[00:17:26] <legoktm>	 model versions are strings right? http://ores.wmflabs.org/scores/enwiki/
[00:24:32] <YuviPanda>	 legoktm: try sometihng?
[00:24:37] <YuviPanda>	 legoktm: for grrrit-wm
[00:25:47] <legoktm>	 YuviPanda: I just +2'd a patch
[00:25:54] <YuviPanda>	 legoktm: uh oh
[00:25:57] <YuviPanda>	 didn't show up here
[00:26:08] <YuviPanda>	 info: Sent message from extensions/ORES to #wikimedia-dev
[00:26:10] <YuviPanda>	 and to -feed
[00:26:12] <YuviPanda>	 sigh
[00:27:04] <legoktm>	 :/
[00:48:01] <YuviPanda>	 legoktm: try now
[00:48:16] <grrrit-wm>	 (CR) Legoktm: "test" [extensions/ORES] - https://gerrit.wikimedia.org/r/247034 (https://phabricator.wikimedia.org/T112856) (owner: Awight)
[00:48:23] <legoktm>	 yaay
[00:48:26] <legoktm>	 oh god
[00:48:28] <legoktm>	 colors
[00:48:30] <legoktm>	 we need colors
[00:48:31] <YuviPanda>	 haha
[00:48:36] <YuviPanda>	 hmm
[00:48:40] <YuviPanda>	 it's also gone back to its old name
[00:49:12] <legoktm>	 halfak: when you're back, /msg chanserv set #wikimedia-ai mlock -c
[00:49:18] <legoktm>	 please :)
[06:32:46] <grrrit-wm>	 (CR) Paladox: "@Legoktm thanks." [extensions/ORES] - https://gerrit.wikimedia.org/r/247022 (https://phabricator.wikimedia.org/T112856) (owner: Awight)
[12:54:28] <halfak>	 legoktm, {{done}}
[15:04:00] <legoktm>	 thank you :D
[15:48:18] <legoktm>	 YuviPanda: will url routing for ores in prod be the same as it is in labs?
[17:10:09] <halfak>	 legoktm, that's a good question
[17:10:30] <halfak>	 YuviPanda, reported that they'll want us to drop <wiki> from the URL pattern. 
[17:10:39] <halfak>	 And use the domain to define the context of scoring. 
[17:11:04] <halfak>	 That's a bummer since it will substantially change the URL pattern that a centralized ORES server understands. 
[17:11:22] <halfak>	 e.g. en.wikipedia.org/ores/damaging/342342323
[17:11:47] <halfak>	 vs. services.wikimedia.org/ores/enwiki/damaging/234234234
[17:12:14] <halfak>	 YuviPanda, may know more. 
[17:12:48] <YuviPanda>	 legoktm: yeah, I'm not fully sure...
[17:13:39] <legoktm>	 hmm
[17:13:52] <YuviPanda>	 let me try talk to bblack
[17:14:23] <legoktm>	 I still don't like that we're putting services on the main domain
[17:14:39] <legoktm>	 I'd much rather prefer ores.wm.o/$wiki/...
[17:14:41] <YuviPanda>	 legoktm: it's good for SPDY / Http2 connection pooling
[17:14:57] <legoktm>	 YuviPanda: clients should not hit it directly though
[17:15:03] <YuviPanda>	 legoktm: JS
[17:15:04] <halfak>	 idealistic +1 to legoktm 
[17:15:18] <halfak>	 But yeah, I get that performance/caching issues may disagree. 
[17:15:26] <YuviPanda>	 I don't think all uses of ORES will be from PPH
[17:15:29] <YuviPanda>	 *PHP
[17:15:37] <legoktm>	 well right.
[17:15:49] <legoktm>	 We should have a MW API endpoint for JS then
[17:16:04] <YuviPanda>	 adding php into the works for no real reason there :P
[17:16:10] <YuviPanda>	 again perf hit...
[17:16:19] <YuviPanda>	 and complexity hit
[17:16:53] <YuviPanda>	 and then you'll need to add support to the mw api for every thing the backing API supports, if you want a mw-api-ish design for the mw api rather than just a proxy
[17:17:14] <legoktm>	 right right
[17:17:28] <YuviPanda>	 so ideally we'll have a layer that's not ores do the rewriting
[17:17:36] <YuviPanda>	 so ores still sees things in its current url structure
[17:18:15] <YuviPanda>	 that could be uwsgi, actually...
[17:18:20] <YuviPanda>	 but I gotta talk to bblack
[17:18:21] <halfak>	 +1
[17:18:23] <YuviPanda>	 and he isn't around atm
[17:18:29] <halfak>	 I think that is the best solution. 
[17:33:48] <legoktm>	 halfak: can I pm?
[17:34:00] <halfak>	 Yessir
[21:57:27] <YuviPanda>	 legoktm: halfak I've filed https://phabricator.wikimedia.org/T116088