[00:00:09] kaldari: for your benefit (join #-operations when you can): [00:00:10] alright, the order: manybubbles does his small sync, then rdwrer does the VE bit, then jgonera and kaldari do their bit [00:19:50] jgonera: https://wikitech.wikimedia.org/wiki/How_to_deploy_code#Case_1b:_extension_changes [00:23:44] (03PS1) 10JGonera: Regression: Make upload progress visible again [extensions/MobileFrontend] (wmf/1.23wmf16) - 10https://gerrit.wikimedia.org/r/117352 [00:24:20] (03CR) 10Kaldari: [C: 032] Regression: Make upload progress visible again [extensions/MobileFrontend] (wmf/1.23wmf16) - 10https://gerrit.wikimedia.org/r/117352 (owner: 10JGonera) [00:25:17] (03Merged) 10jenkins-bot: Regression: Make upload progress visible again [extensions/MobileFrontend] (wmf/1.23wmf16) - 10https://gerrit.wikimedia.org/r/117352 (owner: 10JGonera) [00:29:15] (03PS3) 10Jdlrobson: Hygiene: Remove unused function in QA tests [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117007 [00:32:52] (03PS3) 10Jdlrobson: Fix nearby pages [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117126 [00:33:04] jdlrobson: StevenW: yeh believe kaldari and jgonera doing it now [00:33:05] [4:21pm] StevenW: Kiss them both for me. [00:33:05] [4:21pm] StevenW: With tongue [00:33:15] tongue kiss for you guys [00:33:44] (03CR) 10Jdlrobson: [C: 04-1] "Needs rebase." [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/116912 (owner: 10Jdlrobson) [00:34:58] (03PS2) 10Jdlrobson: Hygiene: Kill that 'I am on the User login page' rule in QA [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/116900 [00:38:24] (03PS1) 10JGonera: Don't load mediawiki.ui.button by default in Minerva [extensions/MobileFrontend] (wmf/1.23wmf17) - 10https://gerrit.wikimedia.org/r/117356 [00:39:07] (03CR) 10Kaldari: [C: 032] Don't load mediawiki.ui.button by default in Minerva [extensions/MobileFrontend] (wmf/1.23wmf17) - 10https://gerrit.wikimedia.org/r/117356 (owner: 10JGonera) [00:40:23] (03Merged) 10jenkins-bot: Don't load mediawiki.ui.button by default in Minerva [extensions/MobileFrontend] (wmf/1.23wmf17) - 10https://gerrit.wikimedia.org/r/117356 (owner: 10JGonera) [00:54:48] (03PS5) 10Jdlrobson: Include revId in MobileWebEditing schema [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/116139 [00:57:32] (03CR) 10Mattflaschen: "This doesn't need to be done in SkinMobileWML, does it?" [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117334 (owner: 10Kaldari) [01:24:34] (03PS1) 10Dr0ptp4kt: Don't show search thumbnails for W0 unless on mdot. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/117364 [01:29:35] (03PS1) 10Dr0ptp4kt: Allow callees such as Wikipedia Zero JS to hide search thumbnails. [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117367 [01:31:10] jgonera, got a couple minutes to review these two ^ ^^^ ? [01:31:32] jgonera, if not today then tomorrow [01:32:45] correct, jgonera, if not today then tomorrow ;) [01:34:15] jgonera, i'm a little wiped out and am going to wrap up. not sure i'll physically be in the office, but rather online. ttyl [01:38:24] (03PS7) 10Jdlrobson: Hygiene: Group together main menu navigation tests [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/116692 [01:38:26] (03PS1) 10Jdlrobson: Hygiene: Address QA FIXME [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117370 [01:38:28] (03PS1) 10Jdlrobson: Hygiene: Kill RandomPage in favour of ArticlePage [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117371 [01:40:05] (03CR) 10Jdlrobson: [C: 04-1] "Use the value of mw.config.get( 'wgImagesDisabled' )" [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117367 (owner: 10Dr0ptp4kt) [01:55:17] (03PS2) 10Jdlrobson: Proper edit handling in case of redirects where page does not exist [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/116912 [12:36:16] (03PS1) 10Zfilipin: QA: Fixed Selenium tests that fail during login at Cloudbees [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117407 [13:15:02] (03CR) 10Siebrand: [C: 04-1] "Per earlier comment. This should be merged with icons instead of +/-. https://gerrit.wikimedia.org/r/#/c/108300/2/MobileFrontend.i18n.php" [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/108300 (owner: 10Theopolisme) [13:22:11] (03CR) 10Siebrand: [C: 031] "L10n reviewed." [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117012 (owner: 10Jdlrobson) [14:43:09] (03PS1) 10Amire80: Simplify a message [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117417 [14:55:34] (03CR) 10Siebrand: [C: 032] Simplify a message [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117417 (owner: 10Amire80) [14:56:10] (03Merged) 10jenkins-bot: Simplify a message [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117417 (owner: 10Amire80) [15:02:01] (03CR) 10Cmcmahon: [C: 032] QA: Fixed Selenium tests that fail during login at Cloudbees [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117407 (owner: 10Zfilipin) [15:02:38] (03Merged) 10jenkins-bot: QA: Fixed Selenium tests that fail during login at Cloudbees [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117407 (owner: 10Zfilipin) [17:15:20] (03PS1) 10Zfilipin: QA: Fixed Selenium tests that fail at Cloudbees [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117431 [17:16:39] (03CR) 10Cmcmahon: [C: 032] "paired with Zeljko on this" [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117431 (owner: 10Zfilipin) [17:17:14] (03Merged) 10jenkins-bot: QA: Fixed Selenium tests that fail at Cloudbees [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117431 (owner: 10Zfilipin) [17:34:44] (03PS1) 10Jdlrobson: QA: Assure login step visits the Main Page [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117432 [17:34:50] (03CR) 10jenkins-bot: [V: 04-1] QA: Assure login step visits the Main Page [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117432 (owner: 10Jdlrobson) [17:34:51] ^ chrismcmahon [17:36:58] (03PS2) 10Jdlrobson: QA: Assure login step visits the Main Page [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117432 [17:37:50] chrismcmahon: I notice you used visit LoginPage before - but it's probably better to simulate a login through the left menu? Otherwise the login page gets visited twice… [17:38:41] jdlrobson: talking with Zeljko atm, lets see how the build works out first [17:38:57] chrismcmahon: sure let me know if i can help with anything [17:47:24] (03PS3) 10Jdlrobson: Hygiene: Kill that 'I am on the User login page' rule in QA [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/116900 [18:26:59] awjr: when I try to access https://wmf.zendesk.com/requests/3141 it says "Request not found: You do not have access to request #3141. It may have been solved or deleted." Is the ticket already closed? [18:27:45] kaldari: that is weird, when i look at it, it says it is awaiting my response [18:27:49] maybe I just don't have permission [18:27:57] that could be [18:28:00] oh well, I'll just reply in email [18:28:02] kaldari: i forwarded the text though via email [18:28:03] yeah [18:28:22] i figure we should hear everyone's thoughts first and the response to chip with what we agree we want [18:49:39] wait a second... we don't redirect tablets to mobile at all ATM, why should we make an exception for Kindle? [18:50:45] brion: would u have time for a quick hangout? [18:52:19] brion: *re localize string stuff* [18:52:50] mhurd: sure [18:53:30] brion: oh, drama on wikitech-l! I hadn't noticed [18:53:54] mhurd: i don't see or hear you :) [18:54:49] MaxSem: kindles can *not* handle desktop site well :) [18:54:57] they are more like phones than iPads [18:55:03] with a really shitty slow browser [18:55:07] unless you mean kindle fires [18:57:09] brion: my interwebs went crazy there... [18:57:17] whee [19:00:17] (03PS1) 10Jdlrobson: Story 1714: Revert back to using VisualEditors toolbar buttons [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117446 [19:02:31] mhurd: thanks for heading the stand up. were there any action items post ? [19:03:12] tfinc: kenan to catch up with max about the account creation API fix he's working on [19:04:57] mhurd: I just moved our editing meeting by 30m, I've another meeting in the current timeslot [19:05:16] (03CR) 10Jforrester: Story 1714: Revert back to using VisualEditors toolbar buttons (032 comments) [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117446 (owner: 10Jdlrobson) [19:06:04] jdlrobson: Is https://gerrit.wikimedia.org/r/#/c/117447/ good to merge right now? [19:06:07] mhurd: did that come through? [19:06:13] hangout seems a bit wonky :D [19:06:21] brion: my connection crapped out again [19:06:25] heh [19:06:39] nope :( [19:06:41] mhurd: so in summary -- new .lproj folders will get committed by the localization team [19:06:57] we just have to add new ones to the project as they get created so they get picked up [19:07:07] other than that, we only have to touch the qqq and en [19:07:36] translators handle the rest on translatewiki.net and the team runs the conversion to produce the files and commits them into git [19:15:48] YuviPanda: ah. cool [19:15:53] mhurd: \o/ [19:15:59] tfinc: what yuvi said :) [19:18:01] mhurd: huh ? [19:18:45] awjr: i think this patch https://gerrit.wikimedia.org/r/117432 will fix the remaining browser tests issues [19:20:30] tfinc, ping re https://wikimedia.mingle.thoughtworks.com/projects/mobile/cards/1750 we need either a testing phone for this or engineering contacts in DoCoMo [19:20:40] ok jdlrobson, im in the middle of migrating bingle/bugello to eqiad [19:20:41] with exception of VisualEdtiro toolbar test (fix on way) [19:20:49] so i cant really look now [19:20:53] MaxSem: can you use a DoCoMo emulator ? [19:20:54] but should be able to later today [19:20:59] MaxSem: they exist on Windows [19:21:15] i saved some just in case we needed them [19:21:18] ^ chrismcmahon help with above would be much appreciated [19:21:42] tfinc, the trick is that it also has to use their services for web <-> i-mode translation. the emulator doesn't exhibit this bug [19:21:42] i'm also more then happy to buy an iMode phone but they work on a different frequency so they will need to have wifi [19:22:10] MaxSem: effectively give me a phone that i can buy that works on a frequency we have [19:22:18] MaxSem: or lets fine someone in japan to test for us [19:22:21] i'm happy with either [19:22:34] hmm, will it even use that proxy over wi-fi if requests are not coming from their infrastructure [19:22:53] brion: i think i'll make a quick pass at wranging view controllers' strings into the qqq & en localizabe.strings - sound ok? [19:22:55] MaxSem: we'd have to explore that. i'm less knowledge about i-mode [19:23:00] mhurd: go for it [19:23:05] i've only used it through emulators [19:23:17] (03PS1) 10Jdlrobson: QA: Fix VisualEditor tests [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117450 [19:23:18] MaxSem: can you reach out to the Wikimedia Japan community ? [19:23:23] brion: k. big strings patch coming soon :) [19:23:30] MaxSem tfinc> twitter WikipediaMobile [19:23:39] MaxSem: we have someone on staff who organized the one and only wmf conference [19:23:44] that or just go to social media for it [19:23:46] jdlrobson: exactly [19:24:16] (03PS2) 10Jdlrobson: QA: Fix VisualEditor tests [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117450 [19:24:22] haitham works for us and organized that conference and speaks Japanese [19:24:36] he may know people [19:24:51] tfinc, the problem is that the bug is not even in phone or our infrastructure but in their proxy so I guess finding someone in Japan is indeed the most promising venue [19:25:10] MaxSem: so i say either use twitter or contact haitham [19:25:22] heck. do both [19:27:05] MaxSem: does that give you enough to work with ? [19:27:22] yep [19:48:08] (03PS1) 10Kaldari: Making 'Contents' in TOC localizable in mobile [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117456 [19:54:48] kaldari - mingle fail. have been working on that too eek [19:55:48] (03CR) 10Jdlrobson: [C: 032] Making 'Contents' in TOC localizable in mobile [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117456 (owner: 10Kaldari) [19:56:28] (03Merged) 10jenkins-bot: Making 'Contents' in TOC localizable in mobile [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117456 (owner: 10Kaldari) [20:21:21] (03CR) 10Cmcmahon: [C: 032] QA: Assure login step visits the Main Page [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117432 (owner: 10Jdlrobson) [20:22:48] (03Merged) 10jenkins-bot: QA: Assure login step visits the Main Page [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117432 (owner: 10Jdlrobson) [21:11:29] dr0ptp4kt: https://github.com/montehurd/apps-ios-wikipedia/tree/i18n [21:15:22] (03CR) 10Kaldari: [C: 04-1] Update attribution (031 comment) [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117013 (owner: 10Jdlrobson) [21:22:33] MaxSem: i saw your patch for the docomo issue - can you move the card in mingle? [21:22:49] jdlrobson: still seeing failures :( [21:22:59] awjr: which ones [21:23:30] 26?! [21:23:30] wtf [21:24:00] im curious to see if the firefox tests on betalabs are any different but i presume they wont be [21:25:05] login seems to be failing.. [21:25:19] all have a dependency on being logged in - checking now [21:25:34] https://saucelabs.com/jobs/23a91cf689f0482699512ab0b514d489 < password issue [21:25:43] chrismcmahon: did MEDIAWIKI_PASSWORD change? [21:25:50] huh [21:26:30] awjr, done [21:26:45] ty MaxSem [21:27:01] jdlrobson: no, a few tests are still not picking up password after Z and I made changes earlier today, but some are. it is conflicts between env.rb's and such [21:28:04] awjr: i still have a host of patches that haven't been merged due to qa that may also remedy some existing known problems [21:28:39] e.g. https://gerrit.wikimedia.org/r/117450 [21:28:47] (03PS3) 10Jdlrobson: QA: Fix VisualEditor tests [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117450 [21:29:39] sec jdlrobson [21:29:57] awjr: also this serious bug didn't get picked up by bingle it seems https://bugzilla.wikimedia.org/show_bug.cgi?id=62294 [21:30:04] o_O [21:30:07] will look shortly [21:30:58] jdlrobson awjr it boils down to how the mobile repo deals with sauce browsers got pretty funky over time. We thought we'd fixed all of it, but turns out we only got partway there. [21:31:01] we [21:31:23] we'll be sharing more sauce-browser code with the shared gem, the way the Good Lord intended [21:31:23] chrismcmahon: can you elaborate? [21:32:56] awjr: this cleared up about half the issues https://gerrit.wikimedia.org/r/#/c/117431/ (I was hoping it would fix all) [21:35:53] chrismcmahon: do you have an idea of how to resolve the outstanding issues? [21:37:10] awjr: in a general sense, yes. [21:37:32] ok - eta and/or any way we can help? [21:37:40] awjr: probably monday [21:39:56] ok - chrismcmahon are you anticipating this will fix all existing failures? if not, what is the signature of the tests it would fix so we can focus on fixing the others? [21:41:38] awjr: I'm anticipating all the tests will load a target page correctly and have proper links to sauce jobs by early Monday morning. [21:42:22] awjr: in the meantime, running locally vs. beta/test2 should be accurate, this is a connect-Jenkins-to-Sauce-properly-finally set of updates [21:42:39] ok gotcha chrismcmahon, thanks [21:45:06] awjr: honestly, I am like three steps away from turning the VE build green for the first time in forever, I rely on Zeljko's expertise for the Jenkins/Sauce stuff. I know in general what's going wrong, but it's better to pair on that stuff. [21:49:34] getting the VE build green is going to be a huge victory chrismcmahon, preemptive but non-jinxing congratulations [21:52:17] (03PS6) 10Kaldari: QA 1521: Switch from VisualEditor to wikitext editor [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/111253 (owner: 10Jdlrobson) [21:52:18] (03PS5) 10Kaldari: Story 1521 WIP: Allow switching from VisualEditor to Wikitext editor [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/111367 (owner: 10Jdlrobson) [21:52:20] awjr: did you notice how much code we removed from MF? there's been amazing progress in the framework itself. MF being the oldest repo-specific browser tests means that the cruft is a little higher than the others, but we'll knock it down [21:52:27] (03CR) 10jenkins-bot: [V: 04-1] QA 1521: Switch from VisualEditor to wikitext editor [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/111253 (owner: 10Jdlrobson) [21:52:33] :D [21:57:22] (03PS7) 10Kaldari: QA 1521: Switch from VisualEditor to wikitext editor [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/111253 (owner: 10Jdlrobson) [21:58:40] (03CR) 10Kaldari: "recheck" [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/111253 (owner: 10Jdlrobson) [21:59:05] awjr: chrismcmahon VE tests are failing because of this - https://gerrit.wikimedia.org/r/117450 [21:59:14] i just need a review arggh! [21:59:41] jdlrobson: i am in the middle of it [21:59:45] ok cool [21:59:52] but i am having to make some changes to my local test set up, so will take a efw [21:59:53] *few [22:00:08] but cursorily, LGTM - i just want to test it [22:00:09] sorry to keep going on i'm just keen to see them green again - i've been tearing my hair over it for the last 3 days [22:00:11] np [22:00:17] i totally understand :) [22:02:06] blargh it is amazing how much email can back up in only 2 days [22:02:44] it looks like the password issue that chrismcmahon was aware of is the culprit for most of them though [22:03:31] jdlrobson: im getting a failure in the ve tests wiht your patchset [22:03:37] orly? [22:03:40] what's the failure? [22:03:42] Failing Scenarios: [22:03:42] cucumber features/editor_ve.feature:14 # Scenario: I can edit a page using VisualEditor [22:03:47] it's failing on the toast notification [22:03:52] also locally or on production? [22:03:56] locally [22:04:02] so could be something weird with my set up [22:04:05] it passes on beta labs - what does the toast say? [22:04:08] i have been trying to clean things up locally [22:04:09] (03PS6) 10Jdlrobson: Story 1521 WIP: Allow switching from VisualEditor to Wikitext editor [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/111367 [22:04:22] Then I see a toast notification # features/step_definitions/common_article_steps.rb:73 [22:04:22] timed out after 5 seconds, waiting for {:css=>"div.toast"} to become present (Watir::Wait::TimeoutError) [22:04:22] ./features/step_definitions/common_article_steps.rb:74:in `/^I see a toast notification$/' [22:04:22] features/editor_ve.feature:23:in `Then I see a toast notification' [22:04:24] jdlrobson: ^ [22:04:56] im trying manually [22:05:23] the submit step seems to never complete for me when i edit in VE [22:05:33] sigh [22:07:07] awjr: mmmmmmmmmmmmm so that sounds like Parsoid might not be up to date [22:07:59] jgonera kaldari: heads up VE code got merged so toolbar is an odd state - can you take a look at https://gerrit.wikimedia.org/r/117446 to correct it? [22:08:43] hm oh yeah my parsoid is a bit out of date [22:09:57] damn it i git pull'd for parsoid and restarted the server but now im getting parsoid error 'couldnt connect to host' [22:10:25] jdlrobson: the code for the test looks fine, im going to merge. if it f's something on betalabs we can revert [22:11:05] i am bummed that every time i get my local instance happily working with all the components it stops working a day later :( [22:11:26] (03CR) 10Awjrichards: [C: 032] QA: Fix VisualEditor tests [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117450 (owner: 10Jdlrobson) [22:11:51] awjr did you run git submodule update [22:11:56] aye [22:12:13] dr0ptp4kt: https://www.mediawiki.org/wiki/Manual:Hooks/MakeGlobalVariablesScript [22:12:18] jdlrobson: wait, for which repo? [22:12:22] i did for ve [22:12:23] or https://www.mediawiki.org/wiki/Manual:Hooks/ResourceLoaderGetConfigVars [22:12:30] awjr: yeh VE was the one [22:12:34] yeah [22:12:38] i think parsoid got moved as well [22:12:40] everything broken always. [22:13:26] (03Merged) 10jenkins-bot: QA: Fix VisualEditor tests [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117450 (owner: 10Jdlrobson) [22:14:17] awjr: sigh :/ [22:15:09] hopefully they'll go green this weekend [22:15:46] there's a test pending on betalabs, hopefully the change is picked up before it runs and we can see what happens [22:15:56] fingers crossed! [22:51:30] jdlrobson: thinking about adding this to r32 http://www.ebay.com/itm/Ghibli-Totoro-Tree-Vinyl-Wall-Decor-Decal-/190717659331 thoughts ? [22:52:14] we need the bus stop one.. [22:52:30] http://i153.photobucket.com/albums/s222/goranosponsa/anime/studioghibli/totoro/Totoro_Bus_stop.jpg < can we get that in decal form? [22:52:48] http://www.artfire.com/uploads/product/3/253/4253/8304253/8304253/large/ghibli_totoro_-_satsuki_and_mei_bus_stop_wall_art_applique_stickers_fd6472e4.jpg < ooohh [22:53:08] jdlrobson: then just add http://www.ebay.com/itm/Ghibli-Totoro-Head-Vinyl-Wall-Decor-Decal-/160867358506 [22:53:17] jdlrobson: how much would it cost for someone to custom print us one of those ? [22:53:20] http://www.ebay.com/itm/Ghibli-Totoro-Satsuki-and-Mei-Bus-Stop-Wall-Art-Applique-Stickers-/191013915619 [22:53:22] (03PS1) 10Dr0ptp4kt: Instruct JavaScript modules in MobileFrontend to not show images. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/117615 [22:53:42] (03Abandoned) 10Dr0ptp4kt: Allow callees such as Wikipedia Zero JS to hide search thumbnails. [extensions/MobileFrontend] - 10https://gerrit.wikimedia.org/r/117367 (owner: 10Dr0ptp4kt) [22:53:47] https://www.etsy.com/uk/listing/156659156/totoro-inspired-mei-kusakabe-soot?ref=sr_gallery_31&ga_search_query=totoro+wall+decal&ga_search_type=all&ga_view_type=gallery aahhhhaha [22:53:57] https://www.etsy.com/uk/listing/151669286/totoro-inspired-soot-sprites-wall-art?ref=sr_gallery_34&ga_search_query=totoro+wall+decal&ga_search_type=all&ga_view_type=gallery ahahahashshasha [22:54:09] too much good stuff on etsy [22:54:12] (03Abandoned) 10Dr0ptp4kt: Don't show search thumbnails for W0 unless on mdot. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/117364 (owner: 10Dr0ptp4kt) [22:54:12] must escape etsy [22:54:12] https://www.etsy.com/uk/listing/152413926/totoro-inspired-totoro-umbrella-wall-art?ref=sr_gallery_33&ga_search_query=totoro+wall+decal&ga_search_type=all&ga_view_type=gallery [22:58:33] yurik, you around? [22:58:41] nope [22:59:03] * yurik is hunting rabbits [23:00:29] yurik, be very, very quiet. [23:00:41] yurik, would you please review and +2 https://gerrit.wikimedia.org/r/#/c/117615 ? this fixes the search regression [23:00:51] vawy vawy quiet i am [23:02:54] dr0ptp4kt, has someone suggested we do it this way? [23:03:04] jdlrobson, yes [23:03:41] this was in apposition to an approach using m.emit with a singleton that gets memoized on first search key tap [23:03:44] jdlrobson, it looks good to me from the zero side, but i think you should +2 it :) [23:04:33] dr0ptp4kt, no idea what you meant :( [23:04:38] jdlrobson, jgonera, would you guys please quickly peek at https://gerrit.wikimedia.org/r/#/c/117615/ and +2 ? [23:04:50] (03CR) 10Yurik: [C: 031] Instruct JavaScript modules in MobileFrontend to not show images. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/117615 (owner: 10Dr0ptp4kt) [23:05:04] dr0ptp4kt: yurik question - Called right before Skin::makeVariablesScript is executed. Ideally, this hook should only be used to add variables that depend on the current page/request; static configuration should be added through ResourceLoaderGetConfigVars instead. [23:05:26] according to docs that what it does and i think it will place it in the page HTML itself [23:05:31] does that have any caching implications for oyu? [23:06:03] the site (zero. vs m.) is a param we vary on [23:06:15] but within zero will it sometimes be true and sometimes not be? [23:06:21] dr0ptp4kt, btw, space is missing on line 380 ;) [23:08:20] yuirk, where's line 380? [23:08:29] jdlrobson, no, zero means always no images [23:08:38] m. on the other hand - always with imges [23:08:38] ok then it looks good to me [23:08:43] oh, i see, hang on [23:08:55] thanks jdlrobson, that is the correct hook from what i see [23:08:59] (03CR) 10Jdlrobson: [C: 031] Instruct JavaScript modules in MobileFrontend to not show images. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/117615 (owner: 10Dr0ptp4kt) [23:09:00] yurik, i'll fix the space [23:09:10] thx jdlrobson [23:11:25] (03PS2) 10Dr0ptp4kt: Instruct JavaScript modules in MobileFrontend to not show images. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/117615 [23:11:40] ^ yurik, i think jdlrobson wants you to do the +2 :) [23:12:03] (03CR) 10Yurik: [C: 032] Instruct JavaScript modules in MobileFrontend to not show images. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/117615 (owner: 10Dr0ptp4kt) [23:12:05] (03Merged) 10jenkins-bot: Instruct JavaScript modules in MobileFrontend to not show images. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/117615 (owner: 10Dr0ptp4kt) [23:12:13] jdlrobson, thanks again! [23:12:16] yurik, thx for the +2 [23:12:24] yurik, got time to look at another? [23:12:29] or not so much? [23:12:31] sigh [23:12:33] sure [23:13:56] yurik, https://gerrit.wikimedia.org/r/#/c/113655/ ... i validated the Set-Cookie in prod (it's easy enough to simulate with an intercepting proxy in non-prod, or just emit it somewhere in code) [23:20:11] dr0ptp4kt, i don't think we should respond with "set-cookie" if the cookie is already set [23:20:30] also, what happens if they switch connection? do you ever clear it? [23:23:17] yurik, i don't think it would turn out too bad in practice, but if we don't set the outbound cookie on the origin response, we will have some objects in the cache that don't contain that cookie and some that do, even though all of them should have that cookie [23:24:02] if they switch connections to another w0 provider, that's a different page varied on a different x-cs [23:24:13] that is, if they're on a different w0 provider. [23:24:20] good point, wasn't considering cache [23:24:44] if they're on a different w0 provider, then.... [23:24:52] i guess maybe that's an edge case to deal with! [23:25:07] but hold on, if the client already has the cookie set, what mechanisms do we have to clear it? [23:25:09] if they're not on a w0 network, of course it doesn't matter because they don't get the js in the first place, so no harm [23:25:33] yeah, i think we probably need to to do a set-cookie for the other carriers. one minute [23:25:43] kaldari: https://gerrit.wikimedia.org/r/#/c/117617/ [23:26:05] dr0ptp4kt, i mean: on w0 net, they get "set-cookie", later they navigate to another page and the cookie is still set [23:26:33] with your change, i think we would have to be setting that cookie for ALL responses [23:26:55] yurik, yeah, you're talking about a transition from one w0 carrier to another w0 carrier, where the two carriers have different support for https zero-rating [23:27:12] no, that will actually be ok i think [23:27:39] if they start on a w0 carrier that zero-rates https, the would get the cookie with the 'tls' value in it. [23:27:57] if they then swiched carriers, they'd still have that cookie in the browser [23:28:07] but hang on [23:28:29] ... [23:28:42] yeah, so in that case, they have the cookie present, which shows contributory features, which is a mistake. [23:29:22] but which js code checks if the cookie is set? what if they visit from a non-w0 net, do they get the js that checks the cookie? [23:29:50] it seems you have to set the cookie regardless [23:31:00] so if its not present, set cookie's expiry to "1" [23:31:26] yurik, you're saying in order to correct the current patch we would need to set the cookie for all responses, correct? [23:32:52] dr0ptp4kt, i'm saying that you should do "else setcookie('zeroopts',...) with expiry = 1 [23:33:04] need to check syntax [23:33:07] yurik, we are in agreement. i'll amend the patchset shortly. [23:33:23] yurik, i was just double checking if setting the cookie in both cases be compatible with https://gerrit.wikimedia.org/r/#/c/115669/6/templates/varnish/zero.inc.vcl.erb . it will be compatible, phew. [23:34:52] dr0ptp4kt, what do you think it terms of local browser caching - if the user revisits the same page (or visits a different one), after switching between w0 and regular, will it affect anything? [23:35:47] yurik, i think the ua will make an extra edge hit pretty regularly iirc. so it should get a new version of the page due to cache varying on x-cs [23:36:12] i think most pages have caching headers that make for a lot of varnish hits. granted, many of them are 304 not modified [23:36:30] but in that case, it would not be a 304, because of the x-cs varying [23:36:36] as i understand [23:36:42] hope it works :) [23:37:05] yurik :D [23:39:56] yurik, just testing amend before i post it [23:43:28] (03PS11) 10Dr0ptp4kt: HTTPS UI for applicable Wikipedia Zero configurations. [extensions/ZeroRatedMobileAccess] - 10https://gerrit.wikimedia.org/r/113655 (owner: 10Yurik) [23:43:37] yurik, ok, updated [23:44:51] dr0ptp4kt, i would rather delete the cookie - this way they won't send it to us for each of the requests [23:45:14] besides, check your logic [23:45:23] you are not setting it when its non https [23:45:32] https://gerrit.wikimedia.org/r/#/c/113655/11/includes/PageRenderingHooks.php [23:53:28] yurik, is "if ( !enableHttpsForSure && !isHttps && !M.settings.getUserSetting( 'zerodontask', true ) )" insufficient for this purpose? or is the point here to add a little extra defense server side? either the operator supports https zero-rating, in which case we always want to send the header, or it does not, in which case we do not want to set it (and the inbound x-cs would not be present, so the extension wouldn't be called). are you [23:53:29] thinking ahead for esi here? [23:53:54] i'm cool adding it in, but was trying to think how it might play out with caches and such [23:54:40] dr0ptp4kt, ?? i'm talking about the link above line 240 -- you are setting the cookie only if enableHttps is true [23:56:25] yeah, i got you there. i'm talking about the javascript that examines the cookie value. if an operator supports https zero-rating, the x-cs header will be passed in over https and their enableHttps flag will (assuming no mistakes) be set to true. so that block of code should be setting the cookie correctly. [23:56:40] (just like it would over http) [23:59:04] does that make any sense? if not, maybe a quick videoconference?