[00:00:45] (03CR) 10Catrope: Use linear model instead of getNode() in MWImageModel (031 comment) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140183 (owner: 10Mooeypoo) [00:01:46] Krenair: Doubt or know? :-) [00:02:34] Doubt. I don't think I've ever messed with these parsing functions before [00:03:05] * James_F nods. [00:03:07] Few have. [00:03:31] (03CR) 10Catrope: Use linear model instead of getNode() in MWImageModel (031 comment) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140183 (owner: 10Mooeypoo) [00:06:31] oh foo [00:10:17] James_F: btw, what's the situation with VE in MW 1.23.0? There's a user having issues with it on wikitext-l [00:10:59] gwicke: VE master is not compatible with 1.23 [00:11:11] there's a bug for that. [00:11:28] gwicke: mediawiki.skinning.content.parsoid dependency => VE master depends on MW 1.24alpha. [00:11:34] gwicke: Oh well. [00:11:39] James updated VE's REL1_23 recently [00:11:41] Also, thanks for the reminder; will respond. [00:11:42] I don't know exactly what that was for [00:12:11] James_F: thanks! [00:13:04] might also be good to mention this temporary gotcha in the VE setup instructions [00:14:51] (03CR) 10Mooeypoo: Use linear model instead of getNode() in MWImageModel (031 comment) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140183 (owner: 10Mooeypoo) [00:18:29] Yeah. [00:29:27] I'll probably end up asking for help later (aka tomorrow) in #mediawiki-core [00:30:01] Krenair: Do you want to revert that patch from master before the train? [00:30:23] (03PS3) 10Mooeypoo: Use linear model instead of getNode() in MWImageModel [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140183 [00:30:55] hmm... This version doesn't horribly break VE completely on beta, just reverts to previous behaviour... It should be fine in production (as it is in beta) [00:31:05] OK. [00:31:23] Krenair: "Beta" and used by several thousand people a day. [00:31:44] Okay, 'as it is in deployment-prep' [00:31:49] :-) [00:32:00] I was referring to beta.wmflabs.org, not the status of VE [00:32:10] Sure, but it gets cut to MediaWiki.org in the morning. [00:32:10] Which means that fixes will have to get back-ported. [00:32:15] Which is fine, just highlighting. :-) [00:32:19] Maybe it's a good idea to revert the patch... Not sure. [00:32:38] wfDebugs in production won't make you friends with Ops, to be sure. [00:32:39] It should be fine, but that's what we thought before beta broke. [00:32:42] :-D [00:32:51] Heh, okay. Let's revert it then [00:34:28] (03PS1) 10Alex Monk: Revert "Update displaytitle on save (try #2)" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140621 [00:34:38] (03CR) 10Catrope: [C: 032] Revert "Update displaytitle on save (try #2)" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140621 (owner: 10Alex Monk) [00:35:03] Also I could have just left the Target file alone, but meh. [00:35:29] g'night [00:36:20] (03Merged) 10jenkins-bot: Revert "Update displaytitle on save (try #2)" [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140621 (owner: 10Alex Monk) [00:40:26] RoanKattouw: https://dl.dropboxusercontent.com/u/17195534/VisualEditor-REL1_23.tar.bz2 [00:41:57] RoanKattouw: SHA1(VisualEditor-REL1_23.tar.bz2)= 7cd90271f91b8ded9d6aae3f9166a11d3c7d0ee9 [00:49:07] Krinkle: Your OOJS test suite doesn't work for me in IE11-pretending-to-be-IE8 [00:49:09] or IE7 [00:49:30] we don't support compatibility modes [00:49:32] or quirks mode [00:49:34] Sorry, IE10-pretending-to-be-IE8, *-IE7, and "IE10 Compatibility Mode" [00:49:36] OK [00:49:38] Let me try actual IE8 [00:49:44] where did you get IE11? [00:49:46] or 10 [00:50:05] Krinkle: We got him a laptop so he could test. [00:50:07] Office IT gave me a laptop that dual-boots Windows 8 with IE10 and Windows 7 with IE8 [00:50:19] I probably want that to be 11 and 9 instead [00:50:23] Instead of 8-with-11 and 7-with-9. [00:50:24] Yeah. [00:50:27] browserstack? crossbrowsertesting? saucelabs? we have accounts on all three [00:50:38] it boots faster than you can pick up that laptop and turn your head [00:50:46] and requires no maintenance [00:51:01] Well it's certainly faster than waiting for Windows updates to complete running [00:51:20] Which is what Windows 8 is doing to me now [00:51:21] Krinkle: But copy-and-paste testing doesn't work. [00:51:25] That [00:51:30] Krinkle: As we discussed several times. [00:51:34] And actual hardware is faster than a VM over a network [00:51:36] James_F: It does when you use the VNC viewer instead of HTML5 canvas [00:51:50] afaik Ed didn't try that originaly and it worked when he did [00:51:57] Huh. [00:52:00] Now you tell me. :-P [00:52:02] then it\s no different then booting VMware locally [00:52:10] i did tell you [00:52:16] Anyway, we've got it now, so… [00:52:27] and for the test suite you dont need a clipboard [00:52:39] (other than plain text for the url, and that one works in HTML5 canvas actually) [00:52:48] Krinkle: Roan has it for getting VE-on-IE working. [00:53:07] Krinkle: It just so happens that your work on OOjs coincides with him having it for the first day. [00:53:13] whatever floats your boat :) [00:54:32] getting this on my screen in under 3 seconds is hard to beat though [00:54:34] http://i.imgur.com/3xsH9aF.png [00:55:06] 3VisualEditor: Template-defined templates (e.g. "{{ {{TNTN:Extension…}} …}}") cause "TypeError: title is null" on edit. - 10https://bugzilla.wikimedia.org/66724#c2 (10James Forrester) (In reply to Ed Sanders from comment #1) > Looks like we don't support evaluated template names: > {{ {{TNTN:Extension}} > |.... [00:55:10] Krinkle: And yet 9 months have gone by without VE working on IE… ;-P [00:55:34] In all seriousness though, how do you get VE on there? Do you install a webserver and git and all? or open the port on your other laptop? Or testing wmflabs? [00:55:43] (on the windows machine) [00:56:04] Well, so right now [00:56:08] The WMF network is broken, so my laptop is on WMF2 [00:56:19] It's refusing to get on WMF-Guest, probably because both it and my laptop are broken in subtle ways [00:56:34] I'm pretty sure OIT doesn't want the Windows box to be on the WMF network (they put it on -Guest) [00:56:43] So I have them running on the same network via my mifi :D [00:56:58] That way I can use my Linux laptop as a web server and my Windows laptop as a client [00:57:01] Looks like a solution looking for a problem honestly :P [00:57:13] I wanted to run the oojs tests locally, which would have been possible if I could have installed npm [00:57:18] But OIT hasn't given me the admin password [00:57:40] A problem I'm sure we'll have one day. But it would take a lot for one to try that route. I'd exhaust the cloud options first. Both because of speed and convenience (local tunnel). [00:57:58] Strange how it's easier to have a data centre cross the world connect to my server than the laptop right next to you [00:58:03] Eventually I want Rummana to use this box to test IE [01:00:20] I know i'm often late with ideas that are already obsolete with something better in place, but have you (plural) tried using crossbrowsertesting VNC or browserstack canvas to test IE regularly? It's quite nice. If it doesn't work well, fine, but I can highly recommend it. As well so you have all the different versions of IE and windows right there. (eg. when someone reports a bug). I hear it's [01:00:21] near impossible to have one windows install with multiple versions of IE. Oh, right, dual boot. Well, that's still time consuming. [01:00:41] It's dual-boot, yes. [01:00:45] And yes, time-consuming. [01:01:03] 3VisualEditor: Template-defined templates (e.g. "{{ {{TNTN:Extension…}} …}}") cause "TypeError: title is null" on edit. - 10https://bugzilla.wikimedia.org/66724#c3 (10Gabriel Wicke) Simple test case: {{{{echo|echo}}|foo}} The Parsoid output looks correct to me. [01:01:05] and unlike some of my other ideas, this isn't possible in theory after some work. It can be used right away. [01:05:33] 3VisualEditor: Template-defined templates (e.g. "{{ {{TNTN:Extension…}} …}}") cause "TypeError: title is null" on edit. - 10https://bugzilla.wikimedia.org/66724#c4 (10Gabriel Wicke) Hmm, actually there is an issue in the Parsoid output for this test case, but it's not likely to be related to this bug. {{{{ech... [01:05:56] James_F: I've been using RedScreen.app for these past 2 weeks. It's great. [01:05:57] Krinkle: It doesn't work on IE8, there are intermittent errors [01:06:08] oo.getHash() fails for certain cases but it fails intermittently [01:06:12] Other times the whole thing runs fine [01:06:30] RoanKattouw: Which windows version? [01:06:34] 7 [01:07:03] "Unexpected call to method or property access." oojs.js line 318 (which is return JSON.stringify(... [01:07:07] index.html or index.jquery.html [01:07:09] ok, so index.html [01:08:14] Krinkle: It's good? [01:08:33] James_F: Yeah. I'm gradually increasing the red ness so I don't freak out [01:08:43] I'm at 40% and it's pretty relaxing (in a good way) [01:08:44] Krinkle: :-D [01:08:55] Krinkle: I worry that I'd find it stressful. [01:09:09] Krinkle: What is index.jquery.html ? [01:09:31] loads jquery.js + oojs.jquery.js instead of oojs.js [01:09:54] And what's oojs.jquery.js? [01:09:59] seriously? [01:10:07] :P [01:10:22] https://github.com/wikimedia/oojs/commit/12e7c360c1cce74759dc1b6a1a071508e4e3a4d8 [01:10:24] I thought you merged that [01:10:35] Ah, Ed and James. [01:10:41] https://gerrit.wikimedia.org/r/#/c/133855/ [01:11:15] If your project uses jQuery, use the optimised `oojs.jquery.js` build instead. This build assumes jQuery is present and omits various chunks of code in favour of references to jQuery [01:11:17] Oh, it replaces isPlainObject [01:11:29] yeah, and later things like Promises maybe [01:11:53] Or at least abstract it in a smaller way (we could wrap around $.Deferred or use it direcrtly, whatever works) [01:12:20] RoanKattouw: I can reproduce it in browserstack. Smells like some weird security bug maybe [01:12:48] RoanKattouw: Hm.. can you try something? In testrunner.js those doc.write's, remove those and have them be in index.html instead. [01:13:36] Last time I was investigiating the bug for VE that looks like this it seemed IE's weird extra origin policy relates to document write. I feel like it assigns it (sometimes) an anonymous origin or somethign, allowing its code not to be called. [01:13:43] from outside [01:14:43] Hmm OK [01:14:45] Trying that [01:15:34] Yeah I don't seem to be able to reproduce it now [01:15:34] Also, it seems somehow cache is increasing the odds. I can't reproduce the error after I get a pass. I find myself restarting the browser (stop browserstack worker, and click start again) to get a new go at it. [01:16:02] I guess this is because JSON.stringify is polyfilled from a document.written URL? [01:16:11] I guess? [01:16:23] Actually no [01:16:28] IE8 doesn't need JSON polyfill [01:16:36] doesn't use [01:16:49] But it does use document.write stuff [01:16:54] Because it fails supportsES5 [01:17:32] >>JSON.stringify+'' [01:17:32] " [01:17:32] function stringify() { [01:17:32] Krinkle: (string) 'function stringify() { [native code] }' [01:17:34] [native code] [01:17:36] } [01:17:38] " [01:17:49] Yeah that's what I get too [01:17:53] (IE8 dev tools) [01:17:55] It doesn't seem to polyfill it [01:18:05] Because typeof JSON.stringify === 'function' [01:18:18] And earlier because typeof JSON == 'object' [01:18:34] No that just prevents it from setting JSON = {}; [01:18:47] It doesn't actually prevent anything else from happening [01:18:57] Which then clears out the previous non-object if there was one (IE has a habit of host objects that deny being an object) [01:19:06] so anything after would definitely be undefined [01:19:11] it's relevant that way [01:19:19] That's not happening though [01:19:23] I know [01:19:32] Well -- or maybe it is, sometimes [01:19:33] ? [01:19:38] Let's try that [01:19:44] Yeah, find the failure and debug it then [01:19:48] 3VisualEditor / 3Data Model: VisualEditor: Template-defined templates (e.g. "{{ {{TNTN:Extension…}} …}}") cause "TypeError: title is null" on edit - 10https://bugzilla.wikimedia.org/66724#c5 (10James Forrester) 5NEW>3ASSI p:5Unprio>3Normal (In reply to Gabriel Wicke from comment #3) > Simple test cas... [01:19:49] Go for it [01:22:03] just in case, "Rerun" and "No try catch" can help [01:22:05] ./wikimedia/oojs/core/test/?testNumber=10¬rycatch=true [01:22:33] 3VisualEditor / 3ContentEditable: VisualEditor: Select-all and typing makes document significantly out-of-sync in Firefox - 10https://bugzilla.wikimedia.org/52171#c5 (10James Forrester) 5REOP>3ASSI (In reply to Jonathan Haas from comment #4) > The underlying bug doesn't seem to be fixed. > > To reproduc... [01:22:38] Meh [01:22:38] Even when the error does happen, JSON.stringify is native [01:22:39] Also [01:22:47] noglobals sometimes fails [01:23:06] "Introduced global variables: 0" [01:23:18] makes sense [01:23:19] iframe [01:23:21] window[0] [01:23:23] And then later "Deleted global variables: 0" [01:23:25] window.length [01:23:33] Oh, because it's async? [01:23:35] funny it shows up though, that's a race condition [01:23:39] it shouldn't be [01:23:51] but I know how to mitigate it [01:23:53] patch incoming [01:24:19] The qunit-fixture housing the iframe is reset in teardown, but IE might not be garbage collecting it in time [01:24:50] // This was previously broken when we used .constructor === Object [01:24:51] // oo.getHash.keySortReplacer, because although instances of Foo [01:24:53] // do inherit from Object (( new Foo() ) instanceof Object === true), [01:24:54] // direct comparison would return false. [01:25:11] That comment is near the failing test [01:25:40] ooooh also [01:25:51] You see the "this test can only be done in a browser environment" thing? [01:25:58] Yes? [01:25:58] That's failing in this case at least [01:26:08] I haven't compared successful runs [01:26:18] But on failing runs, the expected # of assertions there is 2, not 3 [01:26:31] which means that doc = this && 'window' in this && this.window.document was falsey [01:27:02] (03PS12) 10Krinkle: Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) [01:27:57] (03CR) 10Krinkle: "Detach the test iframe. Should help IE clean up the iframe so that window.length / window[0] isn't dirty according to QUnit's noglobals t" [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [01:28:23] (03CR) 10Krinkle: "(it was already being cleaned up by qunit-fixture reset, but IE might be too dumb)." [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [01:28:26] Also, amazingly, manually running the test in the console afterwards succeeds [01:28:37] always? [01:30:10] Hm.. you reckon its trigger by document write (e.g. the one for es5-shim or object-create)? [01:30:17] what if you comment out the json2 one but keep the others [01:30:34] If doc.write causes it, I'll use conditional comments instead [01:30:42] though it means it won't be loaded in old non-IE browsers [01:31:26] Well let me explore a little bit more [01:31:42] Because the thing that fails is always the complex one with an object constructed by a function [01:31:48] And there's a comment about typeof stuff [01:32:23] So additionally, keySortReplacer is calling polyfilled functions from inside itself [01:32:37] Object.keys() is polyfilled [01:33:06] As is Array.isArray() [01:33:08] wait... window in this? we dont use that, we use if ( global.document ) [01:33:21] oh, someone added that in core.test.js [01:33:25] and another iframe [01:33:27] Yeah it's in core.test.js [01:33:28] interesting [01:33:31] I have no idea why [01:33:37] Or what it's trying to do [01:33:49] this != global object in there [01:33:56] this == test context [01:34:05] probably some copypasta [01:34:17] that test never runs anywhere ever, right? [01:34:28] Yeah it's just broken [01:34:32] k [01:34:32] It doesn't run in Chrome either [01:34:34] So ignore that [01:34:37] I'll do a sep comit [01:40:15] It also sometimes fails on util.test.js line 90 [01:40:23] oo.isPlainObject( new IframeObject() ) [01:42:42] When this happens the console behaves very strangely [01:42:49] statements involving IframeObject simply aren't run [01:43:14] I mean, statements involving the construction of an IframeObject [01:43:29] Things like typeof IframeObject === 'function' and IframeObject !== Object do run and return true [01:44:15] And so what happens is the console shows an error, and the corresponding assert is never run, and QUnit complains about 7 expected by 6 run [01:48:23] yeah [01:50:10] (03PS1) 10Krinkle: test: Fix broken getHash for iframe.Object [oojs/core] - 10https://gerrit.wikimedia.org/r/140632 [01:50:23] increases number of tests run by 1 [01:50:49] RoanKattouw: Since that test should get the same iframe fix for IE8 as the other, I'd prefer this one be merged first :) [01:54:25] Krinkle: Is 39d751d the merge ID? Doesn't return in gerrit. [01:54:34] It's pre-gerrit [01:54:36] (03PS2) 10Catrope: test: Fix broken getHash for iframe.Object [oojs/core] - 10https://gerrit.wikimedia.org/r/140632 (owner: 10Krinkle) [01:55:00] (03PS3) 10Krinkle: test: Fix broken getHash test for iframe Object [oojs/core] - 10https://gerrit.wikimedia.org/r/140632 [01:55:09] Gosh, pre-gerrit, huh/ [02:01:24] (03PS13) 10Krinkle: Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) [02:02:11] (03CR) 10Krinkle: "Added dependency on I4920c5a72e7ad5 so that that test is also ensured to pass in IE8 and migrated along." [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [02:02:15] (03PS14) 10Krinkle: Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) [02:04:55] James_F: https://releases.wikimedia.org/VisualEditor/VisualEditor-MediaWiki-REL1_23.tar.bz2 [02:08:01] RoanKattouw: Are you sure? [02:08:15] RoanKattouw: When I do 'notrycatch' and have IE8 dev tools open when I do hit the race condition it is inside basic usage [02:08:16] not complex usage [02:09:01] I have gotten it there in the past [02:09:05] But I can't get that to happen now [02:09:14] Does the basic usage case involve any polyfilled functionality? [02:09:40] I guess all getHash calls use Object.keys() and Array.isArray() [02:09:56] Also, note that some of the "basic" cases use getHashObject() [02:10:30] Indeed, only the ones getHash uses in keySortReplacer [02:14:36] 3VisualEditor: VisualEditor: Clear the 'search memory' after making a link - 10https://bugzilla.wikimedia.org/66821 (10WhatamIdoing) 3NEW p:3Unprio s:3normal a:3None Steps to reproduce: 1. Open a page in VisualEditor and make a link: [[Foo]]. Easy, right? Select text, press Command–K, and hit ret... [02:16:56] RoanKattouw: Sounds liek it might be a bug in IE's JSON implementation [02:17:13] it somehow not being able to callback into userland or sepcifically the object being looped over [02:17:18] e.g. http://stackoverflow.com/a/17342752/319266 [02:17:26] they unset JSON in IE8 and use json2 instead [02:18:09] maybe add this to inside the !supportsES5 [02:18:12] if ( global.JSON ) { [02:18:12] global.JSON = undefined; [02:18:12] } [02:18:31] OK let's try that [02:20:59] Seems to work [02:22:00] Same here, though race condition.. [02:22:10] Can't reproduce it [02:22:18] 3VisualEditor / 3Editing Tools: VisualEditor: Clear the 'search memory' after making a link - 10https://bugzilla.wikimedia.org/66821#c1 (10James Forrester) 5NEW>3RESO/FIX p:5Unprio>3Normal This is now fixed with the changed way that the search interface triggers on link creation (and about a dozen o... [02:22:34] With a race condition it's hard to be sure [02:22:56] (03CR) 10Catrope: [C: 032] test: Fix broken getHash test for iframe Object [oojs/core] - 10https://gerrit.wikimedia.org/r/140632 (owner: 10Krinkle) [02:23:27] (03Merged) 10jenkins-bot: test: Fix broken getHash test for iframe Object [oojs/core] - 10https://gerrit.wikimedia.org/r/140632 (owner: 10Krinkle) [02:24:02] (03PS15) 10Catrope: Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [02:24:19] Krinkle: There ---^^ that forces polyfilling of JSON if !supportsES5 [02:25:14] (03PS16) 10Krinkle: Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) [02:25:50] (03PS17) 10Krinkle: Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) [02:26:00] ar, ok. Added your comment [02:26:09] (03CR) 10Catrope: Implement support for ES3 browsers (tests passing in IE6-8!) (031 comment) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [02:26:54] Krinkle: Once this merges, we should do another release of OOjs (but let's not pull it for VE/MW until tomorrow) – want to help me/Roan remember how to do this again? [02:26:57] (03PS18) 10Krinkle: Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) [02:27:11] Yep [02:27:15] (03CR) 10Krinkle: Implement support for ES3 browsers (tests passing in IE6-8!) (031 comment) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [02:27:24] Cool. [02:27:38] v 1.0.10 here we come. [02:28:18] (03CR) 10Krinkle: "did a conditional assignment instead of unconditional (delete JSON would be better but that's unreliable with globals in IE). Not much dif" [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [02:28:51] (03CR) 10Catrope: [C: 032] Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [02:29:24] (03Merged) 10jenkins-bot: Implement support for ES3 browsers (tests passing in IE6-8!) [oojs/core] - 10https://gerrit.wikimedia.org/r/140436 (https://bugzilla.wikimedia.org/56341) (owner: 10Krinkle) [02:31:39] (03CR) 10Catrope: [C: 04-1] Use linear model instead of getNode() in MWImageModel (031 comment) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140183 (owner: 10Mooeypoo) [02:35:21] (03CR) 10Catrope: [C: 032] Update VE core submodule to master (1f16037) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140602 (owner: 10Jforrester) [02:36:19] 3OOjs: OOjs: Document features that need a polyfil in ES3 and recommended a shim - 10https://bugzilla.wikimedia.org/56341 (10James Forrester) 5PATC>3RESO/FIX [02:36:33] 3OOjs: OOjs: Support ES3 - 10https://bugzilla.wikimedia.org/66804 (10James Forrester) 5PATC>3RESO/FIX [02:36:58] (03Merged) 10jenkins-bot: Update VE core submodule to master (1f16037) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/140602 (owner: 10Jforrester) [02:38:03] 3OOjs: OOjs: Document features that need a polyfill in ES3 and recommended a shim - 10https://bugzilla.wikimedia.org/56341 (10Roan Kattouw) [02:38:15] Krinkle: OK, so… doing the 1.0.10 release now. [02:38:40] Krinkle: I may screw up. :-) [02:38:58] James_F: what're you doing? [02:39:25] Krinkle: Following the instructions in README.md. [02:39:35] cool [02:39:43] checked them yesterday when you mentioned it, should be up to date [02:41:15] Umm. [02:41:27] " 371 code style errors found!" [02:41:29] … [02:42:06] Did we change something about indentation? [02:42:57] are you on the right branch and ran npm install first? [02:43:29] Yes. [02:43:35] jscs runs pre merge, so there's no style errors naturally. [02:43:46] * James_F retries. [02:44:34] "git remote update && git checkout -b release1.0.10 -t origin/master && npm install && npm test" [02:44:43] Same issue. [02:45:24] Krinkle: AAARGH so IE also throws errors for ve.elementTypes.void [02:46:25] I don't know what that is [02:46:52] James_F: Just to verify, is origin what it should be? [02:46:58] and what about git status [02:47:30] jdforrester@local:~/Documents/git/oojs$ git remote -v [02:47:30] origin ssh://jforrester@gerrit.wikimedia.org:29418/oojs/core.git (fetch) [02:47:30] origin ssh://jforrester@gerrit.wikimedia.org:29418/oojs/core.git (push) [02:47:42] cool [02:48:04] Ooooh. [02:48:15] I had an old test file lying around somehow. [02:48:34] James_F: Put git status in your PS1 :) [02:48:45] Krinkle: Yeah yeah. [02:48:55] OK, now it passes without errors. [02:49:02] yay [02:49:25] So I got wikimedia coding style upstreamed in node-jscs, and they made a minor release within a week too, [02:49:32] Nice. [02:49:34] so here I am trying to update to it, turns out the new release is broken. [02:49:36] xD [02:49:40] like, just broken [02:50:07] Will try again in a few days, they're working on it [02:50:31] will reduce our jscsrc files a fair bit [02:50:45] eventually down to 1 line, but we might not be consistent yet. [02:51:08] * James_F nods. [02:51:12] Not one line. [02:51:17] Inline. [02:51:23] jscs --style=wikimedia :-) [02:51:33] well, no, we'd want it to work in our editors as well [02:51:44] argumentless execution [02:52:22] (03PS1) 10Jforrester: Tag v1.0.10 [oojs/core] - 10https://gerrit.wikimedia.org/r/140636 [02:52:24] and so that grunt can find it , too [02:52:25] True. [02:52:29] ^^^ [02:53:27] Lol, I just heard a voice on English radio and was like.. wtf why am I hearing the voice of Douglas Reynholm from the IT crowd [02:53:43] (aka Shithead of the Year) [02:54:04] is voice is really like that (or he was doing that character for hte commercial maybe) [02:54:46] Krinkle: Please review https://gerrit.wikimedia.org/r/140636 :-P [02:55:16] (03CR) 10Krinkle: [C: 04-1] "sorry.." (031 comment) [oojs/core] - 10https://gerrit.wikimedia.org/r/140636 (owner: 10Jforrester) [02:55:48] Krinkle: Good catch. ;-) [02:55:53] >> 1.0.1 === 1.0.10 [02:55:53] Krinkle: SyntaxError: Unexpected number [02:56:00] oh, right [02:56:09] >> 1.1 === 1.10 [02:56:09] Krinkle: (boolean) true [02:56:11] yay [02:56:22] We're doing release versioning, not numerical versioning. [02:56:30] (03PS2) 10Jforrester: Tag v1.0.10 [oojs/core] - 10https://gerrit.wikimedia.org/r/140636 [02:56:38] And certainly not IEEE floating poubt duble versioning [02:56:43] Indeed. [02:56:44] or we'd go from 0.1, 0.2 to 0.30004 [02:56:51] * James_F laughs. [02:57:00] Krinkle: It's me complaining about the fact that 'void' is also a reserved word [02:57:03] And the VE code uses it [02:57:10] Fortunately I don't think we'll target old enough IE versions [02:57:11] RoanKattouw: which IE is this for? [02:57:15] IE8 apparently [02:57:25] (and IE10 in compat mode but screw that) [02:57:32] We don't support IE8 so hey. [02:57:35] Right, that fails for a different reason than super [02:57:36] s/d/w/ [02:57:36] Real IE10 is happy with foo.void [02:57:42] Krinkle: Really? Different how? [02:58:05] void is a 'real' reserved word (actual keyword) and in ES3 those are simply not allowed in dot position by standard spec. [02:58:14] which is silly and got removed in ES5 [02:58:18] Aha OK [02:58:24] Krinkle: But works anyway in ES3 browsers? [02:58:26] So for a non-ES5 browser this is actually reasonable [02:58:33] Yep [02:58:57] same was true for older FF and Safari, while it's harmless to allow it, engines pretty followed the spec there all. [02:59:23] James_F: nope, afaik it didn't work in any ES3 browsers [02:59:49] Hmm. [02:59:51] But afaik we're not going to support ES3 in VE, right? [03:00:06] IE9 and above would take a lot already [03:00:34] more or less the overlap in features. in ES3 browsers ES3 is probably least of our worries. [03:00:44] Though I'm interested in what Roan is running into so far. [03:00:46] * James_F nods. [03:01:14] So I'm running the VE core test suite [03:01:33] Right, functionally I guess it could pass. [03:01:43] Krinkle: Are you testing https://gerrit.wikimedia.org/r/140636 some more? [03:01:46] Though if not, it'll probably be due to DOM [03:01:52] It's a clusterfuck but after patching ve.createDocumentFromHtml() to translate '' to '' , I get only one failure in onCopy (to do with attribute order in HTML serialization), and a crash in the paste tests [03:01:59] missed the update James_F [03:02:08] Krinkle: No worries. :-) [03:02:09] (03CR) 10Krinkle: [C: 032] Tag v1.0.10 [oojs/core] - 10https://gerrit.wikimedia.org/r/140636 (owner: 10Jforrester) [03:02:12] Yay. [03:02:34] The createDocumentFromHtml thing is because if you feed an empty string to IE's HTML parser, you get a document back whose .documentElement, .head and .body are all null [03:02:37] (03Merged) 10jenkins-bot: Tag v1.0.10 [oojs/core] - 10https://gerrit.wikimedia.org/r/140636 (owner: 10Jforrester) [03:02:41] That obviously broke lots of tests [03:02:49] Krinkle: I'm trying to remember the oojs.github.io steps (they're undocumented). [03:04:31] James_F: Right. Clone the repo, cd /tmp, npm install oojs, cp node_modules/oojs/dist/oojs.js /git/oojs.github.io/releases/oojs-123.js [03:04:33] (from memory) [03:04:51]