[00:30:22] hey [00:30:26] anyone from the team currently about? [00:37:22] * Krenair pokes legoktm and mlitn [00:48:50] 3Collaboration-Team-Current, 10Echo, 6Editing-Department, 6Performance-Team, and 3 others: Fivefold increase in render-blocking CSS size for logged-in users - https://phabricator.wikimedia.org/T112401#1640064 (10ori) There is still a 140 kB (compressed) difference in the size of static assets served to log... [00:59:54] ori, IANAD, but anything I can do? if not, roan will be amongst the first awake, in Euro timezone. [01:00:21] oh, I see the phab comment now. [01:02:06] I'll email the team list [01:08:20] * quiddity goes to look for groceries/dinner [01:23:29] quiddity: around? [01:23:45] James_F: ? [01:24:04] I am, for about six seconds. [01:24:06] Why? [01:24:29] I'm going to revert Echo [01:24:45] On top of the performance issues, we also have a UBN in Safari that we never figured out [01:24:57] Eurgh. [01:25:24] And I don't see any more quick fixes for reducing the JS payload [01:25:59] What's the UBN? [01:26:58] https://phabricator.wikimedia.org/T112552 [01:27:17] Hmm, interesting. Looks like a gadget issue. [01:27:24] (It loads fine in PHP.) [01:27:37] Oh. [01:27:42] That's Krinkle's doing. [01:27:47] eh? [01:27:47] TypeError: performance.mark is not a function. (In 'performance.mark('mwLoadEnd')', 'performance.mark' is undefined) – TypeError: performance.mark is not a function. (In 'performance.mark('mwLoadEnd')', 'performance.mark' is undefined) [01:28:02] https://gerrit.wikimedia.org/r/#/c/238367/ [01:28:02] no, that's different too [01:28:51] I repro'd on my macbook and it was unrelated to performance.mark [01:29:01] some CSS styles weren't loading [01:29:16] [15:27:37] And for .oo-ui-textInputWidget input[type="search"]::-webkit-search-decoration too [01:29:24] Hmm. [01:29:26] that's what he traced it down to I think [01:29:41] That's… not fun. Does the affect other things using OOUI on Commons? [01:29:48] There's this little thing there called UploadWizard… [01:30:00] if UW broke I think more people would have complained [01:30:10] Yes, hence my confusion. [01:30:18] Though Safari users are relatively few. [01:30:40] when I tested it you couldn't even get to the search bar [01:30:49] Vector + Commons only [01:30:58] but we could all reproduce [01:31:20] I can get to the search bar, but only via tabbing. [01:31:29] :| [01:31:46] And I get the same issue in Monobook, BTW. [01:31:50] ugh wtf [01:32:03] so that plus perf I think pushes it over the edge [01:32:08] * James_F sighs. [01:32:14] You'll get so much shouting. [01:32:24] I know [01:32:26] The performance issue is minor in comparison. [01:32:30] The main thing is the UBN. [01:33:01] What's the JS payload of one of the major wikis' gadgets? [01:33:23] OK, yeah, Krinkle's deployed and it's still broken. [01:33:40] Lots of CSS warnings, but that's normal. [01:34:05] OOUI takes top 3/5 in my mw.loader.inspect() on enwp, and I have twinkle+navpopups+other stuff [01:34:09] And they all appear equally whether or not you're logged in. [01:34:17] Hmm. Echo's icons are now black in Monobook. [01:34:31] I see I walked in in the middle of something. Probably this. I will be silent. [01:34:32] Deskana: since thursday [01:34:35] Deskana: That's intentional. [01:36:19] If you want to argue with ori that the perf isn't a UBN, okay. But I have no ability to debug the Safari thing. Today was the third? time I'd ever opened up Safari :/ [01:37:04] Sure. [01:39:32] James_F: is that a "Sure, revert"? Or "Sure...?" [01:39:42] The former. [01:40:00] 6Collaboration-Team-Backlog, 10Echo, 7Browser-Support-Apple-Safari, 5Patch-For-Review: Echo looks messed up on Commons for Safari, old Firefox - https://phabricator.wikimedia.org/T112552#1640141 (10Jdforrester-WMF) a:3Legoktm [01:40:23] ok [01:40:42] Note that the train rolls out tomorrow morning. [01:40:48] What are you going to do for that? [01:41:42] I'll ask twentyafterfour to use wmf21 for the new branch [01:42:05] Kk. [01:45:01] James_F: are you on the e2 list? [01:45:14] legoktm: No. [01:46:52] I cc'd [01:48:07] legoktm: You posting an update on MW.org? [01:49:12] uh, where should I? [01:50:44] https://www.mediawiki.org/wiki/Topic:Sobaaiky57zs7py7 I guess. [01:59:01] legoktm, I'm around, but in a post-gigantic-meal-haze. If you write it, I can post it, if desired. (brb in 15 min, cleanup) [02:00:46] (03PS1) 10Legoktm: Revert to 1.26wmf21 state [extensions/Echo] (wmf/1.26wmf22) - 10https://gerrit.wikimedia.org/r/238375 [02:00:50] (03CR) 10jenkins-bot: [V: 04-1] Revert to 1.26wmf21 state [extensions/Echo] (wmf/1.26wmf22) - 10https://gerrit.wikimedia.org/r/238375 (owner: 10Legoktm) [02:01:53] (03PS2) 10Legoktm: Revert to 1.26wmf21 state [extensions/Echo] (wmf/1.26wmf22) - 10https://gerrit.wikimedia.org/r/238375 [02:03:58] (03CR) 10jenkins-bot: [V: 04-1] Revert to 1.26wmf21 state [extensions/Echo] (wmf/1.26wmf22) - 10https://gerrit.wikimedia.org/r/238375 (owner: 10Legoktm) [02:08:30] (03CR) 10Legoktm: [C: 032 V: 032] Revert to 1.26wmf21 state [extensions/Echo] (wmf/1.26wmf22) - 10https://gerrit.wikimedia.org/r/238375 (owner: 10Legoktm) [02:20:01] quiddity, James_F: it's reverted now [02:22:14] Yeah. :-( [02:22:56] legoktm, A) <3 for the afterhours work, B) are you able to write the update message? If you post it onwiki at mw.o, I'll send a note to wikitech-ambassadors, and add a note at the en:WP:VPT thread. [02:26:18] https://etherpad.wikimedia.org/p/omg-echo-doom [03:02:48] legoktm: thank you for making a tough call. Let me know if I can help. [03:08:49] You'll get so much shouting. [03:08:50] where? [03:21:49] ori, primarily at enwiki and mediawiki, by the people who dislike the change and want a userpreference to turn it off (see WP:VPT) [03:23:19] although, the vote halfway down the Signpost, has inspired me with the optimistic turnout. See sidebar's blue&green WMF logo, at https://en.wikipedia.org/wiki/Wikipedia:Wikipedia_Signpost/2015-09-09/News_and_notes [03:24:55] * quiddity wanders off to collapse. 11 hour days are less than fun. [05:48:57] ori, are you, by any chance, around? [06:16:06] (03PS1) 10Mooeypoo: [wip] Only load ext.echo.ui if the user clicks the echo badge [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 [06:16:13] legoktm, ^^ [06:17:32] mooeypoo: we'd need some kind of progress/loading indicator... [06:18:11] legoktm, yeah, I was going to do that next. I just wanted to make sure this is a good start. How do I inspect the payload ori was talking about? [06:18:33] it theoretically should solve the issue, as it won't load the problem-files unless they're needed [06:18:37] but how do I test this [06:18:46] use "mw.loader.inspect()" in your console [06:18:58] it'll output a table of all the RL modules that are currently loaded sorted by size [06:19:05] * mooeypoo nods [06:20:04] legoktm, ""ext.echo.init""1.7 kB"" and ""ext.echo.logger""1.3 kB"" are the only"Echo" anything i see there before I click the button [06:20:07] that seems reasonable [06:20:24] woohoo [06:20:37] So this means it works, I would assume. Is it reasonable that the payload is sent when the user needs it? We will try to minimize that too, but... I assume that's fine? [06:20:45] I believe so yeah [06:21:07] Aiight. So I'll just add something to display loading process now... which... is funny because that usually requires ooui [06:21:08] ha [06:21:09] uhm [06:21:18] I guess I can add a standalone spinner... eh. [06:21:20] also, if we're going to wait for oojs-ui to load, we should fire the API request to fetch the notifications at the same time [06:21:32] Hm [06:22:17] That would require getting the echo.dm first, loading it + firing the API then loading the UI and feeding it the dm. [06:22:21] that's not a bad idea. [06:22:22] there's jquery.spinner I suppose [06:22:39] yeah isn't the point of ooui'fying to move away from jquery ui? [06:22:57] it's technically not jquery.ui [06:22:58] but yeah [06:23:03] We should sort something out upstream in how ooui loads and what it automatically gives core... a spinner would be nice :\ [06:23:11] but yeah, I'll use that for now [06:23:54] legoktm, the "ext.echo.dm" module has no css files. Is that one safe to load on each page load? [06:24:10] OR... I can have it load on click, fire the api, then ask for the ui to load [06:24:14] but that seems a bit meh [06:24:39] can we make the API request without the dm? [06:25:13] I guess I can have it ready to accept a result [06:25:47] It's absolutely not ideal, though. It also manages the promise to make sure you don't double-fetch and such [06:26:11] I think we should avoid loading anything on every page that isn't necessary to render the badge [06:26:22] and the basic init click handler [06:26:37] Yeah [06:26:46] Okay, fair enough. [06:33:40] mooeypoo: I assume we'll need a similar fix for mobile? [06:38:14] legoktm, no, actually, mobile only loads if the popup opens [06:38:22] Also, it's not yet OOUIfied [06:38:39] we have a fix in the making, but even that loads only after the overlay opens [06:49:07] legoktm, meeehhh this is ugly :( I really dislike the spinner. [06:50:09] (03PS2) 10Mooeypoo: [wip] Only load ext.echo.ui if the user clicks the echo badge [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 [06:50:13] legoktm, ^^ [06:50:22] meh, the spinner is REALLY bad [06:58:19] (03PS3) 10Mooeypoo: [wip] Only load ext.echo.ui if the user clicks the echo badge [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 [07:00:15] legoktm, okay, now everything only loads on click, the API request is fired immediately on click and is "fed" into the notifications widget for processing. There's an indicator for it all (looks marginally better but not great) [07:00:41] legoktm, I hope this is good enough to at least solve *this* issue. We can concentrate on upstream stuff later... ;\ [07:01:04] (03PS4) 10Mooeypoo: Only load ext.echo.ui if the user clicks the echo badge [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 [07:55:29] 6Collaboration-Team-Backlog, 10Flow, 7Design: Editing toolbar(s) for Flow (VE and/or wikitext toolbar) - https://phabricator.wikimedia.org/T78346#1640689 (10Trizek-WMF) On [[ https://fr.wikipedia.org/w/index.php?title=Sujet:Sox3xbkro37l4fta&topic_showPostId=sox3xblgm25yjf9q#flow-post-sox3xblgm25yjf9q | this... [08:01:45] 6Collaboration-Team-Backlog, 10Flow, 7Design: Make source/wikitext icon consistent - https://phabricator.wikimedia.org/T97451#1640696 (10Trizek-WMF) On [[ https://fr.wikipedia.org/w/index.php?title=Sujet:Sox3xbkro37l4fta&topic_showPostId=sox3xblgm25yjf9q#flow-post-sox3xblgm25yjf9q | this feedback ]], a user... [08:38:56] (03PS1) 10Matthias Mullie: Add artificial primary key to flow_wiki_ref and flow_ext_ref [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238393 (https://phabricator.wikimedia.org/T109676) [08:40:20] (03CR) 10Matthias Mullie: "Please test this patch in the deployment plan order as proposed in commit msg :)" [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238393 (https://phabricator.wikimedia.org/T109676) (owner: 10Matthias Mullie) [08:40:34] 6Collaboration-Team-Backlog, 10Flow: Flow: Prettify thread permalink URLs/Titles (they are not human readable!) - https://phabricator.wikimedia.org/T59154#1640784 (10Qgil) Would this task be suitable for #Possible-Tech-Projects and #Outreachy-Round-11? [08:44:03] 3Collaboration-Team-Current, 10Flow, 5Patch-For-Review, 5WMF-deploy-2015-09-15_(1.26wmf23), 7WorkType-NewFunctionality: Fix toggling Flow to false in BetaFeatures (opt-out) - https://phabricator.wikimedia.org/T111830#1640785 (10matthiasmullie) Yes, it has been merged, so it looks like it didn't fix the i... [08:57:55] RoanKattouw_away: the one thing i figured is that exception 12 for querySelector is possibly something that is somewhat vague in the spec. [09:02:31] (03CR) 10Catrope: [C: 032] Split out .mw-echo-alert styles back into a separate module [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238204 (owner: 10Legoktm) [09:03:21] (03CR) 10Catrope: [C: 04-1] Don't show orange alert if you are viewing your user talk page (031 comment) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238194 (https://phabricator.wikimedia.org/T107655) (owner: 10Legoktm) [09:18:57] 6Collaboration-Team-Backlog, 10Echo, 5WMF-deploy-2015-09-15_(1.26wmf23): Special:Notifications has rendering issues on mobile - https://phabricator.wikimedia.org/T112191#1640814 (10Catrope) >>! In T112191#1639551, @Jdlrobson wrote: > Misleading patch hasn't fixed this problem. Echo team please let me know if... [10:18:26] 6Collaboration-Team-Backlog, 10Flow, 10MediaWiki-General-or-Unknown, 7Design: Encourage users who are currently ignoring their talk pages to read them - https://phabricator.wikimedia.org/T110048#1640950 (10Trizek-WMF) I get a feedback from someone who has Flow enable on his talk page. [[https://fr.wikiped... [10:21:22] legoktm, mooey|away: Thanks for working on this stuff after I went to sleep [10:21:49] legoktm: And thanks for doing the revert. I should have made that call once I saw the Safari breakage [11:19:40] (03PS1) 10Matthias Mullie: Clear title caches after moving old talk page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238414 (https://phabricator.wikimedia.org/T111830) [12:46:38] RoanKattouw_away: https://bugs.webkit.org/show_bug.cgi?id=149160 [12:46:55] filed the issue about ::-webkit pseudo elements with querySelector [12:47:41] Thanks [12:48:19] I'll write a patch for mw.loader.inspect that tolerates exceptions from querySelector [12:49:42] (03CR) 10Catrope: [C: 032] Clear title caches after moving old talk page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238414 (https://phabricator.wikimedia.org/T111830) (owner: 10Matthias Mullie) [12:50:33] RoanKattouw: of note, FF also fails on these. But Chrome doesn't [12:51:29] (03PS1) 10Haritha28: Add GENDER support [extensions/Thanks] - 10https://gerrit.wikimedia.org/r/238426 (https://phabricator.wikimedia.org/T56979) [12:51:40] As Krinkle said yesterday, saying Safar is the new IE6 is too generous, it's more like IE5.5 [12:51:57] But if FF also fails, then we can't behind "oh it's just Safari but that browser is crap anyways" [12:52:01] *can't hide behind [12:52:51] (03Merged) 10jenkins-bot: Clear title caches after moving old talk page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238414 (https://phabricator.wikimedia.org/T111830) (owner: 10Matthias Mullie) [12:53:19] (03PS2) 10Haritha28: GENDER support for Thanks [extensions/Thanks] - 10https://gerrit.wikimedia.org/r/238426 (https://phabricator.wikimedia.org/T56979) [12:58:54] (03CR) 10Catrope: [C: 04-1] "This message isn't being used any more, it should just be removed." [extensions/Thanks] - 10https://gerrit.wikimedia.org/r/238426 (https://phabricator.wikimedia.org/T56979) (owner: 10Haritha28) [13:04:58] (03PS3) 10Haritha28: GENDER support for Thanks [extensions/Thanks] - 10https://gerrit.wikimedia.org/r/238426 (https://phabricator.wikimedia.org/T56979) [13:23:15] thedj: https://bugzilla.mozilla.org/show_bug.cgi?id=1204880 [13:27:22] RoanKattouw: right [13:27:42] I also have a patch [13:27:45] RoanKattouw: i have a phab ticket ready for this specific issue btw. [13:27:51] Didn't we have a phab task about ... yes please [13:28:06] https://phabricator.wikimedia.org/T112650 [13:28:42] Oh that's why I couldn't find it in the search, you'd only just filed it [13:29:47] yeah, i'm a bit slow, i'm dealing with a ton of Apple iOS 9 crap right now :) too little work to not write tickets for wm.org, too much work to do it quickly :) [13:33:09] RoanKattouw: btw, for kicks, Chrome also exceptions on vendors != webkit [13:33:34] Oh great [13:37:45] (03CR) 10Gilles: [C: 032] Also defer the markRead() master updates [extensions/Echo] - 10https://gerrit.wikimedia.org/r/237531 (https://phabricator.wikimedia.org/T92357) (owner: 10Aaron Schulz) [13:37:48] (03CR) 10jenkins-bot: [V: 04-1] Also defer the markRead() master updates [extensions/Echo] - 10https://gerrit.wikimedia.org/r/237531 (https://phabricator.wikimedia.org/T92357) (owner: 10Aaron Schulz) [13:38:22] (03CR) 10Gilles: "Needs manual rebase, gerrit's won't do" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/237531 (https://phabricator.wikimedia.org/T92357) (owner: 10Aaron Schulz) [13:52:35] 6Collaboration-Team-Backlog, 10Flow, 7Easy: "Are you sure" question when converting namespace to Flow with convertNamespaceFromWikitext.php - https://phabricator.wikimedia.org/T104026#1641410 (10Devirk) a:3Devirk [14:19:27] 3Collaboration-Team-Current, 10Flow, 7WorkType-NewFunctionality: Search for localized {{#useliquidthreads}} as well in find/replace in LQT→Flow conversion - https://phabricator.wikimedia.org/T110723#1641472 (10SBisson) Should we also use the localized magic word when we add it to a page to disable LQT `{{#us... [14:20:07] 3Collaboration-Team-Current, 10Flow, 7WorkType-NewFunctionality: Search for localized {{#useliquidthreads}} as well in find/replace in LQT→Flow conversion - https://phabricator.wikimedia.org/T110723#1641475 (10SBisson) a:3SBisson [14:55:17] 6Collaboration-Team-Backlog, 10Echo: Revert echo back to one button on personal bar - https://phabricator.wikimedia.org/T111871#1641581 (10Paladox) Well it seems it has gone back to one button but this desgn is much better you have suggestion. If you know how to can upload a patch with that change please. [15:12:05] 6Collaboration-Team-Backlog, 10Echo: Revert echo back to one button on personal bar - https://phabricator.wikimedia.org/T111871#1641632 (10Stryn) >>! In T111871#1641581, @Paladox wrote: > Well it seems it has gone back to one button but this desgn is much better you have suggestion. If you know how to can uplo... [15:19:31] (03CR) 10Catrope: Add artificial primary key to flow_wiki_ref and flow_ext_ref (032 comments) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238393 (https://phabricator.wikimedia.org/T109676) (owner: 10Matthias Mullie) [15:49:55] (03CR) 10Legoktm: Don't show orange alert if you are viewing your user talk page (031 comment) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238194 (https://phabricator.wikimedia.org/T107655) (owner: 10Legoktm) [15:51:02] (03PS2) 10Legoktm: Don't show orange alert if you are viewing your user talk page [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238194 (https://phabricator.wikimedia.org/T107655) [15:59:08] RoanKattouw: so the safari thing is an upstream bug? [15:59:14] Pretty sure [15:59:16] But I can't repro [15:59:18] or is mw.loader.inspect() failing unrelated? [15:59:29] I have a patch for that one [16:09:17] (03CR) 10Kaldari: [C: 04-1] "Message should be removed, per Catrope." [extensions/Thanks] - 10https://gerrit.wikimedia.org/r/238426 (https://phabricator.wikimedia.org/T56979) (owner: 10Haritha28) [16:32:37] 6Collaboration-Team-Backlog, 10Echo, 5WMF-deploy-2015-09-15_(1.26wmf23): Special:Notifications has rendering issues on mobile - https://phabricator.wikimedia.org/T112191#1641988 (10Jdlrobson) All information is in the description of this task. You and ignore the other task. Just take a look at the screenshot... [16:35:21] 6Collaboration-Team-Backlog, 10Echo, 5WMF-deploy-2015-09-15_(1.26wmf23): Special:Notifications has rendering issues on mobile - https://phabricator.wikimedia.org/T112191#1642000 (10Jdlrobson) [16:35:42] stephanebisson, heyhey, you around? [16:37:06] 6Collaboration-Team-Backlog, 10Echo, 5WMF-deploy-2015-09-15_(1.26wmf23): Special:Notifications has rendering issues on mobile - https://phabricator.wikimedia.org/T112191#1628451 (10Jdlrobson) An updated screenshot shows you've fixed it somewhat ( I don't know which patches removed the underlines but it looks... [16:37:15] mooeypoo: yep [16:37:39] stephanebisson, so, I came back from my vacation to a bit of an Echo crisis. Can I toss the edit summary widget back your way? [16:37:51] mooeypoo: sure [16:38:03] Also, you're absolutely right about the tests being in the same repo, I was being a bit over zealous thinking it would make review easier [16:38:13] err the same commit [16:38:32] well, it could, but I really use them together [16:38:53] stephanebisson, I think it's actually done. I left it [wip] because I was worried that the automatic base message wasn't implemented but you pointed out it is, so I am 99% sure we covered everything... [16:39:03] yeah, no, you're right, I sould also be better about running them every time [16:39:30] mooeypoo: I'm not sure we handle all the cases for resolve/reopen [16:39:34] I'll test [16:39:56] stephanebisson, that was my worry. I *think* we are according to the code, but...yeah. Please do. [16:40:05] and thanks! I'll go work on re-echoing echo :\ [16:45:04] 6Collaboration-Team-Backlog, 10Flow: Missing 'View changes' for entries on User talk pages with Flow enabled (page-level diff) - https://phabricator.wikimedia.org/T112353#1642016 (10Etonkovidova) [16:52:12] (03PS1) 10Sbisson: [WIP] Use localized version of LQT magic word [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238493 (https://phabricator.wikimedia.org/T110723) [16:53:09] 3Collaboration-Team-Current, 10Flow, 5Patch-For-Review, 7WorkType-NewFunctionality: Use EditorWidget for topic summary - https://phabricator.wikimedia.org/T111102#1642038 (10SBisson) a:5Mooeypoo>3SBisson [16:55:22] (03CR) 10Mooeypoo: "Are you sure? My tests showed a difference when I took the (even though it had stuff in it) outside the
... the styles looked exa" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238017 (https://phabricator.wikimedia.org/T112218) (owner: 10Mooeypoo) [16:55:37] (03CR) 10jenkins-bot: [V: 04-1] [WIP] Use localized version of LQT magic word [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238493 (https://phabricator.wikimedia.org/T110723) (owner: 10Sbisson) [17:17:08] 3Collaboration-Team-Current, 10Echo, 5Patch-For-Review, 7Regression, and 2 others: The new Echo notification snaps back to English numbers - https://phabricator.wikimedia.org/T112224#1642120 (10Etonkovidova) Checked in Betalabs. [17:38:46] 6Collaboration-Team-Backlog, 10Echo, 7Browser-Support-Apple-Safari, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22): Echo looks messed up on Commons for Safari, old Firefox - https://phabricator.wikimedia.org/T112552#1642271 (10Legoktm) p:5Unbreak!>3High a:5Legoktm>3None [17:41:51] 6Collaboration-Team-Backlog, 10Echo, 7Browser-Support-Apple-Safari, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22): Echo looks messed up on Commons for Safari, old Firefox - https://phabricator.wikimedia.org/T112552#1642293 (10Legoktm) a:3Catrope We reverted for now, but this is likely still an... [17:43:41] 3Collaboration-Team-Current, 10Echo, 6Editing-Department, 6Performance-Team, and 3 others: Fivefold increase in render-blocking CSS size for logged-in users due to Echo loading OOUI on all page views - https://phabricator.wikimedia.org/T112401#1642306 (10Legoktm) [17:45:31] (03PS5) 10Legoktm: Only load ext.echo.ui if the user clicks the echo badge [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) (owner: 10Mooeypoo) [17:47:53] 3Collaboration-Team-Current, 10Echo, 5Patch-For-Review, 7WorkType-NewFunctionality: Echo icon FOUC (tracking) - https://phabricator.wikimedia.org/T112104#1642333 (10Catrope) a:3Mooeypoo [17:52:48] (03PS1) 10Matthias Mullie: Clear LinkCache when failing to fetch Flow board [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238503 (https://phabricator.wikimedia.org/T111830) [17:55:18] 3Collaboration-Team-Current, 10Flow, 10MediaWiki-Redirects, 6operations, and 2 others: Flow notification links on mobile point to desktop - https://phabricator.wikimedia.org/T107108#1642391 (10DannyH) [18:13:26] (03CR) 10Matthias Mullie: [C: 04-1] Add artificial primary key to flow_wiki_ref and flow_ext_ref (032 comments) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238393 (https://phabricator.wikimedia.org/T109676) (owner: 10Matthias Mullie) [18:23:49] (03CR) 10Matthias Mullie: Add artificial primary key to flow_wiki_ref and flow_ext_ref (031 comment) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238393 (https://phabricator.wikimedia.org/T109676) (owner: 10Matthias Mullie) [18:24:29] (03PS2) 10Matthias Mullie: Add artificial primary key to flow_wiki_ref and flow_ext_ref [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238393 (https://phabricator.wikimedia.org/T109676) [18:26:38] (03CR) 10TheDJ: "You can also double this up. With this change as you have it now, you are now ensuring that the module is loaded." [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) (owner: 10Mooeypoo) [18:36:40] (03CR) 10TheDJ: "hm, the example might not be fully there anymore i notice." [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) (owner: 10Mooeypoo) [18:37:37] 3Collaboration-Team-Current, 10Flow, 7Performance: Flow JS loading / defered loading is very slow - https://phabricator.wikimedia.org/T108313#1642538 (10DannyH) [18:38:52] (03CR) 10Mooeypoo: "But we don't want to load it in every page load, even if it's non blocking. There are many users who won't ever click the echo badge in th" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) (owner: 10Mooeypoo) [18:40:36] mooeypoo: maybe only preload if "messages > 1" ? [18:40:41] eh > 0 [18:40:43] thedj, why? [18:40:48] why not just load it on demand? [18:40:49] because you click will be faster [18:41:03] delay in clicks are freaking annoying [18:41:05] But we want to avoid sending unnecessary payload. [18:41:23] thedj, there's a delay anyways because the popup loads the notifications from the API though [18:41:36] hmm. [18:41:42] that's done alongside loading ooui, so odds are the notifs take longer than the ooui load [18:41:55] so you'll see a slight delay in the popup showing up *loading state* for the notifications [18:42:10] I think that's better than sending the payload unnecessarily [18:42:32] thedj, but we could try and research this more [18:42:50] I wonder if we can test how many users that have >0 notifications click the popup in the same session [18:42:58] yeah, measureing this might be a good idea. [18:43:04] if it's 90%+ then maybe we should preload [18:46:14] 3Collaboration-Team-Current, 10Flow, 10MediaWiki-Redirects, 6operations, and 3 others: Flow notification links on mobile point to desktop - https://phabricator.wikimedia.org/T107108#1642576 (10Josve05a) [18:47:22] how do we measure unread + didn't click? [18:50:06] measure start, measure click, extrapolate to pageviews with start but without click ? [18:54:40] oh, that's easy :) [18:55:08] we can send that to graphite [18:55:15] seems like intersting data regardless [18:55:19] :) [18:56:59] mooeypoo: your patch is still using jquery.spinner? [19:03:28] legoktm, im fixin it up, let me push [19:04:06] (03PS6) 10Mooeypoo: Only load ext.echo.ui if the user clicks the echo badge [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 [19:04:12] legoktm, ^^ [19:06:34] thedj: BTW etonkovidova found out that the Safari bug still happens, when you use the search bar [19:06:37] Which is terrible [19:08:19] (03PS7) 10Mooeypoo: Only load ext.echo.ui if the user clicks the echo badge [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) [19:10:27] But it means I can reproduce things without messing around with mitmproxy [19:36:28] (03Abandoned) 10Sbisson: [wip] Add browser tests to EditTopicSummaryWidget [extensions/Flow] - 10https://gerrit.wikimedia.org/r/237462 (owner: 10Mooeypoo) [19:40:27] 6Collaboration-Team-Backlog, 10Flow, 10VisualEditor: Re-examine the usage of `:` in visualeditor - https://phabricator.wikimedia.org/T112690#1642789 (10Quiddity) 3NEW [19:51:36] (03PS10) 10Sbisson: [wip] EditTopicSummary widget [extensions/Flow] - 10https://gerrit.wikimedia.org/r/236870 (https://phabricator.wikimedia.org/T111102) [19:54:35] (03PS11) 10Sbisson: [wip] EditTopicSummary widget [extensions/Flow] - 10https://gerrit.wikimedia.org/r/236870 (https://phabricator.wikimedia.org/T111102) [20:00:34] 6Collaboration-Team-Backlog, 10Echo, 7Browser-Support-Apple-Safari, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22): Echo looks messed up on Commons for Safari, old Firefox - https://phabricator.wikimedia.org/T112552#1642846 (10Catrope) An issue that I think is related: simplesearch also doesn't di... [20:01:37] (03CR) 10Sbisson: [C: 032] "This doesn't break anything locally. I'm merging so we can try it on beta." [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238503 (https://phabricator.wikimedia.org/T111830) (owner: 10Matthias Mullie) [20:04:18] (03Merged) 10jenkins-bot: Clear LinkCache when failing to fetch Flow board [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238503 (https://phabricator.wikimedia.org/T111830) (owner: 10Matthias Mullie) [20:05:46] (03PS2) 10Aaron Schulz: Also defer the markRead() master updates [extensions/Echo] - 10https://gerrit.wikimedia.org/r/237531 (https://phabricator.wikimedia.org/T92357) [20:15:01] mlitn, legoktm: Is https://gerrit.wikimedia.org/r/#/c/205967/ interesting as a concept for a cleaner removal of LQT? Or is it not a starter? [20:17:27] (03PS12) 10Sbisson: [wip] EditTopicSummary widget [extensions/Flow] - 10https://gerrit.wikimedia.org/r/236870 (https://phabricator.wikimedia.org/T111102) [20:17:32] James_F: What about log entries? [20:17:47] RoanKattouw: I didn't say it solved the whole problem. [20:18:06] RoanKattouw: But killing stub namespaces from e.g. search drop-downs would be a bonus, if nothing else. [20:20:33] Uhm.... [20:26:25] "It's a non-starter" is a fine response. :-) [20:32:14] (03CR) 10Gilles: [C: 032] Also defer the markRead() master updates [extensions/Echo] - 10https://gerrit.wikimedia.org/r/237531 (https://phabricator.wikimedia.org/T92357) (owner: 10Aaron Schulz) [20:34:52] (03Merged) 10jenkins-bot: Also defer the markRead() master updates [extensions/Echo] - 10https://gerrit.wikimedia.org/r/237531 (https://phabricator.wikimedia.org/T92357) (owner: 10Aaron Schulz) [20:48:35] 6Collaboration-Team-Backlog, 10Echo, 7Browser-Support-Apple-Safari, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22): Echo looks messed up on Commons for Safari, old Firefox - https://phabricator.wikimedia.org/T112552#1643002 (10Catrope) This was caused by bad CSS in MediaWiki:Vector.css. Should be... [20:48:56] thedj: ---^^ [20:49:01] I'm sure there's a WebKit bug report in there somewhere [20:50:11] (03CR) 10Mooeypoo: [C: 04-1] "Fixing a small issue, don't merge." [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) (owner: 10Mooeypoo) [20:50:47] mooeypoo: You are clairvoyant, clearly [20:50:53] I had just opened that URL in my browser [20:51:03] Or, psychic or something, I guess [20:51:32] RoanKattouw: thx, when i have time, i'll see about reporting a webkit ticket. /me hangs another post-it. [20:53:05] RoanKattouw, but I'm definitely not wearing the same tshirt today. [20:53:28] RoanKattouw: so this was in a ,yes [20:54:57] RoanKattouw, $( '.someelement' ).click( function () { ... } ); <-- if '.someelement' is nonexistent, so the jquery object is empty, it's a noop.. yes? [20:55:03] yup [20:55:17] so, I don't need to have another if ( $blah.length ) {...} at all [20:55:20] no no [20:55:26] This is a very nice pattern actually [20:55:34] Return "error objects" on which chained operations are no-ops [20:55:56] I once saw a talk (at OSCON 2011 I think?) by a BBC iPlayer developer about this pattern [20:56:01] error object has click handler? So .click() won't kill anything if it's called on a nonexisting element [20:56:09] I mean "error object" as a generic term [20:56:15] In the case of jQuery it's just an empty set [20:56:25] oh, oh, ok that makes sense [20:56:32] ok that makes it much more convenient, then [20:56:41] $( '.idontexist' ) returns $( [] ), and .click() does for ( i = 0; .... ) which then runs 0 times [20:57:13] RoanKattouw: k, i'll make a nice testcase from your reduction and will file a webkit ticket [20:57:20] Thanks [20:57:57] Yeah it seems to really just be .works { color: green; } @media screen .doesntwork1 { color: red; } .doesntwork2 { color: red; } [20:58:09] In Chrome, .doesntwork1 is ignored but .doesntwork2 is styled [20:58:12] In Safari, neither is styled [20:58:15] In Firefox, .... ?? [20:58:35] RoanKattouw, YGM [20:58:43] Yeah saw [20:58:50] hm. so..... how crappy would it be to just do $( '#pt-notifications-alert a, #pt-notifications-message a' ).click( function () {} ); and use $( this ) to differentiate? it works, but i'm not sure if it's... nice [20:59:03] It saves some code duplication [21:00:04] (03PS1) 10Matthias Mullie: Make it seem like new workflow creation has no page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238601 (https://phabricator.wikimedia.org/T111830) [21:00:43] mooeypoo: That would be perfectly fine [21:05:48] (03PS13) 10Sbisson: [wip] EditTopicSummary widget [extensions/Flow] - 10https://gerrit.wikimedia.org/r/236870 (https://phabricator.wikimedia.org/T111102) [21:06:08] (03PS14) 10Sbisson: EditTopicSummary widget [extensions/Flow] - 10https://gerrit.wikimedia.org/r/236870 (https://phabricator.wikimedia.org/T111102) [21:08:12] 3Collaboration-Team-Current, 10Echo, 7Easy, 5Patch-For-Review, and 3 others: nojs/mw.echo.special.less styles should only be loaded on Special:Notifications - https://phabricator.wikimedia.org/T112407#1643075 (10Etonkovidova) Betalabs http://en.wikipedia.beta.wmflabs.org/w/load.php?debug=false&lang=en&modu... [21:19:45] 3Collaboration-Team-Current, 10Echo, 5Patch-For-Review, 5WMF-deploy-2015-09-15_(1.26wmf23): 'All notifications' and 'Preferences' buttons shoud fill their containers properly - https://phabricator.wikimedia.org/T112188#1643111 (10Etonkovidova) Checked in betalabs {F2607468} - the area around labels is fu... [21:32:44] mooeypoo: Wikitext talk page converted to Flow [21:50:50] RoanKattouw, did we redeploy echo ? [21:51:26] No? [21:51:27] Why [21:51:32] It seems to appear on Mediawiki? [21:51:33] Is it on mw.org again? [21:51:37] yes... [21:51:45] Oh then twentyafterfour did not follow legoktm's request to pin Echo [21:52:11] ah [21:52:13] what do we do? [21:52:23] isn't it sufficiently fixed now? [21:52:26] I've pinged twentyafterfour in #wikimedia-operations [21:52:26] 'cause the lazy loading is not QUITE ready. It's about to be [21:53:00] Was the weird commons/safari uberbug fixed? [21:53:14] I think so, checking now [21:54:09] 6Collaboration-Team-Backlog, 10Echo, 7Browser-Support-Apple-Safari, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22): Echo looks messed up on Commons for Safari, old Firefox - https://phabricator.wikimedia.org/T112552#1643221 (10TheDJ) Reported upstream as https://bugs.webkit.org/show_bug.cgi?id=149185 [21:58:23] (03PS8) 10Mooeypoo: Only load ext.echo.ui if the user clicks the echo badge [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) [21:58:54] RoanKattouw, THERE. It's ready. ^^ [21:59:03] OK, will look later [21:59:05] Busy now [22:00:48] 6Collaboration-Team-Backlog, 10Echo, 7Browser-Support-Apple-Safari, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22): Broken Common.css on Commons breaks seach suggestions, Echo, others for Safari, old Firefox - https://phabricator.wikimedia.org/T112552#1643236 (10Jdforrester-WMF) [22:02:56] James_F, r. [22:03:31] mooeypoo: James_F, aggressively renaming bugs to explain the cause since 2003. :0) [22:03:49] James_F, And doing a geat job at it! [22:05:15] Hm. My jokes aRRRRRRR unsuccessful. So I shall fix it myself. Here, James_F, you lost an R. [22:05:20] 6Collaboration-Team-Backlog, 10Echo, 7Browser-Support-Apple-Safari, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22): Broken Common.css on Commons breaks search suggestions, Echo, others for Safari, old Firefox - https://phabricator.wikimedia.org/T112552#1643242 (10Mooeypoo) [22:05:42] Oh, right. [22:12:52] 3Collaboration-Team-Current, 10Flow, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22), and 2 others: Opt-in: Guided tour on user talk for first visit to new Flow board - https://phabricator.wikimedia.org/T108266#1643269 (10Etonkovidova) Current screenshots from betalabs. Step 1: {F2607587} If a scre... [22:39:46] dannyh, legoktm should the fonts in monobook badge match the icon, and be black? or do I leave the numbers white despite the fact the icon is black? [22:41:23] black on gray isn't terrible [22:41:33] but it is also red and blue [22:42:05] black on blue is bad [22:42:10] black on red is ehh [22:42:15] lets just continue with white? [22:42:47] * mooeypoo nods [22:43:26] also, is your lazy-load patch mostly finished? if so I'll start writing a follow up to measure how many people click when they have unread items [22:43:29] legoktm, I fixed the glaring sizing issue in monobook. There's still the 1px difference but now that the entire widget opens on demand, it's much less obvious. Was there anything else [22:43:36] yeah it's done waiting for review [22:43:41] i'm working on monobook on top of it [22:43:50] OK I'm going to go to sleep [22:44:02] After I review mooeypoo's patch [22:44:05] RoanKattouw, anything else left that you need me to do? [22:44:24] are we going to backport the lazy-load patch or let it roll out with next weeks train? [22:44:25] (03PS1) 10Mooeypoo: Adjust margins for the notification badge in MonoBook [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238628 [22:44:29] legoktm, you remember what bug was associated with it? [22:44:32] ^ [22:44:45] no, I think it just got mixed in the other bugs [22:44:46] mooeypoo: Test https://gerrit.wikimedia.org/r/238194 and merge it if it works [22:45:17] legoktm: Let's decide later, but I'm leaning towards letting it roll [22:45:33] I'm leaning towards being super cautious [22:45:43] I think we solved the glaring bugs, though. [22:46:14] alright [22:46:22] Too bad we can't have a sort of "beta tester" group we can deploy to [22:47:13] TWN runs Echo [22:47:16] 3Collaboration-Team-Current, 10Flow, 5Patch-For-Review, 5WMF-deploy-2015-09-08_(1.26wmf22), and 3 others: Opt-in for Flow on your own user talk page - https://phabricator.wikimedia.org/T98270#1643394 (10Etonkovidova) Checked in betalabs for enabling 'Flow on user talk' option by - a new user with previo... [22:47:20] Yay the Safari bug appears fixed [22:47:26] woooo [22:47:26] At least for search suggestiosn [22:47:50] yaay [22:47:58] See https://phabricator.wikimedia.org/T112552#1643002 for details of what happened [22:48:13] I saw. Yikesers. [22:48:23] Such a small thing... such huge ripple effect. [22:50:44] by the way. am i the only person thinking that our deploy train is too fast now? [22:51:17] No, you're not the only one [22:51:21] i felt much more comfortable when things went out to mediawiki.org first. [22:51:22] I was thinking the exact same thing [22:51:23] It came up a lot these past few days [22:51:25] yes [22:51:29] Well they do go there first [22:51:31] But only for 2 days [22:51:36] YES. I think RoanKattouw and I talked about this, too. [22:51:54] BTW that would not have helped with the Safari issue as it was specific to Commons [22:52:09] And it would not have helped spot the performance issue since the graphs aren't split out by deployment group yet [22:52:26] (03CR) 10Mooeypoo: [C: 032] "WFM" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238194 (https://phabricator.wikimedia.org/T107655) (owner: 10Legoktm) [22:52:26] It could have helped with some of the FOUC stuff [22:52:47] yeah some of the jumpiness complaints [22:53:29] so now that the lazy load is done and monobook is somewhat adjusted, should I still work on the
  • widget to prevent jumping *after* load, even though it's probably not as noticeable now ? [22:54:14] (03CR) 10Catrope: [C: 04-1] "Looks good, minor comments inline" (034 comments) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) (owner: 10Mooeypoo) [22:54:28] mooeypoo: Nah don't worry about that for now [22:55:57] (03CR) 10Catrope: Only load ext.echo.ui if the user clicks the echo badge (032 comments) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238389 (https://phabricator.wikimedia.org/T112401) (owner: 10Mooeypoo) [22:57:45] (03CR) 10Catrope: [C: 032] Make it seem like new workflow creation has no page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238601 (https://phabricator.wikimedia.org/T111830) (owner: 10Matthias Mullie) [22:58:20] (03PS1) 10Legoktm: Measure how likely users open flyout if they have unseen notifications [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238632 [22:58:24] mooeypoo: ^ [22:58:27] bbl in 10 [22:58:57] OK I'm going to sleep for real now [22:59:06] I have to leave the house at 8 tomorrow and it's already 1 [22:59:27] I chose a crazy week to work remotely :S [23:01:28] (03Merged) 10jenkins-bot: Don't show orange alert if you are viewing your user talk page [extensions/Echo] - 10https://gerrit.wikimedia.org/r/238194 (https://phabricator.wikimedia.org/T107655) (owner: 10Legoktm) [23:01:32] (03Merged) 10jenkins-bot: Make it seem like new workflow creation has no page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/238601 (https://phabricator.wikimedia.org/T111830) (owner: 10Matthias Mullie) [23:22:56] 6Collaboration-Team-Backlog, 10Flow: "There are no older topics" should not be on a new Flow pages - https://phabricator.wikimedia.org/T112714#1643528 (10Etonkovidova) 3NEW [23:58:54] 3Collaboration-Team-Current, 10Echo, 5Patch-For-Review, 5WMF-deploy-2015-09-15_(1.26wmf23), 7WorkType-NewFunctionality: Echo waits to show you messages on second load - https://phabricator.wikimedia.org/T112186#1643588 (10Etonkovidova) Checked in betalabs. There is a slight issue with Alerts - if there...