[00:06:51] <YuviPanda>	 halfak: can you give me push rights to the halfak/MW-Utilities repo?
[00:07:41] <halfak>	 YuviPanda, {{done}}
[00:07:54] <YuviPanda>	 so both in anticipation of moving this to prod sooner than latter
[00:08:00] <YuviPanda>	 and because pip has been a bitch
[00:08:07] <YuviPanda>	 I think we should move to using debian packages to do dependencies
[00:08:11] <YuviPanda>	 and I'm starting work on that now
[00:08:36] <YuviPanda>	 halfak: see body of https://phabricator.wikimedia.org/T107493 I edited it
[00:09:15] <halfak>	 What does this move entail?
[00:09:24] <halfak>	 I'm not sure I'm ready to stop working with pip
[00:09:28] <halfak>	 Can we support both?
[00:10:06] <YuviPanda>	 I don't know what this move entails yet
[00:10:12] <YuviPanda>	 https://github.com/halfak/Mediawiki-Utilities/pull/38 for the debian packaging of mw-utils
[00:10:48] <YuviPanda>	 halfak: I think this will mean that version requirements on requirements.txt will get more strict (using jessie packages where available), and we should stop specifying transient dependencies
[00:11:05] <halfak>	 So, why did you remove the install requirements from setup.py?
[00:11:07] <awight>	 YuviPanda: are you suggesting we hand roll a deb package for scipy?
[00:11:08] <YuviPanda>	 halfak: you can still use pip for local development and what not, as long as the versions match we can just use debs for deployment
[00:11:16] <YuviPanda>	 awight: there's already a scipy package
[00:11:19] <YuviPanda>	 we're already usign it
[00:11:32] <awight>	 oh, I thought it was the wrong version... /me goes cross-eyed
[00:11:51] <YuviPanda>	 halfak: because then running setup.py attempts to download them from the internet and that becomes a big no-no in debianland. I can clarify in commit message if needed
[00:12:05] <YuviPanda>	 awight: we version pinned it to jessie but the latest change might have broken it
[00:12:41] <halfak>	 YuviPanda, can we not re-write the setup.py file and still do this?
[00:12:46] <YuviPanda>	 not sure
[00:12:50] <awight>	 I'm a little distraught about the Travis situation... Looks like we can't use miniconda to accelerate the build, cos of more version hell
[00:12:55] <halfak>	 A LOT of people are using mw from pip right now
[00:13:08] <YuviPanda>	 oh right we can't break it for pip users I agree
[00:13:13] <YuviPanda>	 will this break it for pip users?
[00:13:17] <halfak>	 Yes
[00:13:18] <halfak>	 Terribly
[00:13:29] <YuviPanda>	 oh yeah then that's fine I'll find a way to not break it for pip users
[00:13:40] <YuviPanda>	 halfak: it's py3 only right?
[00:13:46] <halfak>	 Yes
[00:13:51] <YuviPanda>	 cool
[00:13:54] <halfak>	 :)
[00:15:08] * halfak maxes out ores-compute with new model building. 
[00:22:42] <YuviPanda>	 halfak: fixed
[00:22:51] <halfak>	 The PR?
[00:23:06] <YuviPanda>	 yeah
[00:23:11] <halfak>	 Thanks for pulling out argparse.
[00:23:18] <YuviPanda>	 :D yw
[00:23:25] <YuviPanda>	 halfak: apparently debuild doesn't like the spaces?
[00:23:39] <halfak>	 Spaces?
[00:23:49] <YuviPanda>	 around the >=
[00:23:54] <halfak>	 Oh.  Weird.
[00:23:59] <halfak>	 Goddamn regex
[00:24:04] <YuviPanda>	 yeah
[00:24:15] <YuviPanda>	 anyway this shouldn't cause problems for pep people
[00:24:30] <halfak>	 Agreed.  Seems that I can still install
[00:24:34] <halfak>	 {{merged}}
[00:24:50] <YuviPanda>	 halfak: cool
[00:25:03] <YuviPanda>	 halfak: so I saw commits go by that were dropping yamlconf. is that going to happen or are you going to keep using it?
[00:25:30] <halfak>	 Keep using it.  revscoring doesn't need it.
[00:25:37] <YuviPanda>	 ok
[00:25:39] <halfak>	 You don't like yaml config files?
[00:25:57] <halfak>	 I should really clean up the yamlconf package.
[00:26:16] <YuviPanda>	 halfak: no I love them, I'm just wondering if you need the extra package vs just using the yaml module :)
[00:26:21] <YuviPanda>	 since now I've to create another package
[00:26:37] <YuviPanda>	 I'm not going to offload *this* into anyone else yet, since it's brutal getting started on this
[00:28:09] <YuviPanda>	 halfak: is 'deltas' your library?
[00:28:34] <halfak>	 Yup
[00:30:04] <awight>	 YuviPanda: I'm happy to help package if u need
[00:31:22] <YuviPanda>	 awight: w00t! https://phabricator.wikimedia.org/T107493 and subtasks
[00:33:11] <YuviPanda>	 halfak: what are we using pylru for?
[00:33:28] <halfak>	 It's an optional in-memory cache that replicates the interface of redis.
[00:33:33] <halfak>	 For a smaller installation
[00:33:33] <awight>	 YuviPanda: can do.
[00:33:35] <YuviPanda>	 hmm
[00:33:42] <halfak>	 Yeah.  Best LRU option I found.
[00:33:57] <halfak>	 I use it when doing localdev.
[00:34:02] <YuviPanda>	 halfak: was https://docs.python.org/3/library/functools.html#functools.lru_cache not good enough?
[00:34:26] <YuviPanda>	 awight: I can cajol some members of the ops team into uploading some of the packages I think :D stopit, pylru definitely, yamlconf maybe
[00:34:30] <halfak>	 Regretfully not.  I just wanted the cache and the decorator wasn't going to work.
[00:34:34] <YuviPanda>	 halfak: ok
[00:34:36] <YuviPanda>	 that's fine
[00:34:47] <halfak>	 YuviPanda, I promise that I really tried to use the stdlib stuff.
[00:34:51] <YuviPanda>	 halfak: :D
[00:34:56] * halfak hates new dependencies on under-maintained projects
[00:35:02] <YuviPanda>	 :D yeah, they all seem ok
[00:36:49] <YuviPanda>	 awight: assign the ones you want to yourself? :)
[00:38:27] <awight>	 that works.
[00:38:52] <YuviPanda>	 awight: sweet :)
[00:38:55] <YuviPanda>	 awight: thanks a lot
[00:39:12] <YuviPanda>	 awight: we'll also want to try to pin the versions in requirements.txt to the ones that already exist in jessie+backports
[00:39:19] <awight>	 wait 'til I actually do anything useful ;)  Sorry to be short, I'm trying to kill some CR on my way out the door for the day
[00:39:31] <YuviPanda>	 awight: 'tis ok! thanks for helping :D
[00:39:47] <awight>	 ok, and I'll let you know if I get anywhere with the stupid Travis py module versions
[00:39:51] <YuviPanda>	 +1
[00:39:59] <halfak>	 AHhH!
[00:40:01] <halfak>	 Spam!
[00:40:06] <YuviPanda>	 haha
[00:40:10] <awight>	 :) beams at my pet monster
[00:43:50] <YuviPanda>	 halfak: can I get push access to yamlconf and deltas too?
[00:47:30] <halfak>	 YuviPanda, how come you want push access?
[00:47:51] <YuviPanda>	 halfak: I just want to push to a branch and then do a PR
[00:47:53] <YuviPanda>	 instead of forking
[00:47:56] <YuviPanda>	 if that's ok
[00:48:02] <YuviPanda>	 ?
[00:48:02] <halfak>	 Sure.  It's fine.
[00:48:07] <YuviPanda>	 cool :)
[00:48:10] <halfak>	 Just wasn't sure you knew you could fork :P
[00:48:12] <YuviPanda>	 halfak: and I've a package for yamlconf already now
[00:48:19] * halfak gets stats love for more forks
[00:48:20] <YuviPanda>	 :D I do, gerrit hasn't corrupted me that much no
[00:48:26] <YuviPanda>	 yeah but my fork list is a mess :D
[00:48:34] <YuviPanda>	 so I try not to fork unless I'm contributing code :D
[00:48:43] <halfak>	 :P
[00:49:28] <YuviPanda>	 halfak: done
[00:49:32] <YuviPanda>	 halfak: https://github.com/halfak/yamlconf/pull/1
[00:54:36] <halfak>	 First pull request!
[00:57:16] <YuviPanda>	 halfak: indeed!
[00:58:01] * YuviPanda is standing in line at security
[00:58:22] <YuviPanda>	 halfak: for a country that has produced some amazing things the USA also seems to have produced the TSA :)
[00:58:35] * YuviPanda offers his shoes to the TSA gods
[00:58:55] <halfak>	 They don't want them.  They just want you to take them off and step in wet things with your socks.
[00:59:52] <YuviPanda>	 Heh
[06:04:55] <legoktm>	 okay.
[06:05:05] <legoktm>	 YuviPanda, halfak: http://i.imgur.com/fafJqvg.png
[06:08:04] <YuviPanda>	 legoktm: haha
[06:08:13] <YuviPanda>	 Should probably not run on non mainspaxe edits
[06:08:31] <legoktm>	 oh no, I faked the data
[06:08:41] <YuviPanda>	 Oh
[06:08:54] <legoktm>	 that's the MW extension I just wrote
[06:09:00] <YuviPanda>	 legoktm: wow nice!nn
[06:09:20] <YuviPanda>	 Is this with or without the rc_id joined table?
[06:10:05] <legoktm>	 it's joining with a "ores_scores" table that I populated with fake data
[06:10:25] <legoktm>	 the table is just ores_rc (rc_id) and ores_score (the actual score)
[06:10:44] <legoktm>	 and I'll probably add a column to mark an edit as reverted
[06:11:07] <YuviPanda>	 Nice!
[06:11:28] <YuviPanda>	 legoktm: in ores_scores? I thought there is already a place for it somewhere else
[06:12:13] <legoktm>	 I don't think there is...
[06:12:40] <YuviPanda>	 where were you thinking of putting that column?
[06:12:45] <legoktm>	 in ores_scores
[06:12:55] <legoktm>	 just a tinyint 0 or 1
[06:12:58] <YuviPanda>	 Krinkle was talking about there being a flag for 'patrolled' that you can mark
[06:13:07] <legoktm>	 yes, patrolled is different
[06:13:09] <YuviPanda>	 And is shareable across tools
[06:13:14] <legoktm>	 actually, we can just use that
[06:13:17] <YuviPanda>	 Shouldn't ores just book into that
[06:13:19] <YuviPanda>	 Hook
[06:13:20] <YuviPanda>	 Yes
[06:13:22] <YuviPanda>	 We should
[06:14:49] <legoktm>	 ok, it'll already "just work"
[06:15:00] <legoktm>	 because RC already has a filter for "Hide patrolled edits" (my local wiki doesn't)
[06:19:10] <YuviPanda>	 legoktm: ah nice.
[06:19:23] <YuviPanda>	 legoktm: so we add another checkbox to show only ores predicted ones?
[06:22:13] <legoktm>	 yeah, I'm not sure about that
[06:23:08] <legoktm>	 we should just replace the silly "Hide ORES" thing I have right now with "Only show ORES-predicted bad edits" or something
[06:23:39] <legoktm>	 YuviPanda: also, what's the best way to store floats in mysql? right now I'm using varchar :P
[06:28:16] <YuviPanda>	 legoktm: ha-ha not sure
[06:28:29] <YuviPanda>	 Doesn't mysql have a float type?
[06:32:35] <legoktm>	 yes but the docs weren't obvious so I just use varchar
[06:32:36] <legoktm>	 https://dev.mysql.com/doc/refman/5.0/en/floating-point-types.html
[06:32:40] <legoktm>	 used*
[06:32:41] <YuviPanda>	 Ha-ha ok
[06:34:55] <YuviPanda>	 legoktm: we can ask in #wikimedia-databases
[06:35:50] <legoktm>	 we could, but I'm now wondering if it makes sense to store the full float value
[06:35:53] <legoktm>	 is that useful?
[06:36:04] <legoktm>	 can we just store the percentage we show to users? (3 digits max)
[06:36:51] <legoktm>	 we could even use tinyint then
[06:39:51] <YuviPanda>	 legoktm: hmm, idk - why not store the full float?
[06:40:00] <legoktm>	 is there a use for it?
[06:40:07] <legoktm>	 		static $thresholds = array(
[06:40:07] <legoktm>	 			'low' => 0.8,
[06:40:07] <legoktm>	 			'medium' => 0.87,
[06:40:07] <legoktm>	 			'high' => 0.94,
[06:40:07] <legoktm>	 		);
[06:40:18] <legoktm>	 I just copied those from helder's script
[06:40:21] <YuviPanda>	 hmm
[06:40:35] <legoktm>	 I don't anticipate we'll have more granular thresholds...will we?
[06:40:36] <YuviPanda>	 you can store 0.87 but this feels like one of those things that is going to bite you in the ass later on
[06:40:40] <legoktm>	 yeah.
[06:40:43] <YuviPanda>	 so unless there are actual problems storing float we should no
[06:40:56] <legoktm>	 does laziness count as a real problem? :P
[06:41:32] <legoktm>	 I'll look into it tomorrow, I also need to fix enhanced RC
[06:41:59] <YuviPanda>	 ah ok
[06:42:04] <YuviPanda>	 legoktm: I asked springle anyway
[06:42:09] <legoktm>	 thanks
[06:42:38] <legoktm>	 fun fact: the frontend rendering of watchlist/RC happens in 4 different classes, all with different hooks!
[06:43:33] <legoktm>	 MatmaRex fixed the backend stuff last year so there's only one hook for the db query, but frontend is a mess >.<
[06:43:46] <YuviPanda>	 :(
[06:43:46] <YuviPanda>	 boo
[13:09:35] <halfak>	 \o/  Just saw the screenshot from vagrant
[13:09:54] <halfak>	 Also, ORES should run on all namespace edits.
[16:03:41] <halfak>	 o/ aetilley
[16:03:43] <halfak>	 sorry fopr late
[16:03:47] <halfak>	 Will be there in a minute
[16:31:18] <YuviPanda>	 halfak: all namespaces? Including js edits?
[17:00:48] <halfak>	 YuviPanda, yes.  Then again, I expect those to be rare
[17:03:04] <halfak>	 aetilley`, add_badwords_ratio
[17:22:00] <YuviPanda>	 halfak: you're probably in a meeting but the email you sent sounds awesome
[17:29:52] <legoktm>	 https://gerrit.wikimedia.org/r/229423
[17:31:12] <YuviPanda>	 \o/
[17:31:36] <YuviPanda>	 legoktm: I think we should use decimal or double, whichever is faster?
[17:58:53] <halfak>	 YuviPanda, :D  Woot!  I'm glad to have you onboard.  Do you want to help us write up the proposal?  This stuff gets filed in academic repositories and therefor counts towards your academic cred.
[17:58:56] <halfak>	 Not sure if you care.
[17:59:15] <halfak>	 We can also just pull you in for the discussions of what technology to set up for the workshop
[17:59:27] <legoktm>	 YuviPanda: yep
[18:15:17] <awight>	 YuviPanda: uuf, so I charged into the shallow end, worked on a couple of deb packages, and ended the night by reading the full task description, where it says the two I did are probably optional.
[18:15:32] <YuviPanda>	 awight: yeah but that's ok - we probably need them for now anyway
[18:15:36] <YuviPanda>	 awight: I did the yamlconf tho :(
[18:17:02] <ToAruShiroiNeko>	 halfak if not busy I did a bit of work on meta
[18:17:04] <ToAruShiroiNeko>	 https://meta.wikimedia.org/wiki/Category:Revision_scoring_as_a_service/Word_lists
[18:17:17] <ToAruShiroiNeko>	 I tiraged problems in a category structure
[18:18:20] <awight>	 YuviPanda: hey, unassign yourself from the other deb subtasks unless you're actually working on them!
[18:18:30] <awight>	 I couldn't tell whether you were squatting...
[18:18:46] <YuviPanda>	 awight: I'm not squatting any of them now
[18:18:53] <awight>	 k!
[18:18:55] <YuviPanda>	 I'll leave them all to you for now
[18:18:59] <awight>	 d'oh
[18:31:11] <halfak>	 :d
[18:31:29] <YuviPanda>	 awight: wait, I shouldn't be assigned to any of them now
[18:31:46] <YuviPanda>	 halfak: I don't know how I can help (re: workshop). Never been to any of these before
[18:35:20] <YuviPanda>	 halfak: so not sure what I can do. maybe help with adding datasets to quarry?
[18:37:07] <halfak>	 +1.  We'll need to write-up a plan for the day and some justification for the systems we want to test out with participants.
[18:38:03] <YuviPanda>	 right
[20:06:37] <madhuvishy>	 halfak: can i move our 2pm thing to 3?
[20:06:48] <halfak>	 Yes
[20:06:49] <madhuvishy>	 YuviPanda seems to be caught up with something
[20:06:59] <YuviPanda>	 thanks
[21:20:42] <halfak>	 YuviPanda, where do I find mwapi's code?
[21:20:56] <YuviPanda>	 halfak: python-mwapi?
[21:20:58] <halfak>	 Oh!  python-!
[21:21:02] <halfak>	 https://github.com/yuvipanda/python-mwapi
[21:21:05] <halfak>	 :)
[21:21:07] <YuviPanda>	 https://github.com/yuvipanda/python-mwapi
[21:21:08] <YuviPanda>	 yes
[21:22:51] <halfak>	 Cool. I've got some time, so I'm digging into the minimalistic API proposal
[21:24:35] <halfak>	 YuviPanda, how would you feel if I submitted a PR with a special 'query()' method that supports continuation?
[21:24:50] <YuviPanda>	 halfak: quite happy!
[21:25:00] <halfak>	 Cool.  I'll look into it. :)
[21:44:45] <halfak>	 YuviPanda, I just filled out some feature requests.  Please review to see if there's anything you don't like.  https://github.com/yuvipanda/python-mwapi/issues
[21:45:04] <YuviPanda>	 halfak: all seem good to me
[21:45:24] <halfak>	 Cool.  I'll get to work at my first opportunity :)
[21:45:48] * halfak is working towards an 'mw<something>' revolution
[21:46:01] <halfak>	 mwreverts, mwxml, mwapi, mwpersistence, mwoauth, etc.
[21:46:42] <YuviPanda>	 nice
[21:47:49] <YuviPanda>	 madhuvishy: the thing I was helping just finished :( looks like I'll have to do this remote and then come in
[21:52:40] <madhuvishy>	 do the wikilabels thing remote?
[21:53:07] <madhuvishy>	 YuviPanda: why come in after? Would tomorrow work better? I'm gonna be here
[21:53:18] <YuviPanda>	 madhuvishy: I need to get a bag from office :)
[21:53:41] <YuviPanda>	 madhuvishy: we can do tomorrow too. I just feel bad for dragging you and halfak around
[21:54:05] <madhuvishy>	 I'm here, and it doesn't make a difference. poor halfak though
[21:54:10] <halfak>	 No worries.
[21:54:13] <YuviPanda>	 sorry
[21:54:16] * halfak is a rich man
[21:54:19] <YuviPanda>	 I just got pulled into supporting a deploy somewhere
[21:54:20] <madhuvishy>	 ha ha
[21:54:22] <halfak>	 In the things that matter
[21:54:27] <madhuvishy>	 :)
[21:54:34] <YuviPanda>	 and then it turned out I wasn't needed at all, but I was around till the end of it anyway
[21:55:21] <halfak>	 Darn people pulling you into deploys WAIT A SECOND....  other people ... yeah those other people ;)
[21:55:29] <YuviPanda>	 haha :)
[21:55:37] <YuviPanda>	 halfak: well I wouldn't mind if I actually ended up doing things :)
[21:55:45] * halfak wants to do things
[21:55:50] <halfak>	 But I can wait.
[21:55:59] <halfak>	 No trouble.
[22:01:00] <halfak>	 YuviPanda, madhuvishy, just to check.  Are we meeting now?
[22:01:09] <madhuvishy>	 No?
[22:02:38] <YuviPanda>	 we can if you guys are ok with doing it remotely.
[22:02:45] <YuviPanda>	 but ok if not and can do it tomorrow
[22:03:05] <halfak>	 Remotely is all I ever do
[22:03:13] <halfak>	 madhuvishy, ?
[22:03:49] <madhuvishy>	 halfak: since YuviPanda is in SF I'd prefer to do it in person
[22:04:04] <YuviPanda>	 ok then!
[22:04:30] <halfak>	 OK.  Tomorrow it is.
[22:04:43] <halfak>	 madhuvishy, I'll leave it to you to find the slot. ;)
[22:04:55] <madhuvishy>	 halfak: okay :)
[22:06:21] <madhuvishy>	 YuviPanda: will it take an hour?
[22:06:26] <madhuvishy>	 or 30 minutes will do?
[22:06:42] <YuviPanda>	 madhuvishy: let's schedule an hour and hope for 30mins? :D
[22:07:33] <madhuvishy>	 halfak: are you playing Squash at 1.30?
[22:12:23] <halfak>	 :C
[22:12:28] <halfak>	 Was planning on it.
[22:12:48] <halfak>	 Nettrom is offline or I'd ask him if he would be able to tomorrow.
[22:12:56] <madhuvishy>	 YuviPanda I moved it to 1-2 tomorrow. halfak we will hopefully be done early, even if not you should go play squash :)
[22:13:04] <YuviPanda>	 ok
[22:13:06] <halfak>	 I could move it to an hour earlier if that works.
[22:13:15] <halfak>	 12:30 - 2PM
[22:13:28] <madhuvishy>	 sure!
[22:13:58] <halfak>	 Done
[22:14:04] <madhuvishy>	 halfak: okay now it's at 2.
[22:14:18] <halfak>	 Looks good.  Except for your PDT-centricness :P
[22:14:29] <madhuvishy>	 halfak: sorry :)
[22:15:28] <halfak>	 It's OK.  I usually assume PDT when WMF people don't specify.
[22:15:32] <halfak>	 UTC for everyone else.
[22:15:49] <halfak>	 The members of this channel cover a lot of timezones.
[22:15:55] <halfak>	 :)
[22:16:06] * YuviPanda wonders what TZ he's in now
[22:16:13] <YuviPanda>	 anyway, off to office now for real
[23:23:23] <awight>	 YuviPanda: Hey uh, you ever have a problem where pybuild creates empty .debs as soon as you configure it to build both a python2 and a python3 package?  I'm seeing this in two different repos...
[23:25:24] <YuviPanda>	 awight: huh, no... although I'll admit to not having built multisupport ones
[23:25:25] <awight>	 Scary.  It seems to be the debian/control file that determines whether I see this glitch or not.  Without the python3 build dependencies and package block, the python2 deb is normal
[23:25:29] <awight>	 k
[23:28:12] <awight>	 gah.  https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=761028
[23:49:38] <awight>	 YuviPanda: python3 setup.py --command-packages=stdeb.command debianize
[23:50:09] <YuviPanda>	 Yeah that's what I used to use before pybuild
[23:50:26] <awight>	 wow.  I was building these stupid things by hand
[23:50:27] <YuviPanda>	 Imo pybuild is simpler in the long run and maybe easier to get into Debian
[23:50:57] <awight>	 Ah i see, it doesn't use dh_pybuild rules
[23:51:27] <awight>	 Does pybuild create packaging files, too?
[23:51:42] <YuviPanda>	 As in?
[23:51:47] <YuviPanda>	 The Debian/ folder?
[23:51:51] <YuviPanda>	 No I just copy them around
[23:51:55] <awight>	 k