[20:00:03] TimStarling: any chance you could look at https://gerrit.wikimedia.org/r/#/c/308310/ ? It's on old patch for sanity-purging important WAN cache entries. [20:04:16] Fatal error: request has exceeded memory limit in /srv/mediawiki/php-1.29.0-wmf.7/includes/profiler/SectionProfiler.php on line 219 [20:04:18] Heh, fun [20:04:30] OOM hits while profiling something that's probably pretty heavy in memory [20:51:34] Warning: Parameter 1 to Wikibase\RepoHooks::onShowSearchHitTitle() expected to be a reference, value given in /srv/mediawiki/php-1.29.0-wmf.8/includes/Hooks.php on line 186 [20:51:42] Yay, people changing hook parameters before updating all extensions [20:51:44] :D [20:58:13] aude: I have a patch up for this ^ cuz it's spammy https://gerrit.wikimedia.org/r/#/c/332819/ [20:58:29] But I'm not clear on the build process right now to get that deployed. Can you help/advise? [21:06:00] addshore maybe? ^ [21:06:46] *waves* *reads* [21:07:38] hipsterpanda: whats version of wikibase / wikidata is currently deployed? [21:08:09] wmf/1.29.0-wmf.8 [21:09:05] 1.29.0-wmf.8 changed the hook signature, but didn't update all extensions. So far this is the second I've spotted [21:12:06] Cool, I'll try and push the patch through the stages now :) [21:15:27] tyvm [21:15:42] Noticed on group1 today, pretty spammy, best to squash before group2 (even though it's technically harmless) [21:23:30] hipsterpanda addshore i filled https://phabricator.wikimedia.org/T155668 [21:23:35] ack! [21:23:38] I saw in -ops ;) [21:23:43] please update there accordingly :) [21:23:46] and thanks [21:46:06] ostriches: there's a bug open to revert the hook signature changes [21:46:16] I think it should stay [21:46:23] Just fix the broken extensions [21:46:37] They're impossible to fix in a non-breaking way. Going back means reverting the fixes to extensions [21:47:17] you know, we have an actual policy on how to make breaking changes now :) [21:47:29] there was no need to remove the & in the first place anyways [21:48:50] https://phabricator.wikimedia.org/T155625 [21:49:25] * ostriches shrugs [21:49:31] Just as hard to move backwards as forward [21:49:36] * ostriches doesn't care a ton, though [21:49:44] [21:49:38] (CR) Reedy: "Changing the docs broke your extension?" [core] - https://gerrit.wikimedia.org/r/332667 (owner: Smalyshev) [21:50:28] * anomie sees T155625 is filed by mwjames with his usual bad attitude and immediately doesn't care [21:50:42] "This breaks our extension [0], and I'm really getting tiered of this." [21:50:47] Apparently he's turning into a cake [21:52:39] MediaWiki would be so much nicer if hooks didn't exist [21:52:42] Fucking hooks [22:00:21] legoktm: Actually, that's a different hook from what my bug is on [22:01:08] I'm talking about ShowSearchHitTitle, which was broken back in Nov (If78cb0c29ae394f16e465c15a8e8246c1b56dcea) [22:01:37] ostriches: There's a reason we're refusing to provide hooks in VE. Learning from MW's mistakes. [22:01:37] o.O what [22:01:48] The bug mwjames filed and is complaining about is SpecialSearchResults [22:02:13] ostriches: If78cb0c29ae394f16e465c15a8e8246c1b56dcea only got merged in the latest branch [22:02:19] the first patchset is from Nov [22:02:26] Ah, misread [22:02:29] So yeah, that just broke [22:03:34] I think they changed multiple hook definitions in this branch in refactoring (hard to blame, the hook calls were moved around) [22:03:40] Yeah [22:06:15] legoktm: https://gerrit.wikimedia.org/r/#/c/332878/ [22:06:26] addshore: I'll just abandon the patch to Wikibase, sorry [22:06:53] okay! :D [22:09:58] I'm trying to find what caused T155625 [22:09:59] T155625: Change to SpecialSearchResults hook breaks extension - https://phabricator.wikimedia.org/T155625 [22:10:01] It's not quite the same [22:16:15] There's some confusion here [22:16:31] SpecialSearchResults hasn't passed those by reference since...at least 2014 or earlier [22:16:57] Or not [22:17:02] Git blame digging is hard [22:20:25] Ib527fc3a3c39eb2e56985e5d1e4905fc4562353c broke the other hook [22:28:29] legoktm: https://gerrit.wikimedia.org/r/#/c/332884/ is the second hook fix [22:28:42] Once that + the other both land in master, we can backport to wmf.8 [22:28:52] And probably have to revert my "fix" in LQT [22:31:33] ostriches: thank you for cleaning htis up [22:31:34] this* [22:32:08] ostriches: I reverted the LQT change [22:32:12] Ok sweet [22:36:14] Creating backports to wmf.8 for all 3 [22:38:39] Actually, just the two affecting prod [22:38:44] https://gerrit.wikimedia.org/r/#/c/322013/ was master-only, hadn't landed in wmf.8 [22:44:58] https://phabricator.wikimedia.org/T155682 [22:47:00] one thing at a time... [22:47:32] I'm looking into it :) [22:50:29] And with that, the hook warnings disappear [22:50:32] That was fun [22:50:39] Let's do it again real soon [22:51:04] [14:50:33] (PS1) Legoktm: Revert "Use the parser to get each template parameter in the Index: namespace" [extensions/ProofreadPage] - https://gerrit.wikimedia.org/r/332895 (https://phabricator.wikimedia.org/T155682) [22:53:53] ostriches: thanks, will you take care of deploying that? I have to go offline soon-ish [22:54:08] I can yeah [22:54:51] thanks both [23:18:50] AaronSchulz: is there an easy way to re-run jobs that are in the "abandoned" state? [23:19:00] context is T153618 and wikitech jobs [23:19:01] T153618: LinksUpdate::acquirePageLock error with SMW enabled - https://phabricator.wikimedia.org/T153618 [23:22:31] bd808: I guess << foreach ( JobQueueGroup::singleton()->get( 'refreshLinksPrioritized' )->getAllAbandonedJobs() as $job ) { JobQueueGroup::singleton()->push( $job ); } >> [23:23:06] * AaronSchulz volunteers bd808 to make that a script :D [23:23:13] heh [23:23:33] looks like an eval.php script to me :) [23:26:55] I'll just make one [23:32:44] refreshLinksPrioritized is just backlinks correct? Or does it have a hook buried in it that other things may piggyback on? [23:53:55] just backlinks