[03:49:27] 10Quarry: Show desktop notification when a query is done - https://phabricator.wikimedia.org/T124625#1960941 (10APerson) 3NEW [16:24:28] o/ [16:31:30] \o [16:32:44] \o/ [16:34:25] * guillom has no meetings today, plans for a day of uninterrupted Python. [16:35:30] you mean python interrupted only by "where is the documentation for this?" "there is no documentation for this is there" and "oh for gods sakes you call THAT a unit test?!" [16:36:29] * guillom has hot tea and nice soundtrack music and will not respond to the trolling attempts of the holds of iron :p [16:36:58] :D [16:37:01] I think there's also a joke to be made about the GIL... [16:37:12] GIL? [16:37:18] global interpreter lock [16:37:21] guillom, naw, trolling would be "Python? Oh! Slow C!" [16:38:09] heh [16:38:26] but seriously, I am glad you have some time to work :) [16:38:37] I remember programming in C. It was fun but a long time ago. [16:39:25] I don't remember programming in C, but I do remember debugging in C. [17:00:49] I programmed in C at this very job [17:01:08] apergos, indeed! I've seen the C bindings into the geolookup DBs [17:01:13] I am so, so sorry [17:01:22] those dbs and their APIs are a pile of crap [17:01:29] I didn't do those [17:01:34] someone else has those [17:01:36] I had to integrate C++ with them so: lucky you! [17:01:44] they're literally Perl programmers writing C. [17:01:48] hahaha [17:01:55] that would be the worst of both languages [17:01:56] and the issues page is full of interactions like [17:02:14] "so, how do you unprotect and delete an object of this class?" "oh, you can't, just terminate the program so it collects automatically" [17:02:33] meeting time! [17:13:44] hi everyone. [17:13:51] hey schana. [17:13:59] schana: can you check https://phabricator.wikimedia.org/T124511 ? [17:14:08] looking at it now, lzia [17:14:29] perfect, thank you, schana. and if chatting over Hangout helps, please let me know. :-) [17:26:03] g'morning lzia & schana [17:26:34] g'morning halfak. [17:27:32] good morning halfak [17:32:13] * YuviPanda waves at schana [17:32:23] I setup a meeting for us a few days down the line, hope that's ok [17:33:52] YuviPanda: that should be fine [17:34:12] cool :) [17:34:19] feel free to ask me any questions in the meantime! [17:39:33] morning folks [21:48:27] DarTar, I've been thinking about some KPIs for revscoring. http://socio-technologist.blogspot.com/2016/01/notes-on-writing-wikipedia-vandalism.html [21:48:50] I think that the two thresholds that I detail in the blog post are good candidates. [21:49:01] halfak: cool, adding this to my reading list [21:50:02] I'm just finishing the implementation of these statistics in the revscoring library :) [21:50:03] what’s being optimized for when writing these tools would make a really important paper [21:50:13] DarTar, +1 [21:50:24] Hoping to get that in the KDD paper -- though that will be focused on Wikidata. [21:50:44] Might be worth writing a work/process focused paper for CSCW [21:50:47] yeah that’d be nice, but it could also make its own standalone, *meta* paper [21:50:52] yup [21:51:04] Just gotta make sure the ML folks read it ;) [21:51:15] ping Zeynep again [21:51:26] or write it on medium [21:51:34] Hmmm... That's not a bad idea. [21:52:46] I created a medium account with a new year resolution to start writing about stuff that is currently buried in handwritten notes on a couple of Moleskines [21:53:11] you should seriously consider this, especially if it has implications beyond a technical audience [21:53:12] Heh. The "if I only had time and a muse" queue [21:53:17] +1 [21:53:50] well, it’s about selecting appropriate venues for the audience you want to target, right? [21:53:53] * YuviPanda too should write more things [21:54:26] * DarTar nods [21:56:37] DarTar: btw, I'm meeting the Jupyter folks at berkley on thursday for lunch [21:56:45] DarTar: http://etherpad.wikimedia.org/p/jupyter-at-wikimedia is what I'm formulating [21:56:51] oh sweet [21:57:55] YuviPanda: looks great, if you talk to them start gauging interest for a potential joint announcement [21:58:17] I need coffee, brb [21:58:20] for which of the three things? :) [21:58:43] DarTar: I'm always worried about putting anything official sounding on anything :D maybe I should be less worried [21:58:46] * YuviPanda shall contemplate [22:02:19] 10Quarry, 5Patch-For-Review: Cannot download data from a query with Unicode characters in its title - https://phabricator.wikimedia.org/T123031#1963675 (10Krenair) 5Open>3Resolved [22:02:43] YuviPanda: it’s healthy that you worry about it as our Ops person :D but you should not worry if you want adoption and actual users. I think what we did with ORES is the right approach (soft release, incremental adoption, followed by a bigger announcement) [22:02:53] * YuviPanda nods [22:03:38] DarTar: from your perspective, rank these in order of importance: 1. notebooks for internal WMF use (with private data access), 2. PAWS (notebook on labs + publishing), 3. Notebook publishing on meta [22:04:56] 10Quarry, 5Patch-For-Review: Query counter increases but draft query is not accessible when window is closed and query doesn't have a title - https://phabricator.wikimedia.org/T101394#1963690 (10Krenair) 5Open>3Resolved [22:04:58] YuviPanda: 2, 1, 3 [22:05:13] hands down [22:05:31] hmmm [22:05:33] happy to give more context if it helps (or whenever you need to make a case) [22:05:40] Could definitely skip meta if we had a good directory [22:05:41] I'm mostly making a case to myself [22:05:52] to figure out which to prioritize [22:06:05] since I don't expect people not-me to be working on this for a while [22:06:06] yeah, halfak: I really see Meta as icing on the cake at this stage [22:06:16] while 2 is huge [22:06:27] 1 is big too, but smaller in impact [22:06:34] 10Quarry: Number of queries shown in profile is wrong - https://phabricator.wikimedia.org/T86512#1963694 (10Krenair) Sounds like T101394 to me - for some reason sometimes 'None' was shown, sometimes an empty string (which meant you couldn't use it). It no longer shows empty strings. [22:06:44] hmm [22:06:54] because with 2 you could drive a much larger number of users / research projects [22:07:02] hmm [22:07:04] true [22:07:07] and probably you will have to solve harder problems than for the internal use case [22:07:14] {{cn}} [22:07:45] well [22:07:53] I've solved most of the problems already for the external use case :D [22:07:54] in other words, I think 2 delivers the highest value to the Wikimedia movement, if you ask me :) [22:08:01] see? [22:08:02] and it mostly works, just needs some testing and whatnot [22:08:12] I spent last week solving the problems for our internal use case [22:08:34] mostly because I'm *so close* to having the internal jupyterhub setup [22:08:41] and then there I've a captive audience :D [22:08:42] alright, so you basically asked me that question to be able to say that both are ready to launch :p [22:08:45] no [22:08:51] * DarTar kidding [22:08:53] :P [22:09:00] but they both require different bits of work now. [22:09:06] and also different kinds of support from 'upstream' [22:09:06] right [22:09:16] so I guess mostly I'm trying to figure out which to sell them more on [22:09:28] I don't think they care about internal install so much - that's been done before so many times I bet [22:09:35] so I thnk they'll care about our labs install [22:09:37] and meta publishing [22:09:44] and meta publishing will be a big thing for *them* [22:09:44] yeah, that’s unheard of [22:09:48] the former [22:09:52] I doubt about the latter [22:10:04] because nbviewers do already a pretty kickass job [22:10:06] hmm [22:10:09] true [22:10:22] and meta publishing is so specific to our ecosystem that nobody will care out there [22:10:27] but, does having it on meta not confer any additional signalling? [22:10:29] hmm [22:10:31] that's probably true [22:10:48] it does, but smaller than announcing the world’s first open science platform for Wikipedia ;) [22:10:50] we could also have it on wikisource/wikibooks I guess [22:10:53] haha [22:10:55] that's true [22:11:19] DarTar, meta integration is part of the platform [22:11:36] halfak, it’s a feature request [22:11:51] I guess the question is what counts as MVP [22:12:13] and I’d submit that 2. delivers a ton of value without 3., the opposite doesn;t hold [22:12:35] if we have 3 out of the box I won’t complain of course :) [22:13:17] * DarTar really needs to get some caffeine now, brb [22:14:09] hehe [22:14:11] ok [22:14:19] I think I need a good endpoint to announce PAWS [22:14:24] the username bug bothers me [22:21:01] DarTar: halfak ok here's what I'm going to do: I'm going to sprint for a couple of days to try get our internal setup running, and then focus on solidifying labs setup + have a concrete plan for whta MVP should be [22:21:23] YuviPanda: <3 [22:21:37] in my mind, only things missing are: db access, fixing that username bug and switching out my homegrown nbserver code with nbviewer [22:21:53] curious about the feedback from Berkeley too [22:22:13] if you’re in the office on Fri I’ll try and find you [22:22:24] DarTar: these are the jupyter core devs, rather than berkeley people who are running a cluster [22:22:30] yeah [22:22:33] they just happen to be in berkeley I guess [22:22:38] yeah that’s what I meant [22:22:46] sorry, brevity [22:22:47] DarTar: yeah, apparently we made it to their knight foundation grant report [22:22:57] wat [22:23:12] that’s. interesting [22:23:36] yeah [22:23:46] I haven't seen what's in it [23:03:47] halfak: DarTar https://raw.githubusercontent.com/jupyter/jupyter-js-plugins/master/jupyter-plugins-demo.gif is what jupyter people are working on for 5.0 [23:08:08] Yeah! Atom/Sublimelike [23:12:24] yeah [23:25:15] YuviPanda: wow [23:25:25] that’s a pretty impressive gif [23:27:18] inorite [23:27:30] and it'll all be pluggable [23:28:43] it’s all pluggable [23:28:58] halfak: tweet your blog post and I’ll reshare it [23:29:46] teehee [23:41:24] DarTar, OK will do now. [23:44:39] Nooo [23:52:26] ?