[01:32:39] J-Mo: hey [01:33:16] J-Mo: sooo... I've been working on making ipython notebooks more easily accessible to people :) I currently have code that, when you run it on your local machine, opens up an ipython notebook for you, except it is running on toollabs' grid :) [01:33:27] this still requires people to have a toollabs account, but is much nicer, I believe [01:33:34] I'll work on making it super easy to publish them now [01:33:40] (and also cleaning up my script, etc) [01:33:43] legoktm: ^ [01:34:44] and then I'll make it run an R kernel and possibly taunt Ironholds [01:35:59] * YuviPanda rubs hands [02:25:38] halfak: ^ (for when you wake up) [10:20:15] ;-) [10:50:15] http://www.theverge.com/2014/8/19/6044403/derp-research-partnership-launched-by-reddit-twitch-imgur-fark-stack-exchange [10:51:15] EdSaperia, yep, we're aware of it [10:51:43] oh, this is five day old news, eurgh, sorry. [10:51:54] it's okay! [10:52:03] ugh, I should sleep. 6:51am is not a good bedtime [10:52:05] * Ironholds crashes [14:12:26] legoktm: halfak ping when either of you are around? [15:13:05] Hey YuviPanda. [15:13:17] * YuviPanda waves at halfak [15:13:35] halfak: saw my messages in this channel from yesterday night? [15:13:51] I did. [15:14:15] halfak: want to help test? :) [15:14:19] Sure! [15:14:48] halfak: 1. clone https://github.com/yuvipanda/labsbooks 2. pip install -r requirements.txt 3. run start.py [15:14:54] halfak: 2.7 for now, sadly [15:15:06] :((((((((((((((((((((((((((((((((( [15:15:18] Most people are using 2.7 [15:15:30] It's probably better to support that first. [15:15:41] halfak: it is [15:15:43] :) [15:15:52] But they *should* be using 3 [15:15:56] * halfak waves the flag [15:15:57] halfak: although, part of my reason was that paramiko didn't support 3.0, but it apparently does now https://github.com/paramiko/paramiko/issues/16 [15:16:05] halfak: so I'll convert it to be compatible with both later on [15:17:27] halfak: this is very early code, just 3h of work on it so far [15:19:16] * halfak installs paramiko [15:19:39] I need to talk to you about server architecture for revision scoring *as a service* in the near future. [15:20:16] halfak: \o/ YES INDEED [15:20:48] I've got AGWest (leading researcher in vandal detection) onboard. [15:21:11] halfak: \o/ coool. we should build it on labs [15:21:17] Yup. [15:21:34] and puppetize from the start so we can move to prod if later [15:21:39] also puppetizing from the start is good anyway [15:21:43] * halfak needs to learn to puppet now [15:21:52] "if later" [15:21:54] :D [15:21:57] Probably [15:22:01] Hopefully there will be a later [15:22:03] :) [15:22:07] I could do most of the work as well there, yeah. it's not *that* hard, it's just a different paradigm [15:22:50] Can I add you as a collaborator here: https://meta.wikimedia.org/wiki/Research:Revision_scoring_as_a_service [15:22:50] ? [15:22:53] halfak: sure [15:23:01] More names help me get people to sign on. [15:23:03] :) [15:23:49] :) [15:24:23] halfak: anyway, tried out labsbooks? :) [15:25:47] Just finished installing paramiko [15:26:38] halfak: I just pushed a new change, you should pull [15:27:04] Ack... I screwed up my virtualenv. I need to start over. [15:27:09] I haven't worked in 2.7 in a long time. [15:27:40] ah [15:31:17] halfak: you should also pull again, I pushed another change [15:34:33] * halfak just hosed his virtualenv again. [15:34:57] Apparently virtualenv wants to pick up 3.3 when I tell it 2. [15:34:58] Hmm [15:35:54] * halfak forces it to behave [15:36:53] Ahh! This is behaving weird. Apparently it SSHs for me. [15:36:59] it does, yeah [15:37:07] to toollabs [15:37:15] I presumed you had an account, but might be wrong... [15:37:31] not doing anything nefarious :) [15:37:33] I do. Also I get a keyerror looking for 'user' in labs config. [15:37:44] halfak: yeah, pull [15:38:25] halfak: if it runs for a few secs and seems to do nothing, open localhost:9500 [15:38:27] "Already up to date" [15:38:33] halfak: oh [15:38:42] Na. key error and the "y"s to my terminal [15:38:48] halfak: aaah, right [15:38:53] yes barf [15:38:54] oh [15:39:05] halfak: wait, so the 'keyerror' was a 'host not in known hosts'? [15:39:08] or a python key error? [15:39:18] halfak: oh wait, realized what was happening. fix coming [15:39:25] na. looking up 'user' in my ssh config it seems. [15:39:33] getpass.getuser()? [15:39:37] As default value [15:40:02] halfak: pushed [15:40:31] halfak: pull and try again [15:43:23] halfak: any luck? [15:45:06] Yup. Running. [15:45:18] It's a little slow to load up a new notebook, but it worked. [15:45:44] Are we in a shared set of notebooks? [15:48:10] halfak: no [15:48:19] halfak: the kernel is running s your own user, for your own notebook [15:48:28] Gotcha [15:48:48] halfak: I'll work on making them easily publishable. public by default again, with html and pdf converts [15:48:58] halfak: and add on more security, and speed up the proxy as well [15:49:23] halfak: it has numpy, matplotlib, scipy, pandas, pytables [15:49:28] halfak: anything more I should load up? [15:57:57] Main concern for me is not running the notebook on a labs machine as much as being able to publish easily. [16:01:27] Right now, I have a sequence of "lab notebooks" for each research project dumped on meta. [16:01:45] It would be nice if I could have a sequence of ipython labnotebooks that I could just link to from meta. [16:02:15] I'd be find with just clicking publish (or rsyncing) when I'm done working on one. [16:02:44] halfak: indeed. [16:02:55] halfak: Publishing I have a decent idea of how to build [16:03:05] halfak: and then you can 'publish' by simply copying a notebook to your ~/notebooks folder [16:03:29] It would be nice to have subfolders so I can split them up by project. [16:04:12] halfak: yeah, that too [16:04:26] halfak: and they'll be at tools.wmflabs.org/notebooks/halfak/ [16:16:09] Are you imagining a profilish interface like the one you made for Quarry? [16:16:14] YuviPanda, ^ [16:16:37] halfak: exactly. [16:16:43] :) [16:17:08] Except you edit it in your tools account rather than in a browser [16:17:18] It would be nice to have some tight integration with research documents on meta. [16:17:42] halfak: I wanna do the analysis of the edits per country on this so trying to build out the infrastructure [16:17:52] What do you imagine it being [16:18:04] One of the problems I was trying to solve earlier today was calling attention to research projects that are most recently active. [16:18:18] I saw that query ya [16:18:20] it would be nice to have notebook activity count as project activity. [16:18:28] Hmm [16:18:42] Yeah. The query was to demo what I was trying to do in lua templates. I think I've lost hope there. [16:18:48] Would be interesting to tie quarry queries and notebooks into them [16:19:57] Unsure how to tho [16:20:57] Agreed. [16:21:01] On both points [16:21:15] Could add a field... [16:21:39] But that is future! [17:00:06] * halfak goes out to get dinner [22:32:31] Speaking of dinner, this is NEAT: http://drawingcircles.tumblr.com/post/80247113245/the-non-traditionalist-dinner-review [22:32:43] And ultimately is an interesting story about UX experimentation.