[15:34:17] Hey hey science people [16:48:55] _o/ [17:07:12] Ironholds, joining us for standup? [17:08:05] halfak, in another meeting I'm afraid [17:08:11] kk [17:17:51] halfak: I think my hackathon project is going to be jupyterhub on labs. [17:18:03] think you can find someone who’d be interested in that? [17:29:10] YuviPanda, notconfusing [17:29:19] aah, right [17:29:22] will he be at Lyon? [17:29:26] or can we convince him to be at lyon? [17:29:31] Not sure, but it would be nice to bring him out there. [17:29:39] he went to the hackathon last year. [17:29:42] right [17:32:59] Do we have any numbers of the proportion of editors using mobile web / mobile apps? [17:33:09] numbers on* [17:33:38] My understanding is that mobile is fast-growing in readership, but still mostly-negligible in editing. [17:33:48] * guillom pings Deskana who just un-away-ed. [17:33:54] guillom, yo [17:33:59] hey Ironholds [17:34:08] Scott Hale and I have been working on a paper on just this [17:34:15] (also geolocation and a few other tidbits buuut) [17:34:37] lemme get that data for you, I have it locally because paper. 30s? [17:34:48] Sure; it's not urgent =) [17:34:50] guillom: http://mobile-reportcard.wmflabs.org/ has a bunch of numbers. [17:34:55] as well [17:35:10] Thanks YuviPanda! [17:35:34] guillom: including my favorite, ‘number of accounts that were created on mobile apps that are blocked per day’ :P [17:35:42] on enwiki [17:35:48] (I was lobbying for user talk pages) [17:35:52] heh [17:35:54] doesn't appear to have desktop for comparison, though. womp womp. [17:36:07] indeed [17:36:08] doesn't [17:36:59] Android is still obliterating iOS, so that’s good to see :P [17:38:50] As an unrelated side note, let me proclaim that I despise half-baked third-party APIs. I now have to wait for developers to see if they want to add an action to their API, despite it existing in the web interface, and despite it existing for another API module. [17:39:24] "What do you mean, you want to be able to update Foo? You can set Foo when you create a project. Why would you want to update it later?" [17:39:53] Well, because Foo is dynamic, you idiot. [17:40:01] [17:40:03] hahah [17:40:14] that reminds me, I should experiment with setting up RESTbase this evening [17:47:52] Numbers for apps can be crunched for sure, but how comparable they are with our desktop numbers is up for debate. [17:49:08] guillom, 90 day global sample: [17:49:09] > data[,j=list(editors = length(unique(username))), by = "user_type"] [17:49:09] user_type editors [17:49:09] 1: desktop 460465 [17:49:09] 2: mixed 21629 [17:49:09] 3: mobile 115423 [17:49:27] "mixed" is users who have made edits from both platforms in that window [17:49:36] that's logged-in, non-blocked users only [17:49:41] (and filters bots) [17:49:49] Ironholds: Hmmm, thank you. That's much more than I expected [17:50:14] Ironholds: That's for editor numbers, though, right? Not editing volume? (just to be sure) [17:50:24] yep! [17:50:28] if you want editing volume, that's easier [17:50:29] ok, thank you [17:50:37] > table(data$user_type) [17:50:37] desktop mixed mobile [17:50:37] 20672954 2863965 411118 [17:50:42] veeery different ;) [17:51:01] Hmm, yes, that's closer to what I expected [17:51:06] this is a 90-day sample covering, like, december-February [17:51:08] This is useful. Thanks :) [17:51:16] so it may be overrepresenting mobile/mixed because christmas [17:51:19] np! [17:53:04] The reason I was wondering about this is: I'm wondering if I should include more mobile-related stuff in Tech News. I need to balance relevance and length. Until now my policy has been to only mention really important mobile-related stuff, but leave out news of lesser impact because the subscribers don't care as much about mobile. I wanted to see if that assumption was reasonable. [17:57:11] * Ironholds thinks [17:57:28] I don't know. I mean, I'm not sure how much the "meta" spaces are surfaced on the mobile platforms these days [17:57:47] Dan or Maryana can probably comment more usefully on whether people are likely to see the news from those platforms [18:06:22] +DarTar can't get hangouts to work in conf room [18:36:40] DarTar, coming? [18:44:09] kevinator, is Dario around? [18:44:22] I don't see him at his desk [18:44:37] I worked from a cafe this morning, so I haven't seen him all day [18:44:39] * Ironholds nods [18:44:40] thanks [19:45:02] Ironholds: I wonder if it would be better or worse to make the new page file format have a full normalizd URI, like [19:45:13] http://de.wikivoyage.org//wiki/Florence 920 32 [19:45:16] just a thought [19:46:18] ottomata, workable! My only worry would be that it complicates things for, e.g., aggregating or partitioning [19:46:26] because it means you have to split post-hoc orr rely on the full URL [19:46:32] which is a lot of partitions [19:46:57] line[0] + line[1] is easier than "okay, find the second // and then split there". But post it in the thread and we'll see what people think! [19:48:27] oh, Deskana, good timing. Got a sec? [19:48:45] yeah [19:48:57] yeah, it is probably easier to join when you want the uri than it is to split all the time [19:49:13] wait, I got a technical problem right?! [19:49:14] but really, that is kind of the unique (human readable) key that Ids the page, right? [19:49:18] * Ironholds goes to crack open the champagne [19:49:27] yup [20:08:01] Ironholds: I do now! [20:08:04] Ironholds: Sorry, crazy day. [20:08:46] Ironholds: How can I help? [20:09:26] Deskana, so, has ottomata reached out to the apps team about the pageID problem yet? ;) [20:09:35] (if not: we have a request!) [20:09:45] I don't know what you mean by that, so I'm going with no. [20:09:51] heh. Fair enough ;p [20:10:05] so, Ori wrote an awesome extension that passes pageID and namespace into the x_analytics field for requests, right? [20:10:15] and so it means we can resolve URLs perfectly, detect redirects, all this fun stuff. FANTASTIC. [20:10:48] But it's a MediaWiki extension and the way Apps works seems to go around it, so that information isn't passed through for apps requests. How would I go about asking y'all to schedule some time to resolve that and have the apps forward pageID and namespace too? [20:10:56] that way we can get ridiculously-well-resolved pageview data [20:11:39] * halfak --> Vancouver [20:11:42] See you 'round folks [20:11:59] take care! [20:12:17] Ironholds: Hmm. I'd say we should get the tech leads from Apps (Dmitry, Adam), Ori, myself, and whoever from Analytics into a room. [20:12:55] Ironholds: Really this solution shouldn't be in the apps but instead in the API somehow... because otherwise we'll have to reinvent it for each third party consumer in order to have accurate data for them. [20:13:03] Ironholds: But I don't know to what extent that is possible... [20:13:09] Ironholds: i haven't ! i hadn't seen Adam online yet [20:13:10] but he is now! [20:13:23] Ironholds: And it depends what we care about, e.g. all third party consumers or just WMF consumers. [20:13:30] am pinging him now. [20:13:37] no need to schedule meeting yet please. [20:13:49] Deskana, yeah, that makes sense! [22:06:27] #wikimedia-analytics [22:06:32] Does that exist? [22:06:40] It does [22:06:53] * guillom discovers new Wikimedia-related IRC channels every day.