[00:03:32] > 11) Wikibase\Client\Tests\Hooks\OtherProjectsSidebarGeneratorTest::testBuildProjectLinkSidebar_hook with data set "Invalid hook #3, original data is being used" (Closure$Wikibase\Client\Tests\Hooks\Other [00:03:39] TimStarling: Is that what you were looking into yesterday? [00:12:45] Reedy: +1 to what MaxSem said. The software is officially called "Notifications", not "Echo", so if you're making breaking changes to class names anyway it's sensible to fix that too whilst you're at it. [00:31:09] Krinkle: no, yesterday I was looking at https://gerrit.wikimedia.org/r/c/mediawiki/core/+/450433 , it turned out to be a ConfirmEdit test messing things up for Flow [00:32:06] TimStarling: Hm.. that link points to https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/ConfirmEdit/+/465429/ which in turn has the above Wikibase failure. [00:32:52] haven't looked further yet, was just hoping to unblock the travis CI failure due to the TitlePermissionsTest failure from https://phabricator.wikimedia.org/T206130 [00:33:03] I reverted the main change that enabled it, but this patch added another one. [00:33:19] I'll disable it for now as well. Too many patches and all failing.. It's been a week now. [00:34:21] I ended up disabling Wikibase in my testing yesterday, I couldn't get it to work and the ConfirmEdit/Flow bug didn't depend on it [00:34:33] I guess it has its own issues [00:42:09] As someone who's spent six hours trying to get Vagrant/Wikibase/WikibaseMediaInfo to work again locally, yes, they do. [02:21:22] TimStarling: can you have a look at https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/452878/ ? [02:24:59] what are the changes since I reviewed it in PS7? [04:18:43] never mind, I merged it [10:47:50] James_F: Going to do it in a dependant patch, at least for development [10:47:57] Can always squash them before merge [11:04:46] James_F: Also, if you've a better name for the namespace... [11:04:50] Considering echo is a reserved word etc [11:08:39] Hmmm, is there a way to get from a pcache entry to the request that created it ? [16:56:55] addshore: I don't think there is, generally. If the wiki has full debug logging enabled (maybe testwiki has? I'm not sure) then you could look for the log record of writing the cache key [19:06:43] TimStarling: Hm.. am I the only one in the meeting, or did I miss something / am in a different universe? [19:09:57] Oh darn, I'm an hour early. How... [19:32:23] oh is it time zone change time again? [19:33:09] hmm not just yet [19:38:03] soon! [19:50:19] 20:49:25 1) Flow\Tests\Api\ApiFlowLockTopicTest::testUnlockTopic [19:50:19] 20:49:25 Undefined index: umcnjaeblr2bsz70 [19:50:19] 20:49:25 [19:50:19] 20:49:25 /workspace/src/extensions/Flow/tests/phpunit/api/ApiFlowLockTopicTest.php:95 [19:50:19] 20:49:25 /workspace/src/tests/phpunit/MediaWikiTestCase.php:416 [19:50:20] 20:49:25 /workspace/src/maintenance/doMaintenance.php:94 [19:50:28] Why is that failing from a vendor patch to remove something completely unrelated... [19:53:28] hmm [19:53:37] that looks like Aryeh's mystery ConfirmEdit failures? [20:06:01] legoktm: congrats on the debian/ubuntu packages. Now the question is: do we want to build those using our CI? :) [20:06:40] we kinda do right now, but it's a bit broken [20:06:42] legoktm: see also: https://mastodon.social/@federicomena/100872798889050795 [20:07:39] hmm [20:08:22] normally the main problem is that there's not really a "one size fits all" debian package because of dependencies being different across major versions, or other packaging details [20:08:27] I tend to side with Ted more than federico in that discussion topic [20:08:47] so right now I have separate branches for unstable/stretch/xenial/bionic (though over time the differences are going down) [20:09:00] * greg-g nods [20:09:09] https://phabricator.wikimedia.org/T156543 is the most relevant ticket [20:10:56] * greg-g nods [20:13:10] Does anyone feel strongly about renaming extension maintenance files to match the casing of the class inside? [20:35:26] Reedy: we don't typically do that for files with side effects (ie. not just a class), I think? [20:35:46] Yeah, that's what I thought [20:36:04] Switching most of an extension to PSR-4 but having a few autoloaded classes isn't the end of the world [20:36:10] wasn't there an RFC recently to get rid of those kinds of files in favor of a script runner framework? [20:36:31] Not sure if it made it to RFC, but there was definitely discussions about something like that, yeah [20:36:34] autoloading classes which have side effects is scary [20:38:55] a month ago, apparently, task is T99268 [20:38:56] T99268: RfC: Create a proper command-line runner for MediaWiki maintenance tasks - https://phabricator.wikimedia.org/T99268 [20:55:56] brion, Krinkle: Daylight saving time started in Australia on Sunday. Europe ends DST on October 28, and the US ends it on November 4. [20:56:29] heh [20:57:25] Amir1: Is https://gerrit.wikimedia.org/r/461815 fine to backport today, or would you prefer it ride the train next week slowly? [20:59:01] Going on train is fine, I'm around to monitor [20:59:06] Timezones are the worst [20:59:08] Krinkle: ^ [21:00:18] Amir1: OK. If it's not a risk to apply everywhere at once, I can also roll it out sooner, which would be nice to get fixed, but either way is fine, no pressure :) [21:05:42] Krinkle: It should be fine. I will be back from dinner in one hour [21:20:38] https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/Echo/+/465562/20/includes/EchoHooks.php - line 588 [21:20:49] Anyone know why CI is complaining about an undefined class? The use statement is there [21:22:40] some loads aren't being auto enough? [21:22:55] what does class_exist() say? [21:23:20] Running class_exists in the middle of a jerkins job isn't easy [21:23:21] https://gerrit.wikimedia.org/r/#/c/mediawiki/extensions/Echo/+/465562/20/includes/mapper/NotificationMapper.php [21:24:34] oh, you totally can - just slap var_dump() somewhere and see the console output ;) [21:25:26] but my theory is that autoloader isn't configured properly [21:25:35] Well, it's easy} [21:25:38] But it's slow AF [21:26:24] "MediaWiki\\Notifications\\Mapper\\NotificationMapperTest": "tests/phpunit/mapper/NotificationMapperTest.php", [21:26:29] Wonder if I can just run that test [21:26:33] Find something else to do meanwhile 😭 [21:27:56] [a90ad4603835b114fc6ddc52] [no req] Error from line 588 of /var/www/wiki/mediawiki/extensions/Echo/includes/EchoHooks.php: Class 'MediaWiki\Notifications\Mapper\NotificationMapper' not found [21:28:06] php tests/phpunit/phpunit.php extensions/Echo/tests/phpunit/mapper/ [21:28:12] Ok, so easy to reproduce locally [21:28:39] MaxSem: Folder casing :( [21:28:48] ofc it was [21:29:05] wasn't there a test that would catch this? [21:29:17] In core? Maybe [21:29:22] or it's not checking PSR-4 directories? [21:29:23] But I think it's failing before that [21:29:27] Or maybe that [21:31:15] namespace EchoOOUI; [21:31:17] * Reedy wonders why [21:31:49] Universal answer: because it seemed like a good idea at the time [21:31:52] Is that some sort of weird naming scheme? [21:33:21] James_F: I'm presuming I can just move that to MediaWiki\Notifications\OOUI or similar? [21:33:54] Reedy: Sure? [21:34:07] Just checking there's no weird OOUI thing that it has to be in that namespace [21:34:27] I presume it dates from the UI migration to OOUI and was never un-namespaced afterwwards. [21:34:50] Reedy: But if you break it, it's RoanKattouw's problem, not mine. ;-) [21:46:29] Is https://www.mediawiki.org/w/index.php?title=Compatibility&diff=prev&oldid=2917108&diffmode=source true? I thought we were still working on it. [21:47:37] Hmm [21:47:52] I would presume yes 7.2 should work, but it's not guaranteed [21:48:20] The table says that [21:48:37] (same as the text) [22:01:37] James_F: 7.2 is roughly the same as 7.1, it works but there are warnings and some edge case behavior might be broken [22:02:07] Right. It just felt a little bold. [22:06:12] Krinkle: is it deployed? [22:06:34] MaxSem: AutoLoaderStructureTest::testPSR4Completeness [22:07:02] Amir1: It's been in beta for a while, I'm asking if you are comfortable with the change going to production earlier as backport (all wikis at once, today; instead of per-group next week). Only if it's okay with you. I can backport it for you. [22:07:26] Yeah it's fine for me [22:07:39] OK. Does now sound good to monitor impact? [22:08:21] yup [22:09:31] k [22:09:38] I'll ping in ops when it's ready. [22:11:09] cool [22:31:55] Reedy: https://gerrit.wikimedia.org/r/465785 and stack is because I love you. ;-) [22:32:21] Wait for jerkins to catch up ;P [22:32:29] * James_F grins. [22:42:18] I think you broke it James_F [22:42:34] https://integration.wikimedia.org/zuul/ loads of jobs, doing nothing [22:48:11] Reedy: Many, many times. [22:48:59] K.rinkle's (not-)SWAT merge has stolen all the executors (as it should), by the looks of it. [22:50:00] starting to go again [22:57:20] TimStarling: can you recheck https://gerrit.wikimedia.org/r/#/c/mediawiki/core/+/452878/ please? [23:58:06] -'MobileFrontend\ContentProviders\ntProviderFactory' [23:58:06] +'MobileFrontend\ContentProviders\ContentProviderFactory' [23:58:13] Structure tests look to be condensing something...