[03:47:16] hi, may I ask questions about Wikimedia Mobile for Android? [18:13:14] brion, so the module I have problem with is mediawiki.language.data, it keeps saying unknown dependency even though I added 'targets' => array( 'desktop', 'mobile' ) in Resources.php [18:13:27] this worked for all the other modules [18:15:07] jgonera, IRL talking is not geeky enough?:P [18:15:50] it seems Brion's busy with something, I didn't want to disturb him ;) [18:27:11] jgonera: there's a 5-minute http cache timeout on the core resources manifest, check again :D [18:28:30] brion, on my local dev instance? anyway, I disable browser cache and always do Ctrl+Shift+R, and it worked immediately for all the previous modules I added (I was adding them one by one) [18:28:34] brion: I'm gonna run through the UUIDs I have again in the device registry so we have the most up to date and save ourselves yet another build :D [18:28:38] I'll ping you when I'm done [18:28:59] yo team. every two weeks i'll have to attend the ops meeting to give them a heads up about "major" deployments that they need to know/care about [18:29:03] today is that day [18:29:10] i'll mention * GeoDate rollout [18:29:13] GeoData [18:29:17] what else do we have? [18:29:27] MaxSem: awjr brion jgonera jcmish --^ [18:31:21] *login [18:31:37] *editing [18:31:53] *photo uploads [18:32:07] tfinc, I'd like to deploy PageImages ideally this week but other than a formal perf review it shouldn't be major [18:32:08] tfinc: alpha rollout [18:32:11] *psm [18:32:14] osm [18:32:36] watchlists [18:32:44] page history [18:33:00] api usage for uploads [18:33:05] k [18:33:08] OSM will not be withing the next 2 weeks [18:33:13] and account creation [18:33:43] tfinc, but could you ask them to assign a contact person for OSM? [18:34:02] MaxSem: this is short and long term [18:34:15] ah [18:34:33] its the first time that were doing this so we'll see what they need [18:34:36] i rather give more then less right now [18:35:58] hehe, we'll overload them [18:38:05] MaxSem: sure [18:38:15] thats the second piece of me attending this meeting [18:38:22] a quick vehicle to get what we need [18:38:29] what else do we need past an OSM ops person? [18:40:57] MaxSem: what's our GeoData roll out plan for Dec/Jan/Feb ? [18:41:14] i'm putting my notes for ops in http://etherpad.wikimedia.org/MobileOps [18:41:34] MaxSem: i know nothing about PageImages. where can i read up on it? [18:43:12] mmm, I'll need to populate https://www.mediawiki.org/wiki/Extension:PageImages [18:45:58] tfinc, this week: enable GeoData on the rest of Wikipedias, enable automatic updates via job queue (currently enwiki's queue is seriously overloaded so I'm wary about enabling it, but I won't wait forever for it to get empty) [18:47:56] tfinc, when the hardware will be ready: move one-server setup from yttrium, start trying it with replication [18:48:34] we'll announce it officially only after I'll be confident that it performs well under load [18:48:39] (all 6 servers) [18:51:09] hey [18:51:10] hi pubsez [18:51:20] pubsez can you repeat your question here? [18:51:48] I setup mobile frontend, works flawlessly [18:51:56] except I don't see s-maxage in the headers [18:52:15] hmmm it's been a long time since i've looked at that portion of MF code [18:52:22] the desktop version has it, is this by design [18:53:34] pubsez im looking through the code now [18:54:23] yeah it looks like we're not setting s-maxage at all [18:54:35] I have been looking at the code for the last few hours, maybe this will save some time, but it appears that OutputPage's setSquidMaxage must be called with the appropriate maxage [18:56:03] MaxSem: do you know anything about what pubsez is asking? [18:56:08] my understanding of the code is limited, but it appears that this function is not called, thus it does not set the s-maxage [18:56:59] Expires: Sat, 26 Jul 1997 05:00:00 GMT [18:57:11] kinda replaces s-maxage:) [18:57:13] i believe MW should be setting the s-maxage (if it's configured) automagically [18:57:53] you are right [18:58:09] that's what I would expect too, but I guess some code change caused this "bug" ? [18:58:56] I mean code change in the main MW, not mobile frontend [18:58:59] well, if you're looged in you're not supposed to receive cached content [18:59:12] I am not logged in [18:59:50] if you've ever been logged in, you should have a few leftoff cookies [19:00:02] now that I've cleared my ccokies, I see [19:00:03] Cache-Control: s-maxage=300, must-revalidate, max-age=0 [19:00:20] never logged in at all, but let me see [19:01:06] hmm, nope definitely it does not work [19:01:08] same, MaxSem [19:01:25] pubsez sorry is this on one of the wmf projects, or another project? [19:01:42] this is mobile fronted, I believe it is wmf project [19:02:32] "project" as in "wiki" [19:02:39] pubsez: i mean, what is the project you're looking at response headers for [19:02:47] wikipedia? [19:02:57] not, this is my private installation [19:03:08] wikipedia uses varnish for mobile fronted, so no issue there [19:03:29] well, varnish still respects s-maxage, and it still gets set in production [19:04:00] this sounds to me like it is not a problem with MobileFrontend but likely a configuration issue with your setup [19:04:37] you might be right, but because desktop seems to set the squid maxage, I am not sure if it is the setup or something else [19:04:47] let me check the wikipedia too [19:05:09] lack of s-maxage does not necessarily prevent caching [19:06:10] correct [19:06:50] but from here, I can not detect if there is any difference, because the servers reset the s-maxage to 0 in wikipedia [19:07:26] pubsez do you have client caching enabled in LocalSettings? [19:07:44] you mean useSquid? [19:08:25] for squid setup, desktop version works fine, only when I use useformat=mobile it appears that it does not set the s-maxage [19:10:31] oh, i see [19:10:45] you mean this option specifically? wgCachePages? [19:12:14] no, i cannot remember off the top of my head the steps needed to configure caching [19:13:10] i see, no, I am pretty sure about the caching settings, but it could be something else that I am missing of course [19:14:14] pubsez im not sure what the problem is and cannot really dig in to figure it out. but i know that cache control information gets sent in OutputPage, and of course can potentially be overridden elsewhere [19:14:39] OutputPage::sendCacheControl() is where the magic happens [19:14:59] i can't think of anything (or see anything looking quickly in MobileFrontend) that should override what typically happens [19:17:05] pubsez actually is looks like having cookies used for cache variance will cause s-maxage to not get set, which might be what you're running into [19:17:25] I see [19:17:33] that does explain the result [19:17:59] is there a way to disable those cookies, through configuration ? [19:17:59] this shouldn't necessarily prevent caching mind you, just public caching [19:19:10] sure, you can totally modify what cookies are used for cache variance - but we have specific cookies used for cache variance in MF for specific reasons [19:20:39] pubsez: there is a hook called GetCacheVaryCookies that you could use to override/modify the cookies that get set for cache variance. but depending on your use case, you should be careful modifying what gets set in MobileFrontend as you might want up with a polluted cache [19:21:30] thanks a lot awjr [19:21:36] no problem pubsez [19:21:57] for my use, the cache is public, no private caching is needed, so I will follow up on your suggestions [19:22:08] thanks again, I have to leave now [19:23:02] here we go: https://gerrit.wikimedia.org/r/39076 [19:30:07] preilly, could you review ^^^ for me from a performance point of view? two points of interest: [19:30:07] (1) SQL query is pretty trivial and shouldn't cause any problems: SELECT * FROM page_props WHERE pp_page IN (...) AND pp_propname='foo' [19:30:07] (2) API limits are much smaller than for most modules due to File::transform()'s worst-case performance (though avg is pretty nice) [19:43:57] MaxSem: were there other things besides MF changes that you wanted deployed during tomorrow's push? [19:44:20] ideally ^^^ if it looks OK [19:45:46] there is also some pre-antiquated FeaturedFeeds stuff I'd like to move forward with: https://gerrit.wikimedia.org/r/#/c/23108/ https://gerrit.wikimedia.org/r/#/c/23112/ [19:45:48] MaxSem do you want to manage tomrorow's deployment then, or would it be easier if I do it and you deal with PageImages? [19:46:10] I've no problems with deploying tomorrow [19:46:20] MaxSem: ok, awesome [19:46:37] MaxSem: what are we ready to tell Erik about OSM build out ? [19:48:59] tfinc, that I'm working on its puppetization, ETA depends on priorities as my plate has other stuuf in it, too [19:56:03] MaxSem: what else is on your plate right now ? [19:56:37] MF (dynamic sections et al) and GeoData [20:01:40] MaxSem: is there a story/bug about what's happening with dynamics sections now ? [20:02:23] https://bugzilla.wikimedia.org/show_bug.cgi?id=41569 https://mingle.corp.wikimedia.org/projects/mobile/cards/277 [20:03:48] New review: awjrichards; "This looks better to me now (though I haven't tested it yet). I was originally thinking we should ca..." [mediawiki/extensions/MobileFrontend] (master); V: 0 C: 1; - https://gerrit.wikimedia.org/r/38642 [20:06:55] MaxSem: how goes talking to volunteer and the event planners for an OSM event in Feb ? [20:07:27] brion: okay all devices have been checked and updated ready to build when you are! [20:08:22] whee [20:08:59] ehm? we haven't talked about engaging volunteers regarding the hackathon [20:14:19] 'The private key for "Wikimedia Foundation (iOS Distribution)" is not installed on this Mac.' [20:15:07] tfinc: looks like I still can't sign adhoc builds [20:15:36] brion: lame [20:15:39] need me to sing a build ? [20:15:55] yes, or else export your identity to me :) [20:16:32] brion: we don't know each other well enough for me to export my identity to you. i'm not that easy. [20:16:54] * brion wines and dines tfinc's private key [20:17:02] * tfinc knows this phising rouse of brion [20:17:32] nigerian prince needs to update his iPhone apps, yada yada [20:21:38] New patchset: Jdlrobson; "allow switching between watchlist reading view and changes view [beta]" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/38643 [20:46:52] brion: open question to you in https://docs.google.com/a/wikimedia.org/document/d/1NYfjwGp3hbNUC5O7HPEzeSxLquX1_5hycbst87iuHKM/edit [20:48:17] * yuvipanda pokes brion with a C++ [20:50:27] tfinc, I've updated the docs: https://www.mediawiki.org/wiki/Extension:PageImages [20:50:33] MaxSem: thaks [20:51:36] * yuvipanda pokes brion with a C [20:52:25] tfinc: can you literally poke brion about the C/C++ thing if he's in teh office? I'm excited about finishing this up and having it out [20:52:47] yuvipanda: you mean like i did with my message earlier but didn't get a response because he was out to lunch ? [20:53:10] * yuvipanda updates head with new information about brion needing nutrients [20:53:21] we have to feed our brion every few hours [20:53:25] tfinc: exactly like that :) [20:53:32] i thought brion survived on screens [20:53:40] * MaxSem left a comment [20:53:45] just don't get him wet … bad things happen then. or ask him to do all the code review … worse things happen [20:53:50] then [20:54:00] yuvipanda: he hasn't feasted on a new screen in weeks [20:54:15] New review: Siebrand; "L10n looks fine" [mediawiki/extensions/MobileFrontend] (master); V: 0 C: 1; - https://gerrit.wikimedia.org/r/38643 [20:54:30] yuvipanda: these days just a screen isn't good enough. he has to be fed "retina" screens to keep up with his needs [20:57:31] ah [20:57:34] needs indeed. [20:57:53] another comment left [20:58:52] MaxSem: replied to one, i believe tfinc would be able to reply to the other [20:59:01] what do i need to reply to? [20:59:17] my comments [20:59:30] yuvipanda, I was talking about *non*-cordova [20:59:43] MaxSem: 'Having experience with native Android app development' [20:59:50] MaxSem: under 'extra points for' [21:00:30] tfinc: also, my exams are over :) [21:00:36] i've one tomorrow, but i've decided i'm not writing it [21:00:37] so this is a vacancy biased towards apples? [21:00:38] yuvipanda: woot [21:00:45] MaxSem: correct [21:00:45] MaxSem: very much [21:00:53] okay [21:00:59] yuvipanda: lets use those brain cells for something more useful [21:01:09] tfinc: currently reforming the dispue resolution system on enwiki :P (https://en.wikipedia.org/wiki/Wikipedia:DRREFORM) [21:01:17] tfinc: but will get back on commons app tomorrow i think [21:01:25] need to fix up the build so we can get CONTINUOUS INTEGRATION! [21:01:41] tfinc: any progress on the mobile language stuff? [21:02:00] yuvipanda: you mean between me leaving on friday and me showing up on monday ? [21:02:04] nah [21:02:07] :) [21:02:10] oh, I forgot about weekends [21:02:10] i like my weekends [21:02:25] such a first world concept though :P [21:02:34] awjr_lunch: have you grabbed lunch already? [21:02:41] tfinc yep, forgot to update my nick [21:04:48] New review: awjrichards; "from preilly's email, it sounds like the redirect stuff may still be needed:" [mediawiki/extensions/MobileFrontend] (master); V: 0 C: -1; - https://gerrit.wikimedia.org/r/37600 [21:09:13] MaxSem: clarification - when you say 'OSM' you mean 'TileServer'? [21:09:19] yes [21:09:35] ah sweet :) [21:10:59] Change merged: awjrichards; [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/38641 [21:11:16] Change merged: awjrichards; [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/38642 [21:11:40] awjr, the thing is that the TestCanonicalRedirect handler was introduced before BeforePageRedirect so just doing it in one place should suffice [21:12:19] my proposal is to try it live on testwiki - otherwise we'll never be sure about anything [21:13:45] gnite folks [21:13:49] MaxSem: i think that makes sense. i previously had set up multiple domains on my localhost for testing this kind of stuff, but i lost that setup a while back due to a hard drive failure :( [21:14:21] MaxSem: perhaps if we have some time during deployment tomorrow we can try it out? [21:14:22] yeah, and you also need Squid+Varnish to test it fully [21:14:28] yup [21:14:29] good point [21:15:20] awjr: I would think it would still be needed [21:15:26] awjr: but you can try it out [21:15:45] yeah, right now it looks like we're only special casing random page [21:16:19] we don't need to send anything from mobile to desktop site these days [21:16:41] so a lot of special0casing can (and should) go away [21:17:41] MaxSem: that wasn't the issue [21:17:59] MaxSem: it was that search and random would redirect to desktop from mobile without that hook [21:18:19] MaxSem: brion created the hook for that purpose [21:18:40] I know. and now we can simplify this [21:19:21] essentially, I can't think of a condition where we need a HTTP redirect to the desktop site [21:19:34] MaxSem: it wasn't for a HTTP link [21:19:54] MaxSem: the site was auto redirecting users to desktop when they clicked on random or search [21:20:06] MaxSem: the user wasn't trying to go to the desktop site [21:20:07] and TestCanonicalRedirect has its own issues like not redirecting from Main%20Page to Main_Page [21:20:19] preilly, I understand [21:20:40] MaxSem: okay well whatever you do please test it very throughly on production [21:20:53] MaxSem: I have a strong feeling that things will break [21:20:54] that's what I'm proposing [21:21:20] MaxSem: okay great [21:22:29] jgonera: im getting a ton of failures too - once it's done we can compare output [21:22:53] awjr, ok, thanks [21:30:23] jcmish: retrospective! [21:30:26] hangout link is in the invite [21:30:29] etherpad: http://etherpad.wikimedia.org/mobile-retrospective-20121217 [21:30:32] MaxSem ^ [21:32:58] jgonera: time for retrospective [21:34:16] awjr: hangout link? [21:35:02] jdlrobson: it's in the calendar invite [21:35:09] jdlrobson: and here's the etherpad: http://etherpad.wikimedia.org/mobile-retrospective-20121217 [21:35:10] it's not for me [21:35:14] i have a weird problem with invites [21:35:18] this has happened before [21:35:37] ah it shows when i login as wikimedia [21:35:38] weird [21:36:05] awjr: MaxSem: I went ahead and moved 4705 (mingle ticket #286) back to In Testing [21:36:43] jcmish, okie [21:36:46] opps 42705 [21:37:26] http://etherpad.wikimedia.org/mobile-retrospective-20121217 [21:39:27] jgonera: http://etherpad.wikimedia.org/mobile-retrospective-20121217 [21:51:00] preilly, thanks for the review [22:00:28] and thats our scrum masters machine dying [22:01:01] sorry guys. awjr is rebooting [22:01:06] but it sounds like we were done? [22:01:24] one more item [22:01:26] code review [22:01:29] so not done yet Maryana_ [22:18:42] grabbing a bite back in one sec [22:21:29] New patchset: Jdlrobson; "only load photo module once [alpha]" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/39151 [22:28:04] MaxSem: quick question is there an api query that returns all the images by a given user? [22:28:14] https://en.wikipedia.org/wiki/Special:ApiSandbox#action=query&list=allimages&format=json&aisort=timestamp&aiuser=MaxSem&ailimit=10 [22:35:37] thanks MaxSem :) [22:35:55] welcome:) [23:28:40] New patchset: Jdlrobson; "add a donate image page [alpha]" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/39155 [23:35:33] https://gerrit.wikimedia.org/r/#/c/38643/ < awjr MaxSem would appreciate your input and merging on this patch set (3 commits) [23:36:39] * MaxSem looks [23:49:22] MaxSem, jdlrobson: looks good to me - max, unless you have objections i'll merge it [23:50:42] Change merged: MaxSem; [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/38643