[00:18:23] I hate you code. [00:18:27] YuviPanda, you were right [00:18:50] of the many things I'm usually right about, which one is this? [00:18:59] when I first started learning to code many moons ago you said to me [00:19:08] "you'll do fine. Just remember your computer hates you and wants you to fail" [00:19:18] ah that [00:19:21] this week I have spent 3 days debugging my port of the Open Location Code JS implementation [00:19:25] do you know what the bug was [00:19:35] Ironholds: http://www.stilldrinking.org/programming-sucks if you haven't [00:19:40] go on [00:19:48] MAXIMUM_LONGITUDE_ was incorrectly set within the class definition. [00:19:50] Do you know why? [00:20:10] hahaha [00:20:11] go on [00:20:12] because in Liberation Mono [00:20:16] 8 looks kind of like 0 [00:20:18] HAHAHAHA [00:20:18] nice [00:20:23] and you mistype because you're not thinking when making the port [00:20:40] and suddenly your computer things the world is <1/3rd as big as it is and introduces a floating point error [00:20:48] a floating point error that does not actually make it BREAK, NO [00:20:49] heh [00:21:00] because you're using integers to store the resulting value because they're meant to be integers [00:21:17] and so they are instead erroneous floating points that are silently coerced to integer and it quietly, happily, gives you the wrong characters at the end. [00:21:54] and you spend a week trying to work out why 7FG49Q00+ is instead 7FG400000 [00:22:17] nice [00:25:48] Ironholds: I just heard Roan exclaim out loud 'WHY WOULD IT DO THAT THAT IS TERRIBLE' [00:25:53] which summarizes a lot of these [00:27:23] yup [00:27:45] YuviPanda, my SO recently said to me [00:28:11] "I need to keep a running tally of the things you say when you're coding because it's always either 'why did it do that?' or 'there's no way in hell that should work!'" [00:28:27] heh [00:28:28] indeed [00:28:36] do read that article I linked to if you haven't already [00:28:45] Ironholds: did I tell you I am taking the train to Chicago [00:29:49] you did not! Where from? Who are you seeing in Chicago? [00:30:00] and I have read it and thoroughly recommend his first book, And Then I Thought I Was A Fish [00:30:02] his second book is meh [00:30:58] Ironholds: nobody! it's quite insane, but I'm taking the train for the journey and then flying back [00:32:09] okay, and where are you going from and would you like some people to hang out with? [00:32:25] I'm going from SF! [00:32:27] Emeryville [00:32:37] we have no idea where we are going to crash or what we are going to do [00:33:58] we? [00:34:08] and: I know some people in Chicago if you want some names! Aaron Shaw lives there, as does Keilana [01:15:38] Ironholds: I'm meeting keilana [01:30:09] yay! [13:35:59] ottomata: hey [13:36:03] Saw email from Ellery? [13:36:21] yo, oh yes! [13:36:26] it is up for me to respond to [13:36:32] but i don't know the answer so I was hoping you would [13:36:32] :) [13:36:47] So I was thinking of setting up a mountpoint in stat hosts [13:36:53] hmmMM [13:36:56] That would simply rsync a readonly copy [13:36:56] i gueUUUues [13:36:58] oh [13:37:03] rsync ok. [13:37:03] To the labstore [13:37:07] thought you were going to say nfs [13:37:07] That is used for dynps [13:37:28] Eventually it will be served by nfs in labs but not in the 'source' [13:37:44] YuviPanda: how about just a writable rsync module accessible in labs from prod network somehow [13:37:54] then people can push [13:38:05] and it doesn't have to be automated [13:38:28] Because that will need a hole opened up maybe? [13:38:34] Do the stat hosts have public ips? [13:38:35] wouldn't you need a hole anyway? [13:38:37] no [13:38:59] No wouldn't need a hole [13:39:06] Because we already have this setup working [13:39:08] oh [13:39:08] For dumps :) [13:39:14] when you say mountpoint, what do you mean? [13:39:25] Oh that was stupid [13:39:28] Just a directory [13:39:38] /srv/super-public-in-labs [13:39:56] Put things there and they will show up in labs soon would be the interface [13:40:09] hm, why does that not need a hole? [13:40:15] oh because something in labs already rsyncs from prod? [13:40:35] ottomata: no because you are rsyncing to another prod hosts [13:40:39] Labstore1003 [13:40:42] OHhhh [13:40:44] oh [13:40:44] got it [13:40:57] oh, coudl you just set up a writeable rsync module there then? [13:41:00] rather than doing it automated? [13:41:48] ottomata: what do you mean by writeable rsync module! [13:41:49] ? [13:41:59] rsync daemon with a read-only no config setting [13:42:01] um [13:42:04] Also you clearly mean 'rather than doing it ottomated' :p [13:42:22] hadoop dfsadmin –refreshNodes [13:42:24] oops [13:42:25] hadoop dfsadmin –refreshNodes [13:42:29] ack! [13:42:30] pAste! [13:42:33] Haha [13:42:33] https://github.com/wikimedia/operations-puppet/blob/production/modules/rsync/manifests/server/module.pp [13:43:07] So we will have an rsync server in labstore and you can push to it whenever? [13:44:30] ya [13:44:34] you can restrictit if you like [13:44:37] maybe only from analytics network? [13:45:00] So I liked the 'directory that gets synced' idea better because it is super easy to audit [13:45:04] Just do an ls [13:45:16] It is also super easy for researchers to use - just a cp [13:45:26] hm, you can still ls the dir on labstore, no? [13:45:40] cp is easy, but it means they will have to maintain their files on stat1002 in multiple locations [13:45:43] Ya but researchers don't have access to labstore [13:45:54] hm, true [13:45:54] as in [13:46:04] Can't they symlink [13:46:18] i guess if your rsync copies symlinks as files [13:46:19] :) [13:46:27] YuviPanda: the ls argument is a good one [13:46:31] i'm cool with that [13:46:35] Ok :) [13:46:42] we actually do that for a couple of stat100[23] -> stat1001 rsyncs [13:46:49] stat1001 has a public webserver [13:46:53] Right :D so just need to extend that [13:49:36] ya, so, YuviPanda, stat1002 and stat1003 already have a readable rsync modules on /srv that other stat boxes can read [13:49:43] in role::statitistics [13:49:44] ottomata: wanna file a bug and respond? [13:49:46] Right [13:49:47] AHHHHHH [13:49:48] ok [13:50:09] I can help review and stuff but can't own this unfortunately [13:50:19] Too many things and I'm doing a 51h train ride skon [13:50:21] Soon [13:50:29] OK FINE [13:50:30] hehehe [13:53:33] ottomata: :D thanks! [13:54:14] YuviPanda: what is the hostname? [13:54:23] labstore1003.eqiad.wmnet? [13:54:26] Ya [13:58:42] YuviPanda: where should hte cron job go? [13:58:43] in pupet? [13:59:23] ottomata: dunno :D [13:59:30] Somewhere? [13:59:32] :D [13:59:36] uhmmMmm [13:59:59] Am about to go afk for a bit - need to get on Bart [14:00:08] I can look in about 30m if you haven't found it [14:00:29] k [15:10:45] Timezone-appropriate greetings, scientists. [15:39:44] GOod morning guillom ! [15:39:53] hey halfak [15:39:54] tz-apt greetings to everyone else :) [15:41:24] This week has flown by. [15:41:36] Agreed. [15:41:43] Both good and bad. [15:41:56] I barely did anything this week and feel bad about that. [15:42:27] Yes, I'm glad it's Friday because the week-end is almost there and I can rest, but OTOH… what hare said. [15:43:18] "rest" [15:43:20] "work for free" [15:43:42] but this is going to be a good weekend; friend is coming up from VT, followed by party, followed by Lughnasadh [15:43:51] weren't you just in vermont? [15:44:03] yeah, and we drank all their beer [15:44:06] so now they get to drink ours [15:44:07] ironholds! i get to go to central wisconsin in september [15:44:43] drink some New Glarus while you're there [15:45:00] (unless you don't drink beer, that's cool too) [15:45:08] i drink lots of things [15:45:33] And now my article is in the next Signpost, and soon the WM blog and a French news site, and I'm basically like "Can my 15 minutes of fame be over now, let me crawl back to my cave". [15:45:57] yes; this is a fairly ironic thing to become famous over [15:47:04] * Nettrom makes a note to read guillom's article [15:47:25] started reading it, but didn't finish, but was a good and interesting read [15:47:26] hare, I'm sorry [15:47:33] guillom: do you do that thing i do where you do not use superlatives unless you damn well mean it? [15:47:35] oh, Nettrom, you're into suggestion systems [15:47:47] ever heard of Franz Guenthner? [15:47:47] like before i can say something is the best X, i have to run through every instance of X in my mind and do a comparison [15:47:56] Ironholds: name doesn't ring a bell, no [15:47:57] hare: Yup. [15:48:00] okie :) [15:48:05] hooray for being hyper-literal! [15:48:08] hare: Same for "always" and "never" [15:48:13] yes!!! [15:48:57] Which is why I say a lot of "almost" and "mostly". [15:49:37] * guillom should go to the 5th floor for the next meeting. [15:50:00] Ironholds: he sounds interesting, did you read a paper of his? [15:50:23] (noticed he worked for FAST, that's some memories of old Norwegian software) [15:51:12] Nettrom, naw, he got in touch because he wants to build us a thing or ten [15:51:24] I just got off a call with him and he's writing up a proper proposal [15:53:00] oh wow [15:54:27] Ironholds: ooh, curious to know what he's proposing, should be interesting given his background [15:55:32] Ironholds: http://www.fontspace.com/m-fonts/m-1mn [15:56:04] jaufrecht, sorry? [15:56:08] Nettrom, yeah, it should be fun [15:57:15] different 8 and 0 characters [15:57:42] just went through a phase of finding my perfect monospace font. [17:43:40] Howdy, all, I could use some help findng some data - hope this is the right group/channel [17:44:42] SVentura, sure [17:44:43] hi SVentura [17:44:58] if we've got it to hand we can give a link, otherwise you should file a Phabricator task as is usual [17:45:01] Yay! what is the best way/source to find out what type of data/content is most accessed on WP? can we pull it by geographic regions? by Languages wikipedias? what are the most trafficked pages? [17:45:19] what do you mean by "what type of data/content"? [17:45:37] and we can filter by all of those things, yep [17:45:47] @ironholds: topics and subject areas [17:46:05] ah. Not easily, but it is doable [17:46:23] that is not something we automatically extract, though, so it would have to be a data request rather than a quick question and pointer to a link :) [17:46:26] @ironholds: should I fill out a request on Phabricator? [17:46:32] got it [17:46:37] yep! halfak, where do R&D cards go these days? [17:46:54] phab research-and-data [17:47:27] I really need that data in hand before moving on large project so will file a proper request - [17:48:00] should I use the research-and-data? anyone in particular I should ping there? [17:48:24] SVentura, drop it in the backlog there. [17:48:24] it's sventura! [17:49:05] @halfak @ironhold: will do. thank you both! [17:49:21] @ironholds: thank you! [17:49:56] halfak: looks like I'm not getting new labsdb machines afterall :( [17:50:01] * YuviPanda shall start hunting for budget soon [17:50:20] FYI: I wouldn't expect an ad-hoc data request (in general) to be quickly prioritized over our own work, so make sure you argue the merit of the work -- not just that you want it done. [17:50:24] SVentura, ^ [17:50:56] YuviPanda, :( so does that mean it'll take longer to get to the place where we can host an article quality table for access in quarry/labsDB? [17:52:19] halfak: possibly. I haven't spec'd that out yet [17:52:57] Gotcha [17:54:10] halfak: a good chunk of that is also just time and prioritizing, I guess. [17:54:26] I'm trying to prioritize for research work that I'm the bottleneck for [17:54:40] and the article assessment stuff I"m not really, but I just wanted to do it for fun but priorittiesss [17:55:19] YuviPanda, yeah no worries there. Just figuring out how lack of LabsDB machines is going to affect my work. [17:55:25] I don't think it'll be too much of a problem. [17:55:32] I still need to make the dataset to load into the DB :S [17:55:34] no don't think so [17:55:38] heh [20:23:29] halfak: in response to the Trello note you left, when you say "researcher" flag, you mean adding the person in wmf-researcher access group? [20:43:52] leila, nope. I'm talking about the 'researcher' user group on-wiki [20:47:31] can you link to it halfak? [20:47:54] leila, https://en.wikipedia.org/wiki/Wikipedia:User_access_levels#Researcher [20:48:41] https://en.wikipedia.org/wiki/Special:ListUsers/researcher could use some cleanup :) [20:48:44] ooo! I did not know this group exist. thanks halfak. [20:49:28] never put a link in IRC guillom. you have to clean up right after. ;-) [20:49:40] Sure. Dario said he was worried about handing it out for advance API access since it allows you to view deleted page metadata, but I pointed out that such data is already available through labsDB and Quarry. [20:49:40] :D [20:50:04] yes, I see your point halfak. I'll follow up on the card. thank you! [21:12:06] halfak: question about API calls for deleted pages. Making 80K calls requires around 2-3 days. Do you have suggestions on how this can be done faster? [21:25:39] 10Quarry, 7HTTPS: Quarry should be HTTPS-only - https://phabricator.wikimedia.org/T107627#1499757 (10Legoktm) 3NEW [21:36:57] 10Quarry, 7HTTPS: Quarry should be HTTPS-only - https://phabricator.wikimedia.org/T107627#1499816 (10Legoktm) In the meantime... https://github.com/EFForg/https-everywhere/pull/2353 [22:10:21] I [22:10:29] 'm off for the weekend. Have a good one folks [22:10:30] o/ [22:10:45] \o