[00:01:09] What else will it do other than store revision scores? [00:03:11] https://github.com/wikimedia/restbase/blob/master/doc/images/RESTBase.svg [00:03:41] halfak: the first use case will be storing HTML and related JSON metadata (data-parsoid, data-mw) [00:04:15] we are also talking with the mobile app folks about storing their various HTML & metadata variants that are all derived from the regular content [00:04:22] Seems like a very different use-case [00:04:29] Maybe not conceptually [00:04:46] the type of content is different, yes [00:05:19] Seems like the usage would be very different too. [00:05:19] but it's ultimately all storage / caching, typically with some service to call when something is not found in storage [00:05:23] Also the origin of the data. [00:07:08] restbase focuses on the storage, plumbing & API parts [00:07:45] the actual work of generating content is all farmed out to services [00:08:24] Makes sense. I look forward to the implied directory of metadata. [00:09:03] *nod* [14:17:16] Hey Ironholds [14:17:30] morning halfak :). How goes? [14:17:30] How's the morning on the East Coast? [14:17:55] Snowy and dark. I'm working on the League of Legends dataset this morning. [14:18:18] Heavy metal music is a good way to address dark snowy weather. [14:18:19] aha! [14:18:41] eh, it's okay. I'm just watching some teewee and updating the pageviews set [14:18:49] teevee? [14:19:05] oh teewee [14:19:23] "A girl who will settle for low standards." [14:19:26] http://www.urbandictionary.com/define.php?term=Teewee [14:28:22] ^ Ironholds [14:28:26] What is teewee? [14:28:30] a TV! [14:28:33] Oh [14:28:34] heh [14:28:39] Internet has failed me [14:51:47] halfak: what is the League of Legends dataset? [14:52:38] Oh! Ironholds and I wrote a paper recently about a strong regularity in human behavior online. For this, we pulled in datasets from a bunch of different systems. League of Legends was one of them. [14:53:01] The data came from a third-party add-on. Duowon i believe. [14:53:18] was fun! [14:53:26] most fun paper I've written thus far, probably because o'the people. [14:53:32] Emufarmers, http://arxiv.org/pdf/1411.2878v1.pdf [14:54:29] I take pride in the fact that people still seem to like me after writing a paper with me. [14:54:30] :) [14:54:36] Can be a very stressful experience. [14:54:41] :> [14:54:43] > Our own experience with the game suggests [14:54:50] ;) [14:55:41] Emufarmers, hey, HCI people need to unwind too [14:55:59] halfak, yeah, so far you and scott are my favourite to write with [14:56:17] Heather/Brent/Shilad et al weren't stressful but that's because I didn't really have anything useful to contribute other than datasets. [14:56:29] Early next year I plan to find out if I hate Dario ;p [14:56:38] Woot [14:56:52] Wait... we have collaborated with DarTar before :P [14:57:30] yeah, but that falls into the Heather/Brent line above :D [14:57:38] I didn't really do much for AFT5. This time I wrote actual text! [14:57:48] and for Scott's collaboration I did half the research and drafted the rebuttal! [14:57:54] (heck, I'm first author on that paper) [14:58:23] I didn't realize that. I'm really looking forward to it :) [14:58:39] realise what? [14:58:45] You were first author. [14:58:52] aha [14:58:56] it's gonna be...interesting. [14:58:58] It's a milestone as a scholar. :) [14:59:04] I'm hoping it'll prompt a lot of followups [14:59:16] like, the note is just to set the stage for looking at editors through a mobile/desktop lens. [14:59:29] which is what I want my contribution to this field to be, I've decided. [14:59:33] Geography + platform variations. [14:59:40] Say Ironholds, how fast can you geolocate 10k countrys? [15:00:16] 64 microseconds [15:00:21] I told you my code was good ;) [15:00:29] I got to 244 with python [15:00:33] neat! [15:00:44] :) I just switched to pypy. No code change. [15:00:50] mine might actually be a bit faster now; In the initial tests I forgot to unload the dataset at the end of the function call [15:00:53] pypy == Just-in-Time compiler [15:00:55] this inflated memory and (methinks) parse time. [15:01:00] let's see what it does now. [15:01:04] you using median, mean, what? [15:01:27] Method: Run it a few times and fell out a mean. [15:01:32] *feel [15:01:47] ...oh wow. [15:01:53] Actually 244 is more like upper 90% [15:01:57] okay, yeah, so when I said that unloading the dataset would make it faster.. [15:02:00] lower 10% is 230 [15:02:29] https://gist.github.com/Ironholds/475dcc6d9742b4af8338 [15:02:33] wingswednesday! where is this headed? [15:02:46] turkeythursday! [15:03:00] sorry, miliseconds rather than micro [15:03:10] Oh yeah. I assumed as much. [15:03:11] 100 evals, maximum of 14ms parse time [15:03:16] yusss [15:03:21] actually I think I can get it even faster [15:03:28] :p [15:03:32] just waiting on a reply from MaxMind to confirm if my hypothesis is correct [15:03:45] (that the IPv6 files also contain the IPv4 data) [15:03:59] I'm thinking of making this into a generalised rgeoip library and throwing it up on CRAN [15:04:50] Ironholds, sounds like a fine idea to me. [15:04:58] yerp! [15:05:02] that and uap-r [15:05:06] * Ironholds connivsed [15:05:09] *connives. Bah. [15:05:26] also, I want today to be a national holiday [15:05:34] The Day It Was Faster To Do Something In R Than Python [15:06:51] But you aren't doing it in R [15:06:57] Python has C modules too :P [15:07:03] Speaking of which ;) [15:07:08] hahah [15:07:16] so go implement pygeoip! Oh wait, it exists. In pure python [15:07:18] and it's slooooow :D [15:07:27] You could grow your user-base substantially... [15:07:40] Actually, it's pretty goddamn fast. [15:07:41] if I ported it to Python? [15:07:46] oh yeah, impressively fast! [15:07:50] just not as fast as Rcpp ;p [15:08:02] Oh no. Don't port to python. [15:08:08] then..? [15:08:13] Just package it up as a python c module too. [15:08:14] :) [15:08:20] oh, gotcha. Can you do that? [15:08:24] Yup. [15:08:29] huh. I'll look into it! [15:08:35] https://docs.python.org/2/extending/extending.html [15:08:36] Like I need more software to support ;p [15:08:38] <3 [15:08:43] Fair point. [15:08:50] I still need to push the updated WikipediR to CRAN. Bleh. [15:09:29] wingswednesday, what's a word for leftovers that starts with "f"? [15:10:03] fragmentfriday [15:10:20] hmmm [15:10:20] fridgefriday? [15:10:22] soporificsaturday [15:10:29] and finally stomachpainsunday [15:10:47] sundaesunday [15:11:27] I hope y'all have fun :) [15:11:35] I'm going to be sat at home reading. [15:13:24] I'm going to be juggling family. [15:14:06] I'm sorry, Aaron, I can't let you do that. [15:14:13] pew pew [15:15:04] heh [15:18:34] halFAQ, more Twin Cities hip-hop for you - https://www.youtube.com/watch?v=Y_x6M3UXq9M [15:18:42] I really, really love the Doomtree collective. Y'all do some good rap. [15:19:55] or https://www.youtube.com/watch?v=toELT1uWDKU which is a collective song performed on bikes. [15:21:25] * halFAQ clicks [15:21:51] No more FAQ production by HAL? [15:23:02] Q: Open the Pod bay doors, HAL? [15:23:06] A: ... [15:23:35] the rappers are, in order, Sims, Mike Mictlan, Dessa, P.O.S., and Cecil Otter [15:23:45] can I just say how much I appreciate that there's a rapper called Cecil Otter? [15:25:31] I might be missing the significance. [15:25:54] Also, if you're looking for some good winter weather music, I recommend Cult of Luna: https://www.youtube.com/watch?v=vKJebgQJMf4 [15:27:38] oh, I know Cult of Luna! [15:27:41] and it's just a ridiculous name. [15:27:52] Cecil Otter. Someone got into hip-hop with that name and got /signed/. [17:44:21] morning DarTar [17:44:28] howdy [18:08:33] looking at the incoming RA candidates, this joke about stealing ellery from fundraising becomes less humour and more "we should do that"> [18:09:04] * halfak draws the blinds [18:09:09] shhh [18:10:07] heh [18:10:19] no, seriously, one candidate cares about us so much they got the name of the company wrong on their cover letter. [18:12:08] not even in a minor way [18:12:14] they open with saying how much they'd like to work at github. [18:12:41] I'd like to work at github too, seeing their intern salary rate, but that's not why I came here ;p [18:20:40] ^ lol [18:20:51] Are these people making it through screening? [18:25:15] I don't know, actually [18:25:27] and oh joy, it's yet another ass trying to shellshock our servers [18:25:34] I really dislike a chunk of our readers [19:37:18] tnegrin, do we have our 1:1? [19:49:49] Ironholds: yes -- [19:49:59] we have our 1:1 15 minutes ago -- my bad [21:04:16] Ironholds: yt? [22:04:56] Anyone in the office and able to see DarTar from where they are? [22:05:12] * halfak scrolls down the list. [22:05:52] gwicke, any chance you can see DarTar from where you are sitting? [22:06:04] Deskana, ^ [22:06:24] bmansurov, ^ [22:06:33] halfak: I'm working from home today. [22:06:40] Gotcha. Thanks anyway :) [22:06:46] halfak: no, I'm not in SF [22:06:59] * Deskana scratches his head [22:07:00] If anyone can throw something at him and tell him he is supposed to me meeting with me, I'll buy you a beer. [22:07:03] Thanks bmansurov [22:07:18] halfak: I can tell him that for beer [22:07:20] ;) [22:07:23] :) [22:08:00] halfak: I can go looking for him [22:08:08] he's not in my normal line of sight [22:08:12] No worries. Not work you getting up. :) [22:08:14] halfak: Katie Horn said she has no idea where he is. [22:08:20] I've sent the sms's [22:08:26] "K4-713: Deskana: If he's here, he's hiding really well." [22:08:28] :D The great Dario hunt of 2014 [22:09:35] he's apparently in some mobile off-site for the day [22:10:26] that's the info I got from Leila [22:10:38] Thanks gwicke [22:10:45] Much appreciated. [22:10:48] :) [22:10:52] yw ;) [22:55:34] * Ironholds yawns [22:55:47] * YuviPanda hands Ironholds some coffee [22:56:53] I don't drink caffeinated drinks [22:58:00] oh [22:58:02] Ironholds: [22:58:03] surprising [22:58:07] wanna chat for 10? [22:58:10] totally! [22:58:17] just send me an invite :) [22:58:40] https://plus.google.com/hangouts/_/wikimedia.org/okeyes-tnegrin?authuser=0 [22:58:51] done and done