[01:04:28] legoktm: I'm sure you're busy, but I think this patch will fix a stuck global rename -- https://gerrit.wikimedia.org/r/#/c/414117/ [01:23:50] bd808: are there any other special characters we might need to escape in the future? [01:27:36] legoktm: the only other thing special to that function is \n things. I don't think a backslash is a legal username char is it? [01:28:04] \n and $n are both treated as back references in the replacement string [01:29:25] I think there's a backslash in https://www.mediawiki.org/wiki/Manual:$wgLegalTitleChars [01:30:53] bd808: https://test.wikipedia.org/wiki/Special:CentralAuth/VeryU%5CniqueAccount [01:32:39] James_F: https://gerrit.wikimedia.org/r/#/c/414606/ should be an easy to review PHPUnit 6 patch if you have time :) [01:33:13] legoktm: I was just looking at your list of patches. :-) [01:34:28] also https://tools.wmflabs.org/ci/last_run.html should have about 90% of the table filled in for PHP 7.0 on Wikimedia deployed extensions, and there's only one new test failure from CirrusSearch [01:35:01] legoktm: Can you default that page to filter only Wikimedia-deployed? :-) [01:35:32] And intersection has no unit tests? Meh. [01:35:52] I spent about 30 seconds last night trying to get it to support query parameters in JS and quickly gave up [01:36:10] Ha. [01:36:45] I think I'll add a "has failures" filter since that's what really matters [01:36:51] * James_F nods. [01:37:12] Also tsk for using checkboxes when you mean radio buttons with a "neither" option. [01:37:52] I...totally forgot radio buttons existed until now [01:37:57] :-D [01:38:30] I even added in code to uncheck the other checkbox and sigh. [01:38:35] Yeah. [01:38:43] I was surprised and amused. [01:50:15] legoktm: yikes on the \. I'll amend the patch to account for that. [02:13:12] bd808: Or I can. ;-) [02:50:18] James_F: I think we can use https://phabricator.wikimedia.org/T144962 as the PHP 7 tracking task [03:06:29] anomie: I tagged 2.0.0 for utfnormal, but it's also going to need a new css-sanitizer release (patch submitted) [03:26:28] legoktm: WFM. [21:04:25] anomie: Might want to look at https://phabricator.wikimedia.org/T188437 [21:04:57] * anomie looks [21:05:13] Do we just neeed to create the empty actor tables everywhere? [21:06:07] And revision_actor_temp? [21:06:50] actor should be optional I thought? [21:07:33] Reedy: I missed testing $wgActorTableSchemaMigrationStage in one place. [21:07:46] aha [21:08:11] uhh, this stack trace is everything wrong with our hook system [21:08:43] CA, Wikidata and Echo in one package, fighting for a place under the sun [21:09:26] Reedy, legoktm: https://gerrit.wikimedia.org/r/415100 [21:10:09] thcipriani: ^^ [21:10:24] thank you! [21:10:31] I just C+2'd the core patch, cherry picked it and pinged him in another channel xD [21:10:33] cheers anomie [21:13:25] Thanks for the IRC ping, I would have seen it in my email eventually but not as quickly. [21:14:52] I'm guessing there'll be one or two more of those in some weird edge cases [21:18:32] I hope that was the only one. But if there's another, the question to answer is "why is this happening when $wgActorTableSchemaMigrationStage is MIGRATION_OLD?" [21:24:36] anomie: Reedy legoktm patch is live on wmf.23 now, FYI thanks all! [22:42:58] anomie: Is the plan for REL1_31 to ship with wgCommentTableSchemaMigrationStage set to MIGRATION_NEW? [22:43:46] (And if so, are there remaining blockers to that?) [22:49:38] James_F: I think he has some extension updates to do [22:49:52] Right.