[10:26:45] 6Collaboration-Team-Backlog, 10Flow, 10Collaboration-Community-Engagement: Create or gather interactive animations to illustrate how to use Flow features - https://phabricator.wikimedia.org/T130913#2150621 (10Trizek-WMF) [12:55:37] (03PS1) 10Zfilipin: WIP Simplify creating of Jenkins jobs for running browser tests daily [extensions/PageTriage] - 10https://gerrit.wikimedia.org/r/279604 (https://phabricator.wikimedia.org/T128190) [13:31:26] (03PS1) 10Zfilipin: WIP Simplify creating of Jenkins jobs for running browser tests daily [extensions/WikiLove] - 10https://gerrit.wikimedia.org/r/279611 (https://phabricator.wikimedia.org/T128190) [13:46:19] 6Collaboration-Team-Backlog, 10Thanks, 10pywikibot-core, 3Google-Summer-of-Code-2016: Pywikibot Support for Thanks (GSoC Proposal) - https://phabricator.wikimedia.org/T130585#2150938 (10darthbhyrava) >>! In T130585#2150450, @Niharika wrote: > Hi @darthbhyrava, I looked over your proposal and it looks good.... [13:49:33] 6Collaboration-Team-Backlog, 10Thanks, 10pywikibot-core, 3Google-Summer-of-Code-2016: Pywikibot Support for Thanks (GSoC Proposal) - https://phabricator.wikimedia.org/T130585#2150943 (10darthbhyrava) [14:53:41] (03CR) 10Sbisson: Stop counting notifications objects on the client (035 comments) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [14:54:02] (03PS19) 10Sbisson: Stop counting notifications objects on the client [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) [16:10:22] 6Collaboration-Team-Backlog, 10Flow: "Could not format Special:Contribution row" when user edit on Flow page was deleted - https://phabricator.wikimedia.org/T130933#2151274 (10Stryn) [16:49:38] 6Collaboration-Team-Backlog, 10Flow, 10VisualEditor, 10VisualEditor-MediaWiki-Links: [Regression wmf23] On creating an inter-wiki link, the nice regex strips section parts of the link - https://phabricator.wikimedia.org/T112898#2151384 (10Elitre) Yeah. [17:06:00] 6Collaboration-Team-Backlog, 10Thanks, 10pywikibot-core, 3Google-Summer-of-Code-2016: Pywikibot Support for Thanks (GSoC Proposal) - https://phabricator.wikimedia.org/T130585#2151419 (10darthbhyrava) [17:06:18] (03PS2) 10Zfilipin: WIP Simplify creating of Jenkins jobs for running browser tests daily [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279346 (https://phabricator.wikimedia.org/T128190) [17:09:10] (03PS2) 10Zfilipin: WIP Simplify creating of Jenkins jobs for running browser tests daily [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279354 (https://phabricator.wikimedia.org/T128190) [17:20:52] 6Collaboration-Team-Backlog, 10Thanks, 10pywikibot-core, 3Google-Summer-of-Code-2016: Pywikibot Support for Thanks (GSoC Proposal) - https://phabricator.wikimedia.org/T130585#2151448 (10darthbhyrava) [17:26:52] 6Collaboration-Team-Backlog, 10Thanks, 10pywikibot-core, 3Google-Summer-of-Code-2016: Pywikibot Support for Thanks (GSoC Proposal) - https://phabricator.wikimedia.org/T130585#2151477 (10darthbhyrava) [17:27:52] (03PS2) 10Zfilipin: WIP Simplify creating of Jenkins jobs for running browser tests daily [extensions/PageTriage] - 10https://gerrit.wikimedia.org/r/279604 (https://phabricator.wikimedia.org/T128190) [17:35:09] 6Collaboration-Team-Backlog, 10MediaWiki-extensions-PageCuration: Page Curation doesn't check for existing tags - https://phabricator.wikimedia.org/T130943#2151558 (10Ankit-Maity) [17:38:47] 6Collaboration-Team-Backlog, 10Flow: Contributions apparently could have wrong pagination - https://phabricator.wikimedia.org/T108289#2151578 (10Mattflaschen) [17:38:53] (03PS2) 10Zfilipin: WIP Simplify creating of Jenkins jobs for running browser tests daily [extensions/WikiLove] - 10https://gerrit.wikimedia.org/r/279611 (https://phabricator.wikimedia.org/T128190) [17:39:01] 6Collaboration-Team-Backlog, 10Flow: "Could not format Special:Contribution row" when user edit on Flow page was deleted - https://phabricator.wikimedia.org/T130933#2151274 (10Mattflaschen) [17:42:29] 6Collaboration-Team-Backlog, 10MediaWiki-extensions-PageCuration: Page Curation doesn't check for existing tags - https://phabricator.wikimedia.org/T130943#2151558 (10Mattflaschen) The article was deleted, so the diff no longer works. Please describe what happened exactly. Did it add a duplicate tag for one... [17:45:54] 6Collaboration-Team-Backlog, 10MediaWiki-extensions-PageCuration: Page Curation doesn't check for existing tags - https://phabricator.wikimedia.org/T130943#2151558 (10Catrope) Looking at the deleted history, I see that a user manually added a `{{db-copyvio}}` template, and 29 seconds later another user used Pa... [17:50:36] (03CR) 10Catrope: [C: 032] Use strong and em instead of wiki markup bold and italics [extensions/Flow] - 10https://gerrit.wikimedia.org/r/274659 (owner: 10Siebrand) [17:50:54] (03CR) 10jenkins-bot: [V: 04-1] Use strong and em instead of wiki markup bold and italics [extensions/Flow] - 10https://gerrit.wikimedia.org/r/274659 (owner: 10Siebrand) [17:53:27] (03PS4) 10Catrope: Use strong and em instead of wiki markup bold and italics [extensions/Flow] - 10https://gerrit.wikimedia.org/r/274659 (owner: 10Siebrand) [17:54:03] (03CR) 10Catrope: [C: 032] Use strong and em instead of wiki markup bold and italics [extensions/Flow] - 10https://gerrit.wikimedia.org/r/274659 (owner: 10Siebrand) [17:59:37] 6Collaboration-Team-Backlog, 10MediaWiki-extensions-PageCuration: Page Curation doesn't check for existing tags - https://phabricator.wikimedia.org/T130943#2151684 (10Ankit-Maity) Yes, that's what happened. I added a speedy deletion tag using Twinkle (just saying) and the user added another using PageCuration. [18:03:12] (03CR) 10Catrope: [C: 031] Stop counting notifications objects on the client [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [18:04:59] (03Merged) 10jenkins-bot: Use strong and em instead of wiki markup bold and italics [extensions/Flow] - 10https://gerrit.wikimedia.org/r/274659 (owner: 10Siebrand) [18:25:24] matt_flaschen: Re https://gerrit.wikimedia.org/r/#/c/279567/2/includes/SubmissionHandler.php : we have / have had other places where we output "user is blocked" regardless of the real error message. I don't quite understand why you say it's unlikely to be something else. Last time I ran into this it turned out that Flow talk page manager wasn't ipblockexempt [18:25:24] and the generic "user is blocked" error was obscuring that [18:25:29] 3Collaboration-Team-Current, 10Notifications, 13Patch-For-Review, 5WMF-deploy-2016-03-29_(1.27.0-wmf.19): Notification titles should differentiate subpages of the talk page - https://phabricator.wikimedia.org/T129686#2112745 (10Etonkovidova) Checked in betalabs Shows the differentiation between user talk... [18:27:03] (03CR) 10Mooeypoo: "Extremely minor documentation change" (032 comments) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [18:30:14] (03CR) 10Catrope: [C: 032] Remove unused error messages, switch API errors to HTML, and handle that on the client. (031 comment) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279567 (https://phabricator.wikimedia.org/T121137) (owner: 10Mattflaschen) [18:34:02] (03CR) 10Sbisson: Stop counting notifications objects on the client (031 comment) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [18:35:57] (03CR) 10Catrope: Stop counting notifications objects on the client (031 comment) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [18:39:52] (03Merged) 10jenkins-bot: Remove unused error messages, switch API errors to HTML, and handle that on the client. [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279567 (https://phabricator.wikimedia.org/T121137) (owner: 10Mattflaschen) [18:44:30] (03CR) 10Catrope: Stop counting notifications objects on the client (031 comment) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [18:45:48] (03CR) 10Mooeypoo: "This is a small issue that is not really the fault of this commit, but should be known:" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [18:46:41] stephanebisson, are you feeling better? [18:46:51] stephanebisson, can I bug you with bundles? [18:47:18] (03PS1) 10Catrope: jscs: Add "checkTypes": "strictNativeCase" and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279650 [18:49:15] RoanKattouw, no, I meant it was unlikely API consumers were relying on the string 'block'. [18:51:41] RoanKattouw, I tested with the following scenarios, and it seems to give correct errors now: autoblock, regular block, page protection, global block, and tor block. Plus an error not related to permissiont to check regressions (topic title length). [18:52:09] OK cool [18:52:24] mooeypoo: I'm better, but I'm supposed to stay away from bundles for a few days... [18:52:48] stephanebisson, allergies, eh? [18:53:01] mooeypoo: I'm developing one [18:53:01] Bundled allergies are the worst [18:53:10] mooeypoo: so, what's up? [18:53:19] stephanebisson, hopefully we'll cure bundles (and echo in general) in the hackathon [18:53:40] anyways, I'm working on a commit to make the popup properly update itself when reopening, including updating the x-wiki bundles [18:54:01] yep [18:54:32] but I have a big issue with bundled notifications. The code re-fetches the info from the API and then checks if the notification exists; if ti does, it *changes* the data in it (header/content and read/unread status if needed) and if not, it adds it in as a new notif [18:54:47] thing is .. the code checks whether the notif exists according to its ID [18:55:04] And it turns out that if a bundled notification changed, its ID changed as well [18:55:17] I see [18:55:26] so I keep getting a stack of "2 people edited your user page", "3 people edited your user page", "4 people edited your user page" ... [18:55:34] instead of having it change the content / header [18:56:00] Is there any way to recognize a bundle by what its id *used* to be? Can the API maybe also tell me "prevID" or something that I can use? [18:56:02] RoanKattouw: https://github.com/wikimedia/jscs-preset-wikimedia/blob/master/presets/wikimedia.json [18:56:43] stephanebisson, I know that you're working on basically rewriting the way bundles are done, so I'm wondering if what you're working on may fix this [18:57:05] What if every flyout opening was like the first one? Getting the data, creating and rendering without caring about what it used to be? [18:57:41] stephanebisson, sure, but then I have to empty the flyout and repopulate it, which is annoying to the user; they'll have an empty one with a spinner each time, even if there was no change [18:58:09] We did that at first, and then changed it to update stuff. Right now (without the fix I'm working on) it only *adds* new things (not change anything) but that means x-wiki aren't updating either [18:58:32] and it leaves the same bug around, btw... "2 people edited.." "3 people edited..." etc etc will happen now too. [18:59:12] I know, there's a few issues right now with new stuff not showing up [18:59:32] is there a way to empty it just when the new one is ready to be shown so it doesn't flicker? [18:59:43] hm. [19:00:00] or to always open it empty and populate once loaded so each opening is the same? [19:00:15] I can do #2 for sure, but I don't think that's nice for the user [19:00:17] 3Collaboration-Team-Current, 10Collaboration-Community-Engagement, 6Design-Research, 5Design Research and product team collaboration, 7Surveys: Use of Notifications survey - https://phabricator.wikimedia.org/T127257#2151893 (10jmatazzoni) [19:00:20] 3Collaboration-Team-Current, 6Design-Research-Backlog, 10Notifications, 13Patch-For-Review, and 2 others: Invite Users to Take Notifications Survey (Using Notifications Panel) - https://phabricator.wikimedia.org/T128937#2151892 (10jmatazzoni) 5Open>3Resolved [19:00:40] (03PS1) 10Mattflaschen: Clarify comment [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279653 [19:00:56] 3Collaboration-Team-Current, 10Flow, 5MW-1.27-release-notes, 13Patch-For-Review, and 2 others: Special:EditWatchlist/raw gives Exception encountered, of type "Wikimedia\Assert\ParameterTypeException" - https://phabricator.wikimedia.org/T130778#2151895 (10jmatazzoni) 5Open>3Resolved a:3jmatazzoni [19:01:05] it depends, if it would be the same, yes it's annoying. If it would be different, it's also annoying to see it jump from old state to new state [19:01:12] stephanebisson, I can probably do #1 but not as easily... let me try and do that, but if I get into trouble, I'll wait for the hackathon; I'd like to take time in the hackathon to sit with you and roan and figure out how to fix echo front end architecture properly, or at least how to start [19:01:13] (03CR) 10Mattflaschen: [C: 032] "Comment-only change to clarify a recent commit" [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279653 (owner: 10Mattflaschen) [19:01:22] and we just don't know if it's gonna be the same [19:01:39] (03PS1) 10Catrope: jscs: Enable checkParamNames and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 [19:01:41] mooeypoo: absolutely [19:01:42] (03PS1) 10Catrope: jscs: Enable checkRedundantReturns and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279656 [19:01:45] (03PS1) 10Catrope: jscs: Enable checkRedundantParams and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279657 [19:02:08] Hm. Yeah, I guess. it works with thecommit with everything except the bundles, which is annoying, but I think it might be unavoidable. I'll try to get things to update and then flip at the end [19:03:30] stephanebisson, i'm a little skeptical, because the model is tied to the widget (gloat, if you must :p ) so it is going to be an issue doing something completely disconnected, but I am going to try a couple of ideas before I throw in the towel on it and wait for us to sort the architecture better [19:03:43] mooeypoo: with the changes I'm making to bundling, having a bundle Id will be much less possible than before... [19:03:53] stephanebisson, I see. [19:04:34] Okay, I think I have an idea on how to do this, then. I'll give it a try and keep you in the loop. In any case, I am really looking forward to us sorting out the issues soon [19:04:44] ok, I'll brb, getting food [19:04:57] mooeypoo: well, you could easily make each opening the same in ext.echo.init.js [19:06:07] 3Collaboration-Team-Current, 10Notifications, 13Patch-For-Review, 5WMF-deploy-2016-03-29_(1.27.0-wmf.19): Notification titles should differentiate subpages of the talk page - https://phabricator.wikimedia.org/T129686#2151899 (10jmatazzoni) Hi @Etonkovidova, could I please see what one of these looks like i... [19:06:53] 3Collaboration-Team-Current, 10MediaWiki-extensions-GuidedTour: ext.guidedTour gives malformed uri sequence at nowiki - https://phabricator.wikimedia.org/T125793#2151902 (10Mattflaschen) 5Open>3Resolved Okay, please reopen if it occurs again. Hopefully this is the last time I say that. :) [19:07:57] (03PS20) 10Sbisson: Stop counting notifications objects on the client [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) [19:08:09] (03CR) 10jenkins-bot: [V: 04-1] jscs: Enable checkRedundantParams and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279657 (owner: 10Catrope) [19:11:56] (03CR) 10Sbisson: Stop counting notifications objects on the client (031 comment) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [19:15:11] (03Merged) 10jenkins-bot: Clarify comment [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279653 (owner: 10Mattflaschen) [19:16:29] (03CR) 10Sbisson: [C: 032] jscs: Add "checkTypes": "strictNativeCase" and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279650 (owner: 10Catrope) [19:18:32] (03CR) 10Sbisson: [C: 032] jscs: Enable checkParamNames and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 (owner: 10Catrope) [19:27:29] (03Merged) 10jenkins-bot: jscs: Add "checkTypes": "strictNativeCase" and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279650 (owner: 10Catrope) [19:31:05] (03PS21) 10Sbisson: Stop counting notifications objects on the client [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) [19:33:29] (03CR) 10jenkins-bot: [V: 04-1] jscs: Enable checkParamNames and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 (owner: 10Catrope) [19:35:28] RoanKattouw: how do you run jscs locally? When I run `grunt jscs` it fails with 32 nonsense errors. [19:39:53] stephanebisson: Is it up-to-date? [19:40:10] stephanebisson: (I did `npm install -g jscs` a while ago which helps.) [19:40:38] James_F: wait, is it not in package.json under dev dependencies? [19:41:10] stephanebisson: It will also be. [19:41:32] stephanebisson: But I find it helps. [19:43:59] rm -rf node_modules && npm install fixes it [19:44:05] Ha. [19:44:10] So often I find that to be true, yeah. :-( [19:44:30] jscs is there as a dependency of grunt-jscs, having it there as a top level would allow us to control the version [19:44:51] Don't do that. [19:44:54] Please. :-) [19:45:05] many npm package do poor version pining, they will satisfy themselves with any version [19:45:11] We trust grunt-jscs to not break this. [19:45:36] And specifically we trust the maintainers of jscs-preset-wikimedia (i.e., me) to not screw it up too badly. ;-) [19:46:23] grunt-jscs' role is to interface with grunt, it doesn't care which version of jscs you have as long as it is has the interface it expects [19:46:49] It pins quite well. [19:47:13] that's not been my experience just now [19:47:44] 6Collaboration-Team-Backlog, 10Flow, 6Language-Engineering, 10Parsoid, and 4 others: Provide an API flag to suppress auto-generated - https://phabricator.wikimedia.org/T113331#2152014 (10ssastry) 5Open>3Resolved Parsoid now sets autoGenerated: true in data-mw for auto-generated reference... [20:01:15] Generally I just run npm test [20:01:24] With npm install first if it yells at me [20:03:21] (03CR) 10Catrope: "Bogus failure: 19:19:44 npm ERR! enoent ENOENT, open '/mnt/jenkins-workspace/workspace/mwext-qunit/src/node_modules/karma/node_modules/soc" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 (owner: 10Catrope) [20:03:32] (03CR) 10Catrope: [C: 032] "Per SBisson" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 (owner: 10Catrope) [20:16:48] 6Collaboration-Team-Backlog, 10Flow, 6Language-Engineering, 10Parsoid, and 4 others: Provide an API flag to suppress auto-generated - https://phabricator.wikimedia.org/T113331#2152050 (10ssastry) https://www.mediawiki.org/wiki/Parsoid/MediaWiki_DOM_spec#Auto-generated_references_blocks docum... [20:19:25] (03CR) 10jenkins-bot: [V: 04-1] jscs: Enable checkParamNames and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 (owner: 10Catrope) [20:24:31] (03CR) 10Catrope: "20:06:59 25 03 2016 20:06:59.812:ERROR [launcher]: Chrome crashed." [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 (owner: 10Catrope) [20:24:38] (03CR) 10Catrope: [C: 032] jscs: Enable checkParamNames and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 (owner: 10Catrope) [20:25:40] RoanKattouw: I think there's a race condition in MWEchoNotifUser::resetNotificationCount. It reads the counts and schedule a deferred update to write them. [20:26:58] Hmm, yeah [20:27:02] Why does that use a deferred update anyway [20:27:13] clicking "mark as read" quickly ends up leaving a wrong count in echo_unread_wikis [20:27:18] Its callers within the same file are markRead and markUnread which are already supposed to be called from deferred updates [20:27:50] Hmm but there are other callers too [20:28:21] mark as read while visiting a page needs to be deferred but is it required for the API? [20:29:34] No, for the API I don't think it would be because that's a POST request [20:30:18] I don't think any of the callers to resetNotificationCount need the deferred behavior [20:30:24] Now going to git blame why we have a deferred there [20:30:52] Oh, It's Always Been That Way (TM) [20:31:00] It was deferred in legoktm's first commit introducing it [20:32:43] it's not quite one of those "How Did it Ever Worked" because if you don't go too fast it mostly works [20:32:54] Right [20:33:38] Also when Aaron first started sprinkling DeferredUpdates all over the place to ensure invariants that multi-DC needs, I didn't think much of it because I didn't realize the race condition potential [20:33:44] Now I know to be more skeptical [20:33:59] it made troubleshooting my new counter object pretty annoying for a while [20:34:49] (03CR) 10Mooeypoo: [C: 032] "LGTM, and now we need to update Echo on MobileFrontend to use this." [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [20:35:31] (03CR) 10Catrope: "That's https://gerrit.wikimedia.org/r/#/c/278340/" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [20:36:15] stephanebisson: There's also https://phabricator.wikimedia.org/T130853 BTW [20:36:21] Different race condition [20:37:04] (03Merged) 10jenkins-bot: jscs: Enable checkParamNames and make pass [extensions/Echo] - 10https://gerrit.wikimedia.org/r/279655 (owner: 10Catrope) [20:37:09] (03PS2) 10Catrope: Disable submit buttons if we know the user can't edit [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279568 (https://phabricator.wikimedia.org/T108762) (owner: 10Mattflaschen) [20:37:23] Amazing [20:37:32] "However, there is a core variable (wgIsProbablyEditable)..." [20:37:34] yep [20:39:16] so about the other race condition (https://phabricator.wikimedia.org/T130853) the 'mark as read' on click won't be needed soon [20:39:44] Why not? [20:39:54] Because of your target_page changes? [20:39:59] if we keep the target pages around, we know what should be marked as read and we can decide not to if it has been marked as unread explicitely [20:40:05] OK, sure [20:40:34] But the thing is that when the user clicks a notification, we used to rely on target_page to cause that click to mark the notification as read [20:40:50] With marked-as-unread that falls apart, and without that behavior clicking such a notification wouldn't mark it as read [20:40:57] So that's why that's there; I don't like it either [20:41:31] But maybe we could do this server-side instead, with ?marknotifread=123 or soemthing [20:42:10] alternatively, we could always mark them as read on click (and not deferre the update count) and do less onPersonalUrl [20:43:06] but "?marknotifread=123" would be processed in the same request as onPersonalUrl, right? [20:45:05] Hmm yeah we could always mark read on click but that breaks marking as read when following a link from an email [20:45:26] Although I do have to say that the magical part of target_page where just visiting a page organically marks multiple notifs as read feels evil to me [20:45:46] (03Merged) 10jenkins-bot: Stop counting notifications objects on the client [extensions/Echo] - 10https://gerrit.wikimedia.org/r/277609 (https://phabricator.wikimedia.org/T129726) (owner: 10Sbisson) [20:45:56] ?marknotifread=123 would be processed exactly the same way that target_page is now, you'd just get the event ID that you're going to mark as read from the query string instead of from the DB [20:45:57] we could still do the onPersonalUrl business but in many cases it would have nothing to do and would do less counter adjustments... [20:46:28] So how about we kill target_page completely and move to ?marknotifread=123 for everything [20:47:00] RoanKattouw, yeah, I added it (for GettingStarted orignally, IIRC). It's a funny name, but useful. [20:47:05] and kill the "mark as read on visit"? [20:47:12] I think that'd let us do mark-as-read-on-click, regardless of if the notif is really new or was marked as unread, it would allow us to support links from emails too, and visiting a page wouldn't magically mark multiple notifs as read [20:47:18] Yeah, kill "mark as read on visit" [20:47:27] I haven't heard anyone speak up in favor of that feature though, and I'd like to [20:47:38] I don't feel it's useful but perhaps it is for reasons that I don't see [20:47:58] Killing it also appeals to me because it would simplify the code quite a bit [20:47:59] we should certainly explore it, less is more! [20:48:02] I think it's useful, but I don't know if it's wroth the trouble [20:48:06] worth* [20:48:07] RoanKattouw, why is mark as read on visit evil? [20:48:09] But yes, exactly, let's explore [20:48:21] For topics at least, it seems useful. [20:48:22] * RoanKattouw adds "file phab task for killing mark as read on visit" to TODO list [20:48:32] If you visit a topic page, why would you want to still see unread notifications about that topic? [20:48:32] For topics it could be [20:48:41] Same for user talk probably. [20:48:46] Even non-Flow. [20:48:48] Although topics could be big, and notifications highlight new replies [20:49:00] For boards I don't think it's as useful, because boards often are big [20:49:07] Just visiting the board doesn't mean you've seen everything [20:49:11] It could be for many things; if we consider unread notifications as actionable, then by visiting the page they refer to, there's no action that should be taken anymore, and they are, by definition, read. [20:49:20] And in complex discussions, just having visited the page doesn't mean you've noticed everything there is to notice [20:49:30] Yeah, it could be debated for boards. [20:49:33] It might be useful to tweak its focus, a bit, then [20:49:34] right [20:49:37] One of the ways I use notifications is to tell me what's new (and that I should care about) in large discussions [20:49:39] I find it a bit evil for mentions, you can easily see the page without seeing the mention and mentions are direct communications that you should not miss [20:49:48] Yeah, exactly [20:49:59] But you still have that notification in your list, it's just marked as read [20:50:16] "You visited VPT therefore you have seen all of the mentions of your name on VPT" is not always true [20:50:45] (Context for those who don't know what VPT is: https://en.wikipedia.org/wiki/WP:VPT ) [20:52:26] stephanebisson, AFAICT, there is none for mention. [20:52:35] target-page is not used. [20:54:54] Oh good [20:54:57] Probably not for Thanks either [20:55:56] I was mentioning it "in theory" but obviously someone thought about it already. [20:59:11] mooeypoo: looks like the changes to notifications may have broken our browser tests - https://integration.wikimedia.org/ci/job/mwext-mw-selenium/5122/console [20:59:29] they seem to inconsistently fail [20:59:52] im guessing race condition [21:00:34] oop no.. feature broken http://en.m.wikipedia.beta.wmflabs.org/wiki/Headings#/notifications [21:00:38] load.php?debug=false&lang=en&modules=jquery%2Cmediawiki&only=scripts&skin=minerva&version=1%2FaBimRT:178 TypeError: this.unreadCounter.update is not a function TypeError: this.unreadCounter.update is not a function(…) [21:01:02] jdlrobson: feel free to revert while I investigate [21:01:16] stephanebisson: i see https://gerrit.wikimedia.org/r/278340 hasnt merged yet [21:01:27] is it possible the Echo change was a breaking one? [21:01:49] e.g. https://gerrit.wikimedia.org/r/#/c/277609/ [21:01:56] jdlrobson: yes, my bad [21:02:19] I made the corresponding change in MF but I didn't notice the original change created a bad state [21:03:45] but now, if we revert the change in Echo and https://gerrit.wikimedia.org/r/#/c/278340 merges, it will be broken as well :( [21:04:30] is the merge cancel because of the rebase? [21:05:07] stephanebisson: looks like that change will fix the issue [21:06:44] yay for browser tests [21:06:59] It should. If it doesn't, we revert everything to turn this Friday night situation into a Monday morning solution. [21:07:51] jdlrobson: I'm impressed by your browser tests. I was never able to make the Flow ones reliable. [21:12:49] stephanebisson: @integration tag is the key [21:12:53] they run per commit :) [21:13:05] surprising how quick they get fixed when they break and stop merges :) [21:13:27] i'd be happy to pair with you at some point around Flow's/Echo's if that would be useful [21:13:49] sure [21:14:00] the @integration tests tend to be more reliable than the ones that run on the beta cluster https://integration.wikimedia.org/ci/view/Mobile/job/browsertests-MobileFrontend-SmokeTests-linux-chrome-sauce/ [21:14:09] they send too many false positives [21:14:11] the problem with the Flow ones is that they are not really broken, they are unreliable [21:15:34] so @integration doesn't run on beta? [21:15:39] it runs locally? [21:16:55] I think there's a few browser tests already in either Echo or Flow that run from 'check experimental' [21:17:16] But we never got around to testing those and making them non-experimental [21:17:59] (03CR) 10Catrope: [C: 04-1] Disable submit buttons if we know the user can't edit (032 comments) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279568 (https://phabricator.wikimedia.org/T108762) (owner: 10Mattflaschen) [21:18:11] * RoanKattouw goes to get lunch [21:23:43] stephanebisson: https://www.mediawiki.org/wiki/Reading/Web/QA [21:23:47] stephanebisson: in particular https://www.mediawiki.org/wiki/Continuous_integration/Browser_tests [21:24:08] if the Flow tests are unreliable they probably would benefit from some better Given statements [21:24:20] for instance we had some tests that would only pass because they ran after the test before [21:27:43] quite possible [21:27:48] the patch is merged [21:28:38] jdlrobson: is the notification overlay working as expected? [21:51:42] RoanKattouw, and 'check experimental' is broken last I checked (Paladox would very much like to help fix that, but I'm kind of CI-ed out for a little while after last round with the release branches) [21:54:21] 3Collaboration-Team-Current, 10Notifications, 13Patch-For-Review, 5WMF-deploy-2016-03-29_(1.27.0-wmf.19): Notification titles should differentiate subpages of the talk page - https://phabricator.wikimedia.org/T129686#2152267 (10Etonkovidova) Mention on Flow user talk sub-pages - see the screenshot below {F... [22:26:43] 3Collaboration-Team-Current, 10Notifications, 13Patch-For-Review, 5WMF-deploy-2016-03-29_(1.27.0-wmf.19): Notification titles should differentiate subpages of the talk page - https://phabricator.wikimedia.org/T129686#2152336 (10Etonkovidova) @jmatazzoni - reverts on Flow boards do not trigger Notification... [23:01:22] 6Collaboration-Team-Backlog, 10Flow, 10Notifications: Revert notifications for Flow edits - https://phabricator.wikimedia.org/T130969#2152397 (10Mattflaschen) [23:01:35] 3Collaboration-Team-Current, 10Notifications, 13Patch-For-Review, 5WMF-deploy-2016-03-29_(1.27.0-wmf.19): Notification titles should differentiate subpages of the talk page - https://phabricator.wikimedia.org/T129686#2112745 (10Mattflaschen) >>! In T129686#2152336, @Etonkovidova wrote: > @jmatazzoni - reve... [23:04:30] mooeypoo, do you know why the QUnit runner is RTL?... [23:05:53] RTL? [23:07:31] matt_flaschen: thx for T130969 Revert notifications for Flow edits. I am looking now at link notifications and see that [23:07:31] T130969: Revert notifications for Flow edits - https://phabricator.wikimedia.org/T130969 [23:07:42] matt_flaschen: Links notifications are sent only when links are made between wikitext pages in mainspace. (?!) [23:08:40] matt_flaschen: should be another ticket? Cannot believe it first - was checking/re-checking - seems to be true [23:09:26] etonkovidova, no, that is deliberate. [23:10:11] matt_flaschen: I see [23:10:23] Changing it would probably cause a lot of unwanted links from various places. E.g. see https://en.wikipedia.org/w/index.php?title=Special%3AWhatLinksHere&target=Earth&namespace=2 [23:10:46] RoanKattouw, yeah, it's very weird. When I run all the QUnit tests at once, it starts LTR then goes RTL. [23:11:02] 3Collaboration-Team-Current, 10Notifications, 13Patch-For-Review, 5WMF-deploy-2016-03-29_(1.27.0-wmf.19): Notification titles should differentiate subpages of the talk page - https://phabricator.wikimedia.org/T129686#2152433 (10Etonkovidova) @jmatazzoni Links notifications are sent only when links are made... [23:11:22] etonkovidova, e.g. people have lists of pages they approved, wikiprojects have lists of pages they plan to work on, etc. [23:11:26] improved, not approved. [23:11:55] matt_flaschen: ok :) [23:14:24] (03CR) 10Paladox: "check experimental" [extensions/Flow] - 10https://gerrit.wikimedia.org/r/279490 (owner: 10L10n-bot) [23:14:46] (03CR) 10Paladox: "check experimental" [extensions/Thanks] - 10https://gerrit.wikimedia.org/r/279510 (owner: 10L10n-bot) [23:15:34] The directionality thing might be some stuff in GuidedTour (though it is supposed to restore properly...) [23:16:49] 6Collaboration-Team-Backlog, 10Flow, 6Language-Engineering, 10Parsoid, and 4 others: Provide an API flag to suppress auto-generated - https://phabricator.wikimedia.org/T113331#2152444 (10GWicke) > If you want to receive the new version right away (without waiting for the HTML to turn over in... [23:18:58] I'm probably going to head out about 5:30 Pacific. [23:19:28] 3Collaboration-Team-Current, 10Flow, 10Thanks: Flow failure with Thanks 'check experimental' - https://phabricator.wikimedia.org/T128628#2152446 (10Paladox) I'm now getting some new errors. 23:16:50 2) ApiFlowThankTest::testRequestWithoutToken 23:16:50 Flow\Exception\InvalidDataException: User has no id and... [23:30:47] 6Collaboration-Team-Backlog, 10Thanks: Failing test ApiRevThankTest::testValidRequest UsageException: Revision ID is not valid - https://phabricator.wikimedia.org/T130973#2152481 (10JanZerebecki) [23:46:20] 6Collaboration-Team-Backlog, 10Notifications: Advanced User preferences for cross-wiki notifications - https://phabricator.wikimedia.org/T114917#2152526 (10jmatazzoni) [23:48:48] 3Collaboration-Team-Current, 10Notifications, 7I18n, 13Patch-For-Review, and 3 others: User names and other content language-y things in notifications should be bidi-isolated - https://phabricator.wikimedia.org/T127407#2042930 (10Etonkovidova) 1) Checked for usernames used previously for {T106741} ET((V... [23:55:51] matt_flaschen, where is the QUnit runner RTL ? [23:56:49] matt_flaschen, it sounds like one of the tests flip the body tag (for some reason) and perhaps doesn't flip it back [23:57:02] is this in our extensions, or is this happening when running the entire QUnit stack in mw ? [23:59:49] mooeypoo, I'm not positive, but I think it's GuidedTour. It is supposed to be restoring, but doesn't seem to be working right.