[00:16:34] 3Parsoid, VisualEditor, VisualEditor-Tables: Entering a single '-' character in a table results in a new row - https://phabricator.wikimedia.org/T88318#1013344 (10ssastry) p:5Triage>3Normal [00:16:58] 3Parsoid, VisualEditor, VisualEditor-Tables: Entering a single '-' character in a table results in a new row - https://phabricator.wikimedia.org/T88318#1009345 (10ssastry) Thank you. We should be able to reproduce this with a test case and fix this now. [00:17:43] (03CR) 10Esanders: [C: 04-1] "IE9&10 support neither MutationObservers or DOMNodeInsertedIntoDocument. I'm convinced its pretty much impossible to mock such an event in" [oojs/ui] - 10https://gerrit.wikimedia.org/r/176681 (https://phabricator.wikimedia.org/T64174) (owner: 10Prtksxna) [00:19:18] (03CR) 10Esanders: "High priority" [oojs/ui] - 10https://gerrit.wikimedia.org/r/188487 (owner: 10Esanders) [00:19:34] (03CR) 10Esanders: "(fixes MenuLayout in FF)" [oojs/ui] - 10https://gerrit.wikimedia.org/r/188487 (owner: 10Esanders) [00:27:35] (03CR) 10Catrope: "recheck" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187610 (owner: 10Ryasmeen) [00:27:50] (03CR) 10Catrope: [C: 032] [BrowserTest] Test for checking General Settings for media [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187610 (owner: 10Ryasmeen) [00:30:18] (03Merged) 10jenkins-bot: [BrowserTest] Test for checking General Settings for media [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187610 (owner: 10Ryasmeen) [00:30:52] (03PS1) 10Esanders: Fix BookletLayout#toggleOutline to use MenuLayout method [oojs/ui] - 10https://gerrit.wikimedia.org/r/188491 [00:33:05] RoanKattouw: do you know if you are planning on doing something around displaying new line marks in VE [00:33:22] I mean, when you have for instance text followed by new line and a category [00:36:12] We're not planning anything around that right now [00:44:35] No plans at all^M^M [00:45:32] RoanKattouw: https://gerrit.wikimedia.org/r/#/c/188297/ looks good to me. Should I merge it, or wait for someone in the VE team proper? [00:45:46] (03PS1) 10Catrope: Improve references test data [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188494 [00:52:21] ori: +1 it but get someone else on the team to merge it [01:00:57] (03CR) 10Esanders: [C: 032] Word character class regex [unicodejs] - 10https://gerrit.wikimedia.org/r/185103 (owner: 10Divec) [01:01:49] RoanKattouw, https://gerrit.wikimedia.org/r/#/c/188487/ [01:01:52] (03Merged) 10jenkins-bot: Word character class regex [unicodejs] - 10https://gerrit.wikimedia.org/r/185103 (owner: 10Divec) [01:04:48] edsanders: https://gerrit.wikimedia.org/r/188494 [01:06:57] (03PS2) 10Catrope: Fix BookletLayout#toggleOutline to use MenuLayout method [oojs/ui] - 10https://gerrit.wikimedia.org/r/188491 (https://phabricator.wikimedia.org/T88471) (owner: 10Esanders) [01:07:05] (03CR) 10Catrope: [C: 032] Fix BookletLayout#toggleOutline to use MenuLayout method [oojs/ui] - 10https://gerrit.wikimedia.org/r/188491 (https://phabricator.wikimedia.org/T88471) (owner: 10Esanders) [01:10:51] (03Merged) 10jenkins-bot: Fix BookletLayout#toggleOutline to use MenuLayout method [oojs/ui] - 10https://gerrit.wikimedia.org/r/188491 (https://phabricator.wikimedia.org/T88471) (owner: 10Esanders) [01:13:01] (03CR) 10Ori.livneh: [C: 031] "edsanders: merge?" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188297 (https://phabricator.wikimedia.org/T88259) (owner: 10Catrope) [01:14:26] (03CR) 10Esanders: [C: 032] Improve references test data [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188494 (owner: 10Catrope) [01:16:21] (03Merged) 10jenkins-bot: Improve references test data [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188494 (owner: 10Catrope) [01:16:38] (03CR) 10Esanders: [C: 032] Make API module's LinkCache data transmission format more efficient [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188297 (https://phabricator.wikimedia.org/T88259) (owner: 10Catrope) [01:18:22] (03Merged) 10jenkins-bot: Make API module's LinkCache data transmission format more efficient [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188297 (https://phabricator.wikimedia.org/T88259) (owner: 10Catrope) [01:27:55] RoanKattouw, so I think I need to make a separate mw.wikiEditor api that loads on every edit/view page so we can log things like the user clicking the edit link and share the logging code [01:29:03] (03CR) 10Esanders: "While open you can't open find and replace. Trevor is rewriting the window manager then will add auto close functionality." [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/188280 (owner: 10Esanders) [01:36:39] RoanKattouw, I don't like it much, but actually VE does this with ext.visualEditor.viewPageTarget.init [01:38:14] Why not use mw.track directly? [01:39:07] it does use mw.track [01:39:56] it also does some juggling of previous Edit events, a session, etc. [01:47:55] Oh, OK [01:48:08] Yeah so like the equivalent of ve.init.mw.trackSubscriber.js, roughly? [01:50:36] 3VisualEditor: [[Linkhttps://en.wikipedia.org/wiki/Link|Link]] is always wrong - https://phabricator.wikimedia.org/T88502#1013480 (10Whatamidoing-WMF) 3NEW [01:53:43] RoanKattouw, yes, seems so [01:55:47] 3VisualEditor: [Regression pre-wmf16] Media Settings dialog is broken - https://phabricator.wikimedia.org/T88503#1013488 (10Ryasmeen) 3NEW a:3Mooeypoo [01:58:39] 3VisualEditor: [[Linkhttps://en.wikipedia.org/wiki/Link|Link]] is always wrong - https://phabricator.wikimedia.org/T88502#1013497 (10Catrope) I have seen this anecdotally a few times, but only in the context of VisualEditor **removing** this kind of stuff. [02:02:01] 3VisualEditor: [[Linkhttps://en.wikipedia.org/wiki/Link|Link]] is always wrong - https://phabricator.wikimedia.org/T88502#1013498 (10Whatamidoing-WMF) I found this because the removal was tagged, but the creation of the problem was back a few revisions in the edit history. [02:07:57] 3VisualEditor-MediaWiki, VisualEditor: The undo link in the page history leads to the source editor and not VisualEditor - https://phabricator.wikimedia.org/T78550#1013501 (10Risker) Suggest that specifically the undo link should lead to the editing interface selected in the user preferences or, in the absence o... [02:13:32] 3VisualEditor: Inserting comment before reference ( tag) seems to work but does not - https://phabricator.wikimedia.org/T85556#1013511 (10Etonkovidova) Sems to be happening only when a line starts with a Citation. The added comments comments might be saved(or not as in the described issue), but they won't... [02:14:00] 3VisualEditor: Inserting comment before reference ( tag) seems to work but does not - https://phabricator.wikimedia.org/T85556#1013512 (10Etonkovidova) p:5Triage>3Normal [02:19:51] RoanKattouw, https://phabricator.wikimedia.org/T88503 [02:24:03] RoanKattouw, I've realised that the hidden field hack won't work for the edit button event [02:24:14] as it's just a link [02:28:34] Ugh right [02:29:18] You could call eventlogging from the click handler of the link, I think that works because EL uses a synchronous way to log [02:29:23] But you should really ask Ori about that [02:29:29] (03CR) 10Catrope: [C: 04-1] Use CSS overriding trick to support RTL in menu layouts (031 comment) [oojs/ui] - 10https://gerrit.wikimedia.org/r/188487 (owner: 10Esanders) [02:29:37] I missed the context. What's the question? [02:29:42] (03PS2) 10Esanders: Use CSS overriding trick to support RTL in menu layouts [oojs/ui] - 10https://gerrit.wikimedia.org/r/188487 [02:30:02] (03CR) 10Catrope: [C: 032] Use CSS overriding trick to support RTL in menu layouts [oojs/ui] - 10https://gerrit.wikimedia.org/r/188487 (owner: 10Esanders) [02:30:33] (03PS3) 10Catrope: Use CSS overriding trick to support RTL in menu layouts [oojs/ui] - 10https://gerrit.wikimedia.org/r/188487 (https://phabricator.wikimedia.org/T88503) (owner: 10Esanders) [02:30:46] (03CR) 10Catrope: [C: 032] Use CSS overriding trick to support RTL in menu layouts [oojs/ui] - 10https://gerrit.wikimedia.org/r/188487 (https://phabricator.wikimedia.org/T88503) (owner: 10Esanders) [02:31:11] ori: Krenair needs to fire an eventlogging event when the user clicks the edit button for the wikitext editor [02:31:13] Which is just a link [02:31:52] So when the user clicks it the browser will navigate away and trying to do async stuff from the click handler will probably not work [02:32:08] we also need to remember that this was done on the next page [02:32:18] so the timing for ready can be calculated [02:34:29] right [02:35:01] (03Merged) 10jenkins-bot: Use CSS overriding trick to support RTL in menu layouts [oojs/ui] - 10https://gerrit.wikimedia.org/r/188487 (https://phabricator.wikimedia.org/T88503) (owner: 10Esanders) [02:40:45] ori? [02:57:25] Krenair, RoanKattouw: EventLogging will use sendBeacon() when it is available, and an Image element when it is not. [02:58:06] One possible approach for getting a comparison of VE and WE load times going is to only log from clients that implement sendBeacon, even though it's not needed for VE, which is in-page [02:58:53] that way the data will not be totally representative because it will only originate from users with very modern browsers, but the comparison between VE and WE would still be valid. [03:01:44] ori, so how will this allow us to recall the timestamp of the most recent 'init' action, for example? :/ [03:02:38] Krenair: let's zoom out for a second -- what are you trying to do? [03:03:16] log Schema:Edit entries from WikiEditor [03:03:16] so it can be compared to VE [03:03:27] one of the things that needs to be done is capturing when the edit button was clicked [03:03:46] you can get that on the edit page itself via the navigation timing API [03:04:02] performance.timing.navigationStart [03:04:07] and then recall this timestamp on the edit page to calculate how long it took to get ready [03:04:33] i'm talking about the edit page [03:04:47] "https://dvcs.w3.org/hg/webperf/raw-file/tip/specs/NavigationTiming/Overview.html : ""This attribute must return the time immediately after the user agent finishes prompting to unload the previous document. If there is no previous document, this attribute must return the time the current document is created." [03:05:04] ok [03:05:13] so we'd just leave those events out in WikiEditor? [03:05:23] ? [03:06:02] oh, we could both log the init event and use that to calculate for ready. ok [03:06:26] I think we're talking across each other a little. Why are we instrumenting the old wikitext editor? I know we had committed to doing this at some point in the past, but I don't think it makes sense. Did someone (James?) explicitly ask you to work on this, or are you just being proactive here? If the latter, I think we might want to think twice about whether it's really worth the trouble. [03:06:48] https://phabricator.wikimedia.org/T88027 [03:08:37] if you have other things to work on, I'd suggest waiting until I have a chance to take this up with Roan / Trevor / Ed / James tomorrow [03:08:47] if not, go for it [03:09:38] I'm certainly not planning to do any more work on it at this hour :) [03:11:35] hah, right [03:11:38] * ori is a PST chauvinist. [03:12:09] Can anyone tell me where oo-ui-toolbar is referenced and setup in VE? [03:14:22] A search yields only browser tests, libraries, and CSS [03:20:48] hmmm... ve.init.Target.prototype.getToolbar ? [03:21:53] Krenair: looks like that's in /lib/ve/src/init/ve.init.Target.js [03:22:46] yep [03:24:56] Krenair: Looks good. Thanks. [14:44:22] 3VisualEditor-ContentLanguage, VisualEditor: VisualEditor: a line started with Japanese input disappear when I switch to and start non-Japanese (latin) input - https://phabricator.wikimedia.org/T74508#1014577 (10whym) @Jdforrester-WMF: do you have any idea which change fixed it? We'd want to link it here, if pos... [15:57:31] (03CR) 10Ricordisamoa: "@Catrope: thanks." [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187618 (https://phabricator.wikimedia.org/T87420) (owner: 10Mooeypoo) [16:07:01] (03PS1) 10Jforrester: Tag v0.6.6 [oojs/ui] - 10https://gerrit.wikimedia.org/r/188568 [16:07:25] Krenair: Can I get a +2 on https://gerrit.wikimedia.org/r/188568 ? Need to do an urgent OOjs UI release for the wmf16 cut. :-( [16:11:51] 3VisualEditor-MediaWiki, VisualEditor: Extra space before the start of the citation/template (not multiple transclusion) dialog contents - https://phabricator.wikimedia.org/T78045#1015053 (10Jdforrester-WMF) 5Open>3Resolved [16:16:44] 3OOjs-UI: MediaWiki theme: Checkboxes and radios look oversized on desktop - https://phabricator.wikimedia.org/T86003#1015066 (10Jdforrester-WMF) [16:17:59] 3OOjs-UI: TextInputWidget autosizing no longer works on Firefox - https://phabricator.wikimedia.org/T88212#1015071 (10Jdforrester-WMF) [16:20:00] 3OOjs-UI, VisualEditor, VisualEditor-InterfaceLanguage: [Regression pre-wmf16] Media Settings dialog is broken - https://phabricator.wikimedia.org/T88503#1015075 (10Jdforrester-WMF) a:5Mooeypoo>3Esanders [16:20:30] 3OOjs-UI, VisualEditor: BookletLayout dies on open - https://phabricator.wikimedia.org/T88471#1015079 (10Jdforrester-WMF) [16:22:40] 3OOjs-UI: MediaWiki theme: Checkboxes and radios look oversized on desktop - https://phabricator.wikimedia.org/T86003#1015094 (10Jdforrester-WMF) [16:23:09] 3OOjs-UI, VisualEditor: BookletLayout dies on open - https://phabricator.wikimedia.org/T88471#1015105 (10Jdforrester-WMF) 5Open>3Resolved [16:23:19] 3OOjs-UI, VisualEditor, VisualEditor-InterfaceLanguage: [Regression pre-wmf16] Media Settings dialog is broken - https://phabricator.wikimedia.org/T88503#1015106 (10Jdforrester-WMF) 5Open>3Resolved [16:48:29] (03CR) 10Mooeypoo: [C: 032] Tag v0.6.6 [oojs/ui] - 10https://gerrit.wikimedia.org/r/188568 (owner: 10Jforrester) [16:49:27] Krenair: Never mind, mooeypoo's done it. :-) [16:51:21] (03Merged) 10jenkins-bot: Tag v0.6.6 [oojs/ui] - 10https://gerrit.wikimedia.org/r/188568 (owner: 10Jforrester) [16:53:46] (03PS1) 10Jforrester: Update OOjs UI to v0.6.6 [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/188578 [16:54:30] (03CR) 10Mooeypoo: [C: 032] Update OOjs UI to v0.6.6 [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/188578 (owner: 10Jforrester) [16:56:17] (03Merged) 10jenkins-bot: Update OOjs UI to v0.6.6 [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/188578 (owner: 10Jforrester) [17:01:31] (03CR) 10Jforrester: [C: 032] "Not any more it isn't…" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187634 (owner: 10Mooeypoo) [17:03:47] (03Merged) 10jenkins-bot: Adjust to ooui's show/hide css change [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187634 (owner: 10Mooeypoo) [17:30:36] Corruption alert: visualeditor-needcheck on cswiki: https://cs.wikipedia.org/?diff=12207560 [17:30:36] Corruption alert: visualeditor-needcheck on frwiki: https://fr.wikipedia.org/?diff=111550391 [17:30:36] Corruption alert: visualeditor-needcheck on frwiki: https://fr.wikipedia.org/?diff=111555965 [17:30:36] Corruption alert: visualeditor-needcheck on frwiki: https://fr.wikipedia.org/?diff=111556073 [17:30:36] Corruption alert: visualeditor-needcheck on frwiki: https://fr.wikipedia.org/?diff=111556128 [17:30:36] Corruption alert: visualeditor-needcheck on frwiki: https://fr.wikipedia.org/?diff=111556210 [17:30:36] Corruption alert: visualeditor-needcheck on zhwiki: https://zh.wikipedia.org/?diff=34200449 [17:30:37] Corruption alert: visualeditor-needcheck on zhwiki: https://zh.wikipedia.org/?diff=34206378 [17:30:37] Corruption alert: visualeditor-needcheck on ukwiki: https://uk.wikipedia.org/?diff=15568361 [17:30:38] Corruption alert: visualeditor-needcheck on ruwiki: https://ru.wikipedia.org/?diff=68367478 [17:30:38] Corruption alert: visualeditor-needcheck on enwiki: https://en.wikipedia.org/?diff=645593885 [17:30:42] Ick. [17:37:27] yikes [17:41:29] James_F, I was busy at the time, sorry [17:41:31] just saw IRC [17:41:40] Krenair: No worries. :-) [17:41:58] in History.md why is 0.6.4-0.6.6 marked as 2014? [17:42:08] Oh, whoops, are they? [17:42:46] Because I wrote it too quickly. :-) [17:53:41] hi, i am trying to implement OOjs UI on my page and get this error: Call to undefined method OutputPage::enableOOUI() [17:54:08] Could someone please help with this? I am totally new to OOjs UI [17:54:14] :) [17:54:32] how old is your clone of MW core? [17:55:23] Krenair: is that question for me? [17:55:35] yes [17:56:04] Krenair: i installed in october, but keep on updating usually [17:56:19] what commit hash? [17:57:22] Krenair: the version page says "1.25alpha (c05b4c9) 04:06, 19 December 2014" [17:57:44] that function it requires was added a week ago [17:57:52] https://gerrit.wikimedia.org/r/#/c/182875/ [17:57:57] oh, updating again :) [18:20:30] (03PS5) 10Mooeypoo: Use css class rather than jQuery show/hide [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/187617 [18:29:46] (03PS6) 10Mooeypoo: Use css class rather than jQuery show/hide [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/187617 [18:29:57] (03CR) 10Mooeypoo: Use css class rather than jQuery show/hide (033 comments) [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/187617 (owner: 10Mooeypoo) [18:33:27] (03PS6) 10Mooeypoo: Use css class instead of jquery show/hide [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187618 (https://phabricator.wikimedia.org/T87420) [18:40:40] (03CR) 10Mooeypoo: Use css class instead of jquery show/hide (036 comments) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187618 (https://phabricator.wikimedia.org/T87420) (owner: 10Mooeypoo) [18:42:03] (03PS7) 10Mooeypoo: Use css class instead of jquery show/hide [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187618 (https://phabricator.wikimedia.org/T87420) [18:44:30] (03CR) 10jenkins-bot: [V: 04-1] Use css class instead of jquery show/hide [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/187618 (https://phabricator.wikimedia.org/T87420) (owner: 10Mooeypoo) [18:52:23] 3OOjs-UI, VisualEditor, VisualEditor-InterfaceLanguage: [Regression pre-wmf16] Media Settings dialog is broken - https://phabricator.wikimedia.org/T88503#1015506 (10Ryasmeen) Verified in Betalabs [18:59:11] James_F: https://phabricator.wikimedia.org/T88027 blocker? [19:15:43] Krenair: Hey. Having talked about the instrumentation with Ori, we think that for timing data instead of trying to measure it (and coming up with approximations and confusion), instead we should just return '0' for all timing measurements on the WikiEditor instrumentation. Does that work for you? [19:16:12] ok... [19:16:19] can we still get useful data from that? [19:16:51] Yeah, the funnel numbers (how many people went ready -> saveAttempt) are the most valuable stuff anyway. [19:16:54] I guess the timestamp of the event being logged along with the session id should be enough to calculate those later.. [19:17:14] Well, network latency isn't constant. [19:17:22] So for sub-second events it'd be very noisy. [19:26:32] 3VisualEditor: VisualEditor: Cannot enable or disable formatting of lines starting with semicolon markup - https://phabricator.wikimedia.org/T73721#1015655 (10Etonkovidova) p:5Triage>3Normal [19:29:19] hmm, we changed the story points thingy values? [19:29:51] MatmaRex: Yeah, TrevorParscal changed the scale to be something saner. [19:30:09] i would recommend using words rather than magic numbers, but oh well… [19:31:37] (we already have #Easy and #Epic, why not add a couple steps inbetween?) [19:32:00] (but ah, agile, scrum, smarter people than me, &c.) [19:34:01] 3VisualEditor: Adding items to lists sometimes generates spurious line feeds - https://phabricator.wikimedia.org/T74479#1015679 (10Etonkovidova) p:5Triage>3Low [19:35:16] James_F, so what is 8 then? [19:36:28] MatmaRex: Except you can't put words into that field, in the Phabricator extension that we have installed. [19:36:42] Krenair: 8 == "will take a few days to do". [19:36:59] James_F: sounds like a bug^H^H^H feature request to me! [19:37:23] didn't we write that extension ourselves, anyway? or WMDE or something? [19:37:40] also, we stick just about anrything else in "projects". could stick this as well. [19:37:42] meh [19:38:37] MatmaRex: WMDE probably did. [19:38:52] MatmaRex: Anyway, aren't you meant to be too busy to work with us? ;-) [19:39:21] yes. i only come here to grumble. [19:39:26] :-) [19:39:35] ;) [19:39:45] Krinkle: Can https://gerrit.wikimedia.org/r/#/c/188294/ be merged please? [19:40:37] (03CR) 10Jforrester: [C: 032] Followup a92661781: fix attribute resolution [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/188294 (owner: 10Catrope) [19:40:59] 3VisualEditor, VisualEditor-Performance: Unnecessary opacity animation for

on load - https://phabricator.wikimedia.org/T88590#1015707 (10ori) [19:41:04] RoanKattouw: ^ [19:41:33] ori: It is deliberate. [19:41:51] ori: Is it really blocking? [19:41:53] it's costly. how important is it? [19:42:16] it's not blocking, but it happens during the busiest time for ve initialization [19:42:39] what is the purpose of it? [19:42:40] It's mildly important to help the user focus on the right bits of the page. We already have complaints about read and edit mode being too alike. I could be convinced to scrap it. [19:42:42] (03Merged) 10jenkins-bot: Followup a92661781: fix attribute resolution [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/188294 (owner: 10Catrope) [19:43:00] James_F: does it need to be *animated*? [19:43:29] I don't think it does [19:43:41] AFAICT it's initially opacity: 1 but covered by a semi-opaque overlay [19:44:18] ugh, i attached the wrong screenshot [19:44:45] 3VisualEditor, VisualEditor-Performance: Unnecessary opacity animation for

on load - https://phabricator.wikimedia.org/T88590#1015713 (10ori) [19:44:46] better [19:44:55] Hmm, no, that doesn't seem to be true [19:45:06] ori: Does the opacity animation happen immediately after clicking edit? [19:45:09] no, wrong again, i clearly need coffee [19:45:14] i.e. during what is normally the network phase of the load process? [19:45:39] 3VisualEditor, VisualEditor-Performance: Unnecessary opacity animation for

on load - https://phabricator.wikimedia.org/T88590#1015701 (10ori) [19:45:40] (03PS1) 10Jforrester: MWTransclusionDialog: Add a little margin to the addTemplateFieldset [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188614 [19:45:46] RoanKattouw: not sure, let me see [19:45:49] it's easy to repro btw [19:45:52] i can come by and show you [19:46:00] OK please do [19:46:08] (03PS1) 10Jforrester: Update VE core submodule to master (55e3589) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188616 [19:46:55] Ooh I was wrong [19:47:04] Instead of an overlay, it's just applying opacity: 0.6 to a bunch of things [19:47:06] yes, it happens very early on [19:47:36] 3VisualEditor, VisualEditor-Performance: Unnecessary opacity animation for

on load - https://phabricator.wikimedia.org/T88590#1015718 (10Krinkle) It's a CSS animation though, not jQuery. And with no layout change impact afaik (only opacity). The transition is specified in CSS as taking 200ms, but that shou... [19:48:42] ori: We saw this one earlier this week I think. [19:49:41] The animation helps a little in transitioning things smoothly and with persistence. We can discuss from a design perspective, but I'm curious whether it's notable from a perf perspective with it having no layout impact and being on the css thread. [19:50:24] We could reduce the animations though by merging a few of them. [19:51:07] 3VisualEditor, VisualEditor-Performance: Unnecessary opacity animation for

on load - https://phabricator.wikimedia.org/T88590#1015721 (10Catrope) What's weird about this opacity transition is that `opacity: 0.6;` is applied two different ways: - To `h1#firstHeading` and `div#siteSub` through `class="ve-ini... [19:51:45] 3VisualEditor, VisualEditor-Performance: Unnecessary opacity animation for

on load - https://phabricator.wikimedia.org/T88590#1015722 (10ori) >>! In T88590#1015718, @Krinkle wrote: > I believe the transition is desired for UX and (other than general C/GPU cost) shouldn't be interfering with the javascript e... [19:53:28] ori: does that apply to transition as well (as opposed to animation) [19:58:26] 3VisualEditor, VisualEditor-Performance: Unnecessary opacity animation for

on load - https://phabricator.wikimedia.org/T88590#1015737 (10Krinkle) @Catrope Yeah, we should consolidate those. Maybe we can use an overall class and use descendant selectors instead of adding it to individual elements in JavaScri... [19:58:38] Krinkle: that's slightly trickier to test, but I don't see why that wouldn't be the case. And besides, even if there was explicit intent behind some of these transitions, what actually happens does not seem to be entirely intentional. [19:59:44] Krinkle: btw, re $.fn.show / hide -- I had a (non-ve-specific) idea, which is that we override them in MediaWiki to toggle a class that we define in our core CSS [20:00:10] basically do what jQuery would like to do but cannot because of an inability to enforce such a standard across the board. [20:00:15] 3OOjs-UI: [Suggestion for improvement] The difference between the background colors for disabled button and selected button is not much visibly apparent now - https://phabricator.wikimedia.org/T88592#1015740 (10Ryasmeen) 3NEW [20:03:51] ori: I'm not sure. hide() might work, but show() requires intimate knowledge [20:04:12] you'd have to hardcode the default display values for each element. And then deal with the fact that it can no longer be overridden. [20:04:31] e.g. div.my-div { display: inline-block; } .mw-show { display: block !important; } [20:04:52] 3OOjs-UI, VisualEditor: BookletLayout dies on open - https://phabricator.wikimedia.org/T88471#1015763 (10Ryasmeen) Verified the fix [20:05:35] nod [20:14:02] ori, James_F: What about the session ID? [20:14:18] what about it? [20:14:37] that needs to be carried across pages presumably [20:15:38] I'm really not sure. One ought to consider both legal and technical implications. Nuria, milimetric, or halfak on #wikimedia-analytics would know. [20:18:07] Krenair: I'm having a hard time piecing together context from above, but feel free to ask in -analytics [20:19:11] milimetric, https://phabricator.wikimedia.org/T88027 [20:19:54] "A string of 32 alphanumeric characters, unique to the current page view session; used for grouping events." [20:20:22] on the editingSessionId description at https://meta.wikimedia.org/wiki/Schema:Edit - meaningful in VE, not so much for WikiEditor [20:20:24] Krenair: ok, got it [20:20:53] it's meaningful for wiki editor, and I don't think very different from a legal / technical point of view [20:21:25] really it's a matter of interacting with a server via ajax vs. "plain" requests [20:22:00] but i'll bring it up with others and see if anyone sees anything else wrong with it [20:22:19] milimetric, I mean the "unique to the current page view session" part [20:23:06] WikiEditor does not load in-page like VE, so the edit link click event would have a separate session ID [20:23:42] saveSuccess and saveFailure would also have different session IDs if I kept to that description [20:24:20] Krenair: I know what you mean, but we can change the description of the field to be closer to its intended meaning [20:25:48] sorry, parallel discussion in -analytics [20:25:55] okay, fine, now I need to work out how to best implement keeping that session ID across pages [20:26:08] issues were already brought up with localStorage, can't remember what we said about cookies [20:27:25] Krenair: yeah, from my point of view, as long as the cookie disappears once the edit is done, it matches the VE instrumentation [20:29:10] 3VisualEditor: VisualEditor: adding section for references will suppress the Related Articles one - https://phabricator.wikimedia.org/T55699#1015839 (10neilpquinn) I've attempted to reproduce the two test cases in a sandbox on the Italian Wikipedia. In both cases ([[https://it.wikipedia.org/w/index.php?title=Ute... [20:29:37] this said, we are working on a general opt-out mechanism for this kind of logging [20:29:39] well the user could just disappear half way through, and never hit saveSuccess/saveFailure/abort [20:29:53] but still keep the cookie [20:30:32] 3OOjs-UI: ButtonSelectWidget doesn't unhighlight when an item gets unselected - https://phabricator.wikimedia.org/T88232#1015844 (10Ryasmeen) Verified the fix in Betalabs [20:32:05] milimetric, ^ [20:32:08] Krenair: right, so we should try and figure out what a good expiration time period is, and always expire the cookie when people do the "first" action in the pipeline [20:32:32] first action: expire any old cookies and issue new one [20:32:53] their first action could be either init or ready [20:33:13] any action in the funnel: increase expiration date to some period of time that we think the vast majority will not exceed in a single editing session [20:33:22] or even perhaps saveSuccess/saveFailure, not sure [20:33:30] ? :) [20:33:46] how can it be saveFailure... like if they come to it via api or something? [20:34:05] api won't send them there [20:35:00] the user being faster than the JS here is probably ridiculous, ignore that line [20:35:12] faster than JS? [20:35:22] oh i see [20:35:36] right, maybe Bruce Lee... [20:35:37] yeah so if they got the edit form but submitted before it could log 'ready' [20:35:43] Krenair: so why not do all this server side? [20:36:24] schema description [20:36:40] "saveSuccess: Editor has successfully saved and user has been shown a post-edit status on the client-side [20:36:40] saveFailure: Editor has not successfully saved and user has been shown a save failure reason on the client-side" [20:37:32] Krenair: well, we know in php when they're about to see a failure and what kind of failure, right? [20:37:50] yes [20:38:05] are you saying the time that it takes to render the failure page would be significant for analysis? [20:38:49] no, I'm saying that the schema says I must do this client-side [20:39:17] I don't know if it could be significant or not [20:40:00] let me think about it for a moment. respecting the schema verbatim is less important than making sure 1. this gets done relatively soon so we have a good baseline experience to compare VE to and 2. keeping true to the intent of the schema [20:40:40] James_F: mind inviting nuria to the channel? [20:40:53] this channel? it's open [20:41:09] oh, oops, i gave her the wrong name [20:41:49] hola [20:41:56] hey [20:42:01] nuria i'll paste you the backlog [20:42:13] hello Krenair [20:42:18] let me catch up [20:42:46] Krenair: i was telling millimetric taht i do not think you need a cookie but let me understand better how things work [20:43:58] ok, so I think we might need to do this client side if we want a good comparison between VE and Wiki editor [20:44:06] Krenair: ok, read backlog, can you explain why couldn't you do this on the server side if you are hitting php [20:44:20] milimetric: why? [20:44:21] VE does this client-side [20:44:21] because the page load time is hard to study independently and that factors into the user experience we're trying to track here [20:44:28] the schema says you do this client-side [20:44:54] Krenair: tehy do it client side cause for their app makes sense, it's a client side app, the schema does not care really [20:45:23] *they is everyone here, including Krenair :) [20:45:32] Krenair: doing it server side for you makes sense because you do not need to set up a cookie (note that in their case this issue does not exist) [20:45:38] but the reason to do it client side is because we're trying to capture *user experience* [20:45:57] so nuria, forgetting the technical implementation details [20:46:14] milimetric: there is no difference where the randomid comes from in terms of UX , is there? [20:46:15] what we want is "how does user experience impact editing" [20:46:46] no, but there is a difference if we say "time from init to ready is X" vs "X + page load" [20:47:14] milimetric: for data for the event yes, for where does the randomid come from no [20:47:16] and logging client side in this case gets us, I think, something more accurate to the schema's intent and what's already being captured for VE [20:47:34] so, first order of business - logging should be done client side [20:47:45] milimetric: correct, logging client side does not imply you generate your token client side [20:47:54] milimetric: makes sense [20:47:58] now, that JS on the client needs to have a shared token for the whole session [20:48:20] so how can we generate that, expire it when the session is over (including page close event) [20:48:31] milimetric: it needs to 'access' a token [20:48:49] which you're saying you can inject from PHP? [20:48:53] milimetric: teh token can come on teh page and be server side generated on step #1 of pipeline [20:49:14] milimetric: like fjhhjg [20:49:59] yep [20:50:14] milimetric: the token is passed among steps with the post among with your data, that is why 1) client can retrieve and log 2) you pass it w/o needing to persist it anywhere [20:50:15] ok, so now the question to Krenair is, can we safely identify the "first step" of the funnel? [20:50:20] no [20:50:28] what are the candidates for "first"? [20:50:36] Krinkle: are you up for taking on https://phabricator.wikimedia.org/T88590 ? [20:51:02] Krenair: you could if you assume there is a token, thus, if no token generate one and call it "first step" , add it to data being passed [20:51:07] init (click the edit link) vs. ready (browse to the edit form and wait for WE to load, via the link at the top of the aritlce, or just manual URL typing, or otherwise) [20:51:51] Krenair: I think we can figure out a way to do it in all those cases, so let's do one by one [20:51:52] Krenair, milimetric: I think if talk on hangout might be more clear. Cause regardless of whether is your first step you can do it. [20:52:00] additionally, passing this session ID via the edit link would involve changing action=edit links, I'm not prepared to do that [20:52:06] Krenair: up for a hangout? [20:52:23] Krenair: yes, it will require changing edit links, but that is easier than managing cookies [20:52:32] hangout? [20:54:20] I guess we could do a hangout... [20:54:27] Krenair: as you would need to add cookie creation/expiration code to your pipeline [20:54:38] Krenair: sounds good. [20:54:49] 3VisualEditor, VisualEditor-Performance: Unnecessary opacity animation for

on load - https://phabricator.wikimedia.org/T88590#1015903 (10Krinkle) a:3Krinkle [20:54:55] * ori hugs Krinkle. [20:54:59] milimetric: can you invite Krenair to cave? [20:55:20] done [20:56:00] nuria: we're in the batcave [21:06:41] 3VisualEditor: VisualEditor Q3 Release Criteria - https://phabricator.wikimedia.org/T86867#1015916 (10Jdforrester-WMF) 5Open>3Resolved Moved into https://phabricator.wikimedia.org/project/view/1015/ description. [21:12:39] James_F, ori: Okay so we can just set the .timing properties to 0 because... we only want to know about percentages of users making it through to each step, not timings between them? [21:19:22] Krenair: Yes. [21:20:55] ok that sounds fine [21:20:55] Krenair: Please let us know if you run into any other issues [21:21:16] James_F, what about changing the action=edit link you get at the top of pages to include an editing session ID? [21:21:19] does that sound OK to you? [21:21:26] Yes, that's fine. [21:26:15] James_F, oh this is Skin::editUrlOptions. I'm sure this will be fun.. [21:26:51] Krenair: :-) [21:30:39] oh right, VE does this with SkinTemplateNavigation [21:37:39] 3VisualEditor: API design and documentation - https://phabricator.wikimedia.org/T40855#1015984 (10TrevorParscal) [21:45:28] (03CR) 10Krinkle: "This event handler is not removed when the editor is deactivated, and it's re-created every time the editor is activated. This creates sta" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/182336 (https://phabricator.wikimedia.org/T78694) (owner: 10Alex Monk) [21:46:05] Krenair: Can you add an unbind for it? Perhaps re-use bindHandlers() [21:55:16] RoanKattouw_away: ori: Hehe, this is why browsers suck: https://github.com/jquery/sizzle/blob/master/src/sizzle.js#L218-L227 [21:55:23] a quick speed up for #ID isn't even trivial anymore. [22:20:49] 3VisualEditor: [[Linkhttps://en.wikipedia.org/wiki/Link|Link]] is always wrong - https://phabricator.wikimedia.org/T88502#1016108 (10Jdforrester-WMF) 5Open>3stalled [22:22:31] RoanKattouw: https://phabricator.wikimedia.org/T52546 [22:23:34] 3VisualEditor: Multi-template 'content' bits should not be editable - https://phabricator.wikimedia.org/T52546#1016112 (10Jdforrester-WMF) 5Open>3declined a:3Jdforrester-WMF Need to do HTML templates first; this is premature. [22:25:33] general question (more parsoid than VE I think?) does a single VE node always correspond with an uninterrupted piece of wikitext? [22:26:27] uninterrupted? [22:29:13] ori: James_F: RoanKattouw: fyi, I'm currently working on a commit for mwext-VisualEditor that uses ve-activated and simliar classes for hiding/showing/fading mw-target stuff instead of show/hide/fadeOut() etc. [22:29:18] as in a continuous piece of wikitext, without pieces of wikitext in between that may not belong to that VE node [22:29:50] Some of it resolves the open task I just self-assigned, but it may also overlap slightly with the overal show/hide removal. Just a heads up that for viewpagetarget, I'm including it in this commit. [22:30:09] (instead of using oo-ui-hide or something) [22:38:01] 3VisualEditor: Random bad stuff happening in VisualEditor - https://phabricator.wikimedia.org/T88612#1016165 (10awight) 3NEW [22:39:25] 3VisualEditor: Random bad stuff happening in VisualEditor - https://phabricator.wikimedia.org/T88612#1016173 (10awight) [22:40:47] MartijnH: i'm not sure what you mean, but… probably not? references can be defined in other places than they are shown [22:41:08] 3Beta-Cluster, VisualEditor: Searching for images in VisualEditor with Firefox is painfully slow in Beta Cluster - https://phabricator.wikimedia.org/T87676#1016176 (10Cmcmahon) I am pretty sure that on or around 16 January VE stopped handling this situation gracefully. [22:44:51] 3VisualEditor: Page bump when editing a section in VE - https://phabricator.wikimedia.org/T88613#1016183 (10awight) 3NEW [22:45:09] Thanks MatmaRex. That's not exactly what I mean - I think a formal way of saying what I mean is that if a wikitext character on position i is part of node n, and a character on position j is part of node n, and j > i, then any character in position k where k > i && k < j will also be part of node n [22:45:28] informal ways of saying things are hard [22:46:53] MartijnH: oh. in that case, no, that is not the case for some broken wikitext (usually misnested html), [22:47:02] like a paragraph directly inside a table [22:47:22] which will be hoisted out and placed before the table, usually [22:47:36] thanks. that at least helps me hate Wiktext more :D [22:48:58] MartijnH: https://test.wikipedia.org/wiki/Misnesting_test [22:49:34] (that is actually rather curious, parsoid treats this differently from the mediawiki parser. i'm sure there are other examples) [22:49:47] oh dear [22:52:06] MartijnH: that hoisting is a HTML feature, by the way [22:53:58] MatmaRex: purely out of interest, a spec feature or something useragents do on malformed HTML? [22:54:16] a quick ctrl-f for hoist finds nothing in the W3C draft [22:54:39] yes, i was just looking it up, actually [22:54:42] https://html.spec.whatwg.org/#foster-parent https://html.spec.whatwg.org/#adoption-agency-algorithm [22:54:46] James_F: Does VE need to support IE8 (in addition to IE9)? E.g. can we use css selectors in ve-mw stylesheets that are IE9+ ? [22:55:05] oh, and a separate section for tables: https://html.spec.whatwg.org/#unexpected-markup-in-tables [22:55:09] MartijnH: [22:56:15] mighty interesting, thank you [23:00:48] (i filed https://phabricator.wikimedia.org/T88618 :) ) [23:05:46] I'm glad you can still :) about it :) [23:13:56] Krinkle: Not in VE, but OOjs UI might. TrevorP|Away? [23:15:00] MartijnH: The hoisting issue is basically things like
.... -->
.... [23:15:19] Also happens to things other than , but that was the example we encountered in practice [23:15:46] James_F: Well, I imagine that pattern will propagate. I believe Moriel was already working on a ve-core equivalent of that commit [23:15:52] MartijnH: However, I think your statement is probably true, unless you deliberately create a hoisting case like that in wikitext [23:16:26] thanks RoanKattouw. I'm currently being amazed. The example MatmaRex showed does demonstrate that it can and does happen [23:17:03] Krinkle: What commit? [23:17:18] James_F: The commit in oojs-ui that replaces show/hide with a hidden class add/remove pattern [23:17:39] MartijnH: Oh wow yeah that's broken [23:17:49] Less broken in Parsoid than in the PHP parser, arguably, but still broken [23:18:50] especially by that MediaWiki apparently emulates the hoisting behaviour rather than emit the malformed HTML and let the useragent sort it out [23:19:46] RoanKattouw: according to the foster parent parts of the spec, the PHP parsers behaviour seems to emulate what useragents are supposed to implement [23:19:54] Right [23:20:33] MartijnH: hmm, that might actually just be HTML Tidy doing its thing [23:21:29] PHP parser might as well just be outputting malformed HTML [23:24:02] MatmaRex: PHP parser? Broken?! Never! [23:24:06] Hey sucheta! [23:25:08] James_F, Hi! [23:25:26] Hope you're feeling better. [23:29:32] sucheta, you're alive!! good to see you online! [23:34:22] Oh well. Surviving still. Waking up at 4:30am is new; thanks to the medicines that are messing up my already messed up schedule! [23:35:02] Well that's a change from going to bed at 4:30am ;) [23:35:15] Unless you went to bed at 4:20am [23:40:36] (03PS1) 10Legoktm: extension.json: Add "VisualEditorPluginModules" to "config" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188700 [23:41:06] James_F: ^I just noticed that after the vagrant patch >.> [23:46:39] legoktm: Whoops. [23:46:57] (03CR) 10Jforrester: [C: 032] extension.json: Add "VisualEditorPluginModules" to "config" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/188700 (owner: 10Legoktm)