[05:08:36] Krinkle: https://gerrit.wikimedia.org/r/c/mediawiki/core/+/681240 more cleanups [19:38:54] Krinkle: I amended https://gerrit.wikimedia.org/r/c/mediawiki/core/+/681240 to fix CI [19:45:36] AaronSchulz: ack, taking a look. [19:48:46] I'd probably have opted for null in this case, given we're changing schema anyway. I know that adds to the query, but seems worthwhile to keep it simpler overall than to have to deal with this special thing. [19:49:41] or if we pull the plug and lead the way on dropping native datetime on non-mysql for this schema we can just store a special string like "indefinite" ,str_repeat('z', 14) or str_repeat('9', 14) [19:49:56] consistently everywhere and use that as simple string value [19:50:11] anyway, don't want to increase scope further... but we are making a schema change. [19:50:19] I don't know where the consensus landed on the mwtimestamp thing [19:54:32] Amir1: I can't seem to find any open or closed task about PostgreSQL vs mwtimestamp. checked T191231 , https://gerrit.wikimedia.org/r/c/mediawiki/core/+/634679 , T252313 and T164898 [19:54:32] T252313: Document mapping of SQL data types across DBMS - https://phabricator.wikimedia.org/T252313 [19:54:33] T164898: PostgreSQL schema change for consistency with MySQL - https://phabricator.wikimedia.org/T164898 [19:54:33] T191231: RFC: Abstract schemas and schema changes - https://phabricator.wikimedia.org/T191231 [19:54:49] Doesn't seem to be any mention anywhere of the sticking with native datetime on postgres. [19:55:00] I thoguht maybe there was opposition, but I don't see that either [20:16:35] Krinkle: I though about it...I might change the patch file but the code still has to work for the old schema [20:41:54] dpifke: CW/TW on -staff for you [20:42:11] as in, warning, not "go look" [20:42:51] Ha. :)