[03:45:54] <Krinkle>	 TimStarling: https://github.com/wikimedia/mediawiki/commit/2e33fb3ff667d1b774fd3551b6dc4380665adb49
[03:46:34] <Krinkle>	 Current version was restored by https://phabricator.wikimedia.org/rSVN89223 though
[03:49:33] <TimStarling>	 right
[03:50:01] <TimStarling>	 really a criminal act is the fault of everyone who stood by and said nothing, right?
[03:50:33] <TimStarling>	 speaking of which, CloneDatabase...
[03:51:05] <TimStarling>	 you pass it a Database object, and it clones it
[03:51:35] <TimStarling>	 and then changes the table prefix EVERYWHERE, in configuration, in all connections in known LoadBalancers, etc.
[03:52:15] <TimStarling>	 it doesn't actually change the prefix in the connection object you passed it, unless that object happens to be registered with LBFactory
[03:52:37] <TimStarling>	 whyyy
[03:55:00] <ostriches>	 Ok, CloneDatabase is totally my fault
[03:55:05] <ostriches>	 I was just looking at that crap again
[03:55:08] <ostriches>	 Man I was an idiot.
[04:04:57] <TimStarling>	 I'm trying to factor some setup code out of MediaWikiTestCase so that it can be shared with parserTests.php instead of duplicated
[04:05:43] <TimStarling>	 it seems like setupAllTestDBs() could go, maybe addCoreDBData()
[04:08:20] <Krinkle>	 Once separated we can also write unit tests for them. :)
[04:09:09] <TimStarling>	 I think I can probably fix CloneDatabase, it only has one external caller (an SMW unit test) and that's assuming that it is well-behaved, so making it well behaved should not be harmful
[04:09:42] <TimStarling>	 inside core, it is only used by ParserTest and MediaWikiTestCase
[04:12:08] <TimStarling>	 I don't think it would be helpful to make it so that all known foreign wikis have cloned tables set up in them
[04:14:54] <ostriches>	 Ideally, what we should do is have CloneDatabase use $wgDBadminuser/password to actually setup a second DB so we don't have to play with prefixes.
[04:17:00] <TimStarling>	 e.g. run it on WMF and it would clone about 100000 tables
[04:17:00] <TimStarling>	 so maybe better to throw an exception on foreign wiki access
[04:19:06] <ostriches>	 That too, yes.
[17:44:48] <AaronSchulz>	 anomie: thanks to Kunal merging the DB write duration tweaks, the rollbacks due to user row INSERT taking 6 sec should be gone next train
[18:20:09] <tgr>	 robla: the archcom email has the correct subject but last week's body
[18:20:54] * robla goes to look
[18:21:53] <robla>	 hmm, the copy in my sent box looks correct.  checking archive....
[18:23:35] <robla>	 tgr: is this the message you're seeing? https://lists.wikimedia.org/pipermail/wikitech-l/2016-August/086380.html
[18:24:59] <robla>	 oooooh, I see the mistake I made.
[18:25:38] <tgr>	 uh, I onlye read the snippet at the beginning :)
[18:25:54] <tgr>	 my bad
[18:26:10] <robla>	 the only problem is with this bit of text: " [T589: schema change for page content language][1]"  I put the wrong title
[18:26:11] <stashbot>	 T589: RfC: image and oldimage tables - https://phabricator.wikimedia.org/T589
[19:45:30] <anomie>	 bd808: I have one last comment on <https://gerrit.wikimedia.org/r/#/c/270553/>, otherwise it looks good to me.
[19:53:27] <bd808>	 anomie: I suppose we could, yeah
[19:53:41] * bd808 will amend after lunch is eaten
[21:01:50] <TimStarling>	 RFC meeting starting now in #wikimedia-office: image and oldimage tables
[21:18:09] <bd808>	 thanks for working through that patch with me anomie :)
[21:18:33] <anomie>	 you're welcome
[23:45:17] <tgr>	 anomie: is the text of action=help exposed somewhere in ApiSandbox?
[23:48:16] <tgr>	 well, obviously some of it is
[23:48:57] <tgr>	 what I mean is that everything after examples section (permissions, data types) seems to be lost