[02:19:44] <wikibugs>	 (03CR) 10Reedy: [C: 032] API: Split description messages into summary + additional text [extensions/ORES] - 10https://gerrit.wikimedia.org/r/358724 (owner: 10Anomie)
[02:27:21] <wikibugs>	 (03Merged) 10jenkins-bot: API: Split description messages into summary + additional text [extensions/ORES] - 10https://gerrit.wikimedia.org/r/358724 (owner: 10Anomie)
[09:34:10] <mavrikant>	 halfak:  https://phabricator.wikimedia.org/T164671
[09:34:10] <mavrikant>	 there ise more than 500 new label on trwiki. if extracting old ones is hard we can use these. but will it be enough?
[14:26:39] <glorian_wd>	 halfak: o/
[15:43:20] <halfak>	 o/ glorian_wd 
[15:43:21] <halfak>	 in meetings
[15:43:37] <halfak>	 Until 1930 UTC :( 
[15:43:46] <glorian_wd>	 oh :(
[15:43:48] <glorian_wd>	 halfak: I want to confirm what I understood from what you said
[15:44:18] <glorian_wd>	 regarding to my PR, could I still use DataSources?
[15:44:53] <halfak>	 Only if you are going to re-use one
[15:45:11] <glorian_wd>	 Ok, if I don't want to reuse it, I should not use Datasource right?
[15:45:23] <halfak>	 Right
[15:45:32] <glorian_wd>	 I should use the ordinary 'Feature'
[15:45:34] <glorian_wd>	 Ok
[15:45:40] <glorian_wd>	 I have another question
[15:47:33] <glorian_wd>	 halfak: Someone told me to add more details into https://phabricator.wikimedia.org/T164994. Details such like what we're trying to achieve, how we want to involve wbsgetsuggestions in what we're doing (in revscoring), the steps to achieve that
[15:47:49] <glorian_wd>	 most importantly, I need to tell that I am gonna use 'continuation'
[15:48:30] <glorian_wd>	 halfak: this information will help Daniel and co. to help us
[15:49:03] <glorian_wd>	 halfak: So, I am gonna add this detail to the ticket. Could you help to add or modify the details that I am gonna add?
[15:58:57] <halfak>	 glorian_wd, sure
[15:58:59] <halfak>	 ping me there
[16:03:03] <glorian_wd>	 kk
[16:03:04] <glorian_wd>	 thanks
[17:09:26] <Zppix>	 Hello halfak  hows everything going?
[17:15:03] <halfak>	 Hey!  meetings all day today :( 
[17:15:20] <halfak>	 Getting a little bit of progress on some open tasks here and there
[17:15:27] <halfak>	 Just about to get on my bike to go to the University. 
[17:15:30] <Zppix>	 Ill look at them and see what i can do
[17:15:31] <halfak>	 back in ~40 mins
[17:15:38] <Zppix>	 okay ill hold the fort down :)
[17:16:45] <glorian_wd>	 halfak: I have pinged you on the ticket
[17:16:55] <glorian_wd>	 I hope there won't be any long debate between you and Daniel 
[18:18:24] <halfak>	 o/
[18:18:28] <halfak>	 (in meeting again)
[18:42:55] <glorian_wd>	 halfak: did you read my last message about the ticket?
[18:43:01] <halfak>	 meetings
[18:54:03] <Zppix>	 glorian_wd:  im here if you need something.
[18:54:22] <glorian_wd>	 Zppix: cool. Thanks :D:
[18:54:33] <Zppix>	 np
[19:48:47] <glorian_wd>	 halfak: still in meeting?
[19:59:07] <halfak>	 glorian_wd, OMG just finished
[19:59:20] <halfak>	 I have a two hour block \o/
[19:59:53] <glorian_wd>	 \o/
[20:00:34] <glorian_wd>	 halfak: okay. Firstly, I have pinged you in the ticket. Secondly, I want to ask, if I have a function which returns 'int' and it's reused in other functions, should I use 'Datasource' or 'Feature'?
[20:02:54] <glorian_wd>	 I guess that I have to use Datasource, but I am not sure
[20:11:08] <halfak>	 glorian_wd, feature, probably. 
[20:11:35] <glorian_wd>	 It is fine to use 'Feature' which is re-used?
[20:11:37] <glorian_wd>	 I know it
[20:11:43] <glorian_wd>	 is gonna working
[20:11:48] <glorian_wd>	 thought
[20:11:52] <glorian_wd>	 though*
[20:11:59] <halfak>	 I don't see the ping.  Please link
[20:12:14] <glorian_wd>	 halfak: https://phabricator.wikimedia.org/T164994
[20:12:34] <glorian_wd>	 Ok. If you're thinking it's better to use Feature, I will use it
[20:14:57] <halfak>	 glorian_wd, I'd take out the confusing example and state "We'll use a simple weighted sum based on the probability reported via the wbs_propertypairs table to get signal for completeness.  Items with all high probability statements complete should be more likely to *be* complete than items that lack high probability statements."
[20:24:08] <glorian_wd>	 halfak: my OS was freezed because my browser tabs 
[20:24:51] <glorian_wd>	 halfak: oh the example was confusing? okay
[20:25:27] <halfak>	 Yeah.  Super confusing
[20:27:10] <glorian_wd>	 {{done}}
[20:27:10] <AsimovBot>	 You rule, glorian_wd!
[20:27:17] <glorian_wd>	 thanks AsimovBot :p
[20:27:45] <glorian_wd>	 halfak: I've replaced the example with the sentence you gave above
[20:35:20] * paladox has begun implement branches list in polygerrit :)
[20:36:04] <halfak>	 cool
[20:40:45] <codezee>	 halfak: o/
[20:43:57] <codezee>	 can you tell if there's a way in revscoring to fetch text from a file for extracting features rather than using extractor and fetch online?
[20:44:21] <halfak>	 yes!  Look at the tests anywhere in revscoring. 
[20:44:28] <halfak>	 they use "solve()" directly. 
[20:44:33] <halfak>	 Let me demonstrate :) 
[20:46:21] <halfak>	 https://gist.github.com/halfak/05bf6ffd098c79d76101c45a6a7cf548
[20:46:26] <halfak>	 codezee, ^
[20:47:17] <codezee>	 thanks! looking
[20:51:34] <codezee>	 halfak: just out of curiosity since I'm reading code of revscoring, what was the major motivation behind - https://github.com/wiki-ai/revscoring/pull/307 from an architecture or software engineering point of view?
[20:55:10] <halfak>	 codezee, it's primarily about how we handle model information -- statistics and thresholds. 
[20:55:56] <halfak>	 I want to expose this better to users of ORES so, I've built a framework for storing and accessing model information.  That PR is almost ready.  I've been doing too much paper work to finish it off recently. :( 
[20:56:16] <halfak>	 Related: https://phabricator.wikimedia.org/T162217
[20:56:38] <halfak>	 There were a few other things I could take care of while I was at it.  Like https://phabricator.wikimedia.org/T163711
[20:57:27] <halfak>	 Oh yeah and this too: https://phabricator.wikimedia.org/T160223
[21:02:12] <codezee>	 tbh, the refactor PR was quite daunting when I looked at the files bec I've just started but the whole thing seems interesting :)
[21:02:56] <halfak>	 codezee, yeah.  This was one of those refactors where I finished it twice and started over again before I got it right :) 
[21:03:05] <halfak>	 I've done a few of those in revscoring. 
[21:03:13] <halfak>	 I wish github was better at tracking diffs though. 
[21:03:46] <halfak>	 There are many cases where I only changed 10 lines in a 200 line file and it just says, "I dunno.  He deleted everything and recreated it from scratch."  
[21:03:48] <halfak>	 No :( 
[21:05:12] <codezee>	 :/
[22:33:59] <halfak>	 OK I'm outta here.  Have a good one folks!
[22:34:41] <Zppix>	 o/ bai