[00:31:54] Deskana, re the above thread [00:32:06] we're almost certain it wasn't a crawler, but was instead something/one malicious [00:32:14] or merely stupid [00:32:27] crawlers don't post from a domestic IP nobody has heard of and obfuscate their user agents [14:14:52] morning halfak [14:15:00] hey Ironholds. [14:15:12] Any feedback on the R:Index mock I sent yesterday? [14:15:22] ack, forgot to look at it, I'm afraid. Link again? [14:15:24] https://meta.wikimedia.org/wiki/Research:Index/Sandbox_splash [14:15:27] my buffer is only 100lines [14:15:27] ta [14:15:44] Compare to: https://meta.wikimedia.org/wiki/Research:Index [14:15:51] oh man, I like [14:16:04] it's like you've got a PhD in designing or something! [14:16:17] :P Do you think I went overboard on the top whitespace? [14:16:23] I could shrink it down a bit. [14:16:45] I don't think so? But I have a specific machine with a specific browser and OS and etc, etc [14:17:03] Indeed. It works well down to 800x600 [14:17:09] But on the vertical phone, it looks like poop. [14:17:27] I really wish I could just have a separate mobile CSS [14:18:12] looks good, though :) [14:18:25] I'm just finishing re-re-re-testing the PV scripts [14:18:35] iff I'm right, it will indeed run automatically. [14:18:38] * Ironholds plots [14:18:42] Woot! [14:18:50] related to that, we have PV data up to 28 September of this year in staging.pageviews [14:18:54] * halfak is not sure which PV script this might be [14:18:55] take a look at the structure and complain! [14:19:00] The one producing public data? [14:19:08] actually we could probably make this public [14:19:12] even the per-country breakdowns [14:19:15] ooooooh! [14:19:17] Ooh! [14:19:31] https://github.com/Ironholds/pageviews and look at staging.pageviews [14:19:34] grab some example lines! [14:19:39] Say, one more thing I want to look at. https://meta.wikimedia.org/wiki/Research:Index/Sandbox_grid [14:19:49] Compare with the one I sent before. [14:19:52] Which is better? [14:20:08] I prefer the first one [14:20:14] I think it makes the important navlinks more prominent [14:20:24] Cool. Thanks [14:20:33] also, I just like the spatial propriety of having links outwards in a spoke [14:20:37] * Ironholds has odd design needs [14:20:46] :) Feels like wikipedia.org [14:22:41] that too! [14:22:51] seriously thought, SELECT * FROM staging.pageviews LIMIT 5; [14:23:09] * halfak logs in [14:23:29] dammit, I forgot to capitalise "Desktop" [14:23:35] LAME [14:23:36] :P [14:24:00] This is pretty sweet [14:24:05] metadata, before you ask, contains either the OS for app hits [14:24:09] or the MCC for Zero hits [14:25:12] MCC? [14:25:23] mobile carrier code! [14:25:28] unique identifier for the mobile provider. [14:25:53] Gotcha. [14:26:40] This is a great table. I look forward to a dashboard with these breakdowns. [14:26:55] It looks like this is based on sampled data, right? [14:27:07] yup [14:27:09] for speed, and win [14:27:21] and also the fact that with sampled data I can go back to March 2013. [14:27:33] Gotcha. Takes too long to process the whole thing? [14:27:59] ehh, I didn't actually test. I mean: I could! We could see what happens. Big MR job streaming line-by-line into python for handling [14:28:10] but I was disincentivised by the fact that that gives us August and most of September [14:28:20] Gotcha. [14:28:23] I'd rather have depth and fuzziness than pinpoint accuracy for the last 50-ish days. [14:28:34] +1 [14:28:40] Though, for the future... [14:28:48] yerp! [14:29:06] urgh. So sleepy. [14:29:08] * halfak is not totally sure that will matter. [14:29:26] Anyway, for publishing the data, it seems good to gather it from a sampled log. [14:29:31] yep [14:29:42] That way, there is a lack of scary guarantees about countries with few viewers. [14:29:44] ain't no privacy violation like a high k-value privacy violation [14:29:56] to be sung to the tune of "there ain't no party like my nana's tea party" [14:30:02] T-Shirt [14:30:08] hehe [14:30:16] ooh. This gives me an idea. [14:30:24] * halfak has no idea what unusual british culture you are inflicting upon him. [14:30:39] not British, American! It's a line from a song in Flight of the Conchords. [14:31:37] My friend Alice keeps threatening to lasercut me a hip-flask. I could go for if(k < 2){ ... } [14:31:43] Oh! I think they are Canadian. [14:31:54] they're kiwis, but I'm pretty sure it was on HBO [14:31:58] it had Kirsten Schaal! [14:37:28] Hey Nemo_bis, you around? [14:40:54] halfak: I have 5 min before going out [14:41:16] Nemo_bis, quickly wanted to show you https://meta.wikimedia.org/wiki/Grants:IEG/Revision_scoring_as_a_service [14:41:37] It's a project that, once complete, will allow me to deploy Snuggle outside of enwiki. [14:41:47] *and* have desirability scores [14:42:21] ooh [14:42:33] Please review if/when you have time. :) [14:42:50] user:petrb was interested in some such service too, if you read wikitech-l some months ago [14:43:10] I was talking to petrb about this project a couple months ago. [14:43:15] I'll look for the thread. [14:44:14] This one: https://lists.wikimedia.org/pipermail/wikitech-l/2013-September/072038.html [14:44:15] ? [14:46:00] left link on talk, now out [14:46:07] Thank [14:46:08] see you soon [14:46:08] s [14:46:10] see ya! [15:14:09] Ironholds: I replied! [15:14:54] JetLaggedPanda, huh? [15:15:02] Ironholds: oh, on the ops thread [15:17:22] Ironholds: I'd also suggest creating an RT ticket [15:17:34] just email ops-requests@rt.wikimedia.org [15:17:49] and respond on the thread with link to the ticket [21:05:42] Hey lzia, do you see DarTar around? [21:05:49] ewulczyn, ^ [21:07:18] kevinator, ^ [21:07:44] AFAIK he’s in a Quarterly Review for Visual Editor [21:07:54] he’s not around [21:08:05] Thanks. [21:20:49] hey halfak [21:21:04] joining the hangout now if you’re still around [22:11:37] Poke Ironholds, leila & ewulczyn. Can you tag the topics you are interested in here: https://etherpad.wikimedia.org/p/Meetup [22:14:49] halfak, I can do this later today. I went through the list once but it requires some more time from me to fully comprehend [22:31:48] halfak, I think it's better to do R&D retrospective when you're in town. I don't want to eat up Analytics time for this, so we may want to do this in a late afternoon some day next week, if we decide we'd like to do it. We don't need to plan for it now, though. [22:36:40] I can be convinced. [22:36:49] It seems like it would be good to do it in person, I agree. [22:39:51] I can definitely benefit from that, but I can also spend more time on it tonight and figure it out. but that can delay you, and then me. ;-) [22:52:57] leila, I propose that, if we do a retro next week, that we do it early in the week. [22:53:18] This would afford us the ability to spend time discussing our process more effectively for the rest of the week while we're in person. [23:23:37] halfak: ping. [23:23:43] halfak: you had a 'diff producer' service, right? [23:23:47] haha [23:23:50] poor him