[00:18:26] guillom: I bet you could make catscan do it [14:45:52] o/ [14:48:09] Good morning guillom :) [14:48:20] Your work on references is getting a bunch of notice. [14:48:28] hey halfak. How's is going? [14:48:36] It is? [14:48:50] I have people sending me email saying, "Check out this work that Guillaume is doing on references. It's related to the work that you are doing." [14:48:56] I'm like, "Yeah I know :P" [14:49:09] héhé [14:49:19] "My name is on that page too." [14:49:22] I figure there must be 10 people for every one who sends me that email. [14:49:42] What's reading pages got to do with anything. ;P [14:49:50] :) [14:49:53] reading is so outmoded [14:49:56] skim and assume! [14:50:37] Ok, now I'm relieved. If a bunch of people were reading that page, I'd feel guilty for not spending more time polishing it. But if they're just skimming through it, then I'm good :p [14:51:01] ha [14:51:02] :) [14:52:00] I'm wrapping up work on references, and starting to look into licenses now. [14:52:19] * YuviPanda is jealous [14:53:23] YuviPanda, what are you Jealous of? [14:54:25] I haven't particularly gotten to write code a lot lately. [14:54:31] and I think skills are atrophying [14:55:34] I suppose fabric and puppet aren't /code/ [14:55:56] I'm not writing a lot of code for work-work; the licensing metrics are going to be a lot of SQL queries. [14:56:37] halfak: yeah, it's mostly a lot of 'apply these vague principles already in your head and then beat it with a hammer till it works' [14:56:41] not much new [14:56:49] ok, so that quote can also just mean code :D [14:56:55] but I guess I'm not doing anything new lately [14:57:00] and that's making me feel somewhat itchy [14:57:09] YuviPanda, of you want to do Code code, I want to experiment with implementing some advanced diff strategies in C++. [14:57:32] so that's my other problem - I'm actually quite terrible at code like that [14:58:07] Meh. matter of practice [14:58:16] But what other kinds of things would you want to pick up? [14:58:37] so this is where you were talking :p [14:59:07] ToAruShiroiNeko, yeah. Swing in to say Hi to guillom and ended up talking about extraction work :) [14:59:18] Also, this is the office. [14:59:28] If I had a room at "the WMF", this is where I would sit. [14:59:41] ah [14:59:51] so this is full of cubicals [14:59:58] halfak: I'm not fully sure. I might pick up some Go next [15:00:02] * ToAruShiroiNeko pretends to be working here [15:00:02] And the actual office is mostly empty at this hour :p [15:00:09] although Rust is probably a better fit for diff related work [15:00:30] Yeah, I was talking to some people about Rust. It sounds like a very interesting programming paradigm. [15:00:35] guillom actual office has 100% less Aaron in it so not as interesting I imagine. [15:00:51] Heh. [15:00:51] But it's important to me that I don't venture too far from the scientific computing crowd. [15:00:54] So, python. [15:00:57] and R [15:01:03] And C++ modules for python. [15:01:13] Psssh. [15:01:18] c# ? [15:01:21] +1 [15:01:43] You'd hardly notice me in the office when I do visit. You'd just see a half-blur running between meeting rooms. [15:01:56] halfak unless I am in the meeting room [15:01:56] I could force you to have milkshakes! [15:02:02] Unless you're there at 6am with me. [15:02:08] or that [15:02:16] YuviPanda, :) wouldn't mind making that a visit tradition. [15:02:16] knowing me, that be what I'd plan [15:02:22] +1 [15:02:49] ToAruShiroiNeko, I have a pull request coming that's going to make our revscoring documentation actually useful. [15:03:02] halfak sure, I can look into it [15:03:12] I am very happy today [15:05:08] While I have a lot of smart people around: Do people have ideas about good ways to look into license templates? Right now I feel that the templatelinks table is my best bet (for things like: breaking down licenses by upload date of the files) but it doesn't work for many questions (like "How many licenses are there for this file"). [15:05:24] ToAruShiroiNeko, with any luck, I'll have it submitted in a couple of hours. [15:05:28] No rush to review today. [15:05:33] Basically, I want Structured Data already. [15:05:40] halfak: btw, a simple pip install -r requirements.txt doesn't seem to work anymore for ores :( [15:05:48] scipy fails to install because it can't find numpy [15:05:48] halfak I did fix the problem with my highlight btw [15:05:51] I"ll file a bug report [15:05:54] now I get an audio warning too [15:05:58] instead of lagging for minutes [15:06:00] that's my ORES time for the day, I think [15:06:02] guillom, one license template per file? [15:06:03] it was somehting surprisingly stupid [15:06:16] YuviPanda thats normal [15:06:17] YuviPanda, Inoright. [15:06:24] Seriously! [15:06:27] halfak: 1 to many licenses per file. [15:06:30] which part? [15:06:31] Why can't people manage their dependencies. [15:06:32] you need to install numpy seperately [15:06:46] guillom, but will each license have a template? [15:07:10] YuviPanda, just the numpy scipy issue. [15:07:21] guillom I would propose merging license templates [15:07:34] Couldn't find index page for 'numpy' (maybe misspelled?) [15:07:39] Download error on https://pypi.python.org/simple/: [X509] PEM lib (_ssl.c:2734) -- Some packages may not be found! [15:07:44] No local packages or download links found for numpy>=1.5.1 [15:07:49] I don't think that's an usual problem [15:07:58] halfak: yes. But for example, if I want to count how many files have one license, and how many have two, etc. I can't use the templatelinks table for that. I'd need to look at the dumps. But then, that wouldn't identify redirects etc. [15:07:59] because a pip install numpy works [15:08:25] guillom, what about categories? [15:08:26] guillom can you query what links here pages? [15:08:35] It seems like there should be categories for license types. [15:08:45] ToAruShiroiNeko, good idea [15:08:50] Issue filed! [15:08:53] * YuviPanda waves [15:09:01] I'd not do categories but rather focus on what links here instead as not every license has a custom category [15:09:01] YuviPanda, thanks for your oss citizenship. [15:09:04] https://github.com/halfak/Objective-Revision-Evaluation-Service/issues/46 [15:09:06] halfak: hmm; categories. Interesting idea. [15:09:10] * halfak feels lame for just working around this for a while. [15:09:13] someone should try it out and see if I'm the only one being affected [15:09:33] YuviPanda, this is a scipy bug isn't it? [15:09:36] YuviPanda I had the exact same problem [15:09:39] I don't know [15:09:42] halfak it is [15:09:52] I think, if it is the same problem I had [15:09:53] it could be a scipy bug, a numpy bug or a setuptools bug [15:09:55] err [15:09:56] pip bug [15:10:01] let me check my notes [15:10:26] ToAruShiroiNeko: whatlinkshere is a good idea. For some reason I thought it had the same source as templatelinks, but it's separate. Thanks! [15:10:28] nope I didnt write it down [15:10:40] templatelinks also arent always right [15:11:03] BTW halfak, dunno if you saw this: https://github.com/halfak/mwrefs/pull/2 (it's not urgent) [15:11:10] particularly if whats translcuded hasnt been "purged" yet [15:11:38] guillom, sorry about that. I thought I merged that earlier! [15:11:48] {{done}} [15:12:15] halfak: no problem; I fixed it locally for my needs because it was a simple fix. Just wanted to make sure your version was fixed too :) [15:12:18] Thanks! [15:12:45] thank you [15:12:48] :) [18:13:00] guillom: templatelinks should work except for very similar licenses (where the template structure is the same) [18:23:51] tgr: Thanks :) [18:25:29] milimetric: Do you take bug reports for lyra, or is it still considered a prototype (i.e. it's expected that things not work)? [18:27:01] guillom: sorry if I misled you, but Lyra is developed by University of Washington, and Jeff Heer's students. We were just demoing it. They take bug reports on github [18:28:04] milimetric: oh yeah, I get that; sorry for being unclear. I was wondering if our Labs instance was accepting bug reports. But you're right that I should probably upstream it anyway. Thanks :) [18:28:50] Oh! Is it a lyra issue or is lyra.wmflabs.org having trouble, guillom? [18:29:15] Well I'm not sure; I'm trying in their test instance now to figure it out :) [18:29:35] I suspect it's a bug in lyra itself. [18:29:53] Oh, ok, I'm happy to do any mainte ance on the site if it needs it. What's the issue? [18:30:20] Yup, the issue is also in their test instance, no probably nothing you can fix in our instance. [18:30:51] K. Paste the bug report once you file it, I'm curious [18:30:58] I have trouble loading raw data. [18:31:27] From a url or pasting it in? [18:31:32] pasting it in [18:31:41] Oh ok, what kind of data [18:31:54] That usually works for me, urls don't work [18:32:00] A CSV version of this table: https://meta.wikimedia.org/wiki/Tech/News/Subscribers [18:32:19] With the first column labeled as rting, and the two others as numbers [18:32:25] string* [18:32:43] K, and what happens, it just doesn't see the data? [18:33:41] Paste in what you're trying to paste. I've gotta go pick up my car but i can try messing with it after [18:33:51] At first it was importing something, and the "source" space had navigation arrows but no visible data in it. Now it just does nothing. [18:34:25] milimetric: https://phabricator.wikimedia.org/P735 [18:34:36] I'll try a few things. [18:34:41] I appreciate the help! [18:36:27] Worst case I can always make the graph by hand; I was being lazy and hoping to do it with Lyra so I wouldn't have to read the Graph docs :) [18:37:20] AH. Well I found /one/ issue. If you don't give the dataset a name, it just fails silently. [18:47:11] https://github.com/uwdata/lyra/issues/221 [18:50:09] Oh, interesting; Lyra is made with Angular. [19:46:30] Yeah, I dislike angular :) [19:46:39] guillom: meanwhile I added that paste as a dataset [19:46:53] as long as you name it, parse the first field as string and the others as number, it seems to work ok [19:47:03] what problem was it giving you? [19:47:35] milimetric: The data table doesn't show up for me. It does show up in the UW instance, but not ours in Labs. Let me try again. [19:48:00] weeird :) [19:48:08] angular - go figure. Yeah, I see the data table [19:49:44] Ah, it works now. Ah, yes, because earlier I was also trying to parse the first field as a date, since it's an ISO format. But it's not the most common so I guess it chokes on it. But yeah, as a string it works :) [19:56:24] yeah, that makes sense, I think the date parsing is mostly just new Date(value), nothing fancy, you'd have to convert yourself or patch it with fancy parsing.