[04:37:25] <Krinkle>	 TimStarling: https://gerrit.wikimedia.org/r/#/c/420224/ - also, how do you feel about removing $wgValidateAllHtml ?
[04:51:04] <anomie>	 Krinkle: FYI, since you've been interested in the past, I poked at failing PostgreSQL tests this evening. But most are addshore's RevisionStore and WikiPage tests that are assuming things exist with certain IDs, that are also failing in T183777.
[04:51:04] <anomie>	 I wonder if it'd be worthwhile to make PG's duplicateTableStructure() detect columns using sequences and create temporary sequences to match.
[04:51:04] <stashbot>	 T183777: mediawiki tests fail in mediawiki code coverage test - https://phabricator.wikimedia.org/T183777
[04:51:24] <Krinkle>	 Was just looking at the RevisionStore fix actually.
[04:54:16] <Krinkle>	 anomie: Hm.. not sure I understand, but how are RevisionStore/WikiPage different from other database-backend tests we have? I don't know much about PG specifically, but afaik our temp table copies should be able to copy and insert rows the same as normal, right? And we have things like insertId() as well, which presumably work. In what way is it failing for PG only?
[04:54:43] <Krinkle>	 But yeah, I'd be willing to review a fix for the issue one way or another :)
[15:51:37] <anomie>	 Krinkle: The tests are just blindly assuming there's a page with page_id = 1, a revision row with rev_id = 2, a text row with text_id = 2, and stuff like that. I'm not completely sure why it all works in MySQL, but at least some of it might be fixed by making new sequences for the temp tables.
[19:28:29] <Krinkle>	 anomie: hmm so currently the auto inc continues where the master table happens to be?
[19:28:37] <Krinkle>	 In the temp copy