[00:01:03] yup, thanks superm401 [00:04:26] Project MobileFrontend-mac-iphone build #340: UNSTABLE in 11 min: https://wmf.ci.cloudbees.com/job/MobileFrontend-mac-iphone/340/ [00:04:26] * jgonera: Restore logo-copyright-en.png [00:04:27] * jgonera: Don't call closeNavigation() on touchend if no menu closed [00:04:38] Project MobileFrontend-mac-ipad build #337: UNSTABLE in 10 min: https://wmf.ci.cloudbees.com/job/MobileFrontend-mac-ipad/337/ [00:04:39] * jgonera: Restore logo-copyright-en.png [00:04:39] * jgonera: Don't call closeNavigation() on touchend if no menu closed [00:14:36] awjr: you about? [00:14:43] tfinc: yup [00:15:55] would you be up for a hangout between howie and myself. he's got some ideas for the presentation ? [00:16:48] awjr: --^ [00:17:00] tfinc: can you give me 5 mins? [00:17:04] awjr: sure [00:17:10] cool [00:24:01] tfinc: ok ready [00:25:34] awjr: https://plus.google.com/hangouts/_/bc09e75a8d2161781a3b011d0c3e3afeca0b42c3?authuser=0&hl=en [00:45:47] Scap is done [00:47:23] awjr, which scenario do you use: http://www.mediawiki.org/wiki/Wiki_family ? [00:55:55] jgonera: looking [00:56:57] jgonera: none of them [00:57:21] i basically have two separate wikis using separate code bases and separate databases on the same virtual machine. both accessible from the same domain, but different paths [00:57:36] it's pretty hacky [00:57:43] but it works! [00:57:58] and i think i have the centralauth tables living in one of the two databases (i think in my default wiki instance) [01:00:23] hm [01:02:59] I think I need to set it up with two different subdomains though, else I'm not going to be able to check if I actually fix the problem [01:03:05] oh, well, back to reading [01:07:04] that makes sense [01:07:16] good luck [01:13:02] ok folks im outta here - see you tomorrow! [01:14:25] bye awjr [01:16:20] New patchset: JGonera; "Update cc-by-sa image and menu images" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/53910 [13:43:27] hey ragesoss [13:43:33] hey [13:43:42] ragesoss: what machine do you have? [13:43:49] also do you have the android sdk? :D [13:43:50] all the machines [13:44:02] full sdk, or just adb? [13:44:08] ragesoss: just ab [13:44:10] adb [13:44:14] i just want adb logcat output [13:44:14] yes [13:45:09] ragesoss: can you pastebin that? raw adb logcat output as you try to upload [13:45:24] walk me through the steps? [13:45:39] ragesoss: in your phone settings [13:45:43] ragesoss: do you see 'developer'? [13:45:49] debugging is on. [13:46:27] ragesoss: okay [13:46:38] ragesoss: so 1. connect phone to computer with usb cable [13:46:42] doen [13:46:45] done [13:47:12] ragesoss: on command line, execute '/platform-tools/adb logcat' [13:47:18] and you should see a stream of running text [13:47:43] do you? [13:47:50] if you are on latest android you might have a prompt on your phone [13:47:57] asking if it is okay to grant this adb connecton [13:48:45] stuck on waiting for device... [13:49:26] ragesoss: you have USB debugging on? [13:49:34] ragesoss: do you get a notification saying how it is connected? [13:49:55] yes. lemme try on the other computer [13:50:03] I think it has a cleaner setup [13:50:06] ah ok [13:52:28] gotta go for a bit... [13:52:42] ragesoss: ok [13:52:45] ragesoss: ping when back [13:54:53] back. grrr still not working [13:55:23] I have the debugging enabled notification. [13:55:30] ragesoss: okay. [13:55:36] ragesoss: adb still says 'waiting'? [13:55:56] ragesoss: try adb kill-server, then adb start-server? [13:55:58] then adb logcat? [13:56:48] got it. [13:57:01] ragesoss: works? [13:57:05] yes [13:57:40] now what? [13:57:58] wow my phone is busy under the surface. [13:58:47] ragesoss: sweet [13:58:50] ragesoss: yes it is :) [13:58:55] ragesoss: and this isn't full level usually :) [13:59:18] ragesoss: are you on a linux / os x machine? [13:59:28] linux [13:59:35] ragesoss: you can use 'tee' [13:59:52] ragesoss: 'adb logcat | tee outputfile.log' [13:59:58] ragesoss: will show you output as well as send it to the log [14:00:32] k. what's the normal way to kill the current terminal-only output [14:00:33] ? [14:00:37] ragesoss: so, 1. download nightly, install, 2. start that command, 3. try to upload, 4. once done (success or failure), ctrl-c terminal to stop command 5. send me outputfile.log [14:00:40] ragesoss: ctrl-c [14:05:12] hey, it's working! [14:05:18] ragesoss: oh? [14:05:20] ragesoss: wow :P [14:05:42] you got the two crash reports, I assume. one from yesterday's nightly, and then one from the new one. [14:06:11] but I reopened the app, tried to catch the errors in a logcat... and they didn't happen. Uploaded a set of 3 successfully. [14:06:12] ragesoss: did they fix themselves? [14:06:29] ragesoss: that is highly concerning. something wrong with the API return values :| [14:06:34] the crashes were unrelated to multi-upload. [14:07:51] ragesoss: ah [14:08:01] ragesoss: yes, still concerning. very weird error [14:09:55] okay, worked with 3. but failed with 15. [14:10:04] will try to get a logcat. [14:10:04] ragesoss: faild in the same way? [14:10:06] okay [14:10:22] yeah, same way [14:17:53] ragesoss: that's too intermittent. ugh. [14:18:13] what, the log? as in you need a higher level of detail? [14:18:32] ragesoss: no no, the failure [14:18:41] I just saw the logs [14:18:48] failed every time but one, right? [14:18:54] grepping through [14:31:12] hey brion [14:31:16] heyyyy [14:32:38] ragesoss: interesting. the service just seems to die, for no reason [14:32:45] once it has uploaded the first one [14:33:08] it uploads one? I thought they *all* fail. [14:33:23] ragesoss: well, *one* seems to upload, at least till the end/ [14:33:34] but ultimately fails. [14:33:38] D/Commons (30710): Uploaded 4628447 of 4628447 [14:33:45] I/ActivityManager( 442): Process org.wikimedia.beta.commons (pid 30710) has died. [14:35:04] ragesoss: i need to find out why it is dying [14:35:06] looking at it now [14:35:08] ragesoss: it doesn't die for me [14:35:14] ragesoss: do you have other things running in the bg? [14:35:27] nothing specific. [14:35:37] hmm [14:35:51] I'm sure there are things syncing and such, but no other active background apps I'm specifically using. [14:35:57] yeah, that's okay [14:36:04] possibly related... [14:36:17] I'm usually using Quasseldroid for IRC. [14:36:26] shouldn't be a problem [14:36:33] unless you have like 4 games or something :P [14:36:35] using Gallery intensively often makes it disconnect. [14:36:36] you're not on a low end android [14:36:48] like, memory gets too full of Gallery stuff or soemthing. [14:37:22] yeah, android kills those things [14:37:23] but, I've had multi-upload fail with that running in the background in cases where it didn't disconnect. [14:37:27] but this service is specially marked to not be killed [14:37:29] yeah [14:37:31] this isn't that [14:37:37] this is my multi-threaded code failing somewhere [14:38:14] threading is the devilllllll [14:38:29] or should i say [14:38:41] threadeviling is ding the [14:38:43] :D [14:40:23] so i had to remind the people working on making file upload controls work in firefox os that many web apps expect a file extension on their filenames [14:40:29] they wanted to make the filename just "blob" [14:40:51] :D [14:40:57] extensions are so DOSy [14:41:18] extensions suck. but NOBODY trusts the mime type on file uploads ;) [14:42:02] :D [14:42:03] indeed [14:42:13] brion: i was fixing extensions on the android app a while back. it sitll sucks [14:42:25] yups [14:42:26] mediawiki expects extensions. android doesn't have extensions [14:42:33] and no extremely reliable way to get mimetype [14:42:36] *sniff* [14:43:01] i also noticed that while mediawiki api uploads can sometimes detect the type, upload wizard totally doesn't and rejects files with no extension [14:43:15] brion: it almost always detects the file type [14:43:23] brion: notice android having .jpeg rather than jpg :P [14:43:29] heh [14:43:47] * brion blames DOS for .jpg [14:43:51] brion: how's the account creation coming? [14:43:54] remember .htm files? [14:44:16] YuviPanda: it seems to work, i'm hoping for some productive feedback :) [14:44:24] yes, I do :P [14:44:58] heh i remember whinging that i couldn't compile Java applets on 16-bit windows because it didn't support long file names and the ".java" or ".class" extensions [14:45:21] :D [15:22:12] ragesoss: around? [15:23:57] yep [15:24:08] YuviPanda [15:24:55] ragesoss: do a test. share multiple files, but *do not* open the my contributions view [15:24:59] ragesoss: just monitor from the notifications area [15:25:01] ragesoss: don't tap it [15:25:05] see if that succeeds? [15:25:13] ragesoss: does that make sense? [15:25:17] it does not, based on previous experience. [15:25:27] nothing appears in notifications area until i open thea pp. [15:25:32] except for about a second. [15:25:32] ow [15:25:33] ok [15:25:37] right. ok [15:25:54] sad part is that it completely completes fine for me. eugh [15:26:01] let's try another device, just to make sure [15:26:04] brion: ahey? [15:26:22] ? [15:26:36] brion: can you test bit.ly/commons-beta? [15:26:44] brion: specifically, the multiple upload feature? [15:26:50] brion: works for me, doesn't for ragesoss, so I'm trying to nail down the problem [15:26:55] moment [15:30:32] brion: what devicE? [15:31:56] hmm [15:31:58] nexus 10 [15:32:04] how do i do multiple selection in gallery? [15:32:20] brion: press and hold the first one [15:32:36] once it vibrates, you can tap additional. [15:32:46] it just highlights then ends the highlight. [15:32:58] no vibrate, no context actionbar [15:33:06] brion: press and hold first one for 'long enough'? [15:33:57] crazy. i took some more screenshots and now it works [15:33:59] O_o [15:34:08] :P [15:34:17] brion: how does it look? [15:35:50] YuviPanda: looks ok at first, but it crashes during upload. just sent you a crash report [15:35:58] this was with Picasa images, had a read permission error [15:36:04] lemme try with local screaanshots [15:36:18] wat that is so weird [15:36:33] heh [15:37:46] local screenshots appear to have uploaded, but there's no final notification telling you it's done [15:38:00] brion: how many did you upload? [15:38:12] https://test.wikipedia.org/wiki/Special:NewFiles <- 3 screenshots all went through [15:38:19] https://test.wikipedia.org/wiki/Special:NewFiles <- of 3 bay bridge photos only one got through [15:38:19] brion: did you open the app up again and confirm? [15:38:31] weird [15:39:15] all of mine are going through [15:39:16] aargh [15:40:24] ragesoss: can you try bit.ly/android-commons? (new build) [15:40:32] [WikipediaMobile] brion pushed 2 new commits to master: http://git.io/0po2DA [15:40:32] WikipediaMobile/master 2db9a8a bbondy: Bug 45782 - Windows 8 app: Support context sensitive sharing of HTML [15:40:32] WikipediaMobile/master bf47278 Brion Vibber: Merge pull request #322 from bbondy/master... [15:40:47] Project WikipediaMobile - Nightly builds build #498: SUCCESS in 16 sec: https://integration.mediawiki.org/ci/job/WikipediaMobile%20-%20Nightly%20builds/498/ [15:40:47] netzen: Bug 45782 - Windows 8 app: Support context sensitive sharing of HTML [15:40:51] brion: thanks! [15:41:03] bbondy: thanks for the patches :D [15:41:17] np I'll work on some more soon as I have time [15:41:53] awesome [15:42:42] brion: the picasa ones didn't go through but the local screenshots did, right? [15:42:57] web browsers in metro are a hybrid (both winrt and win32) so it's nice to work on a real native win8 app :) [15:43:56] YuviPanda: right [15:44:05] that's fine [15:44:11] brion: can you try uploading 10 or 15? :P [15:44:18] brion: also how did the UI look for uploading? [15:44:24] not complete, of course, but the concept? [15:44:25] bbondy: heh… wish i could install firefox on my surface RT though ;_; [15:44:33] YuviPanda... working! [15:44:36] YuviPanda: interface looked fine [15:44:41] heh me too :) [15:44:49] ragesoss: bingo! [15:44:55] i'm gonna have to take more screenshots; i wiped this device last month so it doesn't have a lot of pics on it :D [15:44:58] ragesoss: try again? [15:45:02] I think you can jailbreak it but that's no good for the general population [15:45:03] brion: haha :P [15:45:19] YuviPanda:it's on #5 of 16. [15:45:26] ragesoss: sweeet. let's wait :D [15:45:38] maybe they'll only have surface pros in a couple years then problem solved :) [15:46:41] YuviPanda: it failed at #7. [15:46:50] ragesoss: same way as before? [15:46:52] possibly related to me clearing the notification area. [15:47:08] ragesoss: you were able to clear this? :| [15:47:12] ragesoss: this should be unclearable! [15:47:24] is it clearable? [15:47:24] yeah, same as before... when I reopened app, it stays stuck on 'starting' for the one it left off on. [15:47:33] but all remaining ones are Failed. [15:47:43] YuviPanda: it's not now. [15:47:48] ragesoss: but was it? [15:47:49] YuviPanda: with 13 images it's getting a little sluggish typing in the title [15:48:03] but it was when initially populated from the gallery upload, without directly opening the app. [15:48:23] lemme retry to confirm that behavior. [15:48:31] ragesoss: yes, that would explain some things. [15:49:17] YuviPanda: hmm... failed immediately this time. [15:49:59] YuviPanda: not clearable this time, and it's working. [15:50:13] so it is intermittently clearable and not clearable [15:50:26] I did the clear all button, so maybe it was a different notif that was clearable, and it just crashed at the same time. [15:51:04] YuviPanda: 13 screen shots all uploaded ok, only problem was the sluggishness on typing title [15:51:06] possible. but if the notifications are clearable then that explains everything [15:51:12] ok i gotta get ready to head into the office, see ya in a bit [15:51:15] brion: okay, I'll need to work on making that faster [15:51:48] YuviPanda: when it is working, they don't seem to be clearable. When it doesn't work, they typically clear of their own accord. [15:52:04] sure, but typically clear of own accord also applies when they crash [15:52:13] right. [15:52:15] so we dunno if they crash because of clear or clear because of crash :P [15:52:33] I don't think I crashed it by clearing. I suspect that was just another app that I cleared and it happened to crash about then. [15:53:08] Hello, why the upload image button is not available on mobile devices, while appearing in mobile view on desktop version? [15:53:20] ragesoss: ah, ok [15:53:27] mido: which device? [15:53:45] iPod and iPad, both same result [15:54:10] mido: are you logged in? [15:54:27] mido: either way, jdlrobson / jgonera should be online in an hour or so, and will be in a much better position to help than me [15:54:27] yes [15:55:23] ok, I have also another issue with RTL versions [15:55:47] the button with donate image have an overlapping image of a camera with the text [15:55:55] on desktop? [15:56:02] the image should be on the other side of the button [15:56:04] yes [15:56:05] https://ar.m.wikipedia.org/wiki/%D8%AE%D8%A7%D8%B5:DonateImage [15:56:13] ragesoss: try bit.ly/android-commons again? [15:56:15] made a patch [15:56:15] I still can't see that button on mobile [15:56:33] mido: sorry, I won't be able to help much :( [15:56:39] jdlrobson / jgonera should be here in an hour [15:57:45] Thanks anyway :) [15:58:26] ragesoss: i'm heading out to eat. try it and let me know. [15:59:58] YuviPanda: k. seems to be working this time, on 13 files. nope, crashed at #5. [16:00:16] spontaneously, while I monitored in notif area. [16:01:53] ragesoss: augh. [16:02:47] ragesoss: try on your tablet? [16:03:24] ragesoss: and another logcat dump? [16:03:36] ragesoss: i'm going to head out for dinner. I just realized I hadn't eaten since yesterday afternoon [16:04:04] go! [16:05:11] ragesoss: but try on tablet and send me logcat dumps [16:05:15] and also see how it looks on the biggers creen [16:05:20] k [17:09:01] yo yo yo [17:25:02] tfinc, Maryana, awjr, updated stats at http://mobile-reportcard.wmflabs.org/ [17:25:22] dude! [17:25:25] awesome jgonera [17:25:29] jgonera what is '591'? [17:25:39] w00t! [17:25:39] where? [17:25:44] er and 46, 6, and 11? [17:25:50] those appear to be graph names? [17:26:01] well, they appear to the left of most of the graph names [17:26:01] oh, I guess 591 is total uploads in the last 30 days [17:26:12] I just move my cursor over the graph and look at the legend [17:26:38] yeah, Yuvi and I have to work on naming and legend [17:26:48] oh i see [17:26:57] oh good Chip finally took my advice and mailed out [17:27:02] people will now know whats happening [17:27:07] that's the last total, but I don't know why Limn shows it there by default [17:27:19] jgonera: this is great. thanks for doing this :) [17:27:24] tfinc and now maybe it will finally happen! [17:27:42] Maryana, we have to measure our main goal somehow ;) [17:27:45] tfinc: are they going with mac mini because chromebox has been too much of a pita? [17:27:51] yup [17:28:08] oh, one thing i just noticed: why does it say march 13-march 13? [17:34:21] tfinc: I added 'total uploads in last 30 days' to the reportcard [17:34:21] Maryana, I'm wondering the same thing, I have to ask someone from Analytics how to change that [17:34:27] stole jgonera's code so it reports web too [17:34:28] hey jgonera [17:34:34] thanks for fixing the SQL. [17:34:36] hi YuviPanda [17:34:38] * YuviPanda is ORMspoilt [17:35:01] no problem, I just noticed the data points jumped [17:35:17] I mean, when I moved the mouse cursor [17:35:40] ah, right [17:35:51] also SQL sucks :P [17:35:52] and that made me thinking that something was wrong [17:36:00] jgonera: me and ori-l were thinking of using MongoDB for some of these [17:36:05] apparently Limn will not assume 0 for missing days [17:36:13] might do an experimental switch at some point to see how it is [17:36:17] you mean for event logging? [17:36:39] YuviPanda: works fine on my tablet, and the UI is good. [17:36:51] 101 pics uploaded successfully. [17:36:52] ragesoss: so it's just your phone that is causing problems? [17:36:56] ragesoss: the service is dying when it shouldn't [17:37:03] * YuviPanda wonders if it is a cm problem [17:37:05] YuviPanda: file size related? [17:37:11] ragesoss: no, definitely not. [17:37:17] I think all the pics from the tablet were somewhat smaller. [17:37:17] I make it a point to *not* read files into memory [17:37:22] YuviPanda, it seems like a good idea to try out, because now when we change the schema, we would have to join multiple tables to retain historical data [17:37:30] jgonera: yup, that sucks [17:37:45] plus Mongo's website lists logging as one of the best use cases [17:38:30] jgonera: yup, but their 'preferred logging' method mentions bucketing on insert, rather than aggregation [17:38:35] but they do have a nice aggregation framework [17:39:20] YuviPanda: me likes [17:39:21] http://mobile-reportcard.wmflabs.org/ [17:39:58] YuviPanda, what do you mean by aggregation? [17:40:22] jgonera: bucketing on insert, rather than when we run our query [17:40:32] http://docs.mongodb.org/manual/applications/aggregation/ is their 'aggregation' framework [17:40:43] which we'll have to use to get data in the form we want [17:40:43] so [17:41:09] tfinc: basic version of multiple upload is up. ragesoss loves the UI :) [17:41:11] (I think? :P) [17:41:24] YuviPanda: loves is a little strong. [17:41:27] it has performance issues. [17:41:28] YuviPanda: is that why we have so many new uploads as of yesterday ? [17:41:36] tfinc: no, those all go into testwiki :) [17:41:47] oh yes, I know, queries might be more confusing than in SQL [17:41:49] ragesoss: sure, but the concept is nice. [17:41:57] tfinc: I just uploaded a batch of 101 baby photos to test.wiki today. [17:41:58] although our current SQL queries aren't that simple either ;) [17:42:00] ragesoss: :P Was dog slow on 100? [17:42:09] jgonera: true, I bet the Mongo ones will be simpler than this [17:42:24] YuviPanda: wasn't that bad with 100, perhaps because they were small. [17:42:26] jgonera: ori-l suggested just loading everything into memroy and using numpy / pandas, but loading everything into memory makes me feel dirty :) [17:42:32] ragesoss: file size is not a factor [17:42:36] ragesoss: what was it bad with? [17:42:45] YuviPanda, I agree with you ;) [17:43:00] YuviPanda: my phone, with a medium number of ~5mb files. [17:43:10] ragesoss: medium number -> 20? [17:43:18] less [17:43:25] maybe 15 [17:43:33] the thumbs take a while to all load [17:43:46] ragesoss: are you on CM nightly? [17:43:48] and then if I scroll down, and back up, some of the previously loaded thumbnails are gone. [17:44:07] YuviPanda: yeah. same basic behavior on tablet, except not quite as slow. [17:44:21] ragesoss: interesting. all symptoms point to your phone having much less memory than I'd expect from a CM phone [17:44:31] (missing images usually means OOM errors) [17:45:09] tfinc: but I suspect that when it hits the play store, it *will* increase the number of uploads by a fair bit :) [17:46:46] YuviPanda: can we trim the labels on http://mobile-reportcard.wmflabs.org/ to not include the section headings ? [17:46:53] YuviPanda: free ram is kinda low. [17:46:54] like 'unique-uploaders' [17:46:58] 166MB [17:47:02] tfinc: yeah, I should get on that [17:47:07] tfinc: debugging the app with ragesoss [17:47:09] will do after [17:47:10] k [17:47:44] lol Maryana i like that you added an appendix to the preso [17:48:44] hehe, that wasn't me! [17:48:46] maybe howie? [17:49:01] lol [17:49:04] i'd buy that [17:49:14] hey awjr Maryana [17:49:22] whatsup jdlrobson [17:49:27] ill :( [17:49:33] boooo :( [17:49:39] that just means you had a good time in austin :) [17:49:39] that sucks dude [17:49:46] what Maryana said [17:50:03] that's ok, we have box jon [17:50:06] box jon never goes out [17:50:07] i think the problem was trying to do too much - early starts/lots of talks/lots of gigs/late nights/drinking not a good combination :) [17:50:09] lol Maryana [17:50:12] i love that photo [17:50:15] box jon stays at his desk all night [17:50:57] box jon doesn't check in much code though [17:51:00] does box jon contribute to the code review backlog? [17:51:19] one day he's going to have that killer checkin i guess... [17:51:32] Maryana: on the plus side I had an amazing idea last night to contribute to photo uploads! [17:51:41] I'm adding it now to your wiki thing [17:51:42] sweeet [17:52:02] box jon is sadly bereft of amazing ideas [17:52:37] YuviPanda: whats the github repot for the dashboard? alolita wants to see it [17:52:53] tfinc: github.com/limn-mobile-data [17:52:56] err [17:53:01] github.com/wikimedia/limn-mobile-data [17:53:26] tfinc: ^ [17:53:59] thanks [17:54:49] Maryana: behold! http://www.mediawiki.org/wiki/Mobile/2013_strategy_planning#Other_upload_ideas [17:55:29] Maryana: I had this problem yesterday [17:55:55] https://en.m.wikipedia.org/wiki/Goatwhore needed a photo but i didn't see the call to action so it made it harder.. [17:56:16] omg you saw goatwhore?? [17:56:17] i love them! [17:56:21] i saw em in new orleans [17:56:22] :D [17:56:25] jdlrobson: you coming into the office today? [17:56:31] Maryana: i stumbled across them accidentally last night [17:56:34] tfinc: Monday! [17:56:37] they're great [17:56:37] k [17:56:44] when their lead singer is not strung out on heroin [17:56:47] my flight was changed from yesterday during the day to tonight [17:56:50] which, uh, happens time to time [17:57:25] Maryana: eek [17:57:42] that was one of the scariest shows i ever went to in new orleans [17:57:42] so awjr what [17:57:47] so awjr what is the latest on https://gerrit.wikimedia.org/r/#/c/52054/ [17:57:52] a fight broke out and somebody smashed a bottle over somebody else's head [17:58:00] Maryana: wow. Luckily I saw none of that.. [17:58:05] i could imagine it going that way though [17:58:12] true metal authenticity [17:58:16] jdlrobson: i dunno, i don't think my concerns have been addressed. [17:58:21] also awjr waiting for a reply to comment on https://gerrit.wikimedia.org/r/#/c/52663/ [17:58:42] jdlrobson: i wont be able to look at that until next week [17:59:37] awjr: i'm not sure exactly how to address those concerns - as i'm not sure why RL will not take care of that - maybe I should ask Tim Starling for some suggestions? I see he popped up on a config related bug [17:59:54] https://bugzilla.wikimedia.org/show_bug.cgi?id=44133#c5 < jgonera did you see this? It left me a bit confused as we don't surface these variables on desktop do we? [17:59:56] jdlrobson: i dont have time to get into detail there, but the short answer is caching [18:00:14] the bits cache does NOT get varied by device [18:00:40] so setting device-specific variables in PHP that get passed to the JS is not going to work [18:01:21] erg [18:01:23] jdlrobson: tim is less likely to know about RL-specifics. check with trevor, roan or timo [18:01:57] New review: Jdlrobson; "Krinkle/Trevor do you have an ideas on how best to achieve this?" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/52054 [18:29:23] ragesoss: very possibly a memory issue [18:29:24] D/dalvikvm(13237): GC_FOR_ALLOC freed 15K, 6% free 18872K/20024K, paused 17ms, total 18ms [18:29:28] 6% free when it dies [18:29:35] 29% right after [18:29:52] YuviPanda: are you using full-scale images in the grid view, or pre-scaling them down? [18:29:58] brion: I'm scaling them all [18:30:07] shouldn't be too awful then hmm [18:30:09] brion: problem here is that ragesoss is not encountering errors when I'm *displaying* the images [18:30:12] only when uploading them [18:30:15] which is what baffles me [18:30:19] funky [18:30:31] brion: come on over. were in r32 [18:31:07] brion: keep poking about the createaccount api. would be nice to see it go out in net deployment ;) [18:31:18] ok [18:41:09] jdlrobson, I think he means that this: https://gerrit.wikimedia.org/r/#/c/52270/5/includes/MobileFrontend.hooks.php runs also on desktop (it runs when the mobile extension is installed, not necessarily triggered) [18:41:28] mm... [18:41:58] jgonera: i'm refactoring lots of overlay code :) [18:42:14] with tests you'll be happy to hear ;-) [18:42:32] any chance it uses View or is object oriented in some way? ;) [18:42:38] yes of course [18:42:42] cool [18:43:01] although there is a lot of code untangling going on [18:43:10] I can imagine [18:44:56] although not quite as much as i thought there would be [18:51:35] got stacked overlays jgonera :D [18:56:08] YuviPanda: is http://mobile-reportcard.wmflabs.org/ getting updated automatically updated now ? [18:56:21] tfinc: nope. that RT ticket is still slowly winding it's way through ops [18:56:24] :( [18:56:36] I have access, but web drop, python-dev and virtualenv RT. [18:56:38] but is it moving forward ? [18:56:51] tfinc: there were responses on it yesterday. [18:56:54] why is this so hard to instrument ? [18:56:57] so i suppose yeah. [18:57:11] tfinc: it isn't. it's about 5 commands to someone who has sudo on that machine [18:57:50] thats what i thought. what is ops unhappy about ? [18:58:12] tfinc: they were concerned about exposing private data, but stat1001 has no private data, so shouldn't be a problem [18:58:40] jdlrobson, good, that's going to be useful for Vibha's short tutorial story [18:59:04] jgonera: the photo uploader is the only thing that's not compatible with the refactor i did [18:59:14] i'm just untangling that and i'll be done [18:59:19] :( [18:59:22] why is that? [19:00:12] i know why but it's hard to explain :) [19:00:17] btw, how did you solve the problem of templates not being loaded in tests? I addressed this in my test refactoring commit which still fails on PhantomJS [19:00:21] basically a reference to a non existant jQuery object [19:00:28] jgonera: stubs [19:00:58] we should be able to change templates without breaking tests (I'd argue..) [19:01:08] I'm wondering if we should merge it anyway and just stop using PhantomJS runner until we figure this out [19:01:20] I'm talking about this one: https://gerrit.wikimedia.org/r/#/c/50442/ [19:01:29] it solves quite a few problems [19:01:33] tfinc: got a hangout link? i dont see one in the invite [19:01:43] awjr: just added it to the invite [19:01:50] well, in ideal situation changing templates should break tests I guess [19:02:02] groovy [19:03:17] http://mobile-reportcard.wmflabs.org/ <- i'd better put out another iOS beta, it's gone flat :) [19:03:27] oh i'll also switch it to default to uploading to commons..... [19:04:22] jdlrobson, what do you think? ^ [19:04:32] brion: when will it go to app store? [19:04:53] brion: tfinc android crossed 300 :P [19:04:57] i think end of next week or so is target [19:05:00] beating both web and iOS :) [19:05:20] jgonera: i'd rather keep phantomjs - life has been so much better without it [19:05:31] without? [19:05:33] as far as i understand the only gain is not having to update the hooks when adding test files? [19:05:46] and it includes all the templates and messages in tests [19:05:49] life has been so much better with it [19:06:03] jgonera: i don't believe messages and templates are needed in the tests [19:06:32] they encourage bad habits of adding tests that rely on certain things to be in messages/templates [19:07:06] tfinc: https://rt.wikimedia.org/Ticket/Display.html?id=4730 if you think you can help push that along. [19:07:27] e.g. in case of templates $( M.templates.get('foo').render() ).find( '.content' ), true, 'Test the template foo contains a .content element' ) [19:07:36] brion, http://www.mediawiki.org/wiki/Extension:CentralAuth/Walkthrough [19:07:58] i will update that doc after lunch :D [19:08:09] rather strictEqual( $( M.templates.get('foo').render() ).find( '.content' ).length, 1, 'Test the template foo contains a .content element' ); // this is a bad habit as the test ties the changeable templates to a test [19:08:34] jdlrobson, how do you want to test UI events then? [19:08:55] if I bind something to a click of a button, but that button doesn't exist, how do I test if anything works? [19:09:13] stub it. [19:09:19] I could potentially agree about messages (although not 100% sure either), but not about templates [19:09:26] wrong, then you're not testing anything [19:09:41] if you overstub you're not testing your app, you're testing your stubs... [19:09:41] jgonera: well.. if the button is defined in the template itself maybe it shouldn't be [19:10:10] why? so now we'll get empty templates and we'll generate all the HTML using jQuery spaghetti like before? [19:10:12] e.g. use {{{button}}} rather than [19:10:27] I strongly disagree [19:11:01] I fell into this trap of overstubbing and overmocking when I was starting in TDD [19:11:21] I ended up with some tests that passed even though the app was broken [19:11:28] because they were no longer testing the app [19:13:57] jdlrobson, we can discuss it better when you're back in the office [19:14:38] i just don't see this as a big problem at the current moment in time. I think a command line phantomjs runner is much more useful to us right now [19:16:54] jdlrobson, I see it as a problem because I was unable to write any reliable tests for the photo related components because of it ;) [19:17:11] tfinc: milimetric is not optimistic about speed of getting the requests approved :( [19:17:26] and I had a problem of messed up dependencies and missing modules in tests more than once because of not using RL modules in tests [19:17:27] sounds like i need to get CT involved then [19:17:32] let me know what you need [19:17:51] tfinc: i'll shoot you an emaiL? [19:17:54] yes [19:17:58] tfinc: thanks, writing now [19:20:15] jgonera: can a View that inherits from another View run the parent's initialize function? [19:21:01] jdlrobson, no, I should probably do something about it, or you can try implement something, maybe google for "JS class super implementation" [19:21:21] mmm [19:23:58] depending on how much you need it right now ;) [19:26:46] tfinc: sent email. [20:01:31] tfinc: poke CT about that RT when you have the time :) [20:01:35] or anyone at ops, really. [20:13:04] ragesoss: try bit.ly/android-commons now? [20:15:24] ragesoss: *should* work now [20:15:30] some parts of it are more robust now [20:15:52] works so far [20:16:14] YuviPanda: How well do you reckon GPS location is going to work from an aeroplane? :D [20:16:28] Reedy: I'd like to try, actually [20:16:36] Reedy: it *might* work, because it doesn't need data [20:16:44] Yeah [20:17:02] my phoe is struggling to get a gps fix [20:17:13] it can see a few satellites [20:17:16] Reedy: are you in a plane? [20:17:22] yawai [20:17:33] i'm over new mexico currently [20:17:43] Reedy: you should take a detour to tucson [20:17:50] arggg these views are giving me massive headaches [20:17:52] success! set of 14 from phone. [20:17:55] Reedy: try https://play.google.com/store/apps/details?id=com.chartcross.gpstest&hl=en [20:18:00] ragesoss: yay :) [20:18:12] YuviPanda: I use GPSStatus for this [20:18:14] ragesoss: try a bigger one? and this time, adb logcat it so I can monitor the memory situation? [20:18:19] Reedy: ah. how many can you see? [20:18:27] 1/11 for fix/sats [20:18:31] Reedy: and do you see any GLOSSNASS one? [20:18:34] *ones [20:18:57] lets see [20:19:12] Reedy: also, ran SpeedTest? [20:19:51] * Reedy is uploading a photo from over Nevada [20:19:57] can't adb now. tonight I will. [20:20:09] ragesoss: okay. I'll assume your problems fixed for now then :) [20:20:12] Reedy: app? [20:20:13] :D [20:20:17] upload is teh slow [20:20:26] YuviPanda: using your released market version [20:20:30] sweeeeeet :D [20:21:35] https://commons.wikimedia.org/wiki/Special:ListFiles/Reedy [20:21:42] slow internet is being overly slow [20:21:56] Though, I am using my phone as an bluetooth access point.. [20:22:04] https://commons.wikimedia.org/wiki/File:Clouds_over_Nevada_in_an_American_Airlines_757-200.jpeg [20:22:14] saw that :D [20:22:15] whee [20:23:02] grey, grey, grey aaaannnnndddd grey [20:23:11] new mexico is a bit more to see [20:23:21] YuviPanda: what did you do to our Nexus S ? It's power button no longer works [20:23:25] Reedy: It has GPS co-ords [20:23:34] I'm going to have to replace it [20:23:36] so whee [20:23:36] but [20:23:36] 0 meters above sea level [20:23:37] :P [20:23:47] huh ? [20:23:47] Altitude 0 metres above sea level [20:23:52] Yeaah, that's so wrong :P [20:24:00] Reedy: are you sure your plane is still up? [20:24:03] YuviPanda: See where the co-ords suggest I was [20:24:08] looking [20:24:41] Reedy: right in the heart of Miami international airport :) [20:25:07] Haha [20:25:22] i suspect that's where gogo resolves to [20:25:31] Reedy: a particular plane, in the middle of 'island bar and grill' and 'simply books' [20:25:32] tfinc: Testing out the commons uploader from a plane ;) [20:26:34] not funny as this phone cost $400 [20:26:37] and now i have to replace it [20:27:00] tfinc: what? It was mostly fine when I sent it back. [20:27:22] you cant put it to sleep or wake it back up [20:28:10] tfinc: ow. Can you get the power button replaced? [20:30:10] i doubt the labor cost would be worth it [20:30:18] its simpler if i just replace it with a nexus4 [20:30:38] Labor cost would be 400$? [20:31:23] no, but i wouldn't be surprised if it was 50-100 for labor [20:31:41] o_O ok. [20:31:49] Reedy: it might also be your last known location. [20:33:41] YuviPanda: $30 here http://mobiletechvideos.mybigcommerce.com/nexus-s-power-button-repair/ [20:34:59] YuviPanda: Not when I was in SF [20:35:04] I've never been to Miami [20:35:09] tfinc: that sounds better. [20:35:32] tfinc: I think that might be a "known issue" [20:35:47] Seems my galaxy nexus ha[sd] a similar problem now my older sister is using it [20:35:51] * tfinc goes to grab lunch  [20:35:53] It fixed itself but... [20:35:54] Only just over a year old :/ [20:35:54] yes. it happens [20:38:12] yuvi, i can attempt video file uploads via multi upload [20:38:55] ragesock: yes. I know. I need to filter those out [20:39:04] ragesock: it'll fail, since .mp4 will be rejected :) [20:43:21] ragesoss: can you try bit.ly/android-commons again? made a few final tweaks [20:47:12] k. what happens to big set if I switch from WiFi to data in the middle? [20:47:38] (or switch data types if out and about?) [20:48:40] ragesoss: try? [20:48:45] ragesoss: I *think* some should fail. [20:49:03] ragesoss: you should be able to resume them, of course. but some will definitely fail [20:49:05] not all will [20:50:04] New patchset: Jdlrobson; "Rewrite createOverlay to use an Overlay View that is stackable" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/54069 [20:50:26] [android-commons] yuvipanda pushed 3 new commits to master: http://git.io/iV4fBA [20:50:26] android-commons/master 6a99255 YuviPanda: Upgrade version of UIL [20:50:26] android-commons/master cce2932 YuviPanda: Only leave foreground lock when absolutely no more uploads to do... [20:50:26] android-commons/master 95f6559 YuviPanda: Minor phrasing fix [20:50:29] ^ jgonera it's not perfect but it does the job - i'd be interested if you have any ideas to make it even nicer [20:52:05] brion: a luis reply! [20:52:16] [android-commons] yuvipanda force-pushed beta from e87d4a7 to 44ee682: http://git.io/tUPHIg [20:52:16] android-commons/beta 44ee682 YuviPanda: Single squashed commit for converting master into beta [20:52:53] \o/ [20:53:01] yep. fails one, continues next. [20:54:01] ui for retry via notif tap could be better. doesn't send you to the specific failed image. [20:54:12] Project Android-Commons (mobile) - Nightly builds build #136: SUCCESS in 2 min 5 sec: https://integration.wikimedia.org/ci/job/Android-Commons%20(mobile)%20-%20Nightly%20builds/136/ [20:54:12] ragesoss: yup. filing bug [20:54:12] * yuvipanda: Upgrade version of UIL [20:54:13] * yuvipanda: Only leave foreground lock when absolutely no more uploads to do [20:54:13] * yuvipanda: Minor phrasing fix [20:54:14] * yuvipanda: Single squashed commit for converting master into beta [20:55:31] ragesoss: I think taping should consistently open up details view. retry should be from there [20:55:37] so you can do renames too [20:55:42] and a bunch of other stuff [20:57:46] ragesoss: is this the new version? [20:57:48] the new new version? [20:58:36] brion: I see siebrand adding 'preliminary support for iOS app' into gerrit [20:58:37] \o/ [20:58:45] yep we're setting it up now :D [20:58:47] YuviPanda: if we were to kill the contact us form what effect if any would that have on the apps? [20:58:49] ^ Maryana_foods [20:59:09] jdlrobson: well, the app just redirects people to that URL [20:59:16] jdlrobson: so if you kill the URL they'll see a 404? [20:59:19] so if it 404s that's probably bad right [20:59:28] jdlrobson: if you are just removing it from the chrome it wouldn't affect the app [20:59:38] the app itself isn't affected - it just redirects them to the website [20:59:55] so as long as that URL does something sane, the apps should be fine [21:00:01] i'd say just direct app feedback to OTRS [21:00:06] i'm sure the noise will be low [21:00:07] not new new yet [21:00:20] ragesoss: try nightly, it is new new [21:00:23] Maryana: not possible.. the app currently just links to our contact page [21:00:32] app updates are always a problem [21:00:34] ragesoss: and also means you can have the commons app back to upload files you want to commons :) [21:00:40] yeah, app updates are never 100% [21:00:51] hrm, but the apps are still in beta, right? [21:00:56] no [21:01:01] this is the old school phone gap app [21:01:05] ohhh [21:01:06] sorry [21:01:13] One thing that might help is just remove the contact link from the footer [21:01:18] i always get confused btw commons app and wikipedia app :) [21:01:26] the other alternative is to replace it with a link somewhere else [21:01:43] (or make it redirect to the default wikipedia contact page) [21:01:49] we really just need to gut those old things and make new apps entirely :) [21:01:52] e.g. http://en.wikipedia.org/wiki/Wikipedia:Contact_us [21:02:27] Maryana: yup, in about... 6 months :P [21:03:04] * jdlrobson should buy a drink he's been in a coffee shop all day and only bought one thing ;-) [21:04:01] so Maryana your call - redirect to desktop Contact us, just break it off and give a broken contact us experience or leave it in but remove link to page [21:08:02] Maryana: I'm sending a mail to mobile-l [21:08:29] integration.mediawiki down? [21:09:39] YuviPanda: wish you were in SF next weekend i want to get you drunk :/ [21:09:55] jdlrobson: :( me too. soon enough, hopefully [21:10:38] ragesoss: looks like. eugh [21:11:05] ragesoss: use http://integration.wikimedia.org/ apparently [21:11:21] ragesoss: they're setting up a redirect now though [21:13:55] tfinc, there was like an half an hour delay in the interview, or more, so im still interviewing [21:14:03] jgonera: take your time [21:14:08] i'll reschedule [21:14:42] ok, config for the iOS app on translatewiki is pretty much done, it should merge and start showing up at some point :) [21:14:51] New patchset: Jdlrobson; "Remove contact us link from footer" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/54078 [21:15:58] brion: sweet :) [21:16:16] ragesoss: should work now, mutante setup redirects :) [21:16:32] New patchset: Jdlrobson; "Turn off section toggling on footer" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/54080 [21:19:21] grr, office internets [21:19:28] \o/ [21:19:30] hehe liefi [21:20:22] jdlrobson, so vibha is working on #400 (uploads dashboard 1st time user tutorial), and it looks like it's just going to be three images [21:20:38] she should have them done by end of day today [21:21:52] images?! [21:21:54] noooo [21:22:01] bad design - i18n problem surely [21:22:20] (i don't see how images would work without text..) [21:22:30] ever assembled ikea furniture? :) [21:22:52] we've been putting it off for the app for same reason :) [21:23:09] brion: also, you remember the initial android tutorial that pops up? [21:23:23] there'll be a small bit of text on each, but using icons to convey the message is actually better for localization, i think [21:23:28] you'll see [21:23:30] :) [21:23:31] YuviPanda: the one that's like "click here for all your apps" and crap? [21:23:35] brion: yes [21:23:51] brion: while that by itself is crap, the highlighting framework there is good [21:23:55] ah nice [21:24:00] brion: we can extract that and use that for ours [21:24:05] \o/ [21:24:35] brion: you've screenshots of multiple upload on nexus 10? [21:25:21] Maryana: what was the rationale for 'read in another language' when we did it (aside from the fact we weren't exposing that useful feature from the desktop site)? [21:25:52] YuviPanda: here it is with 3 images: https://upload.wikimedia.org/wikipedia/test/5/5a/More_screen_shots_and_stuff_-_10.png [21:25:52] looked similar with 13 [21:26:08] awjr: oh man, we had a host of usability issues with putting it anywhere else. people didn't even know we had the feature when we hid it in various secondary navs. [21:26:22] New review: Siebrand; "Trivial from i18n/L10n perspective." [mediawiki/extensions/MobileFrontend] (master) C: 1; - https://gerrit.wikimedia.org/r/54078 [21:26:33] Maryana: is it fair to say that it was a highly demanded feature from the broader community? [21:26:51] i think so - high demand from both readers and editors [21:26:56] word thanks! [21:27:00] np [21:27:45] brion: ouch. should be larger with just 3 [21:27:49] brion: my packing algo must suck [21:28:38] YuviPanda: well if they were much bigger they wouldn't fit above the keyboard :) [21:28:51] are they just at your maximum size? [21:29:15] ok i've outstayed my welcome in this coffee shop.. anyone need anything before i go off the grid for a bit [21:29:30] you offering us lattes jdlrobson ? [21:29:38] you have to come to austin to get some ;-) [21:29:45] :) [21:45:41] awjr: i have good/bad news for you [21:45:49] bad news : yeti's are going away [21:45:58] awww [21:45:59] good new : their getting replaced by something more awesome http://www.clearone.com/products_chat150 [21:46:05] :D [21:46:13] ooo are they *really* more awesome? [21:46:22] the polycom conf phones are supposed be awesome too :p [21:46:59] awjr: chip is going to invite you [21:47:02] to a hangout [21:47:03] to test [21:47:08] cool [21:47:50] i never liked those polycoms [21:47:53] polycoms suuuuuuck [21:47:54] but then i hate the telephone feature of phones in general ;) [21:48:10] arent phone just mini tablets ? [21:48:36] or are phablets giant tablones? [21:49:07] [Commons-iOS] siebrand pushed 1 new commit to master: http://git.io/U4BkXQ [21:49:07] Commons-iOS/master 6bc540a Siebrand Mazeland: Localisation updates from http://translatewiki.net. [21:49:28] whee [21:49:56] i really need to stop buying canned beer and soda that have the same color schemes on the cans [21:50:03] hahah :P [21:50:07] this is the nth time i've grabbed a soda and had it be beer instead [21:50:13] i mean, wayyy worse things have happened [21:50:21] in fact, it's kind of a win, but still. [21:50:41] that doesn't really sound like a problem [21:50:59] it's only a problem when you accidentally grab the soda instaed [21:51:34] or grab a soda for the road before hopping in the car... [21:52:46] i need to make some diagrams - i've used dia in the past but find it to be something of pita. do any of you know any good linux/os x diagram software? [21:52:57] brion, tfinc, jgonera? ^ [21:53:08] hmm [21:53:20] most folks seem to like omni graffle [21:53:32] omnigraffle is great on mac, but it's not free [21:53:40] http://www.asciiflow.com/#Draw is kinda awesome, but really only good for simple stuff [21:53:55] it can quickly become a pita [21:54:18] hehehe nice [21:54:27] oh damn omni graffle is really expensive [21:56:08] see if tfinc will let you expense it... [21:56:57] brion have you used dia before/ [21:57:14] i've tried it and gone "arrrrgh i hate this software" [21:57:34] * YuviPanda keeps all diagramming to paper [21:57:50] or finds someone else to do them :) [21:58:16] in a pinch you can use inkscape or libreoffice draw, but they're not always clean and nice [22:01:02] awjr, I usually use Inkscape and just draw things myself, but that might be a pain for complicated diagrams [22:02:01] yeah dia is nice if you have three weeks set aside to wrestle with it [22:02:04] someone uploaded https://commons.wikimedia.org/wiki/File:Darkness.jpeg [22:02:08] maybe i'll try inkscape [22:02:12] that's deeeeeeep [22:02:20] i dont know if i make diagrams enough to warrant omnigraffle [22:02:30] but thena gain, maybe i dont make diagrams very often because i've always used dia... [22:02:35] :D [22:02:59] lol YuviPanda [22:06:04] [Commons-iOS] siebrand pushed 1 new commit to master: http://git.io/c8ZMaw [22:06:04] Commons-iOS/master 8c3a2f4 Siebrand Mazeland: Localisation updates from http://translatewiki.net. [22:10:42] Yuvi: big sets still fail early. 2/350 successful [22:10:57] ragesoss: is this on the nightly? [22:11:04] yes [22:14:33] ok update looks good, lemme confirm it works :D [22:15:37] ah i'll have to maintain the message file entries in the project… i'll script that later [22:17:56] yay nl translation works [22:19:44] [Commons-iOS] brion pushed 2 new commits to master: http://git.io/Rb-lYw [22:19:44] Commons-iOS/master b1eade3 Brion Vibber: Switch settings and upload buttons on main screen [22:19:44] Commons-iOS/master 38035ce Brion Vibber: update message file list in project file [22:20:51] ragesoss: try bit.ly/android-commons? [22:20:51] you'll have to uninstall previous one [22:20:54] I changed one bit of code [22:25:29] still chokes at around #2/382 [22:25:51] blkajsdpifkljasdghk'klhasd [22:25:52] grrr [22:25:54] sigh [22:26:22] ragesoss: might be a case of cm nightly? [22:26:22] ragesoss: how old is the nightly? [22:28:09] 2 nights old [22:28:29] ragesoss: eek. sigh [22:28:42] me not being able to repro it sucks [22:28:53] ragesoss: file a bug report? I'll see if it gets fixed in the details work... [22:29:00] if not i'll get back to this and debug it to death [22:29:45] k. I will maybe flash a stock ROM and compare,later. [22:32:07] ragesoss: ok [22:55:19] Maryana: I need a wiki page that explains the privacy policy for event logging for the disable event logging story.. [22:55:37] disable event logging? [22:55:43] oh, you mean opt out [22:55:54] hmm [22:56:04] i don't think we have one specific to eventlogging [22:56:22] there's the general WMF privacy policy, which states that we can collect information on features usage [22:58:33] jdlrobson: http://wikimediafoundation.org/wiki/Wikimedia:Privacy_policy#Purpose_of_the_collection_of_private_information [22:58:58] thanks Maryana [22:59:11] np [22:59:33] no jgonera. sigh. [23:02:09] New patchset: Jdlrobson; "Story 422: Add do not track option to Settings" [mediawiki/extensions/MobileFrontend] (master) - https://gerrit.wikimedia.org/r/54099 [23:02:40] we lost jgonera again? [23:02:54] Maryana: why isn't box jon robson keeping an eye on him?! [23:03:25] there he is :) [23:03:33] New review: Jdlrobson; "Ori-l - can we make this work across the wikipedia sites (mobile and desktop) by picking a standard ..." [mediawiki/extensions/MobileFrontend] (master) C: -1; - https://gerrit.wikimedia.org/r/54099 [23:03:43] right i have a plane to catch [23:03:45] have a great weekend guys [23:04:08] expect more optimism this weekend YuviPanda ;-) [23:04:15] jdlrobson: :D [23:04:22] although i'm ill so i'll have to work hard [23:04:42] * jdlrobson waves [23:04:52] * jdlrobson teleports [23:12:24] awjr, looking at your LocalSettings now, you log your event logging to production? $wgEventLoggingBaseUri = '//bits.wikimedia.org/event.gif'; [23:12:40] ! [23:12:45] maybe... [23:12:51] ;) [23:13:14] what's the name of your wiki? [23:13:17] i wonder why that's there... [23:13:52] jgonera: the main one is mw_testing0 [23:15:22] ok, I don't see anything with such wiki in EL DB ;) [23:15:31] phew [23:15:56] i'm sure there's a perfectly reasonable explanation for that setting on my local instance, i just can't remember what it might be [23:46:22] awjr, do you know how MW uses memcached, how $wgMemc works? [23:46:39] there are quite a few uses for memcached [23:46:45] I mean [23:46:55] I almost configured CentralAuth on separate domains [23:46:56] but [23:46:58] primarily parser cache (i think) and database stuff [23:47:24] oh and i think session storage [23:47:27] the problem is that the token is stored in memcached using $wgMemc->set() and then read using $wgMemc->get() and it's not there [23:47:38] oops [23:47:46] if I do $wgMemc->get() in the same function as $wgMemc->set() it's still present [23:47:47] jgonera: session token? [23:47:58] the central auth login token [23:48:18] can i see some code? [23:48:19] e.g. centralauth:login-token:24a2220fcc86fd038228dea4d2361e3f [23:48:21] well [23:48:28] it's the CentralAuth extension ;) [23:48:41] lemme take a looksee [23:49:02] so i take it you have memcached all setup and configured locally? [23:49:12] CentralAuthHooks.php line 180 sets the key [23:49:26] SpecialAutoLogin.php line 28 tries to read it [23:49:38] awjr, yes, I copied your memcached config [23:49:43] but by then it's gone? [23:49:51] I checked, the key is the same but it's gone for some reason [23:49:53] yes [23:49:59] that is weird [23:50:35] an SpecialAuthLogin returns a page with "Token is invalid or has expired" instead of an image [23:50:46] probably the same thing's happening on testwiki (guessing) [23:51:01] jgonera: out of curiosity what happens if you increase $wgMemCachedInstanceSize? [23:51:43] I made it 10x bigger and it's the same [23:52:03] jgonera: if you disable memcache does it work? [23:52:19] CentralAuth depends on memcached [23:52:30] it won't work, it uses the $wgMemc directly [23:52:31] one sec - brb [23:53:40] if $wgMemc in your case a MemcachedPhpBagOStuff object? [23:54:00] sorry, my wife just got home from a trip [23:54:16] i need to look, i forget how that works [23:55:11] ok [23:58:27] ok sorry, i got hella distracted. so $wgMemc is not necessarily an interface for memcached, it supports other cache backends [23:58:38] im digging now to see where this stuff happens, it's been a long time since i've looked