[00:00:01] (03CR) 10Catrope: "This is for a different bug. The bug this fixes is that if you go to /w/index.php?action=edit&editingStatsId=123, the value from the query" [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205188 (owner: 10Catrope) [00:02:03] (03CR) 10Jforrester: [C: 032] Update VE core submodule to master (952a9bb) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/204943 (owner: 10Esanders) [00:02:11] (03CR) 10Alex Monk: "Well then we're going to run into issues with ShowEditFormInitial I think" [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205188 (owner: 10Catrope) [00:02:55] (03PS1) 10Catrope: Add dependency on mw.user [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205191 [00:03:23] (03PS2) 10Jforrester: demo: Create RTL version of minimal demo [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/204942 (owner: 10Esanders) [00:03:31] (03CR) 10Jforrester: [C: 032] demo: Create RTL version of minimal demo [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/204942 (owner: 10Esanders) [00:03:46] Krenair: What issues would we run into? [00:04:44] OK I can reproduce the statsId stickiness in Chrome [00:04:54] Now that I have "Disable cache" turned off [00:05:07] (03CR) 10Jforrester: [C: 032] Fix placement of linear selection inside a table [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/205012 (https://phabricator.wikimedia.org/T96446) (owner: 10Esanders) [00:05:17] umm [00:05:20] (03CR) 10jenkins-bot: [V: 04-1] Update VE core submodule to master (952a9bb) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/204943 (owner: 10Esanders) [00:05:23] Actually I was wrong. [00:05:27] 10VisualEditor, 10VisualEditor-ContentEditable, 10VisualEditor-Tables, 7Browser-Support-Firefox, and 2 others: Cannot click into a particular table cell on Firefox - https://phabricator.wikimedia.org/T96446#1219206 (10Jdforrester-WMF) 5Open>3Resolved [00:05:37] Oh FFS [00:05:37] I think I've changed how getEditingStatsId works since the version I have in my head [00:05:47] (03CR) 10Alex Monk: [C: 032] Only trust POSTed stats IDs [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205188 (owner: 10Catrope) [00:05:54] (03CR) 10Jforrester: [C: 032] Update VE core submodule to master (952a9bb) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/204943 (owner: 10Esanders) [00:05:57] Why does mw.user.generateSessionId() return a string of length 16 while the server-side equivalent returns a string of length 32? [00:05:59] WTF [00:06:04] 10VisualEditor, 10VisualEditor-ContentEditable, 10VisualEditor-Tables, 7Browser-Support-Firefox, and 2 others: Cannot click into a particular table cell on Firefox - https://phabricator.wikimedia.org/T96446#1217186 (10Jdforrester-WMF) [00:06:08] I seem to recall hearing about that before [00:06:09] generateSessionId is also used for VE sessions, right? [00:06:57] mw.user.generateRandomSessionId [00:07:01] yes [00:07:07] (03CR) 10Jforrester: "Could we have a test for this?" [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/205089 (https://phabricator.wikimedia.org/T96088) (owner: 10Esanders) [00:07:30] So VE sessions have 16-char IDs while WE sessions have 32-char IDs? [00:07:40] * RoanKattouw looks at database to check [00:08:08] The schema, which was set up for VE, says 32-char [00:08:13] :) [00:09:34] (03Merged) 10jenkins-bot: demo: Create RTL version of minimal demo [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/204942 (owner: 10Esanders) [00:09:36] (03Merged) 10jenkins-bot: Fix placement of linear selection inside a table [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/205012 (https://phabricator.wikimedia.org/T96446) (owner: 10Esanders) [00:09:38] (03Merged) 10jenkins-bot: Only trust POSTed stats IDs [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205188 (owner: 10Catrope) [00:09:41] (03Merged) 10jenkins-bot: Update VE core submodule to master (952a9bb) [extensions/VisualEditor] - 10https://gerrit.wikimedia.org/r/204943 (owner: 10Esanders) [00:11:14] ]> select length(event_editingSessionId) as length, event_editor, count(*) from Edit_11448630 WHERE timestamp BETWEEN "20150419160000" AND "20150419170000" GROUP BY event_editor, length; [00:11:20] | 16 | visualeditor | 13189 | [00:11:23] | 32 | wikitext | 105149 | [00:11:32] No other results [00:11:38] Cool story [00:12:18] So, I'll un-WIP my patch because I've now tested it and it works for me [00:12:36] 10VisualEditor, 10VisualEditor-MediaWiki, 10Wikimedia-Site-requests: Kill $wmgUseVisualEditorNamespace - https://phabricator.wikimedia.org/T92797#1219212 (10Krenair) a:3MZMcBride [00:12:41] But it does mean that wikitext session IDs will randomly be length 16 sometimes (when they've been regenerated by a pageshow handler) rather than 32 [00:16:49] (03PS2) 10Catrope: Regenerate editingStatsId on pageshow [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205189 [00:17:41] (03PS3) 10Catrope: Fix find and replace viewport search [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/205089 (https://phabricator.wikimedia.org/T96088) (owner: 10Esanders) [00:18:49] (03CR) 10Catrope: [C: 04-1] Fix find and replace viewport search (031 comment) [VisualEditor/VisualEditor] - 10https://gerrit.wikimedia.org/r/205089 (https://phabricator.wikimedia.org/T96088) (owner: 10Esanders) [00:21:02] (03PS2) 10Catrope: Add dependency on mw.user [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205191 [00:21:10] Oops, crap [00:22:19] (03PS3) 10Catrope: Regenerate editingStatsId on pageshow [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205189 [00:22:28] (03PS3) 10Catrope: Add dependency on mw.user [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205191 [00:27:10] Krenair: ---^^ [00:28:46] (03CR) 10Alex Monk: [C: 032] Add dependency on mw.user [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205191 (owner: 10Catrope) [00:29:41] (03Merged) 10jenkins-bot: Add dependency on mw.user [extensions/WikiEditor] - 10https://gerrit.wikimedia.org/r/205191 (owner: 10Catrope) [00:29:56] RoanKattouw, so... on every page, regenerate the ID? [00:30:20] Well, pageshow only fires when bfcache is used [00:30:26] Technically it also fires on first load I think [00:30:39] ah, ok [00:30:43] But that appears to happen before we attach to it [00:31:23] Also https://gerrit.wikimedia.org/r/205194 fixes the issue of revid being 0 for all client-side events, but I'm not too sure that I did the right thing there [00:32:03] MDN: "The pageshow event is fired when a session history entry is being traversed to. (This includes back/forward as well as initial page-showing after the onload event.)" [00:32:52] Yeah [00:35:30] that makes it sound like this fires on every page? [00:35:36] not just back/forward? [00:37:13] It does [00:37:20] [17:30] RoanKattouw Technically it also fires on first load I think [00:37:23] [17:30] RoanKattouw But that appears to happen before we attach to it [00:52:48] 10VisualEditor, 10VisualEditor-ContentEditable, 10VisualEditor-MediaWiki, 3Editing Department 2014/15 Q4 blockers, 5WMF-deploy-2015-04-22_(1.26wmf3): "Bitdefender Total Security 2015" inserts