[15:22:36] _o/ [15:24:11] 10Quarry: Seeing all of created queries in one page - https://phabricator.wikimedia.org/T98688#1276295 (10Capt_Swing) See also: # https://phabricator.wikimedia.org/T77948 # https://phabricator.wikimedia.org/T88920 [15:24:17] heyo [17:24:46] o/ Nettrom [17:24:52] g'afternoon :) [17:25:46] I've been talking to some researchers at CMU about working with the models you've got behind suggestbot. [17:25:57] Specifically the one that predicts what work an article needs. [17:26:46] It would be good to get that in wikiclass sometime soon. [17:28:09] hey halfak [17:28:24] cool! let me schedule some time for it this week [17:29:08] quite busy grading, and finishing up a blog post about the ICWSM paper [17:29:47] (the latter went live about 30 mins ago, http://grouplens.org/wasted-effort-and-missed-opportunities/ [17:31:47] * Nettrom lunch [17:32:56] whee. I love the R community [17:33:06] our devtools package now contains use_code_of_conduct() [17:33:30] or to put it another way, instituting a standardised CoC for package contributions and discussion standards is now as easily-accessed and given equal weight to "use tests" [18:42:44] Science: http://www.theguardian.com/science/2015/may/05/microwave-oven-caused-mystery-signal-plaguing-radio-telescope-for-17-years [19:33:48] Ironholds: halfak a warning for space in /tmp on stat1002, it’s about to run out of inodes (aka too many small files), just a fyi [19:34:07] yuvipanda, am I using a lot of /tmp? [19:34:45] Looks like it is mostly spark. [19:34:45] someone might be. I haven’t looked [19:34:49] alright. [19:34:57] I’ll leave it to you guys to investigate :D [19:35:01] was just relaying from -operations [19:35:35] kk thanks yuvipanda [19:37:38] Ironholds, it looks like there's ~400 Rtmp... folders you could delete. [19:37:49] Many of them are from last year. [19:38:43] Looks like spark is responsible for most of the tmp folders. [19:50:26] halfak: it’s got 0% space left, so whatever is creating those might have stopped wokring [19:51:35] yuvipanda, thanks. Trying to chase down someone to do something about it now. [19:51:42] halfak: cooll [19:51:55] yuvipanda, is there already a phab ticket? [19:51:58] halfak: nope [19:52:03] k [20:10:14] halfak: woaoaah, you’re in the OFFICEEE! [20:33:58] yuvipanda, yes! :) [20:36:18] yuvipanda: I was surprised by it this morning as well. Even if I knew halfak was going to be here, it felt weird to see the actual live halfak and not the life-size cutout :) [20:37:53] [what's with the life size cutout] [20:38:12] Ironholds: I'd need to ask you about https://phabricator.wikimedia.org/T94637 when you have a moment in the next few days. My question is basically "How hard is it to do it" and my feeling is that since you've provided the query, any idiot who doesn't know R (i.e. me) could probably do it (since Sherry is asking me). [20:38:42] OTOH if it's hard, I'll send her to Neil :) [20:39:37] Neil? [20:39:42] Quinn [20:39:52] Young [20:40:01] James's minio^W product analyst [20:40:33] Also, hi :) [20:43:18] harej, http://imgur.com/8PLZ7tZ [20:43:37] I'm very familiar with the cutout. I just wonder why it exists. [20:43:48] guillom, I don't know this person [20:44:08] i do! [20:44:30] neil quinn is a wikipedia editor cum wikimedia foundation employee [20:44:35] Ironholds: He's nice. [20:44:38] yes, this [20:44:40] anyway [20:44:49] harej, okay! What's their username [20:44:50] ? [20:45:02] https://en.wikipedia.org/wiki/User:Neil_P._Quinn << former DC area wikipedian [20:45:37] ahh [21:16:28] halfak: if I wanted to use https://pythonhosted.org/mediawiki-utilities/_modules/mw/lib/reverts/api.html how would I import it? [21:17:02] (I have it installed) [21:20:53] guillom, missed actually answering thge question; it should be fairly simple, yes :D [21:21:10] Ironholds: ok :) thanks [22:21:53] harej, from mw import api [22:22:01] got it, thanks [22:22:07] session = api.Session("https://en.wikipedia.org/w/api.php") [22:22:12] or better [22:22:23] session = api.Session("https://en.wikipedia.org/w/api.php", user_agent="harej doin' a thing") [22:22:37] :) [22:35:42] halfak: and you recommended that i do 48 hours in my inspection. given a string timestamp in the format YYYYMMDDHHMMSS, how do I take that minus 48 hours? [22:49:08] harej, in python? [22:49:27] Yes [22:50:08] Timestamp("20150101000000") - 60*60*24*2 # (48 hours in seconds) [22:50:21] You can do second-level math with the Timestamp type from mw. [22:50:28] from mw import Timestamp [22:51:05] t = Timestamp("20150101000000") [22:51:19] t2 = t - 60*60*24*2 [22:51:41] t - t2 == 172800 [22:51:49] 172800 == 60*60*24*2 [22:52:22] * halfak does lots of date math [22:52:41] everything should be expressible in seconds since Epoch [22:53:02] * halfak wonders if anyone notices his username on enwiki and his fascination with date math. [22:53:29] hahe [22:54:17] harej, now after all that, you shouldn't need to do date math for revert detection. [22:54:28] You can instead set the revert_window parameter. [22:54:41] e.g. revert_window=60*60*24*2 [22:54:55] I thought in check it asked for a time stamp? [22:55:11] Ack! Looks like the docs are out of date. [22:55:16] Sec. [22:57:42] harej, just updated. See https://pythonhosted.org/mediawiki-utilities/lib/reverts.html#mw.lib.reverts.api.check [22:57:49] note the "window" parameter. [23:05:08] And I want radius 3, check archives yes? [23:06:20] harej, check archives only works with the DB [23:07:14] Okay. [23:08:56] Radius 3 seems to be good to catch vandalism reverts (if that is what you are looking for) [23:59:36] halfak: it looks like it accepts both before and window. Do I just set before to None and then specify a window?