=== rogpeppe1 is now known as rogpeppe [12:08] rick_h_: morning, I filed two holidays [12:08] frankban: cool looking [12:09] frankban: approved [12:10] frankban: thanks for getting them in pre-planning poker so we can schedule cards [12:10] rick_h_: thank you, I guess we are doing the poker today? [12:11] frankban: yes, I've updated the spreadsheet and it should be a bit shorter [12:11] we've got a lot of carry over from last cycle [12:11] so just left those out [12:12] rick_h_: cool, how was the pycon? [12:12] frankban: it was cool. I did a hand ful of talks but talked with a lot of people. Was cool, ran into one of our applicants for the new team and talked with him some. [12:14] rick_h_: cool, heh, interviewing time also at a conference ;-) [12:14] :) [12:16] rick_h_: for trusty charm I'd be inclined to drop legacy server support (not removing the code, just documenting that legacy server only works on precise). What do you think? [12:17] frankban: +1 [12:17] cool [12:47] morning [12:49] jujugui: my irc client is reporting it cannot verify irc.canonical.com. anyone else seeing this? [12:49] bac: not at the moment [12:50] i'm sure they just reissued the cert and it is a-ok. but was curious. [12:51] bac: no, it works well here. speaking of ssl, can we safely assume the openssl in trusty and precise is patched? we use that in our gui charm [12:52] frankban: yes, i'm sure it is. [12:52] frankban: yes, there should be a security update. The issue we might want to chat to IS about is that those aren't forced on deployed charms that I know of [12:52] bac: cool, good to know [12:52] and I wonder how that effects those clouds running out there with an unpatched SSL gui [12:52] frankban: unfortunately, last time i looked, we were not using the upstream patch, so our package still starts with 1.0.1f, which is the bad version [12:53] rick_h_: I guess that depends on the cloud image installed by juju [12:53] so the patched version is 1.0.1f-1ubuntu2 [12:53] bac: ack, yeah that can be confusing [12:54] i'm sure they updated the cloud images [14:00] good morning everyone [14:00] morning [14:01] hatch: if you get time I want to catch up and see where we're at and what other cards we need ot finish state [14:01] hatch: and get tests going that test changes to state, not just the initial parsing [14:02] yeah those tests are going to be coming in the current branch [14:02] I'm hooking up a new navigate event called 'changeState' and it's working really well [14:03] lots of odd legacy stuff though that I need to workaround [14:03] morning Makyo [14:04] rick_h_ will after our standup work? Give me a bit to get back into the groove [14:04] :) [14:26] hatch: rgr [14:26] looks like a good thing for all those lenovo users out there with the massive bricks :) https://www.kickstarter.com/projects/215201435/dart-the-worlds-smallest-laptop-adapter [14:27] rick_h_ how goes the rest of the conf? getting some bookie stuff done? [14:28] hatch: working on it, almost caught up on GSoC pull requests :) [14:29] haha - did you end up picking up someone? [14:29] hatch: yea, we'll have two students. [14:29] got the slots from google and the 21st they're announced [14:31] ahh cool, so how long until it sells to Facebook for 17 billion? [14:31] 10000000000000 years [14:31] hmm, that's a lot of years [14:32] so it's not a very good investment opportunity is what you're telling me... [14:33] guihelp: review needed on https://github.com/juju/juju-gui/pull/236 [14:34] hatch: did you read the dart fine print about macbook support? they have to buy a real apple power supply to cannibalize the connector, passing the costs on to you. [14:36] bac haha I didn't, I don't really find the mackbook ones -that- big [14:36] it would be nice if the cord was thinner though and retracted [14:37] i think i'm going to save my $168 and not have a brightly colored power supply [14:37] I end up twisting the crap out of that skinny cable.....of course the one that's not replaceable [14:44] kadams54: linter is angry with you [14:44] Yeah, I'm fixing [14:44] coolio [14:44] One missing comma and indentation all outta whack, should be easy fixes [14:50] jujugui call in 10 [15:01] Trying to get plugin installed [15:03] Rebooting this computer real quick. [15:05] rick_h_: my fix drops gui-x and gui-y from annotations. is there *anything* in annotations we'd want to present to ES? iow, just drop all annotations. [15:06] bac: no, I don't think anything [15:37] jujugui Since I'm already stopped from the call, I'm going to duck out now, then work through lunch. Have to get the receipt from the plumbing work to the sewer co. to get the $450 bill canceled. [15:38] Makyo: good luck! [15:56] jujugui after a discussion with rick_h_ it's been decided that the state changes and left inspector will be landed before vegas and be separate from the machine view and deployer. If you have any concerns/comments about this let me know [16:33] rick_h_: trying to deploy from branch in trusty charm exits with an error. I think the failure is this one: http://pastebin.ubuntu.com/7250529/ [16:36] frankban: hmm, ok. I'll have to look at the process there. I don't recall the reason/what that was meant to do [16:36] I'd expect that to create a master branch from the origin master [16:36] but I'm assuming we git cloned and have master already? [16:38] rick_h_: the code path seems to be this: http://pastebin.ubuntu.com/7250546/ [16:38] rick_h_: "git branch" on the source directoty just returns develop [16:38] oh [16:39] right, ok. So it's meant for generically checking out a specified branch [16:39] and it's trying to get master [16:40] http://stackoverflow.com/questions/945654/git-checkout-on-a-remote-branch-does-not-work ? [16:40] rick_h_: yes, because the test uses "https://github.com/juju/juju-gui.git master" as source [16:41] frankban: so I'm assuming a git update that's broken something. Wonder if that stackoverflow git remote update [16:41] will help? [16:41] it does a fetch, but not this remote update (not run across that before) [16:42] rick_h_: on that same branch, that does not seem to work [16:42] I get the same error [16:42] and the test worked on precise, so I guess this is a new git change [16:43] frankban: http://www.btaz.com/scm/git/fatal-git-checkout-updating-paths-is-incompatible-with-switching-branches/ [16:43] ? [16:44] frankban: yea, I mean I see a ton of people with this error, tring to parse out what's the root. Seems like it's a lovely unhelpful error [16:44] rick_h_: "git remote show origin" does not include master [16:45] trying pull [16:45] frankban: oh, a git fetch should show that [16:45] and that's done in the snippet before the failing command [16:45] I'd expect it to show [16:45] git branch -a ? [16:47] rick_h_: http://pastebin.ubuntu.com/7250586/ [16:57] frankban: ok, will pull down the charm source and see. Not sure. Something definitely changed but not sure what atm. [16:58] rick_h_ go away it's your time off :P [16:58] :) [16:58] "just say no to juju" [16:58] haha [16:58] heh, hacky hacky [16:58] rick_h_: thanks. EOD, please feel free to send me an email if you find something, otherwise don't worry, I'll take a look at it tomorrow morning [16:58] frankban: will do [16:59] thanks for the update, sucky that git broke something on us [16:59] rick_h_ is it hacking time now at the conf or are there still talks? [16:59] thanks, have a nice evening! [16:59] I remember needing that work because working with a shallow-clone needed extra work [16:59] to be able to pull some branch down [17:00] cya frankban [17:20] oh man do we ever need a code coverage tool [17:33] Hey, has anyone ever figured out what test-prod crashes in Vagrant? [17:38] not yet no....I can't run debug either [17:38] I can only use test-server [17:38] ^ kadams54 [17:38] debug didn't crash for me [17:39] But I'm not entirely sure it finished successfully either [17:39] luuuuucky [17:39] It just stopped running [17:39] oh....yeah that's what it does [17:39] It seemed to get through the majority of the tests, but then it just stopped. [17:39] Prod at least had the decency to crash with a PhantomJS crash error [17:42] haha yeah - tbh I can't figure out the issue, I can run individual tests just fine [17:42] but it seems like running too many tests causes it to fail [17:42] we probably need to fix our single big index file of tests [17:42] but that's just a guess [17:42] oh... and the phantom error is entirely useless too....at least mine was [18:49] so hows everyones day going? [18:49] rick_h_ those MS emails were from you, correct? [18:50] hatch: ? [18:50] MS emails? [18:50] the security code stuff [18:50] oh yea [18:50] because I'm in CA it didn't want to let me into the azure env [18:50] yeah us Canadians are bad eggs [18:50] :P [18:56] rick_h_ so with your vim you use jj for esc....do you not use home row as the arrow keys then? [18:57] hatch: yes, I nop my arrow keys [18:57] so doesn't hitting jj then keep triggering escape while you move around? [18:57] hatch: no, it's only a shortcut in insert mode [18:57] it does nothing in command mode [18:57] but to move down [18:58] so you can't move around in insert mode then? [18:58] no, I jj to command mode and move around using normal vim-isms [18:58] jcsackett: could you review https://codereview.appspot.com/87710043 [18:58] moving around in insert mode is kind of meh [18:59] insert mode is for typing, moving quickly by word/line/paragraph are all command mode [18:59] ahh ok [18:59] makes sense [19:00] jujugui lf a review on https://github.com/juju/juju-gui/pull/237 [19:00] bac: yup. [19:22] rick_h_ the simulator does not create containers in machines - this might be something we'll want to plan for. It's making it a little more work to QA Huw's branch [19:23] hatch: rgr [19:24] hatch: one day thing you think? maybe add a card to our on deck and I'll try to sched dit [19:25] rick_h_ yeah - frankban is probably the best to get it done the quickest because he is the most familiar with the machine db stuff [19:25] anyone else might take a little bit longer because of needing to get up to speed on the formats [19:25] just going by the branches landed [19:26] card created [19:26] project 1 - on deck [19:26] hatch: ty [19:27] do folks know off hand how long it takes for the juju-gui to injest a personal branch? [19:27] arosales: usually 15-30min [19:27] arosales depends on the machine but a couple minutes [19:27] specifically if I upload a branch (in the right format) to LP, when can I expect to be able to find it in the gui? [19:27] ohh [19:27] sorry I missunderstood [19:27] :) [19:28] what rick said :D [19:28] hatch, rick_h_ thanks [19:28] arosales: if you don't see it by then let us know and we can look into it [19:28] arosales: http://manage.jujucharms.com/heartbeat shows the last ingest time at the bottom [19:29] for the too much info version of the answer [19:30] rick_h_, ah nice thanks for that link [19:30] :_) [19:30] :-) [19:30] rick_h_, naw exactly the info I was looking for [19:39] arosales: it is a 15 minute cycle, so best case is 15 minutes, worst case is 30. [19:46] Makyo are you able to qa my branch? [19:47] bac: sorry for the delay. lgtm, with minor comments. [19:47] thanks jcsackett [19:53] bac are able to review/qa my branch? [19:54] hatch: i might can. let me get this thing landed so i can do a release to charmworld this afternoon [19:54] oh ok np [19:55] hatch, kind of heads down at the moment, sorry. Found a big bug in my branch. [19:56] alright - I'm going to branch off of it and keep trucking [19:57] ok hatch on it now [19:57] oh that was quick [19:57] haha [19:57] well, i got it submitted and jenkins will take a while [19:59] bac, ack thanks. So I could catch it every half hour. Not bad -- thanks for the info [20:01] oh ok :) [20:10] running to the airport [20:18] hatch: gone? [20:42] So much for that bug. The method was more complete in my head than on the screen, just forgot to type a line. [20:54] bac thankjs for the review [21:15] rick_h_: for your viewing pleasure: RT 69317 -- all of our resources on canonistack are hosed. [21:16] bac: huh?! [21:17] bac: oh! yea, I meant to bring that up on the call today [21:17] bac: I'm darn tempted to use our azure account to bring up another QA, we can try to use the new dns name we want. [21:18] rick_h_: huh. perhaps that's a good idea. we'd need two, though, if we want jenkins too [21:18] bac: we'll we've got a jenkins install running [21:18] rick_h_: but it is broken too [21:18] we can build onto it with more jobs ci.jujugui.org:8080 [21:18] that's what we run on azure currently [21:18] oh, you mean another [21:18] gotcha [21:19] but then we'd have to maintain it, so not sure if there's plans to bring back the old or if it's on us to get it back running [21:22] * bac -> dogwalk. [21:40] I love git stash [21:42] it's like a mini secret diff storage [21:49] git stash is dangerous [21:49] I've stashed things in there and forgotten about them [21:49] Like a squirrel [21:50] yeah, then it's like a present! [21:50] a surprise present! [21:50] haha [21:50] kadams54 what are you working on today? Got any python hacking projects? [21:51] Yeah, working on docs for my feature flags library [21:51] Also trying to get this branch landed for unit tokens :-) [22:10] Tests fly by with 16gb ram. [22:12] Makyo heh, how long? [22:12] I'll tell you when they all pass :P [22:21] man I -really- wish we could figure out why charmworld is so slow at responding to OPTIONS [22:21] it's almost 1s sometimes [22:35] jujugui hierarchical commits for review/qa: https://github.com/juju/juju-gui/pull/238 [22:38] Makyo I can do it [22:38] hatch, thanks [22:39] It runs the commits in the proper order, have a card to wait on RPC callbacks. [23:04] Makyo where is the test that you modified? [23:04] ""Fixed test that was doing a deep-equal on an object that had changed."" [23:05] or was this an in-code fix? [23:05] just want to make sure you didn't forget a file or something [23:05] hatch, Sorry, that was a squashed commit message. And my comment seems to have gotten lost. Line 207 has no 'parents' attribute. [23:05] I'll re-add the comment. [23:05] It should gracefully handle both an empty array and a missing attribute. [23:06] ahh ok looks good [23:06] lgtm [23:06] ish [23:06] :) [23:06] Yeah. I'm pulling those functions out now, but it's causing it to infloop. [23:06] Hunting that down. [23:07] I started with recursion, but that was really slow, and causing more headaches than it was worth, thus the stack-based approach. [23:08] yeah I like it - as I was reading I was wondering why you didn't go recusion, but this is much easier to read [23:09] It's super fast, and flexible enough for us to have $MAXINT possible levels, as opposed to $MAXSTACK [23:09] But I guess that's maybe premature optimization :) [23:09] haha yeah - I kind of doubt we will ever run into an issue where stack size is causing problems [23:09] that would be one massive bundle [23:09] haha [23:10] Openstack, maybe :P [23:10] I'm not sure about the "Wait until each level of the hierarchy has fully executed" limitation of the level by level execution follow-up [23:10] what if one item in there gets hung up, then the rest of the stack fails? [23:11] We could timeout? [23:11] something to think about anyways [23:11] Yeah. [23:11] Without that, though, we risk trying to create a relation between two services, one of which has failed to deploy, and then you've lost all your changes. [23:12] yeah I'm thinking of the case where you have three "stacks" so no relations between anything, and one machine fails to start, I would still want the other two "stacks" of commands to execute [23:12] maybe 'fails to start' is the wrong thing [23:12] maybe 'juju fails to recognize the command' [23:12] or something... [23:12] :) [23:12] Yeah. [23:13] Those commands would still be in the change set, but it would pause in executing. [23:13] We'll think through it. It's just comments for now. [23:13] yeah - I'm not sure if it's really a real problem or not either [23:14] Yep [23:15] Man, I spent so much of the weekend writing C++ I'm having a hard time getting back into Javascript. [23:15] I keep thinking, "Man, don't I need to dereference that?" [23:21] lol [23:21] C++ eh? what would make you want to dive into C++ again? [23:27] I was playing around with that time lapse software, but it's really old abandonware written in not-very-good C++. I figured learning Qt wouldn't be a bad idea, given how much QML Ubuntu uses, so I decided to reimplement it in Qt from wxWidgets and just steal all the camera-handling code. [23:27] I got it functional by Sunday afternoon, which was nice. [23:27] Took this with it: https://www.youtube.com/watch?v=IlqyT0Jj1Rk [23:30] oh very cool [23:30] good work [23:32] Heh, thanks. [23:32] Good portfolio piece, too, though I have plans to figure out packaging at some point, make it more widely available. [23:32] Currently need QtCreator to build it. [23:32] Or, well, qmake, plus deps. [23:33] Alright, I'm gonna duck out to dinner. Mom's in town. Will re-push after.