[02:49:40] <bd808>	 ori: what's the magic incantation needed to prime the keyholder agent? I need to feed a key into the beta cluster version
[02:51:38] <bd808>	 nm I found it
[02:52:02] <bd808>	 now I just need to find the passphrase in puppet/private somewhere
[04:56:02] <legoktm>	 TimStarling: did you have a chance to look into the extension registration missing skins issue?
[04:57:03] <TimStarling>	 sorry, I haven't
[04:58:35] <TimStarling>	 and I still have to pack for Lyon
[04:58:53] <legoktm>	 do you have any suggestions on where I should start looking? I'm stumped right now
[05:04:09] <TimStarling>	 was there a ticket for this?
[05:06:02] <legoktm>	 https://phabricator.wikimedia.org/T98956 is the nostalgia one, I didn't get around to re-opening it for cologne blue, doing that now
[05:10:12] <TimStarling>	 so it's not just one HHVM server? it is all of them?
[05:10:18] <legoktm>	 yes, all of them
[05:10:26] <legoktm>	 and even CLI like terbium
[05:13:38] <legoktm>	 I took a dump of mw1111's APC, and the key for mw.o (mediawikiwiki:registration:d6ffe62f2ea2244e61c1920b4e68cafc) has no mention of CologneBlue
[05:18:27] <TimStarling>	 and that hash includes all the extension names, right? so CologneBlue must have been in $this->queued
[05:21:03] <legoktm>	 		$key = wfMemcKey( 'registration', md5( json_encode( $this->queued ) ) );
[05:21:06] <legoktm>	 yes
[05:23:17] <legoktm>	 		$this->queued = array_unique( $this->queued );
[05:23:26] <legoktm>	 I think that line is wrong.
[05:23:42] <legoktm>	 $this->queued is array( filepath => timestamp )
[05:23:47] <legoktm>	 and array_unique operates on values
[05:23:57] <legoktm>	 so if two files have the same timestamp...
[05:27:00] <TimStarling>	 ahh
[05:27:59] <legoktm>	 >>> os.path.getmtime('Modern/skin.json')
[05:27:59] <legoktm>	 1431543930.0
[05:27:59] <legoktm>	 >>> os.path.getmtime('CologneBlue/skin.json')
[05:27:59] <legoktm>	 1431543930.0
[05:28:02] <legoktm>	 yep!
[05:28:27] <TimStarling>	 obviously it is not necessary to check for duplicate keys, PHP arrays can't contain duplicate keys
[05:28:54] <TimStarling>	 that's what the array_unique is trying to do, right?
[05:29:18] <legoktm>	 I guess....I have no idea what I was thinking at that time
[05:31:13] <legoktm>	 TimStarling: https://gerrit.wikimedia.org/r/211946 gets rid of that line
[05:32:06] <legoktm>	 thanks for talking this through with me :)
[05:32:37] <TimStarling>	 no problem, sorry I wasn't on top of this earlier
[05:32:58] <TimStarling>	 I gave the change +2
[05:33:47] <legoktm>	 I'll get it deployed during SWAT tomorrow morning
[14:20:50] <^d>	 ori: Thanks (belated reply, just caught past 3 days of IRC scrollback) :)
[15:46:35] <legoktm>	 anomie: if a user on wiki A is logged out, and logged in on wiki B (same farm), and A is on B's CORS whitelist thing, will javascript on A making CORS requests to B be as a logged in user?
[15:46:57] <legoktm>	 ohai csteipp
[15:47:24] <anomie>	 legoktm: I think so. It depends on the browser sending the cookies.
[15:48:56] <legoktm>	 anomie: ok, I was asking in context of https://gerrit.wikimedia.org/r/#/c/211884/
[15:53:03] <anomie>	 legoktm: Yeah, I'm pretty sure you're correct there.
[15:57:01] <csteipp>	 hey legoktm 
[16:09:32] <legoktm>	 psr-7 was accepted
[16:13:09] <bd808>	 aka "the worst way to model HTTP ever"
[16:13:49] <bd808>	 the immutable data zealots won that round
[16:23:55] <bd808>	 do we have a maintenance script that can be used to configure an email address for a user?
[16:24:04] * bd808 is not finding one
[16:24:47] <legoktm>	 bd808: eval.php
[16:24:53] <bd808>	 heh
[16:26:15] <bd808>	 time for the next episode of "fun with puppet hacks" I guess
[16:29:50] <csteipp>	 I hacked the set password script to also set email on my local wiki... I could upload that.
[16:30:29] <csteipp>	 Very much needed when setting up a private wiki for another organization...
[16:31:00] <bd808>	 I want to make it possible to use mediawiki::user in mw-vagrant to create a new user with an email address set; for testing ContactPage
[16:33:51] <Reedy>	 Slightly surprised that no one has committed a script to do it before now
[16:58:40] <anomie>	 Who wants to review my patches to enable ApiFeatureUsage to make sure I didn't screw anything up (and that the docs aren't too out of date)? https://gerrit.wikimedia.org/r/#/c/212012/ https://gerrit.wikimedia.org/r/#/c/212013/ https://gerrit.wikimedia.org/r/#/c/212014/ https://gerrit.wikimedia.org/r/#/c/212015/
[17:01:06] <legoktm>	 anomie: I think your config patch is going to end up turning it off on beta labs
[17:01:49] <anomie>	 legoktm: How so?
[17:03:33] <legoktm>	 anomie: nope, never mind, I forgot to look at InitialiseSettings-labs.php
[17:25:35] <anomie>	 ^d: Care to check I didn't screw anything up in these patches to enable ApiFeatureUsage? https://gerrit.wikimedia.org/r/#/c/212012/ https://gerrit.wikimedia.org/r/#/c/212013/ https://gerrit.wikimedia.org/r/#/c/212014/ https://gerrit.wikimedia.org/r/#/c/212015/
[17:27:40] <bd808>	 anomie: those changes look right to me
[17:43:38] <legoktm>	 yay for crazy race conditions :( https://phabricator.wikimedia.org/T99688
[17:48:31] <anomie>	 legoktm, bd808: https://gerrit.wikimedia.org/r/#/c/207979/ looks good. But it looks like ExtensionRegistry is screwing up, since the config variable is set it doesn't populate the rest of the default fields :(
[17:49:00] <legoktm>	 uh, is it not doing an array_merge?
[17:49:23] <anomie>	 Apparently not.
[17:49:37] * anomie does `vagrant git-update` in case that's it.
[17:51:07] <legoktm>	 oh, extension registry doesn't support 2d or greater arrays for config settings yet
[17:51:16] <legoktm>	 https://phabricator.wikimedia.org/T99257