[03:06:15] 10Flow, 6Collaboration-Team-Backlog, 10Pywikibot-Flow, 10pywikibot-core, and 2 others: Implement Flow support in Pywikibot - https://phabricator.wikimedia.org/T67119#1566198 (10jayvdb) [03:18:45] 10Echo, 6Collaboration-Team-Backlog: "You have new messages" is highlighted in orange whenever you view your talk page - https://phabricator.wikimedia.org/T109997#1566204 (10Krenair) [04:40:31] (03CR) 10Mattflaschen: "I'll merge this after branch cut, unless anyone finds anything. (Beta cluster will deploy the DB changes using update.php)." [extensions/Flow] - 10https://gerrit.wikimedia.org/r/136280 (https://phabricator.wikimedia.org/T107204) (owner: 10Werdna) [04:43:24] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Old references are not removed properly when removing them from the content - https://phabricator.wikimedia.org/T94569#1566271 (10Mattflaschen) >>! In T94569#1560559, @matthiasmullie wrote: > * TopKIndex::removeFromIndex (and probably others) cu... [04:48:14] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 7WorkTypeNewFunctionality: We should update the database with all necessary new templates that Flow needs upon installation - https://phabricator.wikimedia.org/T105421#1566274 (10Mattflaschen) Links in Template:LQT page converted to Flow and Template:... [04:53:36] (03CR) 10Mattflaschen: Create a guided tour for first-time users (032 comments) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/230924 (https://phabricator.wikimedia.org/T108266) (owner: 10Mooeypoo) [08:21:15] 10Echo, 6Collaboration-Team-Backlog: "You have new messages" is highlighted in orange whenever you view your talk page - https://phabricator.wikimedia.org/T109997#1566478 (10Schnark) The same behavior can be seen without Echo, too: {T59840}. [08:38:09] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 7WorkTypeNewFunctionality: Opt-in: Guided tour on user talk for first visit to new Flow board - https://phabricator.wikimedia.org/T108266#1566525 (10Pginer-WMF) >>! In T108266#1529671, @Mooeypoo wrote: > If we want to use the image in the first step,... [09:00:08] 10Flow, 6Collaboration-Team-Backlog: 'Mark as resolved' requires 'Update summary' - https://phabricator.wikimedia.org/T109741#1566550 (10Pginer-WMF) In order to avoid this confusion, the behaviour proposed [[ http://pauginer.github.io/prototypes/flow/close-topic/index.html | in the original designs ]] (try res... [14:07:50] 10Flow, 6Collaboration-Team-Backlog: Some icons in Flow look blurry - https://phabricator.wikimedia.org/T110035#1567359 (10matmarex) 3NEW [14:08:45] (03PS20) 10Sbisson: [WIP] Beta feature Flow on user talk page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/230648 (https://phabricator.wikimedia.org/T98270) [14:18:58] (03CR) 10Sbisson: [C: 032] Don't open two editors when clicking a nested reply link [extensions/Flow] - 10https://gerrit.wikimedia.org/r/232881 (https://phabricator.wikimedia.org/T109748) (owner: 10Catrope) [14:22:53] (03CR) 10Sbisson: [C: 032] After switching editors, move cursor to end [extensions/Flow] - 10https://gerrit.wikimedia.org/r/232785 (https://phabricator.wikimedia.org/T97290) (owner: 10Catrope) [14:23:15] (03PS2) 10Hashar: build: Use npm for jshint, jscs, jsonlint, csslint and banana [extensions/PageTriage] - 10https://gerrit.wikimedia.org/r/230968 (https://phabricator.wikimedia.org/T108779) (owner: 10Jforrester) [14:24:37] (03PS3) 10Hashar: build: Use npm for jshint, jscs, jsonlint, csslint and banana [extensions/PageTriage] - 10https://gerrit.wikimedia.org/r/230968 (https://phabricator.wikimedia.org/T63618) (owner: 10Jforrester) [14:38:26] (03Merged) 10jenkins-bot: Don't open two editors when clicking a nested reply link [extensions/Flow] - 10https://gerrit.wikimedia.org/r/232881 (https://phabricator.wikimedia.org/T109748) (owner: 10Catrope) [14:41:06] (03CR) 10Sbisson: [C: 031] "According to the doc[1], getParserOutput is supposed to be used to render the content on the page. It is clearly not how Flow operates." [extensions/Flow] - 10https://gerrit.wikimedia.org/r/233066 (https://phabricator.wikimedia.org/T98270) (owner: 10Mattflaschen) [14:42:13] 10Flow, 6Collaboration-Team-Backlog, 10Pywikibot-Flow, 10pywikibot-core, 7Tracking: Pywikibot and Flow (tracking) - https://phabricator.wikimedia.org/T110042#1567550 (10happy5214) 3NEW [14:42:29] (03Merged) 10jenkins-bot: After switching editors, move cursor to end [extensions/Flow] - 10https://gerrit.wikimedia.org/r/232785 (https://phabricator.wikimedia.org/T97290) (owner: 10Catrope) [14:42:38] 10Flow, 6Collaboration-Team-Backlog, 10Pywikibot-Flow, 10pywikibot-core, 7Tracking: Pywikibot and Flow (tracking) - https://phabricator.wikimedia.org/T110042#1567565 (10happy5214) [14:42:41] 10Flow, 6Collaboration-Team-Backlog, 10Pywikibot-Flow, 10pywikibot-core, and 2 others: Implement Flow support in Pywikibot - https://phabricator.wikimedia.org/T67119#1012198 (10happy5214) [14:43:24] 10Flow, 6Collaboration-Team-Backlog, 10Pywikibot-Flow, 10pywikibot-core, 7Tracking: Pywikibot and Flow (tracking) - https://phabricator.wikimedia.org/T110042#1567550 (10happy5214) [14:43:54] 10Flow, 6Collaboration-Team-Backlog, 10Pywikibot-Flow, 10pywikibot-core, and 2 others: Implement Flow support in Pywikibot - https://phabricator.wikimedia.org/T67119#1026831 (10happy5214) [14:44:41] 10Flow, 6Collaboration-Team-Backlog, 7Tracking: Bots and Flow living in harmony (tracking) - https://phabricator.wikimedia.org/T107093#1567591 (10happy5214) [14:46:38] 10Flow, 6Collaboration-Team-Backlog, 7Tracking: Bots and Flow living in harmony (tracking) - https://phabricator.wikimedia.org/T107093#1486543 (10happy5214) [14:46:40] 10Flow, 6Collaboration-Team-Backlog, 10Pywikibot-Flow, 10pywikibot-core, and 2 others: Implement Flow support in Pywikibot - https://phabricator.wikimedia.org/T67119#1567601 (10happy5214) [14:49:49] (03CR) 10Hashar: "check experimental" [extensions/PageTriage] - 10https://gerrit.wikimedia.org/r/230968 (https://phabricator.wikimedia.org/T63618) (owner: 10Jforrester) [15:07:14] (03CR) 10Sbisson: [WIP] Beta feature Flow on user talk page (039 comments) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/230648 (https://phabricator.wikimedia.org/T98270) (owner: 10Sbisson) [15:07:26] (03CR) 10Sbisson: [WIP] Beta feature Flow on user talk page (031 comment) [extensions/Flow] - 10https://gerrit.wikimedia.org/r/230648 (https://phabricator.wikimedia.org/T98270) (owner: 10Sbisson) [15:07:40] 10Flow, 6Collaboration-Team-Backlog, 10Pywikibot-Flow, 10pywikibot-core, 3Google-Summer-of-Code-2015: Implement Flow support in Pywikibot - https://phabricator.wikimedia.org/T67119#1567740 (10happy5214) [15:12:01] 10Flow, 6Collaboration-Team-Backlog, 10MediaWiki-General-or-Unknown, 7Design: A solution is needed for the users who just don't want to read their talk pages - https://phabricator.wikimedia.org/T110048#1567753 (10Amire80) 3NEW [15:12:23] 10Flow, 6Collaboration-Team-Backlog, 10MediaWiki-General-or-Unknown, 7Design: A solution is needed for the users who just don't want to read their talk pages - https://phabricator.wikimedia.org/T110048#1567762 (10Amire80) [15:24:29] 10Flow, 6Collaboration-Team-Backlog: Reconsider Flow's use of .mw-ui-hovericon (dropdown menu items that gain progressive/constructive/destructive styling on hover) - https://phabricator.wikimedia.org/T110051#1567825 (10matmarex) 3NEW [15:59:11] I'm going to watch the MW security training meeting on the YouTube stream. [15:59:41] Link is in #wikimedia-staff channel [16:19:43] 10Flow, 6Collaboration-Team-Backlog, 10MediaWiki-General-or-Unknown, 7Design: A solution is needed for the users who just don't want to read their talk pages - https://phabricator.wikimedia.org/T110048#1567963 (10Ciencia_Al_Poder) Just to confirm, when do you mean //read// you really mean //read// and not... [16:29:46] 10Flow, 6Collaboration-Team-Backlog, 10MediaWiki-General-or-Unknown, 7Design: A solution is needed for the users who just don't want to read their talk pages - https://phabricator.wikimedia.org/T110048#1568013 (10Amire80) >>! In T110048#1567963, @Ciencia_Al_Poder wrote: > Just to confirm, when do you mean... [16:44:48] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-25_(1.26wmf20), 7WorkTypeNewFunctionality: Regression on production: 'Cancel' for Nested Replies (>two levels) broken - https://phabricator.wikimedia.org/T109748#1568112 (10Etonkovidova) Checked in betalabs - all is fixed for 8 l... [16:44:48] (03PS1) 10Matthias Mullie: [WIP] Fix references/storage [extensions/Flow] - 10https://gerrit.wikimedia.org/r/233432 (https://phabricator.wikimedia.org/T94569) [16:46:48] (03CR) 10jenkins-bot: [V: 04-1] [WIP] Fix references/storage [extensions/Flow] - 10https://gerrit.wikimedia.org/r/233432 (https://phabricator.wikimedia.org/T94569) (owner: 10Matthias Mullie) [16:48:53] mlitn: hey, you there? [17:04:20] hey stephanebisson [17:04:22] I’m here [17:05:23] mlitn: nevermind, I found the answer to my question this time, thanks [17:07:03] :) [17:11:03] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-25_(1.26wmf20), 7WorkTypeMaintenance: F3. End the occupation - https://phabricator.wikimedia.org/T105574#1568194 (10Mattflaschen) [17:13:48] (03Abandoned) 10Mattflaschen: Add explicit calls to the garbage collector every 100 topics [extensions/Flow] - 10https://gerrit.wikimedia.org/r/231201 (https://phabricator.wikimedia.org/T108601) (owner: 10Mattflaschen) [17:17:58] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Fix new topic notification on wikis with job queue - https://phabricator.wikimedia.org/T107835#1568215 (10Quiddity) p:5High>3Unbreak! [17:18:23] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Fix new topic notification on wikis with job queue - https://phabricator.wikimedia.org/T107835#1505375 (10Quiddity) [17:19:06] 10Echo, 3Collaboration-Team-Current, 5Patch-For-Review, 7WorkTypeNewFunctionality, and 2 others: Split notifications into Alerts and Messages - https://phabricator.wikimedia.org/T108190#1568217 (10Pginer-WMF) I proposed to use blue to highlight messages, to avoid presenting them as a problem that needs imm... [17:21:03] mlitn, I'm going to schedule an evening SWAT for "Get rid of $wgFlowOccupyPages" (https://gerrit.wikimedia.org/r/#/c/228271/1) (the mediawiki-config one). [17:21:14] mlitn, that will mean it will roll out before the Flow one, but that should be fine. [17:21:23] The Flow one is merged, though. [17:24:47] I just saw http://en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA is not working. [17:26:11] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-25_(1.26wmf20), 7WorkTypeMaintenance: F3. End the occupation - https://phabricator.wikimedia.org/T105574#1568253 (10Mattflaschen) http://en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA somehow broke on Beta (http://en.wikipedia.b... [17:27:16] matt_flaschen, filed as https://phabricator.wikimedia.org/T109982 [17:27:58] 10Flow, 6Collaboration-Team-Backlog: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1568266 (10Mattflaschen) [17:27:59] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-25_(1.26wmf20), 7WorkTypeMaintenance: F3. End the occupation - https://phabricator.wikimedia.org/T105574#1568265 (10Mattflaschen) [17:31:23] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1565672 (10Mattflaschen) [17:34:09] Evening all. I'm trying to install Flow on a wiki, but when I have it installed and go to a page with it, I get a database error that implies the tables haven't been added to my database. I've run update.php after including it in Localsettings.php and I can't find an answer online. Anyone have any ideas? [17:44:39] (addendum to the above question. StuckWithTheFlow posted these details, yesterday...) [17:44:41] I've installed Flow on my new test MediaWiki, enabled it in LocalSettings.php and added the NS_TALK namespace to the areas it enables, but when I try and create a page with Flow I get the error "A database query error has occurred. This may indicate a bug in the software." [17:44:41] I've also run update.php and convertnamespacetext too [17:44:41] turning on verbose errors, I get Error: 1146 Table doesn't exist, but since I've run update.php I'm not sure why [17:44:41] then trying to run the Web Updater script, it just has a spinning wheel [17:45:03] thanks [17:46:03] (np. FYI: We just finished a meeting, so people are shuffling between rooms and tasks. It might be a few minutes before anyone checks IRC.) [17:47:57] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1568361 (10Mattflaschen) Looks like it is probably because populateContentModel.php was not run in Beta on the page table (it was in prod; see... [17:48:31] StuckWithTheFlow, which table doesn't exist? [17:51:57] ebernhardson: may I request your assistance... [17:53:56] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1568386 (10Mattflaschen) I ran it on all talk, but I realized this has to be fixed manually, since now the wgFlowOccupyPages code is removed t... [17:55:18] 10Echo, 3Collaboration-Team-Current, 5Patch-For-Review, 7WorkTypeNewFunctionality, and 2 others: Split notifications into Alerts and Messages - https://phabricator.wikimedia.org/T108190#1568401 (10DannyH) Given where we are in the quarter, we don't have time to do any more testing on current Flow functiona... [17:55:22] smw_flow_workflow [17:56:21] more precisely, matt_flaschen Error: 1146 Table 'xxxxxxxxxxxx.smw_flow_workflow' doesn't exist (localhost) [17:57:17] StuckWithTheFlow, well, does it exist? [17:57:30] Maybe under a different prefix? [17:57:40] stephanebisson: sure, whats up? [17:57:50] matt_flaschen: not that I can see [17:57:53] ebernhardson: i'm writing a maintenance script and i have a couple of questions [17:58:02] I have Semantic MediaWiki installed and I'm wondering if that's interfered somehow [17:58:55] ebernhardson: it extends LoggedUpdateMaintenance, it has doDBUpdates for the payload and getUpdateKey that returns a hardcoded string for now [17:59:18] StuckWithTheFlow, MW allows you to specify a table prefix (https://www.mediawiki.org/wiki/Manual:$wgDBprefix), though I don't know how commonly used this is. [17:59:25] ebernhardson: should I expect `php update.php` to run it? [17:59:52] yes, I installed it overall with Softaculous and it wanted me to set a prefix which I did [18:01:05] StuckWithTheFlow, it looks like Flow should support that. I'm not sure why it wouldn't have created the table. Do you see any errors when you run update.php? [18:01:15] There are no flow tables, and I spotted no errors running update.php [18:01:48] 10Flow, 6Collaboration-Team-Backlog: Take out "This description has been modified" line in side rail legal text - https://phabricator.wikimedia.org/T110082#1568448 (10DannyH) 3NEW [18:02:26] matt_flaschen, aren't there Flow-specific update php files to run? [18:02:27] Or is it all in update.php. Sorry, I ran a couple of updaters but I don't remember if they included ones specifically for installation or just to update things from our code [18:02:28] stephanebisson: first run through it should run, yes [18:02:43] stephanebisson: oh, you need to add it in the right hook though [18:02:56] mooeypoo, in certain cases, but the tables are created automatically. Otherwise, the unit tests wouldn't work. [18:03:04] ebernhardson: what hook? [18:03:07] oh that's right. [18:03:07] stephanebisson: in FlowHooks::getSchemaUpdates() [18:03:15] 10Flow, 6Collaboration-Team-Backlog: Tweak LQT->Flow import code to support a mode where LQT is enabled by default - https://phabricator.wikimedia.org/T107634#1568461 (10DannyH) [18:04:07] StuckWithTheFlow, can you ask at the support desk (https://www.mediawiki.org/wiki/Project:Support_desk) if you haven't already> You can also file a bug if you want, but I'm not sure where the problem is. [18:04:26] looking at update.php it hasn't tried to add any Flow tables [18:05:11] it's all semantic mediawiki table updates. I don't know how update.php works so I'm not sure how I could force the Flow stuff. I will ask at the support desk, thanks for your time [18:05:29] ebernhardson: nice [18:07:26] ebernhardson: also, I have big and ugly sql in there... should I try to build it with the $db api like WikiPage.php:986 (getContributors)? Or it is ok for a maintenance script to have sql ? [18:07:53] 10Flow, 6Collaboration-Team-Backlog: Move "Mark as resolved" up in the Flow dropdown menu - https://phabricator.wikimedia.org/T110083#1568470 (10DannyH) 3NEW [18:08:21] stephanebisson: maint scripts often have sql. it can be looked down on because maint scripts are hard to test...but doesn't stop anyone [18:09:29] ebernhardson: what's the right way to interpolate params in sql in php? [18:10:03] stephanebisson: basically just $dbw->addQuotes( $foo ) ... its not particularly beautiful [18:10:33] * ebernhardson would like to get a modern sql layer in the middle there someday ... but such a huge task [18:11:49] ebernhardson: if you have "... where page_namespace = $ns ..." [18:11:59] this is just string concat [18:12:06] there must be a better way [18:12:50] stephanebisson: not in mediawiki (which is where my modern sql issues come in) [18:13:08] stephanebisson: unless its a simple enough query you can pass an array of k/v pairs to Database::select() [18:13:17] array( 'foo' => $unsafeUserDat a) [18:13:46] it's a complex query, 3 tables with a sub-select [18:14:04] yea, you will just have to manually escape, label the variables as such, then build the query [18:14:05] probably doable with $db api but that would be much harder to read [18:14:21] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1568496 (10Mattflaschen) I ran a manual UPDATE for the one page affected (I checked and the Talk:Flow on en_rtlwiki does not exist, plus that... [18:16:37] 10Flow, 6Collaboration-Team-Backlog: Style of new anon edit warning - https://phabricator.wikimedia.org/T110086#1568505 (10DannyH) 3NEW a:3Pginer-WMF [18:17:07] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1568519 (10Mattflaschen) For the remainder (not strictly required for this bug but for consistency with {T104278}, I started: ``` mattflasche... [18:17:41] 10Flow, 6Collaboration-Team-Backlog: Style of new anon edit warning - https://phabricator.wikimedia.org/T110086#1568529 (10DannyH) [18:17:42] 10Flow, 6Collaboration-Team-Backlog: Logged out warning is un-dismissable - https://phabricator.wikimedia.org/T108836#1568528 (10DannyH) [18:34:58] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-18_(1.26wmf19), and 2 others: Keep input focus when switching from VE to wikitext editing - https://phabricator.wikimedia.org/T97290#1568600 (10Etonkovidova) >The cursor should always be at the end of the existing text, not the beg... [18:35:32] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1568606 (10Mattflaschen) Run on all wikis, all namespaces completed. [18:43:31] 10Flow, 6Collaboration-Team-Backlog, 10MediaWiki-General-or-Unknown, 7Design: Encourage users who are currently ignoring their talk pages to read them - https://phabricator.wikimedia.org/T110048#1568650 (10Mattflaschen) [18:46:18] 10Flow, 6Collaboration-Team-Backlog, 10MediaWiki-General-or-Unknown, 7Design: Encourage users who are currently ignoring their talk pages to read them - https://phabricator.wikimedia.org/T110048#1568668 (10Mattflaschen) The old title made it sound like we should optimize for users who just preferred never... [18:51:10] 10Flow, 6Collaboration-Team-Backlog: Editing posts: Keep input focus when switching from VE to wikitext editing - https://phabricator.wikimedia.org/T110093#1568709 (10Etonkovidova) 3NEW [18:52:08] (03PS21) 10Sbisson: [WIP] Beta feature Flow on user talk page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/230648 (https://phabricator.wikimedia.org/T98270) [18:52:41] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-18_(1.26wmf19), and 2 others: Keep input focus when switching from VE to wikitext editing - https://phabricator.wikimedia.org/T97290#1238153 (10Etonkovidova) The issues with keeping focus while editing was filed as a separate issu... [18:54:28] Beta issue is resovled. Should not affect production since we already ran that script there earlier (before occupy pages was merged). [18:56:02] So I think we should be good to remove the OccupyPages config, which I've scheduled for evening SWAT. [18:58:15] Sounds good [19:00:00] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Port remaining editor uses to EditorWidget - https://phabricator.wikimedia.org/T109000#1568818 (10Catrope) [19:00:01] 10Flow, 6Collaboration-Team-Backlog: Editing posts: Keep input focus when switching from VE to wikitext editing - https://phabricator.wikimedia.org/T110093#1568817 (10Catrope) [19:00:30] legoktm: when you have a chance, could you take a look at https://gerrit.wikimedia.org/r/#/c/233066/ to see if there's consequences we're not considering? [19:00:31] 10Flow, 6Collaboration-Team-Backlog: Editing posts: Keep input focus when switching from VE to wikitext editing - https://phabricator.wikimedia.org/T110093#1568709 (10Catrope) Editing existing posts still uses the old editor system; this bug should be fixed once we migrate that to the new editor system. [19:02:29] stephanebisson: Where is your big query with a sub select and stuff? [19:02:41] I may have some tricks up my sleeve so you don't have to manually do escaping [19:02:45] stephanebisson: is 'This is a placeholder in place of a Flow board.' going to be seen by users? [19:04:00] legoktm: I don't think so... it's kind of what I want to know, how is getParserOutput used [19:04:12] legoktm, is it possible that the Echo api doesn't update 'echomarkseen' with alerts ? [19:04:46] RoanKattouw: https://gerrit.wikimedia.org/r/#/c/230648/21/maintenance/FlowUpdateBetaFeaturePreference.php [19:05:53] OK [19:06:25] This bug is killingme. I am sending an api request 'echomarkseen' with params={ actopn: 'echomarkread', format: 'json', uselang: 'en', sections: 'alert', token: '...' } [19:06:31] and I get a response back with 1 unread [19:06:39] .... it's supposed to make it all read :( [19:06:42] So what you can easily do at least is build the outer query using the builder stuff [19:06:58] * RoanKattouw cringes at page.page_title = REPLACE(user.user_name, ' ', '_') ... but sadly that seems necessary [19:07:28] RoanKattouw: I wish there was a better way to join these [19:09:02] mooeypoo: actopn or action? [19:09:23] stephanebisson: Content::getParserOutput is used whenever something wants the HTML version of the content object. callers are stuff like EditPage, WikiPage, DifferenceEngine, etc. But I think Flow overrides most of those. [19:10:10] so if an extension isn't really aware of different content models, it could end up displaying that [19:10:45] (sorry, got distracted) [19:10:49] action [19:10:55] You could do something like [19:10:58] 10Flow, 6Collaboration-Team-Backlog: 'Mark as resolved' requires 'Update summary' - https://phabricator.wikimedia.org/T109741#1568885 (10DannyH) 5Open>3Resolved a:3DannyH The current solution is fine. We tried using the distinction between Keep summary / Update summary buttons, and it was confusing even... [19:12:44] [12:06:26] This bug is killingme. I am sending an api request 'echomarkseen' with params={ actopn: 'echomarkread', format: 'json', uselang: 'en', sections: 'alert', token: '...' } <-- markseen and markread are two different things? [19:12:54] yes [19:13:03] Sorry. [19:13:09] I'm talking about mark read [19:13:20] forget I said anything about markseen [19:13:24] ok :P [19:13:42] * mooeypoo is about to throw her laptop out the window with this bug [19:13:47] stephanebisson: it looks fine [19:13:59] (03CR) 10Legoktm: [C: 031] Use placeholder HTML rather than rendering the board in getParserOutput [extensions/Flow] - 10https://gerrit.wikimedia.org/r/233066 (https://phabricator.wikimedia.org/T98270) (owner: 10Mattflaschen) [19:14:06] I would +2 but i didn't test it [19:14:25] legoktm: thanks [19:14:34] legoktm: how would someone meaningfully test it? [19:15:45] I meant like pull down the patch and load a few random pages and make sure nothing broke [19:16:12] ok [19:16:30] you might be able to cause some breakage by combining TemplateSandbox + Flow [19:17:34] $db->select( array( 'page', 'user' ), 'user_id', array( 'page_namespace' => NS_USER_TALK, 'page_content_model' => CONTENT_MODEL_FLOW_BOARD, "user_id NOT IN (....)" ) ); [19:17:47] stephanebisson: So that --^^ would be a bit nicer and be equivalent to your other query [19:17:59] Replaced the join with a multi-table select with a where cluse [19:18:06] but I would suspect that TemplateSandbox is already broken w/r to Flow [19:18:18] ....but I didn't put the where clause in. Pretend 'page_title = REPLACE(user_name, ' ', '_') [19:18:22] was in there somewhere [19:18:34] I wonder if the NOT IN( SELECT ... ) thing could be replaced with a LEFT JOIN ... WHERE foo IS NULL ? [19:18:47] mooeypoo: so, you send the echomarkread request to mark all alerts as read, then use meta=notifications which gives an unread alert? but only one? [19:19:20] I did test the empty page scenario, but that's one thing, along with regression testing. [19:21:27] RoanKattouw: would it be more secured or optimized? I wouldn't say more readable... [19:21:44] SELECT user_id FROM page JOIN user ON page_namespace=NS_USER_TALK AND page_title=REPLACE(user_name, ' ', '_') LEFT JOIN user_properties ON up_user=user_id AND up_property='foo' AND up_value=1 WHERE page_content_model='bar' [19:21:52] legoktm, no, the response from the api to the 'echomarkread' gives me back a number of unread notifications. The number might not be 0, and indeed it's not - seems the operation fails. [19:21:53] Probably not more readable [19:21:53] BUT [19:22:10] In any case I wonder what the REPLACE will do for optimization [19:22:21] legoktm, but it works fine if I have debugger statements peppered into my code, which suggests to me that this is some race condition [19:22:29] And the placeholder is not supposed to be seen by users, but like legoktm said it could if an extension isn't aware of Flow. [19:22:56] legoktm, RoanKattouw suggested I remove the other request (that is supposedly unrelated) to 'echomarkseen'. Now things work. So... can 'echomarkseen' somewhow invalidate echomarkread ? [19:23:02] In this case the NOT IN thing might be faster because basically nobody will have the pref enabled when you run the script [19:23:47] matt_flaschen: I'm imaginining that there is an extension out there that just calls WikiPage::factory( $title )->getRevision()->getContent()->getParserOutput() on an arbitrary title and uses that parser output somehow expecting wikitext or something [19:24:07] legoktm, well, the output itself is supposed to be HTML obviously. [19:24:14] I'm fairly sure you can break MassMessage by converting "Project:Example" to a Flow page. [19:24:54] legoktm, before the change as well? [19:25:17] yeah [19:25:33] Also, I believe that placeholder HTML could be generated from wikitext. Not sure that matters. [19:27:45] legoktm, when the popup opens, a "echomarkseen" is sent to the api to update the seen time. Now, *if* the widget is setup as "mark all as read on open" (which 'alerts' is) then after the notifications have came back from the API and have rendered, the code also sends a 'echomarkread' with sections: alert to make them all read. [19:27:48] mooeypoo: I'm a little confused...if you mark them all as read, can't you just assume there are 0 unread notifications? [19:28:38] legoktm, if both are sent, echomarkread doesn't seem to actuall apply to anything. It isn't updated. It gives me back a response that there's still 1 unread notification (which is wrong) *and* the next time I reload the page, there's a seen but *UNREAD* alert [19:28:59] wtf [19:29:13] legoktm, however. If I prevent "echomarkseen" from being sent, then things work beautifully. The echomarkread actually marks all notifications as read, gives me back a nice little response, and all is right in the world [19:29:25] except that the alerts load as read but UNSEEN because I didn't update seen time [19:29:31] * mooeypoo cries for echo mercy [19:30:23] legoktm, also, the problem isn't about me assuming. I assumed before, and updated the button label to 0, but then when you refresh the page, you get a gray 1, because there's still an unread alert in there, despite the fact that it's seen. [19:30:26] matt_flaschen: by the way, the change to getParserOutput doesn't remove completely the error "Formatting for wrong user..." [19:30:32] legoktm, in short, this is whack. [19:30:58] this is using your patch right? [19:31:04] legoktm, yes [19:31:19] though I fixed a couple of small things. Let me uncomment 'echomarkseen' and push it [19:32:03] (03PS26) 10Mooeypoo: Split alerts and messages in Echo [extensions/Echo] - 10https://gerrit.wikimedia.org/r/231200 (https://phabricator.wikimedia.org/T108190) [19:32:04] legoktm, ^^ [19:33:33] * legoktm tests [19:34:44] legoktm, I thought it was a race condition within the JS but the fact that regardless of what the JS displays it still remains unread after you refresh the page is telling :( [19:34:49] this is so weird [19:35:27] I think we're not invalidating the cache properly [19:35:53] legoktm, I have to eat. Do you need me for the next 30-40 minutes or can I go eat and pick this back up again when I come back? [19:36:09] nah, go eat :) [19:36:31] ta, I'll be on hangouts if there's anything major. I'll bbiab. [19:58:00] stephanebisson, hmm, can you get a trace of the error that still happens with getParserOutput? [19:58:32] Block.php:152, as usual, I'll get a trace in a few minutes [19:58:47] (03CR) 10Mattflaschen: "Thanks for looking at it. Let's not merge this before branch cut." [extensions/Flow] - 10https://gerrit.wikimedia.org/r/233066 (https://phabricator.wikimedia.org/T98270) (owner: 10Mattflaschen) [19:58:53] (03CR) 10Mattflaschen: [C: 04-1] Use placeholder HTML rather than rendering the board in getParserOutput [extensions/Flow] - 10https://gerrit.wikimedia.org/r/233066 (https://phabricator.wikimedia.org/T98270) (owner: 10Mattflaschen) [19:59:46] Thanks. [19:59:59] Also, is this with the permissions workaround from before? [20:02:28] (03PS22) 10Sbisson: [WIP] Beta feature Flow on user talk page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/230648 (https://phabricator.wikimedia.org/T98270) [20:03:16] matt_flaschen: yes, setting TPMU in the context and using your fix to getParserOutput [20:03:47] mooey|lunch: I'm not sure what exactly I did, but: http://i.imgur.com/1ru8Kd3.png [20:12:09] mooey|lunch: that was after I opened and closed the flyout [20:13:56] a bad value is getting cached somehow [20:13:57] 1440447141.276023 [0 127.0.0.1:60539] "SETEX" "wiki:echo-notification-count:3:1.5" "86400" "1" [20:13:58] 1440447141.282414 [0 127.0.0.1:60539] "SETEX" "wiki:echo-notification-count-alert:3:1.5" "86400" "1" [20:13:58] 1440447141.283278 [0 127.0.0.1:60539] "SETEX" "wiki:echo-notification-count-message:3:1.5" "86400" "0" [20:14:08] those "1"s should be "0" [20:15:23] mooey|lunch: I'll keep looking after lunch. Also the flyout icon is broken in monobook. [20:19:18] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Fix new topic notification on wikis with job queue - https://phabricator.wikimedia.org/T107835#1569040 (10Mattflaschen) a:3Mattflaschen [20:22:10] matt_flaschen: I had forgotten to apply the container->extend fix [20:22:21] matt_flaschen: it does help but it breaks something else... [20:22:41] I see this is the log: InvalidArgumentException from line 232 of /vagrant/mediawiki/extensions/Flow/vendor/Pimple/Container.php: Identifier "permissions" does not contain an object definition. [20:22:58] Odd [20:24:20] stephanebisson, extend is saying that that (before extension) it is not a callable. [20:25:25] stephanebisson, is it possible it's getting double-extended? (Though I actually don't see how that would cause that error.) [20:26:59] stephanebisson: https://phabricator.wikimedia.org/P1919 [20:27:17] If you break in container.php, or add some temporary logging code, we could figure out what it is when extend complains. [20:27:23] stephanebisson: (semi-tested) [20:28:28] legoktm: Yeah I saw the undefined thing when she walked me through it on her machine [20:28:53] She set conditional breakpoints (with the condition being blah === undefined) to figure out where that comes from but then we went to lunch [20:36:14] matt_flaschen: it's expecting a function but $c['permissions'] contains an instance of Flow\RevisionActionPermissions, which seems legit [20:37:57] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-25_(1.26wmf20), and 2 others: Catchable fatal error: Argument 1 passed to FlowHooks::isTalkpageManagerUser() must be an instance of User, null given in /srv/mediawiki/php-1.26wmf18/... - https://phabricator.wikimedia.org/T109063#1569067 [20:38:00] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-18_(1.26wmf19), and 3 others: Catchable fatal error: Argument 1 passed to Flow\Formatter\RevisionFormatter::formatApi() must be an instance of Flow\Formatter\FormatterRow, bool give... - https://phabricator.wikimedia.org/T109537#1569070 [20:38:27] 10Flow, 6Collaboration-Team-Backlog: Editing posts: Keep input focus when switching from VE to wikitext editing - https://phabricator.wikimedia.org/T110093#1569072 (10DannyH) p:5Triage>3Normal [20:39:36] 10Flow, 6Collaboration-Team-Backlog, 7Design: Flow entry fields should not look terrible - https://phabricator.wikimedia.org/T95512#1569077 (10DannyH) [20:39:38] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-18_(1.26wmf19), and 2 others: Keep input focus when switching from VE to wikitext editing - https://phabricator.wikimedia.org/T97290#1569075 (10DannyH) 5Open>3Resolved Looks good -- I moved Elena's ticket about editing posts in... [20:39:46] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-25_(1.26wmf20), 7WorkTypeNewFunctionality: Regression on production: 'Cancel' for Nested Replies (>two levels) broken - https://phabricator.wikimedia.org/T109748#1569078 (10DannyH) 5Open>3Resolved [20:40:09] legoktm, i'm fixing monobook now. Let me know if you have *any* ideas about the bug. I'm at a loss with it :( [20:40:31] 10Flow, 6Collaboration-Team-Backlog, 10Wikidata: Wikidata - Flow bugs and feature reqs (tracking) - https://phabricator.wikimedia.org/T100844#1569096 (10DannyH) [20:40:33] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-18_(1.26wmf19), 7WorkTypeNewFunctionality: Switching between source & VE in Flow entry field should keep the cursor focus in the entry field - https://phabricator.wikimedia.org/T101189#1569094 (10DannyH) 5stalled>3Resolved T9... [20:41:18] matt_flaschen: not sure why but extending earlier in the code helps [20:42:08] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-18_(1.26wmf19), 7WorkTypeNewFunctionality: F4. Add legal text to Flow side rail - https://phabricator.wikimedia.org/T97371#1569101 (10DannyH) 5Open>3Resolved I'm going to talk to Nick + Legal about the small-screen display, a... [20:42:35] stephanebisson, it's because once the id ('permissions' in this case) is read, it freezes the service so you don't have two different objects implementing it. So by moving it up, you put it before it was frozen. [20:42:48] stephanebisson, I thought the location I suggested before was early enough but seems not to be. [20:43:13] matt_flaschen: in 1 case it is not [20:45:18] all working now! [20:45:26] Cool. :) [20:45:31] when can we merge the change to getParserOutput? [20:46:13] I'd prefer tomorrow after branch cut. It shouldn't break anything, but it seems high-Possible Impact. [20:46:19] That gives it more time on Beta Cluster before prod. [20:46:56] sounds good [20:56:04] (03PS23) 10Sbisson: [WIP] Beta feature Flow on user talk page [extensions/Flow] - 10https://gerrit.wikimedia.org/r/230648 (https://phabricator.wikimedia.org/T98270) [20:56:58] (03CR) 10Sbisson: "Depends on https://gerrit.wikimedia.org/r/#/c/233066/" [extensions/Flow] - 10https://gerrit.wikimedia.org/r/230648 (https://phabricator.wikimedia.org/T98270) (owner: 10Sbisson) [20:58:30] RoanKattouw, https://gerrit.wikimedia.org/r/233598 [21:02:31] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Fix new topic notification on wikis with job queue - https://phabricator.wikimedia.org/T107835#1569139 (10Mattflaschen) Probably the cause (slightly redacted by changing user and event ID actual integers to just 'int'), from fluorine: ``` 2015-... [21:02:39] 10Echo, 6Collaboration-Team-Backlog, 6Scrum-of-Scrums, 7Schema-change: Perform schema change to echo_target_page changing from a 1 to 1 mapping between pages and user/notification to a 1 to many. - https://phabricator.wikimedia.org/T94427#1163090 (10Mattflaschen) [21:02:40] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Fix new topic notification on wikis with job queue - https://phabricator.wikimedia.org/T107835#1569141 (10Mattflaschen) [21:04:30] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Fix new topic notification on MediaWiki.org - https://phabricator.wikimedia.org/T107835#1569143 (10Mattflaschen) [21:06:26] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Fix new topic notification on MediaWiki.org - https://phabricator.wikimedia.org/T107835#1569145 (10Mattflaschen) If that is the cause, then it is not jobqueu-specific but rather due to https://git.wikimedia.org/blob/operations%2Fmediawiki-config... [21:07:59] 10Echo, 6Collaboration-Team-Backlog, 6Scrum-of-Scrums, 7Schema-change: Perform schema change to echo_target_page changing from a 1 to 1 mapping between pages and user/notification to a 1 to many. - https://phabricator.wikimedia.org/T94427#1569146 (10Mattflaschen) >>! In T94427#1189486, @Springle wrote: > T... [21:08:15] 10Echo, 10Flow, 6Collaboration-Team-Backlog, 5Patch-For-Review: S12. Flow Notifications are not marked as read by visiting the Topic page - https://phabricator.wikimedia.org/T85064#1569153 (10Mattflaschen) [21:08:18] 10Echo, 6Collaboration-Team-Backlog, 6Scrum-of-Scrums, 7Schema-change: Perform schema change to echo_target_page changing from a 1 to 1 mapping between pages and user/notification to a 1 to many. - https://phabricator.wikimedia.org/T94427#1569150 (10Mattflaschen) 5Resolved>3Open [21:16:40] 10Flow, 3Collaboration-Team-Current, 7WorkTypeMaintenance: addurl CAPTCHA broken (HTML is rendered as text) - https://phabricator.wikimedia.org/T109554#1569180 (10Mattflaschen) a:3Mattflaschen [21:34:34] RoanKattouw, so it seems T109554 is from the new editor rewrite (I was wrongly thinking a ConfirmEdit regression). The HTML comes from the server fine, but it's being treated as text. [21:34:42] Ugh [21:34:45] Which would *normally* be a good idea except ConfirmEdit is a little wonky. [21:35:24] RoanKattouw, to avoid XSSing ourselves unintentionally, I was thinking maybe we could whitelist ntspamfilter and only treat that one as HTML. [21:35:51] Also, do you think it makes sense to have a common base class for the editor widgets (not the editor itself, but the one that handles the network requests). [21:36:04] E.g. BoardDescriptionWidget, NewTopicWidget. [21:36:36] Hmm maybe [21:36:49] You're saying this because you want to centralize the error handling, I'm guessing? [21:38:06] RoanKattouw, yes. [21:39:02] They're all exactly the same, and now exactly the same and about to get more complex. [21:39:43] Yeah [21:39:50] I felt a bit uneasy about putting the error stuff in three times [21:39:56] But I was like "it's only 2-3 lines of code" [21:39:59] But that's about to stop being true [21:40:06] Now it's only 5 lines of code. [21:40:08] And so it begins... [21:40:41] RoanKattouw, I could put up a quick fix that duplicates it (to SWAT out), then a follow up so the arch thinking isn't rushed. [21:40:56] That sounds very sensible [21:44:03] Forgot about HtmlSnippet. That's a nice feature. [21:45:20] matt_flaschen: https://phabricator.wikimedia.org/T109982 - for enwiki I still see the same result for query on _content_model for page and revision tables [21:46:03] etonkovidova, enwiki beta or enwiki production? [21:46:14] matt_flaschen: enwiki beta [21:48:00] etonkovidova, paste bin of what output you get? [21:48:57] 10Flow, 3Collaboration-Team-Current, 7WorkTypeNewFunctionality: Port remaining editor uses to EditorWidget - https://phabricator.wikimedia.org/T109000#1569221 (10Catrope) a:3Catrope [21:49:12] matt_flaschen: it's exactly the same as in the https://phabricator.wikimedia.org/T109982 - 12 rows for SELECT page_id, page_namespace, page_title, page_content_model, rev_content_model FROM page JOIN revision ON rev_page = page_id WHERE page_namespace = 1 AND page_title IN ('Flow', 'Flow_QA'); [21:49:32] mooeypoo: ok, figured it out. [21:49:52] it's a really fun race condition between echomarkseen and echomarkread [21:50:07] etonkovidova, I forget, how to you normally connect to Beta DB? This time, I used sql enwiki. [21:50:28] /s/to you/do you/ [21:50:56] marktraceur: ssh deployment-db1.eqiad.wmflabs [21:51:43] legoktm, "fun" is an interesting term for it, whatever it is [21:52:06] haha [21:52:07] https://phabricator.wikimedia.org/P1921 [21:52:20] so, echomarkseen triggers a user options save [21:52:55] which in turn clears the notification count cache (because your preferences might change echo things, removing or adding some notif types, etc.) [21:53:09] at the same time, echomarkread is marking things as read, and also clears the cache [21:53:31] and depending on which one goes first, a bad value might get cached [21:53:47] But why would resetting twice hurt any values? [21:54:05] resetting the cache would mean having to fetch directly from the DB, no? [21:55:47] but your db read could be out of date [21:56:14] markseen: hit db, read new values [21:56:27] markread: mark things as read [21:56:33] markread: hit db, read new values [21:56:40] markread: set new values in redis [21:56:47] markseen: set new values in redis (that are now out of date) [21:57:00] or something crazy like that. [21:59:48] etonkovidova, yeah, I see the discrepancy. Will look at it after I fix the CAPTCHA thing. [22:00:12] matt_flaschen: ok then :) [22:00:25] RoanKattouw or legoktm, do you know how to read db-eqiad.php/db-labs.php? Specifically I forget which one's slave. [22:00:38] etonkovidova, it depends how the db is accessed, maybe slave lag. [22:00:39] looking [22:00:51] matt_flaschen: Off the top of my head, I think the first listed one is the master [22:01:30] RoanKattouw, okay what's weird if so is that query (see T109982) gets different results if I: [22:01:34] 1. sql enwiki from deployment-bastion [22:01:38] matt_flaschen: hmm... could be that. But I should wait for you not just re-run the script or something? [22:01:49] matt_flaschen: I typically use https://tendril.wikimedia.org/, I don't know if that is set up for labs though. [22:02:02] 2. Connect directly to deployment-db1. [22:02:24] Weird [22:02:36] So it's probably either connecting to a different server, or using a different DB [22:02:36] sql now uses a slave [22:02:41] you need to pass --write to get the master [22:02:43] etonkovidova, the script will actually not do anything (or it shouldn't) since it only changes null rows. You might want to move on to another task for now. [22:02:45] It should be easy to figure out what's going on [22:02:50] Oh what legoktm said sounds very plausible [22:03:09] But also you can use ps ajxf|less to figure out which DB sql enwiki decided to talk to [22:03:09] matt_flaschen: kk - no problem :) [22:03:10] I didn't know that, so that's good to know (and safer). [22:03:17] [Ops] sql --write [22:03:21] However, it still doesn't make sense to me, because the order I did was: [22:03:48] 1. Run the script on talk namespace. Actually did the wrong thing (wikitext) for Talk:Flow_QA since it was too late and the default was already wrong [22:03:50] legoktm, that... is whack. [22:03:59] 2. Fix just that page to flow-board. [22:04:12] 3. Run it on all pages (which shouldn't have affected that page since it was already non-null). [22:04:18] legoktm, also, I'm pretty sure this scenario was in one of my db architecture exams [22:04:23] heh [22:05:01] So master should have flow-board and the slave should have either null, wikitext, or flow-board. But wikitext implies a *lot* of slave lag since it didn't even get to my fix after just talk namespace. [22:05:48] Also the slave (if that's what sql enwiki is) is *right* and deployment-db1 is wrong. [22:05:52] So now you see why I have to come back to this. [22:06:33] matt_flaschen: How did you perform #2 exactly? [22:06:44] Oh, **** [22:07:01] I thought sql was the master, so that makes sense... [22:07:17] (03PS27) 10Mooeypoo: Split alerts and messages in Echo [extensions/Echo] - 10https://gerrit.wikimedia.org/r/231200 (https://phabricator.wikimedia.org/T108190) [22:07:29] legoktm, I updated a small bug I found in JS + the monobook and modern themes [22:07:46] Would be nice if the slaves did not allow updates. [22:07:51] though "modern" is only half supported since we really don't support that skin at all [22:08:09] RoanKattouw, yeah, I did the update after just: [22:08:14] sql enwiki [22:08:17] on deployment-bastion [22:08:20] ouch :/ [22:08:31] legoktm, so, the bug about the markseen and markread are backend, yes? nothing I should do in JS? or should I try and make them take a break from one another before each of them updates ? [22:08:49] matt_flaschen: Yup that's exactly what I suspected [22:09:21] Rerun the update query against the master and pray that replication to that slave doesn't breka? [22:09:28] (I don't think it would break) [22:09:49] mooeypoo: I'm asking Aaron about it in -dev right now, if it's easy we should just fix it in the backend, otherwise we should have the markseen request go after the markread one or something. [22:11:43] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1569297 (10Mattflaschen) So I thought: ``` sql enwiki ``` on deployment-bastion would give me a slave, but apparently it now gives a slave.... [22:13:21] RoanKattouw, alright, wish me luck. It's also unfortunate that it's even possible to run an update on a slave from the prompt. Think I should file a task about that? [22:13:27] Yes you should [22:13:34] slaves should not allow writes from random people [22:13:48] (03CR) 10jenkins-bot: [V: 04-1] Split alerts and messages in Echo [extensions/Echo] - 10https://gerrit.wikimedia.org/r/231200 (https://phabricator.wikimedia.org/T108190) (owner: 10Mooeypoo) [22:14:08] matt_flaschen: Also, "on deployment-bastion would give me a slave, but apparently it now gives a slave.", one of those probably needs to be "master" [22:14:49] Thanks, fixed. [22:18:10] 10Flow, 6Collaboration-Team-Backlog: On small screens, put legal text on the bottom of the page - https://phabricator.wikimedia.org/T110116#1569345 (10DannyH) 3NEW [22:18:26] 10Flow, 6Collaboration-Team-Backlog: On small screens, put legal text on the bottom of the page - https://phabricator.wikimedia.org/T110116#1569353 (10DannyH) 5Open>3stalled [22:20:38] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1569356 (10Mattflaschen) Re-run against master (using sql --write). [22:22:52] etonkovidova: Thanks. [22:27:04] etonkovidova, you can retest that one now. [22:27:21] matt_flaschen: it was quick :) [22:31:34] matt_flaschen: page_content_model='flow-board' for both 'Flow' and 'Flow_QA in beta enwiki - hurray! [22:31:48] :) [22:33:15] matt_flaschen: Hmm did you work on a generic editor wrapper à la ReplyWidget yet? Because I think I might need one [22:33:19] for T109000 [22:33:24] (03PS28) 10Mooeypoo: Split alerts and messages in Echo [extensions/Echo] - 10https://gerrit.wikimedia.org/r/231200 (https://phabricator.wikimedia.org/T108190) [22:33:33] RoanKattouw, no, got side-tracked by the DB thing. [22:33:34] Editing an existing reply probably doesn't even need all of ReplyWidget [22:33:36] OK good [22:33:43] Then we're not getting in each other's way [22:34:02] I can probably get away with using ReplyWidget here actually [22:34:10] It has a config option to turn off the features I don't need [22:34:38] RoanKattouw, do you know why it sets the label to a , not text directly? [22:35:05] It doesn't block me, I just don't get an excuse to use HtmlSnippet. [22:35:09] (plus I'm wondering) [22:35:30] legoktm, okay, so, I made the JS play nice with this race condition bug. We should probably still fix it in the back end, but it should no longer be a blocker for this commit now. [22:35:48] matt_flaschen: Hah, good point! No, I don't know [22:36:07] You're absolutely right, there's no reason to pass in a span there, it should just pass in text [22:36:08] AFAICT [22:36:14] mooeypoo, should probably go ahead and file the back-end one if it's not already. [22:36:34] I thought legoktm filed something [22:37:18] oh, no, wait, that was a paste thing [22:37:27] I don't think there's a bug for it, I started working on a patch though [22:37:56] mooeypoo, do you know (about the at https://git.wikimedia.org/blob/mediawiki%2Fextensions%2FFlow.git/f77dbdc82eb28b197245f22cc30ab115433b6735/modules%2Fflow%2Fui%2Fwidgets%2Fmw.flow.ui.BoardDescriptionWidget.js#L185 )? [22:38:12] 10Flow, 3Collaboration-Team-Current: flow-action-unsupported only content on en.wikipedia.beta.wmflabs.org/wiki/Talk:Flow_QA - https://phabricator.wikimedia.org/T109982#1569431 (10Etonkovidova) In betalabs: ``` [enwiki]>SELECT page_id, page_namespace, page_title, page_content_model, rev_content_model FROM page... [22:39:46] matt_flaschen, ah, yes, so, setLabel() accepts either string or jQuery. If it accepts a string, it uses .text() to it, which "destroys" any classes or stylings or whatever you need in there. In this particular case I, too, used .text() so it was redundant, but it was a force of habit to give label a jQuery object to prevent issues. [22:40:02] matt_flaschen, RoanKattouw i think in this case, though, it was almost right (except for the .text() part on the span, no? ) [22:40:28] mooeypoo, .text() is the safe bet. It's just that ConfirmEdit (but hopefully no others) *has* to be html(), so I'm going to whitelist it. [22:41:19] matt_flaschen, yeah if it's confirm edit, $.parseHTML or something. I thought of doing something like that when I saw that bug in the list but I got extremely destracted by the "split echo" bug [22:41:46] mooeypoo, OO UI has something cool, HtmlSnippet, so you can mark something as safe HTML. [22:41:58] isn't that just in HTML ? [22:41:59] err [22:42:01] PHP [22:42:18] Maybe before? Not anymore. [22:42:23] ooh [22:43:14] wait... what? IT's not doing anything except have a .toString() which delivers the content. How is this safer? [22:43:46] mooeypoo, it's just a marker: [22:43:53] oh oh i see what you mean [22:44:14] I thought it sanitizes or something [22:44:25] No, it's for armoring [22:44:32] Which means protecting content from escaping [22:44:45] That's not armoring, that's handcuffing. [22:45:06] Don't be fooled by the handlebars helper formerly known as "escapeHTML", which (conditionally) did armoring, which is almost the opposite of escaping [22:46:47] Whoever chose to use these words for these terminologies need to read the tales of the knights of the round table [22:46:49] It's kind of like tainting in reverse (http://stackoverflow.com/questions/11520708/data-tainting-in-javascript). HtmlSnippet says, "I think this is safe", whereas taint was "I think this is unsafe, and that unsafeness should propagate". [22:46:58] armor is to prevent something (sharp) from getting in [22:47:07] not to prevent something from going out [22:47:17] You are confusing the foreigner. [22:47:28] Yeah apparently escaping is something sharp [22:47:40] Says the Israeli to the Dutchman [22:47:41] If something sharp escaped we're all doomed [22:47:51] Although saying that "escaping" (=going out) is something that "goes in" is .... well [22:47:57] matt_flaschen, that should be a song [22:48:27] RoanKattouw, well.. to be fair, it might be escaping from the outside... in... [22:48:45] Whatever this "it" is must be introverted [22:49:02] itroverted? [22:50:19] I was going to make the joke that the opposite of 'escape' is 'imcape' but then I remembered that my sample size ('implode/explode') is 'ex' not 'esc' [22:50:21] 10Flow, 6Collaboration-Team-Backlog: Better communicate that resolved topics can be expanded - https://phabricator.wikimedia.org/T103866#1569472 (10DannyH) p:5Normal>3High [22:50:36] Though those seem to be interchangeable in today's spoken English [22:55:28] mooeypoo, have you ever seen the new topic widget just hang in the 'pressed in' state? Working on the CAPTCHA fix. [22:55:41] matt_flaschen: just to double check - rev_content_model - is not important to populate it? [22:55:55] etonkovidova, it is, but the latest revision should be enough. [22:56:14] etonkovidova, you can join from page to revision on rev_latest to see what the latest revision is. [22:56:29] etonkovidova, oh, sorry, page_latest (latest revision ID for this page) [22:57:07] matt_flaschen: thanks! [22:59:08] mooeypoo: https://gerrit.wikimedia.org/r/233630 [23:02:27] mooeypoo, never mind, pressed-in thing isn't happening anymore. [23:02:50] matt_flaschen, I never seen this happen [23:06:38] 10Flow, 6Collaboration-Team-Backlog, 7Accessibility: On small screens, put legal text on the bottom of the page - https://phabricator.wikimedia.org/T110116#1569504 (10Josve05a) [23:06:42] (03PS29) 10Mooeypoo: Split alerts and messages in Echo [extensions/Echo] - 10https://gerrit.wikimedia.org/r/231200 (https://phabricator.wikimedia.org/T108190) [23:12:02] (03CR) 10Legoktm: Split alerts and messages in Echo (032 comments) [extensions/Echo] - 10https://gerrit.wikimedia.org/r/231200 (https://phabricator.wikimedia.org/T108190) (owner: 10Mooeypoo) [23:13:23] RoanKattouw_away, the CAPTHCA also does not work even if formatted, so working on that now. [23:44:24] (03PS1) 10Legoktm: Use db-replicated objectcache for storing last seen time [extensions/Echo] - 10https://gerrit.wikimedia.org/r/233639 [23:44:27] (03PS1) 10Legoktm: Use mw.Api#postWithToken for echomarkseen requests [extensions/Echo] - 10https://gerrit.wikimedia.org/r/233640 [23:44:56] mooey|away: https://gerrit.wikimedia.org/r/233639 will fix it, I made it on top of yours though [23:48:55] Deployed the config change for wgFlowOccupyPages; testing now. [23:49:08] 10Flow, 3Collaboration-Team-Current, 5Patch-For-Review, 5WMF-deploy-2015-08-25_(1.26wmf20), 7WorkTypeMaintenance: F3. End the occupation - https://phabricator.wikimedia.org/T105574#1569657 (10Mattflaschen) [23:50:27] 10Flow, 3Collaboration-Team-Current, 7WorkTypeMaintenance: Run FlowUpdateRevContentModelFromOccupyPages.php in production - https://phabricator.wikimedia.org/T109671#1569659 (10Etonkovidova) Not sure if it's relevant: There is no results - http://quarry.wmflabs.org/query/4877 ``` use enwiki_p; SELECT page... [23:56:45] (03CR) 10Legoktm: "This now makes 6 memcached requests per view:" [extensions/Echo] - 10https://gerrit.wikimedia.org/r/231200 (https://phabricator.wikimedia.org/T108190) (owner: 10Mooeypoo)