[00:00:09] halfak: I can kindof understand that, I guess [00:00:20] Yeah. Me too. [00:00:32] In the end I conceded. [00:12:20] stablemwparserfromhell [00:12:34] mwparserfromstable [00:15:56] :D [00:16:10] makes horsey noises [00:16:21] lol! [00:16:56] Ironholds: bearloga_ got a few minutes for R questions/ [00:16:59] ? [00:17:06] halfak: I'm adding PAWS to R [00:17:09] err [00:17:11] YuviPanda: sure wassup [00:17:11] R to PAWS [00:17:13] YuviPanda, totally [00:17:15] bearloga_: let me come over [00:17:16] wander to my desk [00:17:19] NO, ME [00:17:20] Was gonna ask WTF [00:17:21] ok [00:17:22] lol [00:17:22] (fine, him) [00:17:33] have both! [00:17:35] +1 for R. [00:17:36] at my desk. I'm not moving. [00:17:40] I'll probably use that. [00:17:46] halfak, on the R front, check your email about the upgrade. [00:17:48] I want data.table in python [00:17:59] Ironholds, re. ORES? [00:18:24] halfak, re upgrading to R 3.2.3 on stat1002, which'll mean ggplot 2.0.0 [00:19:28] Oh! Cool. I usually don't R on the server though. [00:19:37] Takes too long to push the pixels back to my screen! [00:20:00] halfak, fair! [00:20:03] they came to my desk [00:20:04] still the king [00:20:41] What!? The pixels go to your desk before mine? Damn bureaucracies. [00:24:01] halfak, no, the YuviPanda :p [00:25:08] halfak: so once I get R installed I can allow you to mix R and Python [00:26:03] That's what I was hoping for. [00:26:30] Once I'm ready for a datatable, I'm ready for ggplot. [00:26:53] Right now, I mix R and python by outputting data into files. [00:28:53] I'm still giddy from the possibility of having R >3.0.2 in the next 24-48 hours [00:30:09] bearloga, what do you get from the upgrade? [00:35:15] halfak: mainly: C++11, ggplot2 2.0.0, rvest (for web scraping), devtools, any package on GitHub but not yet on CRAN that you want to install [00:35:36] Oooh. that last one sounds pretty darn good. [00:35:48] I haven't looked at ggplot2 yet [00:39:09] halfak: oh and parallel computing! that'll be in there too. but especially not having to see "____ is not available for R 3.0.2" when trying to install a package you've been playing with locally and then want to use on stat1002 but oh sorry you can't [00:41:12] halfak: basically any package that's been created or updated since 2013-09-25 is not guaranteed to work, and probably won't. but with 3.2.3 THEY WILL [00:42:22] Always wondered what was breaking all these packages with high regularity [00:42:50] Not enough to google about it. I just needed them to work and when they wouldn't, I didn't really care why. [00:43:11] R tends to enter my flow much closer to THISNEEDSTOBEDONENOW than a lot of other systems I use. [00:43:13] you'd have had it with the notebook stuff when I had finished it [00:43:25] but this is fairly easy anyway (importing a few packages) [00:43:39] bearloga: do you know when you'd want me to do it? I want to block off a few hours afterwards just in case... [00:45:01] YuviPanda: tomorrow morning should be a-ok if Ironholds is ok with that [00:45:21] bearloga: Ironholds can either of you decide and put an event on my calendar? :D [00:46:34] yeah that sounds good [00:46:42] (tomorrow morning) [00:47:07] 11AM? [00:48:28] yup yup [00:49:05] kk [00:49:07] let me create event [00:50:26] done [00:57:15] these days a lot of my life is spent compiling stuff [01:00:16] R doesn't support installing things through https? [01:07:19] o/ Ironholds [01:07:25] berkman talk confirmed for Feb 2nd. [01:07:26] YuviPanda, it...should? [01:07:33] halfak, awesome! Looking forward to hanging out! [01:07:38] apparently not by default, or something of that sort [01:07:38] :D [01:07:38] Gonna take you to my favourite breakfast spots [01:07:51] Now to decide if I'm writing a *new* talk or upcycling an old one. [01:07:52] anyway, I've R setup properly for notebooks, pushing through now [01:07:54] YuviPanda, yeah, not by default because HTTPS support for the mirrors is still not uniform [01:07:58] right [01:08:08] so I'll have to write an .Rprofile thing that points to a specific mirror [01:08:10] and enforces https [01:08:42] I haven't fixed the ability to mix R and python yet but that's next [01:09:44] YuviPanda, use the RStudio mirror [01:09:50] it's got internal geolocation\ [01:10:04] ok! [01:11:41] jesus fucking christ our network is slowwww [01:12:06] YuviPanda, says the operations engineer in the most blatant case of {{sofixit}} ever [01:12:22] Ironholds: I don't even know where the routers in the office are [01:12:42] just unplug anything with wires until it stops [01:12:50] This is why they don't let me in the server farms any more [01:12:54] halfak: I think I'll have working R and R+Python integration (calling into R from Python) by tonight [01:13:19] YuviPanda, there's a pretty good mapper in the other direction too, but don't use rPython [01:13:28] THAT uses JSON as the transport mechanism :/ [01:13:28] http://eneskemalergin.github.io/2015/10/01/R_Magic_with_IPython/ [01:13:34] so that's rpy2 [01:13:38] Damn man. I'm still working on emails and followout reporting from the all-dev-hands [01:13:41] but there's one that uses Rcpp + the Boost Python headers [01:13:54] I almost wrote some code with Amir earlier. [01:14:10] Ironholds: oh, interesting. by other way you mean calling python from R? [01:14:44] halfak: I just prioritize badly. I don't have an actual bed to sleep on tonight yet, for example [01:14:49] because SHINY [01:15:01] plus it's far easier to loom over Ironholds and bearloga in person and get them to test this notebook stuff [01:15:18] YuviPanda, indeed! [01:15:24] If you get an an airplane now, you could be sleeping on my couch by midnight. [01:15:30] :D [01:15:35] halfak, offer taken, SF is weird [01:15:41] * Ironholds wanders to airport [01:15:45] \o/ [01:15:57] (a joke...partly. The bit involving me moving was a joke ;p) [01:16:20] I've done things like that in the past [01:16:21] although Dan Jarratt keeps preaching The Good News About GroupLens to me [01:16:31] however since paying rent in SF I'm too poor to afford these things [01:16:33] lol that's great. [01:16:56] How did you come to meet Dan Jarratt? [01:17:49] YuviPanda, that's the exact reason to do it. Just put your non-bed on AirBNB. [01:18:05] halfak: I don't even legally live where I live so that's not going to work [01:18:05] I'm not even kidding, I bet you could pay for your ticket in a few days even without a bed. [01:18:19] Yeah good point. [01:18:33] * halfak is not positive he has ever stayed in a 100% legal airbnb. [01:18:55] I have been instructed about how to lie to landlords on many occasions. [01:19:06] if I do get kicked out of the country I want to get kicked out of it for something more glorious [01:19:20] halfak: yeah, but I've to get people to lie to the person who asks me to lie to the landlord... [01:19:41] lol [01:20:15] two levels [01:20:46] I hate the fact that I'm network bottlenecked on working [01:20:56] maybe that'll force me to get off the fucking guest network [01:21:44] the bug preventing user accounts with '(') in them from working on PAWS persists despite newer ipython notebook release [01:21:48] I need to track this down later [01:23:22] halfak, twitter, and now we're working on a paper together [01:23:25] I don't know what happened [01:23:31] actually that reminds me, I need to poke him about the paper [01:24:18] :) Talked to notconfusing? He's been their for a while now. [01:24:51] Ironholds knows most of the wiki-side of GroupLens [01:25:10] is that Max? [01:25:21] Yup [01:26:25] I think I've seen him about a lot but never directly met him [01:28:12] * halfak assumes all wiki-tech-scholars know each other [01:30:24] http://www.alyz.com/rpy2.xhtml [01:30:32] examples of using ggplot2 from python via rpy2 [01:33:58] rpy was pretty annoying to use the last I tried. [01:34:06] That was in 2007 [01:34:13] halfak: yeah, this is rpy2 [01:34:15] Only 9 years ago HOLY CRAP [01:34:26] :D [01:34:30] * halfak oils his knees [01:34:34] the code I see looks clean enough but I've never written any R so [01:34:43] omg this push has taken like fucking 30mins [01:34:51] and before it's done I've managed to get rpy2 to actually work locally [01:34:53] jfk [01:40:25] goddamit [01:40:36] turns out debian jessie's packages for R are a bit older than i'd like [01:40:52] Ironholds: bearloga so even if you had waited for otto to upgrade the machine you'd have only gotten the version I had just shown you [01:45:49] YuviPanda, 3.1.1 is still pretty good [01:53:38] wait, stat1002 is showing up as Ubuntu 14.04 trusty to me which has 3.2.3 https://cran.rstudio.com/bin/linux/ubuntu/trusty/ am I just being silly and misunderstanding something? [02:02:43] bearloga: that's the R repo [02:03:02] bearloga: http://packages.ubuntu.com/trusty/math/r-base is the official ubuntu repo [02:03:04] which is older [02:03:06] and what you have now [02:03:44] bearloga: and yes, me upgrading R is basically importing the packages from that URL onto our internal mirror so stat1002 can use it [02:30:30] halfak: as an update before I leave: I've R and rpy2 setup, but need to update to a newer version of R, and also figureout kernel registration properly. still, productive day! [02:30:37] * YuviPanda goes afk now [02:31:16] Have a good night, YuviPanda [02:31:20] Good luck with the bed! [16:35:26] Greetings [16:35:37] * guillom just finished reading http://www.lemonde.fr/sciences/article/2016/01/11/wikipedia-la-connaissance-en-mutation_4845347_1650684.html [16:36:23] hey guillom :) [16:36:44] Hello Ironholds. Still in SF? [16:38:03] I saw there was some tattooing going on over the week-end :) [16:44:52] Ironholds, you'll never be able to take a stats test again withour cheating. [16:45:07] guillom, I am! [16:45:16] halfak, yeah, Mikhail's last words before he went under the gun were [16:45:22] "wait, but what if I go do my PhD?" [16:46:24] PhDs are overrated. Any idiot can get one. I speak from experience :p [16:46:54] guillom, good, I want one :P [16:47:02] heh :D [16:47:16] I'm thinking a couple of years more industry work and then moseying on out of the city to somewhere with a cool programme, but who knows [16:48:57] * halfak has never seen "mosey" spelled out before [16:49:35] Weird language! [16:49:58] hehe [16:52:30] the parental units rather like their PhDs [16:53:05] I do like it when people call me "Dr. Halfak" [16:53:19] No one ever does that seriously though [16:53:38] If you would teach at a college... [16:53:47] ...then he would be Professor Halfak [16:53:52] and still not Doctor Halfak ;p [16:54:06] BDR gets _really_ pissy if you call him Doctor, amusingly [16:54:10] "DOCTOR RIPLEY IS MY WIFE" [16:54:58] Depends on the uni. Where my parents teach is Dr, Br, Fr, Mr, or Ms. Where I went was Professor or Dr depending on what area [16:56:13] Br & FR? [16:56:19] Brother and Father [16:56:23] Broctor and Froctor [16:56:24] Oh [16:56:50] I suppose that it would be Boctor or Foctor [16:56:51] the college had one of the largest bendictine monistaries attached to it [16:57:02] I went to one of those too. [16:57:19] http://www.css.edu/ [16:57:41] the parentals are at http://stvincent.edu [16:59:50] No Sr for Sister? [17:01:56] You should teach in Germany, where you might occasionally be addressed as "Herr Professor Doktor [name]", at least in formal writing. [17:02:07] I never met a nun in my 18 years being on campus [17:02:12] Well that's a mouthful. [17:03:44] Guerillero, interesting. I had the opposite experience. Never met a Brother on campus -- that I know of. Might have mistakenly assumed "Father" status of some. [17:05:48] Guerillero, I feel like generalising to universities from a university with a benedictine monastery providing the teaching staff maay be a scientific mistake [17:12:10] I open IRC and see something about nuns on campus [17:12:31] Fun conversations today [17:12:58] It all started with Ironholds not being able to get a PhD anymore because he tatooed a cheat sheet on his arm. [17:13:03] ;) [17:13:32] ah that [17:15:58] YuviPanda, I'm going to murder sphinx [17:16:06] Also, probably submit a pull request [17:17:31] If a sphinx is half human, is killing a sphinx homicide? [17:17:46] Is it a demicide? [17:17:47] I've dumped 5 hours into a relatively simple "WHY CAN'T YOU FIND MY ATTRIBUTES?" [17:18:19] I've built a simple environment to replicate the problem and I can't [17:18:33] So, now I get to figure out what sort of wonky thing is happening within 'revscoring' [17:18:43] guillom, both, I suppose. [17:19:22] guillom, seems like sphinx should be endangered, so I'd be a poacher too. [17:19:39] halfak: Do you absolutely need to fix it today? Otherwise, sleeping on it usually helps. [17:19:47] (In my c [17:19:52] case) [17:19:58] guillom, this is my 3rd attempt. Slept on it twice already [17:20:01] :/ [17:20:21] Hence the considerations for drastic measures [17:20:29] Tried rubber-duck debugging already? [17:21:21] Even more, I rebuilt the entire thing from scratch to replicate the problem. [17:21:31] Then went back to my first assumptions about what I was testing. [17:21:41] I think I might need a non-rubber duck to talk to about it soon. [17:23:18] btw halfak I'm going to steal some of your bibliography from https://meta.wikimedia.org/wiki/Research:Automated_classification_of_edit_types [17:23:50] \o/ [17:24:09] there is no theft in open knowledge :) [17:24:16] :D [17:24:38] (I suppose one can steal credit, but we have other words for that.) [17:32:38] halfak, is Jimmy Lin a meaningful name to you? [17:33:09] kjschiroo, I don't think so. I meet a lot of people, so I could have forgotten. [17:33:59] okay, altiscale is asking me if I am in his class and I just wanted to make sure I wasn't getting access by some odd way before telling them I'm not. [17:34:06] Oh wait. Looks like I have an old email here that I forgot to reply to from Andre Wiggins [17:34:44] Was about this paper: http://www.umiacs.umd.edu/~jimmylin/publications/Weissman_etal_JCDL2015.pdf [17:35:29] Woops [17:35:37] * halfak goes to a meeting he was ignoring [17:35:47] Jimmy Lin was mentioned in the wikitech-l / wikimedia-l thread "Obfuscating IP addresses on history pages" in April 2015. [17:36:51] About this other paper: http://www.pensivepuffin.com/dwmcphd/syllabi/infx598_wi12/papers/wikipedia/lieberman-lin.YouAreWhereYouEdit.ICWSM09.pdf [17:41:01] bk [17:45:33] I used recommend.wmflabs.org again at a wmit course [17:52:27] kjschiroo, just sent you an email with names to drop [18:02:41] guillom, warning that we just hired another Guillom ;p [18:03:42] I can barely handle one of them [18:04:13] Ironholds: backfilled? [18:04:27] i know its having mysql replicaiton problems, but i'm not sure it needs to be backfilled, the replication is just lagging [18:04:28] ja? [18:05:11] ottomata, from the perspective of a database consumer it's backfilling, but fair point; I'll clarify [18:05:41] mandatory grumble of nobody told me this, I found out by chance when I was doing some work late last night and given automated reporting tools that use those dbs an announcement should probably go out in future ;p [18:06:18] its not something we've managed, jaime usually deals with the mysql replication, and he's on vaca this week [18:06:28] i'm looking at it, and it is very funky custom replication [18:06:40] slave machine is pretty busy, but not sure if that is normal or not [18:08:31] ottomata: thank you for working on the logging of the IPs issue. :-) [18:09:08] jaa, sho thang [18:09:13] lemme know if it works if/when you can find out [18:09:49] yeah, just saw your comment about checking it. will look into it in a bit, I'm sure Tilman is already on it. :-) [18:10:54] good morning all! o/ [18:11:05] hi bearloga [18:11:07] well, with the replication lag atm, it might not be viewable yet :/ [18:11:19] ooki, let's give it some hours ottomata. [18:13:32] Ironholds: I was certain I'd never need to be on the [[Ambiguous names]] pages. [18:13:42] Emufarmers: :> [18:18:31] guillom, my scalar nature is still sacrosanct [18:19:28] Nerd :p [18:19:49] you're only saying that because you're in a deque and not a set ;p [18:21:10] I was told last week I wasn't special anyway. [18:21:31] awww [18:21:34] * Emufarmers hugs guillom. [18:22:36] guillom, by whom?! [18:22:42] Emufarmers: It was during the all-staff. A guest speaker told us "You're not special". [18:22:42] which liar put this idea in your head? [18:22:49] oh, that. yeah, that was true [18:22:54] but as an individual you are special [18:22:59] our group dynamics are just not unique [18:23:41] I don't think that's what he meant. [18:23:58] then screw him [18:24:04] you are special and wonderful and great [18:24:32] The good news is that the new Guillaume will hopefully benefit from me teaching people how to pronounce our name :D [18:25:29] it's gee-yom, everyone knows that [18:25:56] okay, everyone who went to the Lycée Français Charles De Gaulle de Londres knows that [18:26:19] I learned that, how to swear, and the name of the different tenses in French. I can't actually SPEAK french but I can describe French speech nicely ;p [18:26:35] héhé [19:20:08] Ironholds: bearloga where are you two? [19:20:34] YuviPanda, we got dragged into another meeting [19:20:36] YuviPanda: we're in R31 for weekly meeting. It's almost done :\ [19:20:38] bearloga, told you we were wanted ;p [19:20:52] well, I have my portal meeting after this potentially. I'll find out. [19:21:06] ok! [19:21:19] Ironholds: bearloga I talked to ottomata about the upgrade and he won't hate me [19:21:29] YuviPanda: *high five* [19:21:30] so it's all good. I'll wait for you two to show up so we can test and upgrade smoothly [19:21:42] ^___^ [19:21:57] cool! [19:23:13] YuviPanda, yeah I'll be in this meeting :/ [19:23:25] Ironholds: do you know for how long? [19:31:04] YuviPanda, another 30 minutes [19:34:55] Ironholds: ok. we upgraded it on stat1003, bearloga is testing it right now, I'll upgrade on 1002 and 1001 after confirmation that it's all good [19:36:39] there's a 1001? [19:38:50] yeah [19:41:28] YuviPanda: i doubt its needed on 1001 [19:41:37] 1001 is mostly a webserver for datasets.wikimedia.org and stats.wikimedia.org [19:41:39] and maybe a few other things [19:41:43] most folks don't have access there [19:42:43] ottomata: ok [19:58:06] ottomata: btw, I've upgraded R on both machines and bearloga is checking stat1002 now [19:58:31] yeehaw [19:59:32] ottomata: labs is on a lot of fire recently, so I won't get back to notebook stuff just yet [20:09:15] Oooh. I just found http://yuvi.in/blog/ [20:09:22] And ORES downtime gets mentioned! [20:09:49] YuviPanda: aye ok, just saying that icinga has a host down alert for an21 [20:09:52] because of that IP thing [20:13:40] thank you again YuviPanda! :D [21:01:44] « Chapter 3 begins: "Science is the systematic use of theoretical and empirical methods to try to increase understanding of some set of phenomena or events." If the third chapter begins by defining science, I can imagine why my professor thought the first two were unnecessary. » [21:02:05] https://mako.cc/copyrighteous/groups-interaction-and-performance-by-joseph-edward-mcgrath [21:02:57] lol [21:06:10] * guillom ponders ordering the book for the WMF library. [21:08:32] * guillom adds to list. [21:10:45] ottomata: yeah, that needs a facter fix. file abug? [21:53:50] o/ ragesoss [21:54:10] We're finally picking up a bug I talked to you about at WikiConfUSA [21:54:12] For ORES [21:54:26] Specifically, to allow you to hard-code features/data for a particular scoring. [21:55:01] So that you could find out how the score changes when you, e.g., add another reference or category link. [21:55:17] I'd like to talk to you about how you'd like to provide this data. [21:58:04] halfak: hi! [22:05:46] I don't know. FirstI'd like to be able to see the actual feature set (in as human-understandle a form as you've got) for a given revision. [22:05:56] Is that part of the plan? [22:11:14] oh right, you proposed doing that with a ?features flag in issue #100 [23:00:32] ragesoss, hey! [23:00:36] Sorry to disappear. [23:00:48] I was just about to got to a meeting and forgot to check the clock. [23:01:24] So, yeah, I can give you a example set of features for an article quality prediction. [23:56:10] ragesoss, will have to get back to you. When i do, I'll have some example features and I'll lay out the proposed GET/POST arg patterns. :) [23:56:23] halfak: sounds great