[00:00:05] <James_F>	 bd808: I commented on https://gerrit.wikimedia.org/r/#/c/314490/ that maybe that caused it?
[00:01:00] <James_F>	 bd808: But https://gerrit.wikimedia.org/r/#/c/307073/ merged, so race condition sounds plausible.
[00:09:13] <bd808>	 the failing test grabs a user, changes some settings, saves, and the tries to verify the save settings on a new user instance. That looks more like a race than something realted to tim touching parser tests.
[00:09:35] <bd808>	 I'll look at the guts of saveSettings and see if it has a differed or something
[00:10:57] <bd808>	 it looks like User::saveSettings tries pretty hard to be immediate and atomic
[00:11:36] <bd808>	 It actually should throw an exception if the db write fails
[00:12:08] * James_F nods.
[00:15:00] <bd808>	 the failure only happening on hhvm and not php5 seems suspicious though
[00:15:53] <bd808>	 ah ha! User::clearSharedCache uses onTransactionPreCommitOrIdle()
[01:10:39] <jackmcbarn>	 Glaisher: ping
[22:02:09] <TimStarling>	 RFC triage meeting starting now on #wikimedia-office