[07:04:28] morning ori__ [07:04:36] hi average [07:04:38] :) [07:04:40] hello nuria [07:04:51] morning nuria [07:06:45] hey humans [07:06:59] hey Ironholds [07:07:04] hey oliver [07:31:57] hey Oliver [07:32:05] hmm, too late to the party, am I not [08:04:12] YuviPanda: never [08:04:31] ori__: my underscores are larger than yours! [08:04:43] it's what you do with them [08:05:17] python name-mangling, for class-private identifiers, for example [08:05:17] hmm, I can't put them in Java code [08:05:28] so I am not really using them much [08:05:45] so yeah, you do have the advantage here, ori__ [08:06:09] are you coming to the summit? [08:07:08] ori__: no, sadly am not. [08:07:21] although I've not really been involved in enough architecturey things... [08:07:45] ori__: I'll know in about a month when I can move there though. [08:07:52] that's a bummer [08:07:57] that you're not coming, i mean :) [08:08:06] yeah [08:08:10] next year, maybe [08:09:12] ori__: I'll spend some time with the hhvm patch later today and respond to erik's comments. [08:09:36] been sleeping on time, which means I'm also eating things and also spending time walking around in the outside (Ingress!). less time to hack on random things [08:10:15] * YuviPanda withdraws from the underscore competition [08:11:16] YuviPanda: what's your secret? [08:11:56] ori__: spent a week progressively fucking up my sleep cycle until it came out the other side and became normal :D [08:12:37] took that week off too [08:28:49] time off? bastard [08:29:05] * Ironholds is going to be eligible to start accumulating holiday time in a month or so. woo? [08:29:27] Ironholds: i have a long reply to the UA thing on officewiki, will post shortly [08:29:32] ori__, cool :) [08:29:40] I'll be asleep, but I'll get to it in the morning [08:30:04] my near-future focus is likely to be working on the filters in things like varnishkafka/pageview [08:30:27] if there's one thing maniacally triaging the requestlogs is for, it's identifying weird shit and then telling others about it ;p [08:30:32] so, UA data is likely to make an appearance [08:32:10] alright, on that note, sleeptimes. laters, all [08:36:04] I had a conversation with ori about UA Friday, will also post today [08:36:19] (while you guys are asleep, my day just started it) [08:37:38] Ironholds: heh, not an employee, so easier [08:37:56] Ironholds: i actually had 6 days off, but that gave me about 11 days 'real' because of christmas and things [08:38:16] hey nuria [08:38:20] nuria: what's your TZ? [08:39:00] TZ? ...*ahem* what's that? [08:39:46] nuria: timezone :) [08:40:43] when people say 'my day just started' can either mean they're in a different timezone... or they wake up at 8pm (I used to do the latter :) ) [08:40:58] ahhhh, sorry. i am in europe but I try to work late to overlap a big part of my day with east coas time [08:41:09] *east coast [08:41:19] ah, nice :) [08:41:36] I'm in India, so used to work *really* late to overlap with SF... [08:41:56] can someone invite me to -staff [08:42:14] i am in the channel list by nick [13:16:54] [travis-ci] develop/6f91a8b (#152 by nuria): The build passed. http://travis-ci.org/wikimedia/limn/builds/16864781 [14:31:23] morning [14:33:01] holaaaa [14:43:14] goooddeemeoororningngng [14:57:01] Hi. [14:57:16] Someone seems to have fun at kicking me out of the hangout invite. [14:57:20] I cannot join again :-( [15:01:24] invtied you again? [15:01:49] Thanks ... still rebooting the google machine ... just in case... [17:51:07] (CR) Milimetric: [C: 2 V: 2] For Bug 59843 [analytics/wikimetrics] - https://gerrit.wikimedia.org/r/106775 (owner: Milimetric) [18:07:53] qchris_away: if you come back tonight and have 15 mins to talk can you ping me? [18:18:57] ori, thanks for the UA concerns writeup; it all seems totally reasonable (both the concerns and the solution) [18:19:12] out of interest, have we actually got/picked a UA library, or is that on the to-do list? [18:29:08] * DarTar still has to catch up with UA, I spent my Sunday going through data retention stuff. ori, Ironholds: reading you [18:30:27] Ironholds: both nuria and average have tried different things. I think the requirements aren't 100% clear yet, so a tool choice might be premature [18:31:06] I know dclass and different regex collections have been used [18:31:11] and I believe what ori suggested is a simpler, custom, predictable regex approach [18:32:38] For what it's worth, if we can get a custom simple regex to sufficiently mangle the UA in compliance with the data retention guidelines, then that's the best way to go. The reasons are well put by ori on-wiki [18:55:23] milimetric, agreed. I think once you break it down to that it's pretty simple [18:55:39] the trick is going to be running them in the right sequence, of course (IE has the annoying 'Firefox' tell) [18:58:57] I added some comments on wiki, maybe I don't get the subtle implications of pre vs post processing here [19:00:02] isn't a "custom simple regex to sufficiently mangle the UA in compliance with the data retention guidelines" a totally orthogonal problem to the pre vs post-processing discussion we're having? [19:00:06] milimetric: ^ [19:01:10] DarTar, I interpreted Milimetric's comment to refer to how we get things from 'raw UA' to 'processed UA with enough information to allow for reworking if it turns out our UA recognition script screwed something up' [19:01:23] oh, DarTar, I was just answering Ironholds's question about whether we decided to use a specific tool [19:01:32] we're both wrong! :P [19:02:07] understood, but ori brings up a similar point about a "simple, generic regex" [19:02:23] yes, I think that's the right logic when thinking about what tool to use [19:03:00] my only question is, does such a regex exist. From the little I know, it's not possible [19:03:02] but I could be wrong [19:03:22] what worries me is, assuming it existed today, will it still do the job in a year from now? [19:03:52] yeah, that is to say, there's very good reason for other libraries to get updates and change the way they interpret UA [19:03:55] the whole point being that this rule should change as little as possible [19:04:00] this, by no means, is a simple problem :) [19:04:12] yea, that's ideal [19:04:25] so is that orthogonal to whether we do it pre/post? [19:04:33] I don't think so... all these things are inter-related [19:04:54] if it's a simple enough tool that doesn't sacrifice any analysis power, then let's do it as close to the source of the data as possible [19:36:44] milimetric: quick question: is there a way to select a single series in a limn graph (and toggle off the other ones) via some URL parameter magic? I guess the answer is no but I thought I'd double check [19:39:17] sorry DarTar, the answer is no [19:39:34] kk [19:39:55] i had suggested a little (only) link show up next to legend items you're hovering over [19:40:07] if you're cool with that, it would be fairly easy [19:53:01] hey halfak, should i disable module storage data collection this afternoon? [19:53:20] or should we accumulate some additional data just in case or something [19:53:56] ori: If you have any concerns, disable. If it's no trouble, I'd appreciate it if you kept it running just in case we want some more of the data. [19:54:31] no concerns, but i haven't looked at the data at all [19:55:08] i could also reseed the bucketing algo so we get a fresh cohort, so we can compare initial loads [19:55:32] leaving it running as-is for a few more days is also fine [19:56:08] Let's let it go as is. I'm aiming to have something for us to look at on Wednesday morning. [19:56:16] We can re-assess at that point. [19:56:19] Sound good? [19:56:31] sounds awesome. thanks. [19:56:35] :) [20:08:22] ottomata: heey [20:08:25] yoo [20:08:31] sorry, I was like.. asleep [20:08:40] now up again [20:10:11] aayyeee, s'ok [20:10:20] i'm about to go meet up with charles, soooo, we'll work on this tomorrow? [20:20:56] hey. one question about the pageview data here: http://reportcard.wmflabs.org/ [20:21:04] does this include API calls? [20:21:07] in any way [20:22:14] especially, does it include any impressions from the wikipedia app? [20:35:13] dennyvrandecic: Which graph of those on the reportcard are you referring to? [20:41:08] pageviews [20:41:28] i just send an email to analytics with the same issue [20:42:09] ok. [20:42:33] I'll leave it to the creator of the reportcard then :-) [20:43:22] is erik here? [20:43:42] No ... but he reads the lists. [20:43:45] m, b, or z? [20:43:56] ah, z. [20:44:31] back in a bit [20:45:07] thanks, then the list is correct :) will wait for z. to answer [20:52:58] hi dennyvrandecic -- it's toby [21:38:36] hi tnegrin [21:39:17] sorry was in other chat [21:48:44] hi tnegrin [21:49:03] hi denny [21:49:33] I think we're good [21:49:44] ok, just wanted to check back [21:50:18] thanks! [22:00:19] join #wikimedia-analytics [22:00:27] alrigth! [22:00:34] hey everybodddyyyyyy Dwan == Charles [22:00:41] Yeah, sorry about the confusing nick [22:00:44] I'll change it [22:00:50] tnegrin, average, qchris, nuria ^^ [22:01:15] hello! [22:01:22] hi Toby [22:01:44] welcome to the IRC -- it's where it all happens [22:01:55] Hi Dwan! [22:02:02] are you in some cool coffeehouse with ottomata? [22:02:07] yes actually [22:02:12] we are in the trendiest possible coffee house [22:02:14] check out the name [22:02:19] Urban Vintage [22:02:22] what could be more hip!? [22:02:29] 4 barrel ;) [22:02:32] psshhhh [22:02:32] smells kind of hipster.... [22:02:38] that might be more hip [22:02:39] welll hmm [22:02:45] someone once described this cafe to me [22:02:52] as feeling like you were inside a martha stewart catalog [22:03:03] ok, taht is not hipster [22:03:05] *that [22:03:08] I'm 30% cooler for even being here [22:03:11] would you like to purchase a giant clothespin for $7? [22:03:12] sounds like my mom's place [22:03:13] because you can! [22:03:21] how much is the toast? [22:03:27] hmmm good q. [22:03:36] i will ask roberto next time I go up to the counter [22:03:44] I suspect it's reasonably priced [22:03:50] roberto and I do pushups together…sometimes...once [22:03:55] in the cafe [22:04:37] uh -- don't give Charles any bad ideas!