:04] thanks [19:36:05] jhobs: be sure to install grunt [19:36:17] you can run `grunt test` in your core repository to pick these up [19:36:29] well my problem is i can't even install jsduck for some reason [19:36:34] haven't been able to for a while [19:36:43] what happens when you run npm install ? [19:36:58] jsduck is a ruby tool [19:37:06] no? [19:37:14] yeah [19:37:24] and I get operation not permitted despite running with sudo [19:37:46] jsduck shouldn't be needed any way to run `grunt test` [19:38:04] jhobs: you can always install everything else [19:41:59] jdlrobson: "unable to find local grunt" despite staring at the gruntfile [19:42:38] jhobs: what happens when you do npm install grunt-cli --save-dev [19:43:06] i'll try with the save-dev option, but I just ran that without it [19:43:13] what's the option do? [19:45:30] jdlrobson: Can we have a little PM chat please? [19:45:46] just for like 15 mins max and then we shall continue the rest by email [19:45:55] d3r1ck: what's up? [19:46:01] just ask :) [19:46:16] jdlrobson: turns out some local npm thing was fucking up [19:46:17] jhobs: it updates package.json [19:46:22] got it working now, thanks for the help [19:46:30] rm -rf node_modules is always my favourite method for those issues [19:46:52] i'll keep that in mind [19:47:13] i've been getting close to wanting to just reimage the whole device some weekend [19:52:08] bmansurov: left some comments and CSS fixes [19:52:14] for the overlay [19:52:20] ok thanks [19:52:20] bmansurov: looking good overall :) [20:14:33] jdlrobson: we've seen some good interactions around the references patch [20:15:07] we should tell folk about what we're planning on doing [20:19:24] (while also being bold, obvs) [20:20:39] jdlrobson, phuedx: either of you know if I have to take some additional jsduck-related action to stop it from failing since this is a new module? It's failing with an error about categories, but before I go digging into more core files I'd like some confirmation [20:21:18] I believe I just need to go add mw.viewport into a file tracking which modules to test, but I'm not certain [20:22:02] jhobs: yep, but i can't remember which file [20:22:11] phuedx: ok, I should be able to find it, thanks [20:22:24] qunit is also failing, but I think it may just be jenkins [20:33:10] dbrant: any opinion on the API options discussion? i think we're blocked if we can't store much data there [20:36:32] niedzielski: i'm processing it :/ [20:48:19] bmansurov: how do i get on staging [20:48:22] i can't seem to ssh into it [20:48:27] i was gonna add other languages for you.. [20:48:40] 1 sec [20:49:03] jdlrobson: ssh reading-web-staging-2.eqiad.wmflabs [20:54:31] bmansurov: http://reading-web-staging.wmflabs.org/wiki/Lang [20:54:34] BOOM [20:54:48] they're recognised at least now.. [20:54:53] not sure why they don't show uo [20:55:05] i think we need to add languages to the sites table [20:55:23] I didn't get a chance to find out which db the site is using [20:55:36] i dumped some data to the wiki database, but apparently it was the wrong database [20:55:39] (showing up in desktop) [20:55:51] http://reading-web-staging.wmflabs.org/wiki/Lang?useformat=desktop [20:55:52] http://reading-web-staging.wmflabs.org/wiki/Lang [20:56:18] weird [20:56:38] i think reloading apache will fix it [22:32:33] dr0ptp4kt: what's the purpose of mcc-mnc header enrichment in the apps as seen here? (1/2) [22:32:37] dr0ptp4kt: https://github.com/wikimedia/apps-android-wikipedia/blob/master/app/src/main/java/org/wikipedia/networking/MccMncStateHandler.java#L14-L36 [22:33:07] dr0ptp4kt: and should some of the logic change in light of our (almost) always using the canonical domain? [22:42:56] mdholloway: it has the same purpose as described in the mailman post. regarding the change, good question - i imagine you're wondering if we need to update something somewhere on varnish or api.php in order for the event logging events to make their way into the event logging database [22:46:11] dr0ptp4kt: i haven't yet read the "longer technical thread" alluded to there, but it looks like there are a couple of separate concerns intertwined there: the banner display as well as event logging. is that accurate? [22:46:42] mdholloway: the api.php code that processes x-mccmnc is https://github.com/wikimedia/mediawiki-extensions-ZeroBanner/blob/ad93217f9a8a184dd171cf68052bc3329749f913/includes/MccMncLogging.php . iirc that server side code would only ever run for api.php requests sent to mdot with the current config. yurik do i understand that correctly [22:46:45] ? [22:49:22] (reading now, i see it actually isn't that long ;) ) [22:57:58] dr0ptp4kt: here's a better question. enriched headers are wanted for all cellular-based connections, and not only those from a potentially zero-rated IP address, correct? so in practical terms we'd want to remove the check for a mobile domain in shouldSendHeader. [22:58:26] dr0ptp4kt: (with whatever changes that would require on the server side for event logging purposes.) [23:01:17] mdholloway: yes, or maybe we should stop sending that header. want me to email about that? it would be easier to just stop sending it [23:03:43] dr0ptp4kt: sure, that sounds good, thanks. i've been updating a patch to implement our mobile-domain-for-zeroconfig-api-only logic but got hung up on this piece. will keep an eye out for your email. [23:40:54] dr0ptp4kt, i guess so, haven't looked at that code in ages )