[00:02:49] (03CR) 10Catrope: Provide move left/right quick action buttons for images (033 comments) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141091 (owner: 10Jforrester) [00:06:31] (03PS1) 10Krinkle: build: Clean up npmignore [oojs/core] - 10https://gerrit.wikimedia.org/r/142466 [00:07:17] 3VisualEditor / 3ContentEditable: VisualEditor: Text-only GeneratedContentNode rerenderings aren't highlighted correctly - 10https://bugzilla.wikimedia.org/67169#c1 (10Ed Sanders) What about after https://gerrit.wikimedia.org/r/140439 ? [00:07:29] (03CR) 10Krinkle: "I've tested this locally by publishing this branch to npm as 'oojs-krinkle'. cd /tmp, npm install oojs oojs-krinkle. Compare the two direc" [oojs/core] - 10https://gerrit.wikimedia.org/r/142466 (owner: 10Krinkle) [00:07:43] (03CR) 10Krinkle: "(I'll unpublish that package once merged)" [oojs/core] - 10https://gerrit.wikimedia.org/r/142466 (owner: 10Krinkle) [00:08:38] (03CR) 10Catrope: [C: 04-1] Provide tools to make text bigger or smaller (037 comments) [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/141617 (https://bugzilla.wikimedia.org/51613) (owner: 10Jforrester) [00:09:13] (03PS1) 10Jforrester: Update OOjs to v1.0.10 [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/142467 [00:10:12] RoanKattouw: https://gerrit.wikimedia.org/r/142465 and https://gerrit.wikimedia.org/r/142467 [00:10:15] (03PS2) 10Krinkle: build: Clean up npmignore [oojs/core] - 10https://gerrit.wikimedia.org/r/142466 [00:12:12] (03CR) 10Jforrester: [C: 031] build: Clean up npmignore [oojs/core] - 10https://gerrit.wikimedia.org/r/142466 (owner: 10Krinkle) [00:17:24] (03PS1) 10Alex Monk: Allow FieldLayouts to have help text [oojs/ui] - 10https://gerrit.wikimedia.org/r/142468 (https://bugzilla.wikimedia.org/51798) [00:27:17] 3OOjs UI: OOjs UI: Dialogs are not accessible - 10https://bugzilla.wikimedia.org/65473#c19 (10Alex Monk) (In reply to Derk-Jan Hartman from comment #0) > They don't restore the focus upon closing to where it was before opening the > dialog. This part was reverted in Gerrit change 141940 I'm not sure it was v... [00:31:49] 3VisualEditor / 3ContentEditable: VisualEditor: Double-click should not open items that shouldn't be opened - 10https://bugzilla.wikimedia.org/67176 (10James Forrester) 3ASSI p:3Unprio s:3major a:3None References defined inside a {{reflist}} aren't editable – but double-click now opens the reference... [00:32:02] 3VisualEditor / 3ContentEditable: VisualEditor: Double-click should not open items that shouldn't be opened - 10https://bugzilla.wikimedia.org/67176 (10James Forrester) p:5Unprio>3High [00:34:49] 3VisualEditor / 3ContentEditable: VisualEditor: [Regression] Typing when a block node is selected should replace the node with a new paragraph of the typed input - 10https://bugzilla.wikimedia.org/67177 (10James Forrester) 3ASSI p:3Unprio s:3normal a:3Ed Sanders Correct outcome: * Select an inline i... [00:35:02] 3VisualEditor / 3ContentEditable: VisualEditor: [Regression] Typing when a block node is selected should replace the node with a new paragraph of the typed input - 10https://bugzilla.wikimedia.org/67177 (10James Forrester) p:5Unprio>3High [00:35:17] Oh Internet Explorer... [00:35:23] mooey: So I found a bug in their *inspector [00:36:49] 3VisualEditor / 3ContentEditable: VisualEditor: Pressing "delete" / "backspace" when selecting a node sometimes does not delete it - 10https://bugzilla.wikimedia.org/67178 (10James Forrester) 3NEW p:3Unprio s:3major a:3Ed Sanders The same node sometimes is deleted and sometimes isn't. Appears to appl... [00:37:02] 3VisualEditor / 3ContentEditable: VisualEditor: Pressing "delete" / "backspace" when selecting a node sometimes does not delete it - 10https://bugzilla.wikimedia.org/67178 (10James Forrester) 5NEW>3ASSI p:5Unprio>3High [00:52:33] * RoanKattouw has started https://www.mediawiki.org/wiki/User:Catrope/IE_bugs to track browser bugs [00:57:17] 3VisualEditor / 3ContentEditable: VisualEditor: Pressing "delete" / "backspace" when selecting a node sometimes does not delete it - 10https://bugzilla.wikimedia.org/67178#c1 (10Erik Moeller) Definitely no clean repro - comes and goes as you click on templates. But once it stops working in a session, it seem... [01:20:57] (03PS3) 10Jforrester: Provide move left/right quick action buttons for images [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141091 [01:21:19] (03CR) 10Jforrester: Provide move left/right quick action buttons for images (035 comments) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141091 (owner: 10Jforrester) [01:22:19] (03CR) 10jenkins-bot: [V: 04-1] Provide move left/right quick action buttons for images [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141091 (owner: 10Jforrester) [01:28:59] (03PS4) 10Jforrester: Provide move left/right quick action buttons for images [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141091 [01:30:21] (03PS1) 10Jforrester: Update VE core submodule to master (aaa8547) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/142473 [01:31:46] RoanKattouw: Also https://gerrit.wikimedia.org/r/#/c/142465/ and https://gerrit.wikimedia.org/r/#/c/142467/ [01:31:52] (03CR) 10Catrope: [C: 032] Update VE core submodule to master (aaa8547) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/142473 (owner: 10Jforrester) [01:32:27] (03CR) 10Catrope: [C: 032] Update OOjs to v1.0.10 [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/142467 (owner: 10Jforrester) [01:33:29] (03Merged) 10jenkins-bot: Update VE core submodule to master (aaa8547) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/142473 (owner: 10Jforrester) [01:34:24] (03Merged) 10jenkins-bot: Update OOjs to v1.0.10 [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/142467 (owner: 10Jforrester) [01:54:49] (03PS2) 10Jforrester: Provide tools to make text bigger or smaller [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/141617 (https://bugzilla.wikimedia.org/51613) [01:55:17] (03CR) 10Jforrester: Provide tools to make text bigger or smaller (034 comments) [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/141617 (https://bugzilla.wikimedia.org/51613) (owner: 10Jforrester) [02:09:41] (03PS18) 10Jforrester: Grunt: Create a distribution files for the ve module [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/137367 (https://bugzilla.wikimedia.org/66117) [02:09:59] (03CR) 10Jforrester: "PS18 is a rebase." [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/137367 (https://bugzilla.wikimedia.org/66117) (owner: 10Jforrester) [02:14:01] James_F: http://jsfiddle.net/k3QV2/ [02:14:42] + /1/ [02:32:27] mooey: If you feel like crying, look at http://jsfiddle.net/k3QV2/7/ in IE vs a real browser [03:22:45] RoanKattouw, .... oy vey [03:23:39] RoanKattouw, uh oh this is has the huge potential of affecting RTL UIs... [03:26:12] mooey: You should see the current issue RoanKattouw's found. [03:26:26] mooey: obliterates the text. [03:26:41] (background-color:xxx is for some reason rendering, umm, foreground.) [03:33:58] mooey: No no. [03:34:05] mooey: It's dir=rtl that makes it break. [03:34:08] :-D [03:38:30] mooey: http://jsfiddle.net/v939T/2/ [03:39:51] http://jsfiddle.net/v939T/3/ for better contrast [03:50:13] O.O [03:50:21] waaaaaat. [03:53:06] James_F|Away, RoanKattouw_away to make this a truly fair test, you should be putting "פו" inside that rtl span. Not that it helps, but at least it's literally RTL. :D [08:04:05] (03PS13) 10Mooeypoo: [wip] Merge Media Edit and Media Insert dialogs [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140813 (https://bugzilla.wikimedia.org/67170) [08:05:52] (03CR) 10jenkins-bot: [V: 04-1] [wip] Merge Media Edit and Media Insert dialogs [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140813 (https://bugzilla.wikimedia.org/67170) (owner: 10Mooeypoo) [09:06:17] (03PS3) 10Zfilipin: [BrowserTest]Added script to run cucumber for various languages and upload it to commons using mediawiki_api gem [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/138377 (owner: 10Vikassy) [09:54:02] 3VisualEditor / 3ContentEditable: VisualEditor: Pressing "delete" / "backspace" when selecting a node sometimes does not delete it - 10https://bugzilla.wikimedia.org/67178#c2 (10Ed Sanders) I can reproduce by clicking directly from one node to another. Looks like a focus issue. [13:23:32] (03PS4) 10Krinkle: test: Add test for compare() with inherited properties [oojs/core] - 10https://gerrit.wikimedia.org/r/141938 [13:23:40] (03PS5) 10Krinkle: test: Clean up test for copy() [oojs/core] - 10https://gerrit.wikimedia.org/r/141943 [13:23:42] (03CR) 10jenkins-bot: [V: 04-1] test: Clean up test for copy() [oojs/core] - 10https://gerrit.wikimedia.org/r/141943 (owner: 10Krinkle) [13:24:31] (03CR) 10Krinkle: "recheck" [oojs/core] - 10https://gerrit.wikimedia.org/r/141943 (owner: 10Krinkle) [13:26:00] (03PS4) 10Krinkle: test: Add test for copy( Object, Function callback ) [oojs/core] - 10https://gerrit.wikimedia.org/r/141944 [13:26:04] (03PS4) 10Krinkle: test: Clean up EventEmitter tests [oojs/core] - 10https://gerrit.wikimedia.org/r/141956 [13:26:14] (03PS4) 10Krinkle: EventEmitter: Remove dead code that claims to prevent double bindings [oojs/core] - 10https://gerrit.wikimedia.org/r/141957 [14:22:29] (03PS1) 10Esanders: Always refocus the paste target when a node is selected [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/142527 (https://bugzilla.wikimedia.org/67178) [15:24:45] Krinkle: Your thoughts on https://github.com/jdforrester/grunt-banana-checker/pull/6 appreciated. :-) [16:03:19] 3VisualEditor / 3MediaWiki integration: VisualEditor: deleting the text from the article and converting it to a redirect page still leaves in categories and Other Stuff - 10https://bugzilla.wikimedia.org/67200 (10Amir E. Aharoni) 3NEW p:3Unprio s:3normal a:3None Deleting the text from the article and... [16:13:32] 3VisualEditor / 3MediaWiki integration: VisualEditor: Blanking the page does not remove meta-data (categories, defaultsort keys, etc.) - 10https://bugzilla.wikimedia.org/61858#c5 (10Bartosz Dziewoński) *** Bug 67200 has been marked as a duplicate of this bug. *** [16:13:32] 3VisualEditor / 3MediaWiki integration: VisualEditor: deleting the text from the article and converting it to a redirect page still leaves in categories and Other Stuff - 10https://bugzilla.wikimedia.org/67200#c1 (10Bartosz Dziewoński) 5NEW>3RESO/DUP *** This bug has been marked as a duplicate of bug 6... [16:18:02] 3VisualEditor / 3Editing Tools: VisualEditor: If some content is selected when inserting a template, replace it with the template with the selection magically inserted as the first(?!) parameter - 10https://bugzilla.wikimedia.org/54186#c4 (10Bartosz Dziewoński) I think a better way to solve this would be to... [16:18:17] 3VisualEditor / 3MediaWiki integration: VisualEditor: Blanking the page does not remove meta-data (categories, defaultsort keys, etc.) - 10https://bugzilla.wikimedia.org/61858#c6 (10Amir E. Aharoni) (In reply to James Forrester from comment #3) > On this one I'm minded to WONTFIX, but I could be argued arou... [17:15:36] (03CR) 10Catrope: [C: 032] Always refocus the paste target when a node is selected [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/142527 (https://bugzilla.wikimedia.org/67178) (owner: 10Esanders) [17:17:09] (03Merged) 10jenkins-bot: Always refocus the paste target when a node is selected [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/142527 (https://bugzilla.wikimedia.org/67178) (owner: 10Esanders) [17:21:37] (03CR) 10Catrope: [C: 04-1] "Haven't really reviewed this in depth, just happened to notice a few minor things" (033 comments) [oojs/ui] - 10https://gerrit.wikimedia.org/r/142468 (https://bugzilla.wikimedia.org/51798) (owner: 10Alex Monk) [17:25:59] (03CR) 10Catrope: [C: 04-1] "Code is looking good, haven't tested yet" (032 comments) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141091 (owner: 10Jforrester) [17:27:17] 3VisualEditor / 3Data Model: VisualEditor: Editing a link spanning a comment splits the link - 10https://bugzilla.wikimedia.org/52127 (10James Forrester) [17:41:50] >> typeof'foo'; [17:41:51] Krinkle: (string) 'string' [17:41:55] >> typeof/foo/; [17:41:55] Krinkle: (string) 'object' [17:42:01] >> typeof[/foo/]; [17:42:01] Krinkle: (string) 'object' [17:42:11] switch ('2') { case'2': console.log(2); break; default:console.log('d'); } [17:42:14] >> switch ('2') { case'2': console.log(2); break; default:console.log('d'); } [17:42:14] Krinkle: undefined; Console: 2 [17:42:24] fascinating how keywords don't require spaces [17:56:46] (03CR) 10Esanders: "What happens if I delete all the data in [[Foo|barquux]], am I left with an annotated comment?" [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/141063 (https://bugzilla.wikimedia.org/52127) (owner: 10Catrope) [18:01:38] btw, TrevorParscal, is there a place where I can see how VE's toolbar works in setting up the buttons properly? My attempts to do it on my own failed (and the demo page uses a different method that doesn't suit my needs, it seems) and I"m a little stuck with my grad project with this. [18:01:52] It's not urgent, but I've been going in circles, I want to try and get into how VE's toolbar works this weekend. [18:04:08] mooey [2014-06-26 20:51:26] James_F|Away, RoanKattouw_away to make this a truly fair test, you should be putting "פו" inside that rtl span. Not that it helps, but at least it's literally RTL. [18:04:10] mooeypoo: You can edit that jsfiddle yourself, so please do :) --^^ [18:04:42] RoanKattouw, I did, and this was more of a jest, really, it works the same way. But I thought it was only fair to use rtl word in an rtl example :D [18:05:13] also, RoanKattouw, I have a quick question for you about one of your comments if you have a second [18:05:26] woah, VE's toolbar is weird :o [18:05:59] MatmaRex, yeah, I got really confused trying to replicate the general idea outside VE. [18:07:07] mooeypoo: OK sure [18:07:13] RoanKattouw, in one of your comments, you suggest I store the deferred object of 'getFileRepos' so the next time the method is called, the promise is returned without requesting again [18:07:18] 3VisualEditor / 3ContentEditable: VisualEditor: Clicking on Edit link beside a section heading does not take user to that specific heading rather takes to the beginning of the article - 10https://bugzilla.wikimedia.org/65928#c10 (10ryasmeen) 5RESO/?>3VERI Verified the fix in production. There is a differ... [18:07:49] but I'm doing that with storing its result, aren't I? I'm not sure I understand the difference. I check if this.sources is filled first and only ask for the api call if it isn't, so the call will only happen once anyways. [18:07:53] RoanKattouw, What am I missing? [18:08:14] ,..... wait, I think I just understood it while re-reading your comment. [18:08:35] ok, forget it, I think I got it. I didn't think about the option of a second call going out while the first is not yet resolved. [18:08:44] is this really considered a good idea? D: https://github.com/wikimedia/mediawiki-extensions-VisualEditor/blob/master/modules/ve-mw/init/targets/ve.init.mw.ViewPageTarget.js#L1145-L1170 [18:09:33] 3VisualEditor / 3ContentEditable: VisualEditor: Cursor is appearing in the read mode in Firefox after canceling the edit in VE - 10https://bugzilla.wikimedia.org/67106#c1 (10ryasmeen) This is happening in production now, so environment is not just Betalabs now. [18:09:38] mooeypoo: Don't store the result, store the *promise* [18:09:45] Because it contains the result, and also the pending state [18:09:55] RoanKattouw, yeah, I just realized that. Gotcha. I didn't think of that. [18:19:18] 3VisualEditor / 3ContentEditable: VisualEditor: [Regression] The previous line is also getting highlighted while adding a reference list/image - 10https://bugzilla.wikimedia.org/65706#c13 (10ryasmeen) 5RESO/?>3VERI Verified the fix in production [18:41:33] 3VisualEditor / 3Mobile: VisualEditor Mobile: The highlight for images are appearing in wrong position after tapping on any image with Safari in iOS - 10https://bugzilla.wikimedia.org/67063#c6 (10ryasmeen) The highlight for images appearing in wrong position is now fixed ! :) However the tapping on link or i... [18:42:48] 3VisualEditor / 3Mobile: VisualEditor Mobile: Second tap on a reference navigates away from (closes) VE on iOS Safari - 10https://bugzilla.wikimedia.org/67002#c5 (10ryasmeen) 5RESO/FIX>3REOP This bug did not get fixed by : Use getClientRects()[0] instead of getBoundingClientRect() https://gerrit.wikime... [18:43:03] 3VisualEditor / 3Mobile: VisualEditor Mobile: Second tap on a reference navigates away from (closes) VE on iOS Safari - 10https://bugzilla.wikimedia.org/67002 (10ryasmeen) 5REOP>3ASSI [18:51:03] 3VisualEditor / 3Mobile: VisualEditor Mobile: Tapping on a link redirects to the link target page in Safari when the keyboard is hidden - 10https://bugzilla.wikimedia.org/65326#c6 (10ryasmeen) This bug did not resolved by : FocusableNode: Use getClientRects()[0] instead of getBoundingClientRect() https://g... [18:51:33] 3VisualEditor / 3Mobile: VisualEditor Mobile: Tapping on a link redirects to the link target page in Safari when the keyboard is hidden - 10https://bugzilla.wikimedia.org/65326#c7 (10ryasmeen) did not get* [18:59:02] 3VisualEditor / 3Mobile: VisualEditor Mobile: Second tap on a reference navigates away from (closes) VE on iOS Safari - 10https://bugzilla.wikimedia.org/67002#c6 (10ryasmeen) For images, second tap on it opens up the image which might also be the effect of this same bug. [19:06:19] 3VisualEditor / 3Mobile: VisualEditor Mobile: References require double tap to be selected on iOS Safari - 10https://bugzilla.wikimedia.org/66999#c6 (10ryasmeen) 5RESO/FIX>3REOP This bug did not get resolved by Ed's fix. The issue here is, the first tap is showing a lighter highlight which is equivalent... [19:07:03] 3VisualEditor / 3Mobile: VisualEditor Mobile: References/images require double tap to be selected on iOS Safari - 10https://bugzilla.wikimedia.org/66999 (10ryasmeen) [19:07:18] 3VisualEditor / 3Mobile: VisualEditor Mobile: References/images require double tap to be selected on iOS Safari - 10https://bugzilla.wikimedia.org/66999 (10ryasmeen) 5REOP>3ASSI [19:14:17] 3VisualEditor / 3Editing Tools: VisualEditor: Fix tabindexes in the citation dialog - 10https://bugzilla.wikimedia.org/66824#c3 (10Matthew Flaschen) Created attachment 15765 --> https://bugzilla.wikimedia.org/attachment.cgi?id=15765&action=edit Tabbing issue in Firefox I can still reproduce it. I should... [19:39:18] 3VisualEditor / 3Editing Tools: VisualEditor: [Regression wmf1] Cursor jumps to the beginning of the document after adding citations/references - 10https://bugzilla.wikimedia.org/64041#c11 (10ryasmeen) So, this issue did get re-introduced in wmf 10, and then got somehow resolved in wmf 11, Because I am getti... [19:50:16] (03PS14) 10Mooeypoo: Merge Media Edit and Media Insert dialogs [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140813 (https://bugzilla.wikimedia.org/67170) [20:05:58] (03PS2) 10Alex Monk: Allow FieldLayouts to have help text [oojs/ui] - 10https://gerrit.wikimedia.org/r/142468 (https://bugzilla.wikimedia.org/51798) [20:23:47] 3VisualEditor / 3ContentEditable: VisualEditor: The image does not remain highlighted but the areas around it are instead, after making an image frameless-right or basic-right - 10https://bugzilla.wikimedia.org/65716#c13 (10ryasmeen) 5RESO/?>3VERI Verified the fix in production [20:33:39] 3VisualEditor / 3Editing Tools: VisualEditor: Making an image frameless-left/Basic-left and then unchecking the wrap option shows a flash of highlight trying to select the previous line and then goes away - 10https://bugzilla.wikimedia.org/67212 (10ryasmeen) 3NEW p:3Unprio s:3normal a:3None Steps to... [20:35:02] 3VisualEditor / 3ContentEditable: VisualEditor: Block items converted to inline initially show a flash of highlight trying to select the previous line, and then the selection normalises to just the item - 10https://bugzilla.wikimedia.org/67212 (10James Forrester) 5NEW>3ASSI p:5Unprio>3Low s:5normal... [20:38:12] RoanKattouw_away: Krinkle: https://gist.github.com/anonymous/bfda441ee9c94a5d2d80 [20:39:21] Krinkle: So I wanted to pick your brain about how we could reduce the length of things like OO.ui.TextInputWidget.super.prototype.setValue.call( this, value ); [20:39:37] RoanKattouw: I've been thining about the same thing [20:39:42] TrevorParscal: I don't like local variables for that [20:39:44] Mantle's OO framework (used by mobile and Flow) is very basic but it does have this._super() which calls the parent method [20:39:56] I can show you how it's implemented but it's pretty evil [20:40:04] _super is terrible, the only way that can be done is by overriding ht emethod and setting an instance propery [20:40:12] Yup that's how it's done [20:40:30] John Resig that that in 2008 with Klass and Class, [20:40:38] A few ideas [20:40:39] Krinkle: we are trying to come up with a way to make OOJS more attractive to users of this other thing [20:40:45] So I want to migrate them to our EventEmitter and that's probably gonna be an easy sell [20:40:46] * TrevorParscal likes Krinkle's ideas [20:40:53] Because their API is an exact subset of ours [20:41:08] Firstly, abstract creation of the class behind a method [20:41:18] e.g. FooBar = Foo.extend({ }) [20:41:29] OO is also more feature-rich than their code (theirs doesn't have mixins so they have ugly hacks in some places, for instance), but if it regresses in this respect then they may complain about that [20:41:35] that way we're in control of the internals and can change and optimise things the future. [20:41:51] (For reference, https://git.wikimedia.org/blob/mediawiki%2Fextensions%2FMantle.git/HEAD/javascripts%2Fcommon%2FClass.js is their code) [20:41:52] I'm in support of this, what are you feelings about how ExtJs does this? [20:42:02] The main thing I want to move away from is this fragile global scope we use now where you create a naked constructor, then have to mmediately call initclass/inherit/mixin, and then assign prototype thingies [20:42:17] Mantle also uses .extend() BTW [20:42:37] As for super, I'm not sure yet. One idea I saw yesterday is using an argument, I liked it a lot. [20:42:53] the subclass registers it as an override and gets the parent as the first argument [20:43:00] for those methods [20:43:03] ugh [20:43:04] 3VisualEditor / 3Editing Tools: VisualEditor: Deleting a block item highlights the previous line - 10https://bugzilla.wikimedia.org/67213 (10ryasmeen) 3NEW p:3Unprio s:3normal a:3None Created attachment 15767 --> https://bugzilla.wikimedia.org/attachment.cgi?id=15767&action=edit Screenshot Steps t... [20:43:08] I dislike using arguemnts [20:43:22] It necessitates wrapping, doubling function call overhead [20:43:23] it's the only other way of getting it in the scope [20:43:27] It makes .call/.apply more confusing [20:43:34] that's cheap when done right [20:43:46] and no, 1) you woudn't do call/apply yourself [20:43:56] 2) when invoking it from the outside, remember this is not the real method [20:44:02] I guess you would call/apply the wrapper, yeah that makes sense [20:44:07] I just ... don't really like these wrappers [20:44:20] e.g. Foo.extend{ constructor: ..., override: { doStuff: function ( super, a, b,c ) {} } [20:44:27] Also BTW, what I'd really like to be fed in is not the method, but the parent class [20:44:33] Because then you can call other functions and static properties [20:44:36] the actual Bar.prototype.doStuff will be a,b,c [20:44:52] That's what it'd be, yes. [20:44:55] otherwise you lose context [20:44:58] Yeah [20:45:09] super === Foo, not Foo.prototype or Foo.prototype.doStuff [20:45:21] Well [20:45:21] I don't like what this breaks though [20:45:29] super could be Foo.prototype.doStuff.bind(this) [20:45:32] 3VisualEditor / 3Editing Tools: VisualEditor: [Regression wmf9] In Firefox, closing the Language Inspector without making any change throws sel.nativeSelection is null - 10https://bugzilla.wikimedia.org/66734#c12 (10ryasmeen) 5RESO/?>3VERI Verified the fix in production [20:45:46] Why would you need an explicit reference to the parent class? [20:45:53] this.static ? [20:45:54] You can't easily assign functions any more, and we'd have to do something that un-confuses jsduck [20:46:04] Hmm I guess maybe you don't necessarily need it [20:46:05] this.constructor.static you mean? [20:46:24] we don't need .construcutor in between per se [20:46:31] we can just reserve that instance property if we want [20:46:36] it's by reference anyway [20:46:42] That's the static of the eventual object with all inheritance resolved, but if that doesn't work for your purposes your purposes are probably evil [20:47:08] Krinkle: I think super === parentClass.prototype.methodName is probably fine [20:47:38] RoanKattouw: http://i.imgur.com/zTtOoLs.png [20:47:41] how are you going to handle super in async cases? [20:47:42] ignore the other bits, but look at the super [20:47:49] (03PS5) 10Jforrester: Provide move left/right quick action buttons for images [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141091 [20:47:51] TrevorParscal: you call it manually [20:47:57] nothing changes that [20:48:09] TrevorParscal: If it's an argument instead of an instance property, that's fine [20:48:29] so the first or last arg is always super? [20:48:31] (03CR) 10Jforrester: Provide move left/right quick action buttons for images (032 comments) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141091 (owner: 10Jforrester) [20:48:33] First arg [20:48:34] and what about variadic args? [20:48:38] I don't really like it [20:48:42] i hate that [20:48:45] Cant be the last, that's evil. [20:48:47] Also because it introduces wrappers [20:48:59] yes, and confusion [20:49:03] TrevorParscal: before you write it off, beware that it doesn't affect the internal method only the setup [20:49:23] What do you mean only the setup [20:49:25] subclass.prototype.doStuff.toString() == 'function (a, b, c) { .. [20:49:28] no super there [20:49:33] Sure [20:49:41] The thing that actually gets called is a method that takes a,b,c, [20:49:49] But the thing you define is a method that takes super,a,b,c [20:50:03] The thing you pass to the class extension handler, yes. [20:50:18] jduck isn't confused by it, I checked [20:50:18] Well that's the thing where you actually define everything [20:50:24] Really? [20:50:35] Do you not have to define what the super parameter is absolutely everywhere? [20:50:37] Yes, it uses whatever identifer is below it, and the type of value [20:50:43] It doesn't look at parameters. [20:50:53] { doStuff: function ( .. ) { [20:50:56] becomes @method doStuff [20:50:57] he is saying you just don't document it [20:51:03] Right, that too [20:51:06] becase it wouldn't be an argument [20:51:09] because it isn't on the prototype [20:51:14] this seems too magical [20:51:16] you don't cal doStuff(super, a) [20:51:18] in evil ways [20:51:26] It's like python a bit [20:51:29] Krinkle: I know you don't *call* it that way [20:51:30] :o [20:51:31] it's there for you internally [20:51:31] But you do *define* it that way [20:51:40] And jsduck builds documentation from the *definitions* [20:51:53] Krinkle: nooo please don't do that :( it sounds awful [20:51:57] jsduck doesn't do shit with parameters [20:52:00] Wouldn't it document the super parameter automatically, or complain about it not having documentation text? [20:52:04] anything [20:52:05] ever [20:52:08] Oh it doesn't verify them? hah [20:52:15] Anyways yeah [20:52:21] I think Python is evil for taking super as a parameter [20:52:24] Anyway, I made my case. Just one of many ideas. [20:52:31] And I wouldn't like for us to do it either [20:52:36] Python takes this as param, not super [20:52:39] byt yeah [20:52:41] Right! [20:52:43] Which is even worse [20:52:46] Yep [20:52:55] In any case [20:53:03] 3VisualEditor / 3Editing Tools: VisualEditor: Fix tabindexes in the citation dialog - 10https://bugzilla.wikimedia.org/66824#c4 (10Alex Monk) Weird... I can replicate this in Firefox, but not Chrome. I tried copying the inner HTML out of the dialog iframe and into a separate file - it worked there. I also t... [20:53:06] I think we can agree that the way this._super is implemented by Mantle is pretty horrible too [20:53:26] Krinkle: thank you for thinking aboout this [20:53:27] They've already run into issues where they tried to use it from within a deferred callback and couldn't, you have to cache it in that case [20:53:27] RoanKattouw: So just to summarise, why not super as prepended argument in the abstraction layer for overridden methods? [20:53:33] I'm not decided on what to do [20:53:45] Krinkle: Disconnect between definition style and calling style [20:53:47] only that we should consider ways to make OOJS attractive to others [20:53:50] (03PS1) 10Alex Monk: Set tabindex for template dialog info/remove buttons on parent

rather than [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/142718 (https://bugzilla.wikimedia.org/66824) [20:53:57] RoanKattouw: That seems thin. [20:54:04] Also, less importantly, need for a wrapper around every single function [20:54:15] Doubling call stack depth etc [20:54:20] Only overridden functions, and a wrapper that isn't evil is very cheap [20:54:20] Though other solutions have that attribute too [20:54:29] "only overridden functions" makes it worse, not better [20:54:34] Now there are two kinds of methods! [20:54:38] And they are magically differnet [20:54:46] but nobody cares or sees that? [20:54:58] The people that call the methods don't see it [20:55:03] But the people that define the methods do [20:55:07] or read the code [20:55:45] I'm not a fan of abstraction, I"m merely saying, if we want pretty syntax, we have to let go of not liking what's under the hood and making sure it's performant and usable, and not how crazy it is in principle. [20:56:29] eh, to disambiguate, we don't let go of perf/usable, only of the hood. [20:56:44] s/making/make [20:57:34] I think that's reasonable [20:57:44] I mean I feel like this._super()'s implementation is crazy [20:57:58] But from a usability standpoint it is actually fairly nice [20:58:14] also, as a principle, I very much despise references to mixin methods. You mix it in and it becomes part of your class. AFter tht point, reference shall go through 'this' or something like that, not back to the original mixin again. [20:58:21] TrevorParscal: regarding your gist ^ [20:58:35] What if you have to override the mixin method and call the parent? [20:58:50] Then you're doing it wrong :P [20:58:58] No you're really not [20:59:00] That happens all the time [20:59:15] IN the current system I'd argue, it shouldn' be mixed in then but a static helper method you call directly [20:59:24] in the new system, .override() is your friend and you get "super" [20:59:30] Krinkle: I'm with you, it was just a discussion starter [20:59:31] (or rtahter, "original" / "previous") [21:00:00] s/is/could be/ [21:00:17] I'm OK with not keeping references to all the mixins around in the closure scope, I think that's fine [21:00:24] I think the parent is special in a way that mixins aren't [21:00:40] But you can't say you should never access the mixin again, because we have parent calls in overridden methods [21:00:49] I guess with this._super() you could just make that work magically [21:01:08] I just ... it's really nice but the way you have to implement it is just so hideous [21:01:20] Krinkle: Hey here's a thought... how deprecated and/or evil is arguments.callee? [21:01:37] Bah, can't be used in strict mode [21:01:41] very [21:01:46] Come on ES5! That was a useful feature! [21:01:48] 3VisualEditor / 3Editing Tools: VisualEditor: Clicking on "Make full size" switches to Custom tab, keeping the default size and not making it full size on the first attempt for all kind of images - 10https://bugzilla.wikimedia.org/66863#c3 (10ryasmeen) Verified the fix in betalabs [21:02:00] What use is having functions be objects if they can't even access their own properties [21:02:20] did you mean caller? [21:02:25] No no [21:02:29] if you need callee, name your function? [21:02:34] The Function object of the function that is currently running [21:02:41] Right I suppose you could use named functions [21:02:56] :) [21:03:04] can even call it callee [21:03:48] But not multiple times, right? [21:04:02] Can you do foo = function callee() { ... }; bar = function callee() { ... }; etc? [21:04:21] "it does not create a variable in the outer scope (except for IE 8 and below)" [21:04:25] So that suggests you can [21:05:34] So here's another slightly horrible suggestion [21:05:43] Suppose Bar extends Foo [21:06:10] Bar.prototype.methodName = function f( arg1, arg2 ) { /* Call parent method */ f.parent(); ..... }; [21:06:14] yeah, it's locally scoped, like vars, params, this, and arguments. [21:06:18] 3VisualEditor / 3Editing Tools: VisualEditor: Clicking on "Make full size" switches to Custom tab, keeping the default size and not making it full size on the first attempt for all kind of images - 10https://bugzilla.wikimedia.org/66863#c4 (10ryasmeen) 5RESO/FIX>3REOP We need to backport the fix for this... [21:06:33] Then extendClass could set Bar.prototype.methodName.parent = Foo.prototype.methodName; [21:07:01] RoanKattouw: No. Unlike override methods, this would actually be a performance penalty and close to impossible [21:07:13] you can't magically ensure every prototype has a property, right? [21:07:16] it's a dynamic link [21:07:26] Hm.. maybe. [21:07:31] Oh I see what you're saying [21:07:38] It breaks if it's not an overridden method [21:07:42] But that's also true of wrapping [21:08:04] Any problem to do with f.parent being absent in some cases is exactly the same problem you'd have if you wrapped the function [21:08:24] Depends. The override is explicitly called for [21:08:28] Because if you were able to wrap the function, you were also able to set a property on it, and vice versa [21:08:47] we can do that for these as well I guess. [21:08:55] Well, I mean [21:08:57] omg, ryasmeen, that was awesome. Anyone not disinterested in backporting the unimplementation of svg max size? [21:09:01] Right now functions already know whether they are overrides or not [21:09:03] [21:09:10] RoanKattouw: how? [21:09:20] Because they either call their parent or they don't [21:09:39] the class doesn't know though [21:09:50] If the question you're asking is "but if a method that isn't actually an override tries to call its parent, what will happen?", then the answer in all frameworks, including the current one, is "bad things" [21:10:03] or, more practically, "it doesn't" [21:10:35] I think if you use a property like ._super or .parent it creates the intuitive impression that you can call that and thus make this an override method that uses a parent [21:10:57] e.g. not have to state something like overrides: ['doStuff'[ and that only those have that property [21:10:58] mooeypoo: :) [21:11:44] Krinkle: Well if something is not an override method then fundamentally there is no parent [21:11:48] And so it shouldn't be written to call one [21:11:54] just another concept: https://gist.github.com/anonymous/fb08347676a727f399f7 [21:11:54] we could wrap constructors to set parent, mixins properties [21:11:54] and do something with the config or, whatever [21:11:56] By override we mean a method that uses its parent [21:11:56] That is a problem that all frameworks share [21:11:56] then, you have instance specific properties that provide links to the classes that made it [21:11:56] and the mixins have symbolic names [21:11:57] just another concept: https://gist.github.com/anonymous/fb08347676a727f399f7 [21:11:57] we could wrap constructors to set parent, mixins properties [21:11:57] and do something with the config or, whatever [21:11:57] then, you have instance specific properties that provide links to the classes that made it [21:11:57] and the mixins have symbolic names [21:12:00] not jsut any that has a parent [21:12:09] (03PS5) 10Mooeypoo: [wip] MWImageModel unit tests [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/141749 [21:12:51] TrevorParscal: yep, inherit/mixin being properties to the extend method makes sense. I had the same idea. [21:12:58] * Krinkle digs up gist from 2 months ago [21:13:02] 3VisualEditor / 3ContentEditable: VisualEditor: [Regression wmf10] Context menus are appearing and then disappearing right after adding cite options/references/formula/image - 10https://bugzilla.wikimedia.org/66855#c7 (10ryasmeen) Verified the fix in test2 [21:13:29] mooeypoo: What happened now? [21:14:04] RoanKattouw, oh, nono, nothing. ryasmeen suggested one of the patches be backported, and I just got amused with the backporting of an unimplementation. :D [21:14:45] But it was actually about backporting the other patch, so my amusement was more theoretical than practical. [21:14:54] also, I always wanted to use "disinterested" in a sentence. [21:14:55] :D [21:15:02] Trevor apologizes that his IRC is totally broken and he's not getting any of this conversation [21:15:09] TrevorParscal: RoanKattouw: Hm... this.parent or this.super, I think there was a reason that can't work. Because of recursion [21:15:13] mooeypoo: "not disinterested" is a nice litotes [21:15:19] 'this' is always the subclass object [21:15:19] Krinkle: Yeah I told him this [21:15:24] even when you're in the parent class [21:15:25] sorry, my IRC was borked, but it's working now [21:15:26] Krinkle: 'constructor' could be executed in the correct context [21:15:32] But it falls flat when you consider parent calls in methods [21:15:36] my example code is crap, i'm not defending it [21:15:37] Indeed [21:15:51] RoanKattouw, indeed [21:15:53] TrevorParscal: sure, but this is a javascript pitfall that is more important than the example you posted. [21:16:02] jsut making sure you understand it, it's not obvious. [21:16:07] So, let me write a gist with my proposal [21:16:12] in all honesty, I believe that the way we do it now, while perhaps vebose, is technically correct and clear to anyone who understands JavaScript [21:16:13] I actually messed it up in an oojs release once [21:16:30] Which I am not married to either, it's ugly in its own ways, but I think it has arguments going for it [21:16:30] using things like Class.super are already probably overly sugary [21:17:06] Developers who choose a library based on how neat it looks, rather than how correct or capable it is, are probably making a mistake [21:17:28] it's not that we don't have the equivilant of this._super, it's that we have to write more characters, and ours works more reliabily [21:18:24] https://gist.github.com/catrope/ed1ecfefdb1d16784644 [21:19:27] (updated; refresh) [21:19:49] Krinkle said 1) something about performance and 2) something how it would approximately never work [21:19:56] But I don't understand his argument for #2 [21:20:24] My main pet peeve is the global scope fragility. I absolutely hate having to call inheritClass *right* after, or you'll below away the prototype [21:20:37] And calling initClass on the root ones is even worse [21:20:59] Right [21:21:05] I think that's largely a separate issue [21:21:11] Having a Base.extend({} and from there on extending others is more explicit and puts the intent right upfront. Even if it was only for the constructo method, I'd like us to do that [21:21:51] I missed the beginning of this conversation. Is there a specific lack we want to address aside from the grand scheme? [21:21:56] The lengthy identifiers, right? [21:22:04] Well I specifically would like making parent method calls easier [21:22:08] Yeah [21:22:11] Although all this other stuff should also be cleaned up, yes [21:22:13] Also [21:22:25] What would be *really* neat [21:22:31] Is if we can make the syntax close-ish to what Mantle uses [21:22:31] To make migration easier for them [21:22:56] While at the same time, maintaining b/c with old-style OO.inheritClass() code (this should be easy), so migration isn't painful and bidirectionally dependent for us [21:22:58] (and for multimedia!) [21:26:21] Roan's idea is pretty neat [21:26:23] RoanKattouw: We can also just use ES6 [21:26:34] and have resourceloader ship an es6 engine written in es3 and eval that in older browsers [21:26:41] lol [21:26:43]