[00:02:06] * @throws DBUnexpectedError [00:02:06] * @throws Exception [00:02:06] * @throws FatalError [00:02:06] * @throws MWException [00:02:12] * Reedy blinks [19:02:56] Nikerabbit: Hello. I'm a GCI student. I'm working on this TODO: https://doc.wikimedia.org/mediawiki-core/master/php/ExifBitmap_8php_source.html#l00224 . What am i supposed to do? [19:03:31] https://doc.wikimedia.org/mediawiki-core/master/php/classExifBitmapHandler.html#a1ecd3c14e00b9dfc119b98b727f77a2d [19:47:02] He found N by git blame or what? [20:27:52] AaronSchulz: Question: In beta, https://doc.wikimedia.org/mediawiki-core/master/php/classTitle.html#aef45388a9f9b788c7c5d30104097cc63 is returning false (no next revision) when I'd expect there to be one (it returns the revision ID in my local environment), is this because I'm querying a slave? Is this a race? [20:37:26] urandom: where are you calling it? [20:37:50] AaronSchulz: https://github.com/wikimedia/mediawiki-extensions-EventBus/blob/master/EventBus.hooks.php#L178 [20:38:03] TL;DR from a TitleMoveComplete hook [20:38:29] in my local testing, that produces the revision that represents the move itself [20:41:48] urandom: are you on vagrant? [20:41:54] nope [20:42:45] i do have vagrant setup, but i haven't been use it for this [20:42:51] s/use/using/ [20:46:19] urandom: in prod, you'd need to use the GAID_FOR_UPDATE flag since the slave (a) would not have seen something committed a few ms ago and (b) even if it did, it would probably be in same REPEATABLE-READ snapshot (the slave read "transaction" goes from first query to request end normally) [20:47:42] would be nice if the hook just have the revision [20:48:01] it's cleaner to do that than read back and re-discover what the request just wrote (if possible) [20:48:55] AaronSchulz: ok, that makes sense [22:39:35] AaronSchulz: comments welcome: https://gerrit.wikimedia.org/r/#/c/259604/ [22:39:48] something tells me that's too naive