[00:00:46] howief: Sample data with bugfix http://pastebin.com/m7ed1cef5 [00:01:25] looks great! [00:01:36] so in = total that signed up within timeperiod [00:01:42] Yes [00:01:56] out = total that signed up within timeperiod, but left within the interval [00:02:02] Yes [00:02:05] this is great data [00:02:14] Time period = START <= t < START+24hrs [00:02:22] Interval = 7 * 24 hrs [00:02:28] it'll really help us track how changes we make affect retention [00:02:30] For this data, but these parameters are configurable [00:02:36] awesome [00:02:41] it looks great [00:02:51] OK so the way this works is you tell me what data you want with which parameters etc., and I get it for you :) [00:03:04] nice! [00:03:16] that parameters you have should get us started [00:03:50] do you think you can run the top 10 language wikis from inception until today? [00:04:00] (the most recent week's data won't be comparable, but that's ok) [00:05:19] Inception being? [00:05:25] hmmmm [00:05:31] Also, do you want the top 10 separate or summed? [00:05:38] how about Aug 1? [00:05:39] (Summed is harder, haven't implemented that yet) [00:05:41] top 10 separated [00:05:43] OK that's fine [00:05:47] no worries about summing [00:05:55] You can do that yourself too if you want [00:06:02] we may want to do that later, but let's get the top 10 first [00:06:06] OK [00:06:10] *RoanKattouw looks up what the top 10 is [00:06:29] top top 10 are here [00:06:40] Beta_Feedback_Survey [00:06:47] sorry http://usability.wikimedia.org/wiki/Beta_Feedback_Survey [00:06:54] partway down page [00:09:25] So Aug 1 till Dec 4? [00:09:31] Or Dec 1 or what [00:09:46] Let's do Dec 4 [00:09:58] OK [00:10:04] Wait that's tomorrow [00:10:30] *RoanKattouw uses Dec 3 instead [00:10:44] oops - you're right :) [00:11:20] It'd technically work since it's 11 minutes past midnight UTC [00:11:43] But those 11 minutes aren't particularly interesting [00:12:20] OK it's running [00:12:50] great thanks! [00:13:29] http://toolserver.org/~catrope/retention/enwiki.txt other langs still generating [00:16:49] howief: All done, http://toolserver.org/~catrope/retention/ [00:16:55] params.txt lists used params for reference [00:17:03] wow that was fast [00:17:21] thanks for doing this [00:17:24] Actually it runs pretty slow [00:17:34] I could make it a bit faster I guess [00:17:52] seemed pretty fast to me [00:18:08] Yeah that was just 10 wikis, ran in a few minutes [00:18:21] Imagine having to sum numbers for 778 wikis [00:18:26] true [00:18:33] but this isn't something we'll be running every day [00:18:42] we'll probably look at it every week or so [00:18:46] That's true, but I'd still like it to not take an hour [00:19:00] yeah [00:19:03] I'll add a reminder about speeding it up [00:19:19] cool [00:20:27] Getting the total number (in+out) for a timeframe is relatively fast, getting out is slower, getting in is about twice as slow as that [00:20:42] So I wanna move to calculating in with in=total-out [00:21:18] i see [00:22:35] Not sure it matters a great deal compared to reconnecting to the database as much as it is, but WTH, we're not gonna run it often anyway, as you said [00:22:58] yup [00:26:43] interesting [00:26:57] the enwiki retention rates haven't changed a whole lot since the first few weeks [00:27:15] i'm curious to see what the dewiki numbers look like [00:29:49] *RoanKattouw wonders where Trevor went [00:30:48] nkomura1: I won't be able to deploy those fixes tomorrow, as I'll be unavailable starting 3AMish PT [00:31:03] no worries [00:31:26] there were a few questions from trevor about adam's revision right? [00:31:33] they are not urgent revisions [00:31:53] so we can deploy them next week [00:31:57] Yeah [00:31:58] thanks for letting me know though [00:35:04] You may also want to know that my max for December is around 20-25 hours, and I'm at 10 right now [00:36:33] Of course if work needs to be done and I've hit my limit I'll just do it anyway and not invoice it :) but I'll probably skew that time away from usability a little bit [00:36:43] I favor of site requests and a pet project of mine [00:37:44] The good news is we'll have language-specific icons on the prototypes tomorrow [18:14:08] hi RobH [18:14:15] Hiya [18:14:19] hey [18:14:25] a quick request [18:14:36] I sent an email to apple, no reply. I am going to try another contact there about the xserve [18:14:50] okay [18:15:02] and the other server I neglected to order (sorry!) but it will be ordered today (as I am ordering a bunch of search servers) it takes them about a week to get a server to me. [18:15:17] I also have room to install it. I will be emailing mark about the network setup when it arrives =] [18:15:26] great [18:15:41] sorry for the delay, I was a bit burned out pre-vacation. [18:15:50] *RobH is no longer burned out. [18:15:55] no problem [18:16:10] glad to hear you are over being burned out [18:16:20] yea...i needed the vacation badly. [18:16:22] can Ryan_Lane help you with installation or anything? [18:16:40] Nah, once it is up and accessible via the internet, I expect to just give him full access to it. [18:16:51] cool [18:16:52] that way he can setup the OS and such as he sees fit for testing [18:17:13] actually my question was [18:17:19] we are filling up linode quickly [18:17:33] i asked Ryan_Lane to monitor system usage [18:17:55] but he doesn't know the link to the monitoring system [18:17:56] so i thought to ask [18:18:02] i'm looking at nagios right now. Is there any way to look at historican cpu/memory? [18:18:22] err historical [18:18:31] nkomura: i just remembered what the link was :) [18:18:41] good [18:19:27] We do not log or store information on the linodes in cluster. However, that stuff should be accessible via the Linode manager [18:19:32] Ryan_Lane: do you have access to that or no? [18:19:47] RobH: no [18:20:07] how do I get access to it? [18:20:56] let me see if I can find it [18:22:23] Ryan_Lane: i dont have you in my address book, fixing that [18:22:26] what is your email addy? [18:22:56] rlane32@gmail.com [18:23:04] or ryan@ryandlane.com [18:24:43] Ok, emailed you the info for admin on that linode [18:24:50] ok [18:26:02] got it. thanks [18:26:38] very welcome [18:27:20] hmm. i don't see any memory stats. heh [18:27:30] cpu stats are fairly low [18:39:32] TrevorParscal: you around? I'm looking at those RTL css revisions you were questioning yesterday [18:39:44] yes [18:39:57] Ryan_Lane: prototype response has been slow time to time [18:40:04] are you resolved on them? i saw you made a follow up commit [18:40:07] it would be good to look into memory usage [18:40:20] nkomura: there are some network spikes, and occasionally some cpu spikes [18:40:27] i'm quite sure any changes to the rtl stylesheet were made by cssjanus and not by hand [18:40:39] nkomura: I'll run a cron on the system to give me memory stats every minute or so [18:40:48] Ryan_Lane: great [18:40:54] adam_miller: i wasn't commenting on the RTL stylesheet [18:41:06] i have a feeling it is memory as well [18:41:12] it was that you've been adding body.rtl rules [18:41:45] where? I dont see any in the revisions you commented on [18:42:01] it seemed you were unclear on cssjanus, or at least were not using it well [18:42:53] http://www.mediawiki.org/wiki/Special:Code/MediaWiki/59254 [18:43:42] i moved some things out of no flip rules there so they would get flipped, why is that bad? [18:44:48] TrevorParscal: maybe this would be something better to go over in person on monday? [18:44:59] sure [18:45:10] it's not broken right now [18:45:26] but it needs to be done slightly differently [18:45:32] k, lets revisit it next week then and you can educate me a bit [18:45:38] no rush - monday we can get it sorted [18:45:41] :) [18:46:16] he he, I'm sure I won't be educating, probably just nit-picking [18:46:17] he he [18:47:58] well enlightening me further on your very interesting opinions, anyway [18:51:50] ha ha- yes, that sounds about right [18:52:34] Ryan_Lane: prototypes are pretty slow today [18:53:02] it would be great if you can find what is the bottleneck [18:56:31] nkomura: nothing seems abnormal [18:56:56] network too? [18:58:37] nkomura: network stats are low. not even using a Mb. ranging around 30Kb/s [18:59:01] odd [18:59:11] i'm not experiencing any slowness [18:59:19] really? [18:59:36] my access to production and prototypes differs significantlly [18:59:36] nkomura: is the slowness on pretty much everything, or specific pages? [19:00:01] nkomura: production has a lot more cache [19:00:04] even the languages i rarely visit load a lot faster [19:01:00] is there way to increase cache for prototypes? [19:01:15] nkomura: let me look at the current caching options [19:01:31] great [19:02:20] adam_miller: sorry to read about your car [19:02:28] i had to drop off my car for service this morning [19:02:46] cars breakdown whenever we need them the most, don't they? [19:02:52] thanks, the one i test drove today looks like it's going to be the replacement [19:03:13] ha, yeah mine is still drivable, but it needs way too much to pass inspection [19:03:45] i am yet to find out if mine passes the smog check [19:03:52] i am keeping my fingers crossed [19:04:22] we have those in PA now, but I don't think they're nearly as harsh as california's yet [19:05:30] i support the clean air, but i do not like the cost part ;) [19:05:35] what does MIA mean by the way? [19:05:41] missing in action [19:05:48] ah [19:08:05] adam_miller: speaking of action, have we overwhelmed yet? [19:08:40] i have a good amount of action now, but i'm not quite overwhelmed [19:09:42] okay, we will send you more to your way! [19:10:13] was the hand-off from roan about link dialogues clear? [19:10:18] roan won't be online today [19:10:22] totally [19:10:36] great [19:10:37] i collected a list of to-dos from him that i've started working on [19:11:00] super [19:11:23] we'll have more UI refinements we would like to rely on you next week [19:16:55] nkomura: the only caching we aren't doing is squid/varnish. installing squid would probably make things worse :). I can turn on file caching though [19:18:58] let's try file caching if it helps [19:19:03] http://usability.wikimedia.org/wiki/Talk:Prototype - shall I procedd answering questions or stop it? its getting a bit redicilous.. [19:19:11] so after the fresh install, we expect the slowness [19:19:18] as the files are updated? [19:19:24] http://usability.wikimedia.org/wiki/Talk:Prototype#Nothing_new.2C_however. [19:21:12] he uses a screenresolution of 600*800 and complains that vector is worse than the WP design from 2004 [19:21:13] nkomura: fresh install? [19:21:30] when the new build is installed [19:21:40] which we often do on prototypes [19:22:01] possibly [19:22:05] hannes-_-: please refrain from further reply to that thread [19:22:34] actually, likely. yes. [19:23:38] Ryan_Lane: that makes sense [19:24:14] nkomura: ok. but why? do you think roan and I did not answer him in a good way? [19:24:22] enabling it. let me know if there are any issues [19:26:32] hannes-_-: i haven't read everything yet [19:26:46] nkomura: let me know if that helps as well [19:26:46] but i sense the lack of neutrality and objectiveness in the threads [19:26:59] if Q0k likes monobook so much [19:27:09] it would be best to encourage the person to opt-out from the bta [19:27:10] beta [19:27:29] rather than making the argument why he had to use the beta and why we cannot change the beta for him [19:27:41] true [19:27:46] we have 83% retention rate from english wikipedia community [19:28:03] so you could simply direct to the fact and he can opt-out [19:28:12] Ryan_Lane: thanks [19:28:21] you're welcome [19:30:20] nkomura: did you also want me to do that for the sandboxes? [19:30:46] do you mean the file cache? [19:30:49] yes [19:31:04] i haven't experienced slowness using sandbox [19:31:09] ok [19:31:13] but the situation may change [19:31:18] once you populate the data [19:31:22] true [19:31:24] so please go ahead and enable that [19:31:28] ok [19:36:40] enabled. should I send out an email letting people know that's been turned on? [19:37:04] yeah [19:37:11] great suggestion [19:47:43] TrevorParscal: are u there? [19:47:49] yes [19:48:05] I got a qucik question regarding the templatefolding [19:48:28] for the folding-icon: are we limited to the line heigt? [19:48:47] or character-heigt [19:49:12] height [19:49:31] hmm does that question make sense? [19:50:02] you are saying, there will be a box of some sort around the {{templatename}} thing [19:50:08] and that box is going to have icons in it [19:50:40] basically, yes [19:51:01] can the box be higher than the characters (of the code) are? [19:51:18] like [templatename| ^ | > ] where ^ is an icon to launch an editor, and > is an icon to expand [19:51:29] the launch icon should be safe at 16px [19:51:40] excactly [19:51:51] take a look at the icons we use for links on wikipedia [19:51:52] and can this have a background? [19:52:03] like the external link icon [19:52:12] yes, I expect it will have a background color [19:52:17] maybe a border of some sort [19:52:18] ok, i know what u mean [19:52:31] ok cool [19:52:37] so you say it s 16px? [19:52:40] we will try and keep it simple but basically it needs to look like a block rather than string of characters [19:52:47] starting at the baseline.-...? [19:53:00] agreed. [19:53:39] these 16px = starting from the baseline? [19:53:50] or do we have some space underneath the baseline? [19:54:30] http://en.wikipedia.org/skins-1.5/vector/images/external-link-ltr-icon.png is an example... [19:54:46] which is actually a 10x10px [19:55:00] which may be a sign that I'm wrong about the 16px :) [19:55:16] it should be vertically centered to the text [19:55:54] hmm ok. how can we get sure about the excact height? [19:56:01] 10px or 16px is a huge difference [20:06:48] template folding, eh? [20:06:51] I love you guys [20:06:58] excuse me while I bow in worship [20:07:14] coming in now... cyall in a bit (getting lunch on my way in) [20:07:49] hannes-_-: until the thing is actually built, we cannot be sure about it - but 10px sounds safe for now [20:07:53] k [20:07:56] really heading in now.. [20:07:58] cyall [20:08:17] werdna: we are the brave team [20:11:41] brion, werdna: can one of you grant me permissions to do recursive exports from the production wikis? [20:14:05] what permission is that? system administrator or something? [20:15:39] yeah [20:15:48] hmm i think i added it to staff group a while ago, not sure if it's on sysops [20:15:59] you should be able to find it pretty easy if it's not in tho [20:16:10] oh? roan tried adding me to staff last time, and it wasnt in there :) [20:16:21] he ended up adding me to system admin [20:18:08] what's your user name? [20:18:14] I added a recursive-export group [20:18:16] ryan lane (all lower case) [20:18:49] and what wikis do you have an account (I just need one) [20:19:13] mediawiki.org, or enwii [20:19:21] err enwiki [20:21:44] 20:21 <+StewardBot> Werdna changed global group membership for Ryan lane from Staff, sysadmin to Staff, sysadmin, recursive-export [20:21:51] you seemed to already have the right [20:21:52] but oh well [20:21:57] oops [20:22:10] i was probably supposed to be taken out of sysadmin at some point in time [20:22:41] 20:22 <+StewardBot> Werdna changed global group membership for Ryan lane from recursive-export, Staff, sysadmin to recursive-export, Staff [20:22:48] thanks [21:28:11] importing content into the prototypes, there may be some slowness [21:28:30] roger [21:59:28] zhwiki prototype is done. I can't seem to get the mainpage to default to Wikipedia:?????? though. I edited MediaWiki:mainpage. no dice [22:00:57] going get some dinner. away for a bit [22:52:14] mdale? [22:52:20] yes [22:52:42] AMW was working from the new toolbar for a while [22:52:51] but it stopped working since few days ago [22:52:56] was there any change? [22:53:07] i wanted to show it off at Open House last night [22:53:26] hmm [22:53:37] but i couldn't make it work and it does not work on my computer either [22:53:45] yea it appears someone svn updated sandbox 2 :( [22:54:16] hm? [22:54:56] i know you use prototype for js2 [22:55:16] but what does it have to do with sandbox 2? [22:55:24] mwEmbed loads from there [22:55:34] the gadget loads from sandbox2 [22:56:44] shall we create a directory which is independent from sb updates? [23:00:12] hmm [23:02:23] oky try turning on the gadget and reloading [23:02:26] should work now [23:02:29] I revert it [23:02:36] to the correct version [23:02:46] okay [23:05:00] hmm still partially broken.. let me try a different version before Tim started committing stuff. [23:09:41] oky well its "working" but not the latest and greatest [23:10:01] i'll try it [23:10:19] i'd like to discuss how we can incorporate into the usability offerings in the citron release [23:10:42] we will need to find a permanent place so that users can have consistent experience [23:11:09] yea ideally as "part of the software" but have to work that out with Tim somehow [23:11:31] I could request a tool-server account to host it there [23:12:00] that sounds one step closer [23:12:20] yea.. but still kind of hackish ... [23:12:34] why was sandbox 2 updated? [23:12:51] you guys just updated them all at once? [23:12:52] i am not aware any update [23:13:20] it was at some later svn version which is why the mwEmbed gadget broke [23:13:44] what was the time stamp [23:13:51] could you see who updated it? [23:13:57] I don't know I have reverted it now... [23:14:02] roan was updating the toolbar [23:14:10] but i don't see the update in sandbox [23:14:11] yea maybe that did it [23:14:28] if we are just a bit more careful should be oky... [23:14:30] but roan's update was this morning [23:14:35] it was broken much before then [23:14:54] ~ don't know.. should have investigated prior to reverting [23:15:52] at any rate I think Tim will stop breaking things now that he moved it out of the trunk so I can just switch to the "js2-work" which will only contain "working" versions so svn updates should not hurt anything [23:17:00] ok [23:17:38] great to see AMW working again [23:19:19] I saw a note in usability about performance and using file Cache ... we could install squid proxy then would not have to worry about loading the js from the prototype machine for having the beta having an option for "enabling" the mwEmbed gadget [23:19:54] (since squid could easily handle many thousand requests per second on that box) [23:21:04] mdale: will you talk to Ryan_Lane when he is back from dinner? [23:21:09] sure [23:21:45] you know that prototype is hosted by the third party company right? [23:21:49] right [23:22:09] squid proxy can work for external servers too? [23:22:19] i guess that is why you call it proxy [23:23:48] yep its just a webserver front end similar to apache ... but it just sits in front of apache and handles all the static requests. [23:28:44] cool