[01:11:39] <legoktm>	 tgr|away: nope, if you make it non-Wikimedia specific, I'd love that
[02:07:58] <tgr>	 legoktm: cool, thanks
[02:08:33] <tgr>	 is there a reason the deployment of that extension has stalled, or just lack of time?
[03:09:19] <legoktm>	 tgr: issues with deploying to beta (there is no pageview API in beta, nor does it exist for things like testwiki), and me not having time to push it
[03:11:31] <Krenair>	 what kind of stuff has to be set up to get a pageviews api?
[03:12:08] <bd808>	 a cassandra db that is populated somehow
[03:12:59] <bd808>	 seems like we could make a fake one
[03:13:40] <bd808>	 the prod pipeline is fed from hadoop jobs against the webrequest data gathered from varnish
[03:14:56] <Krenair>	 ah
[03:15:06] <Krenair>	 And I imagine hadoop takes some resources to set up
[03:16:00] <bd808>	 well... it should be possible actually to build out a functional system jsut for deployment-prep.
[03:16:15] <bd808>	 not sure how much disk it would need
[03:16:49] <bd808>	 probably not a huge amount though, especially if we cranked the retention down
[03:16:51] <legoktm>	 Last time I asked (half a year ago) analytics said they tried to create one in the past but weren't able to maintain it basically
[03:17:08] <bd808>	 but the analytics code isn't well factored for multiple deployments
[03:19:09] <bd808>	 we jsut need some analytics-minded volunteers to pitch in, but the env, and help refactor everything until it works :)
[03:19:24] <bd808>	 *build the env
[03:19:38] <legoktm>	 Sounds like we found one volunteer!
[03:19:47] <legoktm>	 ;)
[03:19:58] <bd808>	 crap! Did I do it again?
[03:20:22] <bd808>	 I've got a bit of a full plate for the near future
[03:20:46] <bd808>	 it would be cool to have hadoop in tool labs... but that's another story entirely
[03:22:14] <bd808>	 We could replace a *lot* of tools with a nice datamart built from the mediawiki databases and hadoop is pretty helpful in doing ETL work that needs a lot of number crunching
[03:24:31] <legoktm>	 Same :( I filed another extension deployment bug so I marked that one as stalled to not have 3 open simultaneously
[03:24:43] <legoktm>	 ETL?
[03:25:21] <bd808>	 Extract, Transform, Load -- database manipulation jargon
[03:25:32] <bd808>	 or data generally
[03:25:37] <legoktm>	 Ah
[03:25:57] <bd808>	 gather up some stuff, change it somehow, put it in a db
[03:29:12] <bd808>	 legoktm: you need to stop being so productive ;)
[03:31:49] <legoktm>	 :P
[13:13:44] <anomie>	 tgr|away, legoktm, bd808: PageViewInfo could take the same tack as ORES: ORES has a "testing" wiki that scores any revision as a transform of the rev_id (e.g. rev_id 123456 is scored 0.65), which was good enough for me to be able to do the development needed for T143895.
[13:13:44] <stashbot>	 T143895: [Epic] Implement ORES service proxy in api.php - https://phabricator.wikimedia.org/T143895
[16:11:49] <bd808>	 Reedy: does this patch look ok? https://gerrit.wikimedia.org/r/#/c/316981/ -- I'll put it up for swat if I didn't do something dumb
[16:13:52] <Reedy>	 bd808: I already had a patch for that :)
[16:14:19] <Reedy>	 Just defining $wgMWOAuthCentralWiki anyway
[16:15:01] <bd808>	 ah. that one would work too. Do you think its less confusing?
[16:15:25] <bd808>	 I was just running a script on silver and noticed the error
[16:15:47] <Reedy>	 		if ( $wgMWOAuthCentralWiki === false ) {
[16:15:48] <Reedy>	 			// Treat each wiki as its own "central wiki" as there is no actual one
[16:15:48] <Reedy>	 			$wgMWOAuthCentralWiki = wfWikiId(); // de
[16:15:58] <Reedy>	 So setting it to false is indifferent
[16:16:34] <Reedy>	 bd808: WFM. More obvious than my patch, I think