[00:03:00] atgo zacs call is over hangout? [00:03:04] PPena no [00:03:07] oh [00:03:08] you in the office? [00:03:12] can you send me the # [00:03:20] are you in the office? [00:03:21] im WFH [00:03:22] PPena: it's IRL [00:03:29] ahhh [00:03:34] IRL? [00:03:39] ok... hang on. can you just ring zac on hangout? [00:03:56] yeah i can send him the link [00:11:38] (CR) Ejegg: [C: 2] "You get an index! And you get an index!" [wikimedia/fundraising/tools] - https://gerrit.wikimedia.org/r/171977 (owner: Awight) [00:11:48] (Merged) jenkins-bot: add indexes for is_XXXX_donor -BUNT columns [wikimedia/fundraising/tools] - https://gerrit.wikimedia.org/r/171977 (owner: Awight) [00:12:11] I like to move them fast. [00:14:37] Everbody gets a car! [00:14:53] *Everybody [00:14:53] what is this [00:14:54] ? [00:15:14] that one time Oprah gave her whole audience new cars [00:15:51] oh shit [00:15:53] thx [00:15:58] http://www.oprah.com/oprahshow/Oprahs-Entire-Audience-Are-Surprised-with-New-Cars-Video [00:16:54] "it can happen to anybody" [00:16:54] also, bees: https://www.youtube.com/watch?v=xAhuSDRIDHE [00:17:09] Sometimes I wonder where I've been for 10 yrs [00:17:42] doing something worthwhile? [00:17:50] ejegg: K4-713 is fond of recalling when Bill Gates gave everyone in the audience malaria [00:18:07] * ejegg schnork [00:18:26] you mean Vista? [00:19:26] well, glad that didn't turn into everybody pummeling each other [00:20:23] ejegg: https://www.youtube.com/watch?v=ppDWD3VwxVg [00:21:31] oof, it's a bit of a letdown I just discovered [00:22:25] oh man, that was pretty awesome of him. should have held off delivering that last line [00:22:30] AndyRussG: well, looking at the WMDE and mobile stuff, I have no personal preference. I don't think the Cucumber is easily compatible with qunit. [00:22:54] ejegg: for real. the chortling audience is a bit sickening [00:24:07] awight: hmmmm I don't know Cucumber much... I do remember seeing it used for defining user interactions to test [00:24:47] In this case I'd want to like test all the code paths with different sorts of data as necessary [00:24:53] AndyRussG: it's a minor detail, but a really nice language [00:25:02] yeah I've heard! [00:26:03] My main worries with this change would be: 1) some edge case combination of cookies/banners/whatever that we don't think of but might think of and control for when doing the tests... [00:26:41] and 2) accidentally depending on some minor JS language detail that is incompatible on some browsers (like some native object method that isn't totally standardized yet) [00:26:45] AndyRussG: yes for sure [00:27:38] It's just nice having the tests include this readable format which says, "you don't have a cookie, and visit enwiki"... etc [00:28:15] AndyRussG: before I segued though, what were u saying about qunit? [00:28:16] Yeah! we could shoot for that, plus if there's already a banner framework in it... [00:28:23] Meh not much [00:29:01] u said it already works with CI, so can be run automatically with Gerrit changes on Jenkins? [00:29:30] I guess the question is, do we want actual unit testing of each javascript function, or more integration testing, where we measure the end effect on the client? [00:29:52] Oooh well both of courz [00:30:18] And testing on a variety of browserz and OS'z [00:31:08] yeah it looks like the mediawiki-core cucumber tests are performed using Selenium, which supports some of that [00:31:27] Heh I guess integration testing with a mock of our caching/cluster infrastructure would be just silly over-the-top... [00:31:31] https://github.com/wikimedia/mediawiki-selenium [00:31:55] probably yeah. we should just assume certain behaviors and mock that layer however necessary [00:32:19] does ops have some kind of integration testing setup? [00:32:28] hehe, I wish. [00:32:37] I can't even get Jenkins to run locally [00:32:50] apparently there is no staging server, even [00:34:25] AndyRussG: I don't think we have it running in the CI environment, but mediawiki-selenium looks like it will run under Windows [00:35:45] AndyRussG: if it's helpful, I'd be happy to dive into writing whatever Ruby support u need. [00:37:50] awight: thanks! [00:38:05] Jenkins runs on Windows? [00:38:25] mmm, java [00:38:57] we had that setup at my last workplace [00:39:00] And if we go by the book, Java would seem to be Windows? [00:39:03] AndyRussG: hopefully not... this is to just run Selenium [00:39:16] * awight bleaches eyeballs [00:39:30] Ahhhh right for testing windows-based browsers...mmm... [00:39:36] I was wondering why.... [00:39:51] e.g. $env:BROWSER="internet_explorer" # Windows PowerShell [00:40:22] seleniumgrid could let you test on windows while jenkins ran on *nix [00:40:28] * AndyRussG eats peanut butter like Homer Simpson [00:40:45] one hundred invidually-wrapped slices of American cheese... [00:40:45] ...with a thing "that... digs..." [00:41:06] * awight huddles under a desk [00:41:32] the individually wrapped cheese slices are coming!! [00:42:17] one hundred individually wrapped cheese slices on the wall, one hundren individually wrapped cheese slices... [00:42:37] take one down, pass it 'round... [00:55:27] * awight swears viciously about the key destroying Mingle card I am adding [00:57:00] oooh yeah nasty that one [00:57:27] glad VE doesn't do that :p [00:58:43] * awight turns green [01:21:59] I wonder if there's a differece between how you have to code for different JS testing frameworks [01:21:59] QUnit needs pretty atomic functions IIRC [01:27:21] hey andyrussg [01:27:34] just checking in - how's things? [01:27:35] have a great weekend, all! [01:27:35] hi [01:27:57] atgo: not bad, not as far along as I'd hoped tho [01:29:18] ok. i'm talking to k4 right now, and she's pretty concerned about the timeline for this [01:29:18] tomorrow will be a day tho [01:29:18] atgo: OK I understand [01:29:19] yeah... just about making things unstable under load, you know? [01:29:19] so... she's suggesting (strongly) a drop dead date for deploy on Thursday next week [01:29:32] atgo: that sounds extremely reasonable [01:29:47] Can we throw more resources (self) at this? [01:30:12] awight word, yeah. we should be a team on this kind of thing :) [01:30:17] * awight is just spaghetti looking for a wall [01:30:32] so, would it make sense to reserve a deployment window for thursday net week? [01:30:35] watch out for the cheese [01:30:58] hmm tough call, I don't know how quickly they fill up [01:30:59] atgo: I don't think reserving a window is a big deal [01:31:04] yeah we can do it Wednesday... [01:31:06] ok cool [01:32:10] AndyRussG is there more that i can do to help define the former "step 6" now "step 4"? [01:32:39] atgo: awight: one sec lemme pull up the doc again... [01:32:40] atgo: another important question: are you having this important chat over tequila? [01:32:49] haha just sitting in a conf room [01:32:53] blah [01:32:55] nvm [01:34:43] atgo: awight: hmm.. is there more we can do to find out any general requirements for persistent cookie info generally? That's what I wanted to scope out [01:34:52] oh right. [01:35:02] can you start an email thread about it? [01:35:14] awight: step-4-previously-step-6: "We’d be able to have series of campaigns that show first a full-screen banner and then top-screen banners without ever showing users more than one full-screen banner, and without losing banner impressions. Once again, some effort will be needed to figure out exactly how it’ll work." [01:36:11] yes [01:36:24] awight|gone: involves persistent cross-campaign info on client machine, also a few other use cases for something similar [01:36:59] atgo: same recipients as the existing thread, maybe plus whoever else was on other CN threads recently, I guess? [01:37:19] sure... i'm not sure who the stakeholders are on this [01:37:34] K yeah also a good first quesiton [01:39:04] i suppose centralnotice-admins is not the place then...? [01:55:38] AndyRussG i would think not [01:55:44] atgo: K [01:57:22] atgo: regarding working out more details, current-and-previous step 5 needs a bit of UI and feature details filled in [01:57:36] ok... i'm most concerned about the immediate one [01:57:38] that megan really wants [01:57:47] right, 4 [01:58:00] the rest of them, we'll figure out... but k4 is saying we've got a deadline for deployment before december so we can load test, so let's focus on what's right in front :) [01:58:16] logging off for a sec... brb [02:12:51] atgomez: regarding load testing, another test option (in addition to real-world load testing) might be to set up a special labs instance to do that... do you still have K4 with you? in case u want to run the idea by her.... [02:21:50] let's chat about it on monday... [02:21:59] have a great weekend! i have to run home now