=== rogpeppe1 is now known as rogpeppe === gary_poster|away is now known as gary_poster [13:32] rick_h_, I was going to say thank you to Hannah for her email but wanted to confirm that you can add things on allhands. could you confirm? [13:37] gary_poster: looking now [13:37] ty [13:39] rick_h_, radio silence on the CI email. :-/ worried that it means tl;dr [13:39] sigh [13:39] gary_poster: yea, I was thinking of doing personal pings to at least curtis/john [13:39] yeah would be great, thank you [13:45] gary_poster: objective sheet submitted [13:46] awesome, thanks rick_h_ [13:46] thanks for getting it opened up [13:46] morning all [13:46] up early hatch ? [13:47] are you still 3hrs back? [13:47] so on Friday I picked up a digital ocean account, it would be much cheaper to run our CI there than on ec2 - unfortunately it doesn't support Juju until we can use Manual Provisioning in the next version of Juju [13:47] or two I guess [13:47] rick_h_ I'm ALWAYS GMT-6 [13:48] and not cranky about it all! [13:48] lol [13:48] -always- [13:48] is that better? [13:48] :P [13:48] it's 7:48 here [13:48] hatch: yea, I've got a free open source account on rackspace but no juju there either [13:48] hatch: ah, so you're only 1hr back now [13:48] so much better :-) [13:49] if we actually used DST it would make no difference, it's still dark an hour later haha [13:49] heh === BradCrittenden is now known as bac [13:59] hi frankban, using quickstart the gui deploys but then there is a connection refused error when connecting the web socket. have you seen that before? [14:00] bac: is it different from what you reported yesterday? [14:00] yes [14:00] frankban: it is using --no-browser and isn't the auth-connect problem. will paste [14:01] bac: cool thanks [14:01] frankban: http://pastebin.ubuntu.com/6514606/ [14:01] i assume it is the connection to begin deploying the bundle [14:04] * bac changes his ec2 connection password now... [14:04] :) [14:05] bac: it seems so, and it's weird, is this trunk? [14:05] frankban: yes [14:11] bac: what happens if you re-run the same command, reusing the existing env? [14:12] * bac tries [14:12] frankban: it succeeds. [14:12] bac: :-/ [14:13] new Tim Horton k-cups - are almost exactly like in the store [14:13] innnteresting [14:13] that says a bit about what's in the store [14:13] hey gary_poster Happy Birthday! [14:13] :-) thanks hatch [14:14] bac: so my guess is that, even if juju marked the service as started, the guiserver was not yet up and running... [14:15] no problem! Where is the party??? [14:16] yeah, hbd gary [14:16] bac: so a possible solution is that app.connect retries the "juju.connect(api_url)" step several times before giving up [14:16] frankban: ok, i'll file a bug. do you want to try to reproduce? [14:16] bac: I'll do, are you using local or ec2? [14:20] frankban: ec2 [14:21] thanks bac :-) [14:24] rick_h_ while granted it's not the best coffee around, a k-cup is no different than a drip coffee maker albeit with a lot more waste [14:40] marcoceppi: I have a WIP branch with several comments explaining different bits up for your pre-review at https://codereview.appspot.com/36190043/ [14:54] frankban: you may be interested to know i discovered yesterday that with no default-series set juju now creates precise instances. in the past it had defaulted to the OS of the local machine, saucy in my case. this seems like a sane change. [14:56] bac: cool, so we only have to handle explicit default-series [14:59] benji: thanks! I've not used lbox, or codereview, so I'll dig in, in a few mins [14:59] gary_poster rick_h_ https://github.com/arturadib/shelljs you might find this interesting [15:00] cool [15:01] hatch: heh, I'm not generally all for reinventing X with Y stuff. Go JS though I guess. Probably helps with the grunt stuff. [15:01] well it beats trying to write stuff in make/bash :) [15:03] if you spent as much time finding a way around make to just read part of the book about make, you'd be a jedi master :P [15:04] HAH! that doesn't fix it's shortcomings lol [15:05] yet the clones never seem to get its advantages either [15:05] big money big money no wammy's! Cmon launchpad, third time is the charm! [15:05] lol [15:05] <3 that a show that's not been on the air for 20 years is still a know-able reference [15:06] haha yeah it's awesome [15:06] well the only real advantage to bash/make is that you don't need to install any deps :) [15:07] so any clone will never be able to get that advantage, unless we can convince peeps to install node/python by default in the distro hah [15:07] hatch: that statement shows you don't know enough bash/make if you think that's the only advantage :) [15:08] I guess...what's another one? [15:08] multi pass dependency resolution? INI variable overrides? speed? target globbing? [15:09] well you can do those from any higher level language as well [15:09] and for ones which you can't you can always shell out for that one little bit [15:10] can do vs is implemented in X? [15:10] the thing is all the clones start out solving the easy stuff that anyone can do. It's the more powerful useful things I get annoyed with the clones for [15:10] it's not about adding features, just "anything but X" mentality [15:11] right, well some things just haven't been implemented in the higher level languages so in those cases it needs to be shelled out for [15:11] so you have to know shell anyway :) [15:11] right, but that doesn't make it enjoyable to write [15:11] carry on! /me goes back to hacking [15:13] jujugui looking for a review and qa https://codereview.appspot.com/36500044/ plz and thx [15:14] hatch: I can peek at it [15:14] thanks [15:15] it's a pretty easy review/qa just a little tedious to qa [15:16] hatch, would it be easy to also do a better job of passing the Juju Core error message through? [15:16] maybe that's uin the branch already--just going by your cover letter [15:17] gary_poster I think we fixed that in a previous branch, let me take a look [15:17] cool ty [15:18] gary_poster good catch I'll add this to the branch [15:19] ^ rick_h_ [15:19] thanks hatch [15:19] hatch: k, will ignore it for a little bit more then [15:22] rick_h_ I'll have to spin up an LXC instance to test this, it'll be a bit [15:22] hatch: rgr [15:36] so my router reset (for whatever reason) and assigned all new ips across my network :/ [15:37] i've probably been h4x0r3d [15:50] jujugui call in 10 [15:50] ty [15:51] gary_poster: do you have time for a pre-imp after the call? [15:51] frankban, yes, sounds good [15:51] thanks! [15:57] frankban: i filed bug 1257349. were you able to reproduce? [15:57] <_mup_> Bug #1257349: Quickstart fails if guiserver not up yet [15:58] jujugui call in 1 [15:58] bac: not tried yet, thanks for filing the bug [15:58] frankban: new card too [15:59] cool [16:00] benji Makyo ping [16:10] rick_h_ ok the new feature has been added but you cannot qa it as far as I know :) [16:17] hatch: ok, so lbox is done, go review [16:17] wha? [16:18] hatch: so was your last comment means that lbox is done and you'd like that reivew? [16:18] review? [16:19] yes :) [16:19] the feature I added will only show up if there is difficulty in deploying a charm, so I'm not sure how to simulate that [16:19] the name collision feature? [16:20] that one is easy to do, but now the validation is done in the GUI so the server will never get an invalid name [16:20] k, will look at it [16:23] thanks [16:34] rick_h_: I looked at the lander a fair bit I only had small, nit-picky thoughts, so either I'm getting soft, or I like it. [16:35] benji: heh, ok [16:44] cd .. [16:44] oops [16:45] rick_h_: would it be sane for me to do the "update browser tests to use sandbox env" card? [16:45] benji: yes, that should be something we can do purely in the gui functional tests themselves. [16:45] rick_h_: cool, what does it mean? [16:45] benji: it might bread out into a couple of cards, let me know if you want to pre-chat on it [16:46] benji: hangout? [16:46] rick_h_: sure; shall I create one? [16:46] sure thanks [16:46] rick_h_: or benjiyork.com/chat worked for you, right? [16:46] benji: rgr [16:47] rick_h_: hmm, "full" (not sure how that happened). Try this: https://apprtc.appspot.com/?r=56434678 [16:48] rick_h_: ok, I'll make a hangout [16:48] sec, retrying [17:06] hatch: review inbound [17:06] thx [17:10] man, I miss the recently closed tabs ui in the current chrome dev stuff :( [17:13] gary_poster, Makyo: six "envs management" cards created, the ones with my face [17:14] frankban, cool [17:18] rick_h_ that dropdown is gone? [17:19] https://twitter.com/BoredElonMusk/status/407921563011256320/photo/1/large [17:20] lol [17:22] so Makyo is that a real thing with musk? or is it a baseless (though still funny) joke? [17:22] Just a joke, I'm sure. [17:22] bac baseless? PLEASE! [17:23] :P [17:23] Given that his location is set to 'Mars'. [17:23] you never know [17:23] "Apple falls short of analysts expectations, stocks fall" <--- the problem with the stock market lol [17:23] nancy did set ronnie's schedule based on her numerologist [17:25] gary_poster rick_h_ the email from Maarten looks like a perfect usecase for our 'proposed' subordinate for the GH stuff [17:28] hatch: well he replied to the email i sent out with the notes of our work in progress stuff. [17:28] hatch: so he's aware of that already [17:28] oh ok cool - I try to follow emails but there are so many :) [17:29] yea, know what you mean [17:33] man that unit error branch was a 1 line fix and took an hour to qa [17:33] :/ [17:33] lol [17:34] I gota speed up that lxc machine somehow [17:39] umm [17:39] so we don't have any tests for removeUnit in the fakebackend? [17:39] how did THAT happen [17:39] :D [17:41] Magic? [17:42] haha [17:48] lunchtime! [18:08] hatch, rick_h_ agreed that it sounds like a good match [18:09] juju is so awesome [18:09] :) [18:17] rick_h_, should I volunteer us to Maarten? [18:17] I can also help with that if you guys need a grunt [18:18] gary_poster: well I was assuming he's looking for more. He replied to the original CI email and asked some questions and curtis jumped in explaining some use of subordinate charms. [18:18] gary_poster: I'm hesitent because I'm not sure where it's going. [18:23] rick_h_, yeah, I know what you mean, and we want to be done. [18:23] gary_poster: I guess I'm happy to talk about it, but his email seems very centric on setting up a workflow around java based things, auto creating jobs out of github projects, etc, none of which is in our current scope of stuff. [18:24] rt [18:24] ok might reply but will do so carefully. :-) thanks rick_h_ [18:24] if who ever does work on it want to see if we have some common ground I'm happy to coordinate that bit [18:25] gary_poster: rgr [18:25] cool [18:26] ahh I thought it was just jenkins plugins as subordinates [18:26] in the general sense [18:27] hatch: maybe I'm reading more into the email. I'm just trying to be careful :) [18:27] haha, yeah don't go get yourself sucked into writing Java [18:27] :P [18:37] marcoceppi I would like some assistance writing the http hook for my ghost charm so that I can hook it up to apache for load balancing so that I can release a bundle...any idea who I might talk to? [19:05] hatch: me? [19:05] That's right, Santa is gona be Canadian! http://gizmodo.com/canadas-getting-ready-to-claim-the-north-pole-as-its-o-1475643608 [19:05] marcoceppi you want to help? [19:05] I'll take it! [19:05] hatch: http hook should be pretty straight forward [19:06] not sure what you'll nee [19:06] well reading the apache charm it was saying I needed to write some apache directives and set them on relation-change ? [19:07] * marcoceppi becomes disgruntled [19:07] hatch: one sec, let me see [19:08] ok thanks - I could be totally misunderstanding it too [19:19] https://twitter.com/NodeSummit/status/407951852358348800 Go JavaScript! [19:40] weird, juju-quickstart that has a test that fails depending on what shell you run it in. [19:41] rick_h_ any idea if I can set up my work email for the juju project on github somehow? [19:41] due to the way the docstring has line-breaks when --help is run [19:41] bac that's interesting [19:41] it was frustrating. now it is interesting. [19:42] yeah I suppose I didn't have to go through the frustrating part :) [19:42] hatch: http://www.codeography.com/2011/08/05/project-specific-git-author.html ? [19:43] rick_h_ well that is for commit messages and such [19:43] the emails come from gmail [19:43] er [19:43] github [19:43] at least as I understand how it works.. [19:44] hatch: https://gist.github.com/BPScott/1366790 [19:44] hatch: see the link at the top of it [19:44] hatch: a quick search for github select email per project [19:45] oh awesome! [19:45] and now it is fixed [19:45] hatch: that's a good point thouogh. I'll add that to the notes/docs [19:46] rick_h_ somewhat in the same light - it might also be worth a note about setting up your 'user' for the juju git repo so that people don't commit using their personal emails and whatnot [19:57] can we remove python support yet? huh huh can we can we can we plz!!!!!!! [19:57] hatch: huh? [19:57] pyjuju sorry [19:57] :) [19:57] oh, heh [20:02] gary_poster: our call is now? can you invite my canonical me? [20:03] bac, finishing up with mramm [20:04] jujugui looking for a review/qa https://codereview.appspot.com/36690043/ plz and thx [20:05] On it. [20:05] Also, done with appointment. Looking for a card after review. [20:05] thanks Makyo [20:06] most of it is just tests [20:15] and now onto the bug I have been avoiding..... [20:15] *grumble grumble grumble* [20:29] hey gary_poster did you end up picking up that js book yesterday? [20:29] hatch yes [20:30] think it's worth picking up? The descriptions looked a little beginner-ish. I was thinking of picking this up as well http://shop.oreilly.com/product/0636920027713.do [20:30] iunno, it's so hard to pick up a js book these days :) [20:31] reviews are mixed [20:37] hatch, I think the functional one is worth picking up. if you haven't seen someone go deep with functional programming, I think you'll find it fascinating. The beginning is beginnery but it doesn't stay there, and the whole idea is to introduce a different way of thinking from the ground up. For the other one you linked, yeah, reviews seem not so good. [20:39] cool, picking it up now [20:39] now if i could just remember my oreilly information [20:56] hatch, chunk mismatch on your review (doing QA and going by diff for now, just fyi) [20:56] poop ok thanks [20:57] arg the issue I'm experiencing is not even possible! [21:08] bcsaller you look like you're bored and want to give me some info on databinding ;) [21:09] * bcsaller yawns [21:09] how can I help? [21:09] well I have this very odd bug https://bugs.launchpad.net/juju-gui/+bug/1253114 [21:09] <_mup_> Bug #1253114: Unit details left breakout sometimes empty when opening [21:09] as far as I can tell, it's rendering the template, the binding is being setup [21:10] but it's not updating the DOM to the values in the model [21:10] it's very intermittent [21:10] * bcsaller looks [21:11] so I was wondering if you had any insight into what might be causing it, right now I'm console.log debugging hoping to catch a glimpse of something out of the ordinary [21:11] I should mention that the binding IS attached because the DOM will update after the first rendering [21:11] hatch: this is with the simulator running though, it could be removing the unit, maybe adding it back later? [21:12] like if this happens and you turn off the simulator will it ever recover? [21:12] if it happens and I just sit watching the unit details page, it will eventually fix itself on the next delta for that unit [21:13] so this is *only* with the simulator? it could be a fakebackend issue with addUnit then? [21:14] I'm wary of saying its a databinding issue, not because I think that code is perfect, but because this seems like the type of issue that would show up other places [21:15] hmm [21:15] it's almost impossible for me to debug on a real env [21:15] as it appears to happen when the unit changes states [21:23] hatch: short of walking through the debugger with you I don't have any additional insights. I'd be tempted to explore if it is a difference in the deltas coming from the fakebackend [21:25] bcsaller ok no problem I'll keep stepping through thanks though :) [21:27] ok I found something... [21:27] when it doesn't work the delta passed to the updateDOM doesn't contain the attributes for the unit details [21:28] I"m just rambing now feel free to get back to your real work :) [21:49] bcsaller found the issue fyi - if we take this path http://bazaar.launchpad.net/~juju-gui/juju-gui/trunk/view/head:/app/views/databinding.js#L173 then the other bindings aren't updated so it's kind of a race condition. if no modelChangeKeys are provided it works as expected. [21:49] just thought you'd be interested to know [21:50] now...how to fix... [21:50] * bcsaller checks [21:51] hatch: should be an || in the condition, no? [21:51] basically instead of mixing it, it's filtering them out [21:51] == undefined || [21:52] hmm, no, maybe not [21:52] nope [21:52] right [21:52] :) [21:52] I'll fix it, no problem, just pointing it out [21:52] I'm actually surprised that we haven't run into this before heh [21:53] so this -could- have happened on a life env but almost never heh [21:53] yeah, odd [21:54] bcsaller my guess is that originally we wanted to update ONLY those keys but later decided that it should mix it in [21:54] and it was never updated [21:54] but somehow it just 'worked' for the 99.9% time [21:58] Morning [21:59] morning huwshimi [22:01] morning [22:16] hmm - fixed issue, cannot be sure [22:17] that's the crappy thing about race condition bugs hah [22:49] gary_poster since you were the other person to experience this issue think you could review/qa? https://codereview.appspot.com/36850043/