[02:02] jujugui PR on loading relation indicator images https://github.com/juju/juju-gui/pull/218 (testing CI, too, tests fail on Air, can fix on ubuntu if they fail CI) [02:02] s/tests fail/console tests fail [03:32] No luck, will check in the AM === rogpeppe3 is now known as rogpeppe [11:12] frankban: morning [11:12] hi rick_h_ [11:13] frankban: heads up, I talked to wallyworld last night and he says he's tested the joyent provider in trunk (and it's backported to the 1.18 tree) so hopefully that'll work [11:13] frankban: and thanks for replying to his email on the megawatcher updates. Can you create cards in the backlog for us to update where we watch for those ip address changes? [11:14] rick_h_: yes I'll do. https://165.225.150.74/ is the GUI deployed on joyent \o/ [11:15] rick_h_: there is still bug #1300846 which affects the quickstart implementation. I'll check if it gets fixed for 1.18 [11:15] <_mup_> Bug #1300846: Juju crashes bootstrapping joyent

[11:15] * frankban lunches [11:16] frankban: awesome, I just saw a branch from him around the provider and wallyworld admitted it needed a "couple of tweaks to be perfect" [11:17] so not https://codereview.appspot.com/83540043/ [11:17] sorry, so looks like it's in progress and should land somewhere soon hopefully [11:17] frankban: very cool to have another provider working woot thanks for keeping up on that. [11:18] * rick_h_ runs to take the boy to day care back in a few === rogpeppe2 is now known as rogpeppe [11:45] back [12:30] Makyo: mroning, do you have time today for reviewing https://github.com/juju/juju-gui/pull/215 ? thanks [12:30] morning even [12:32] frankban: think he's still a couple hours away [12:32] frankban: I'll be relocating shortly when the cleaners get here but can start going through it if you need a second review [12:34] rick_h_: there is no rush, it was more like a message from the past, thank you [12:34] lol echo echo [12:37] :-) rick_h_: I created the two cards in backlog/deck [12:37] frankban: thank you much [12:54] rick_h_: thanks for the review. re comments: good suggestion [13:10] frankban: thanks for the update :) [13:44] morning [13:44] morn [14:00] hatch__: https://plus.google.com/hangouts/_/canonical.com/go-over when you get a sec [14:24] Are there any Handlebars helpers in the codebase to handle displaying ModelLists (or even ArrayLists)? [14:25] rick_h_ thanks for the review - ugly I know *hangs head* [14:25] it'll get better! [14:25] kadams54 negative, it's always been to specific [14:25] domain specific* [14:25] Like {{#each}} but using object.each() to iterate rather than a for loop [14:25] hatch__: yea, just want to make sure we know things are broken and note them to make sure we get them all fixed [14:26] kadams54 we usually convert it to an array/object then use the native handlebar helpers for looping === hatch__ is now known as hatch [14:26] kadams54: yea, often we have, in the view, a prepForTemplate() or something [14:26] kadams54: that builds a handlebars friendly data structure [14:26] Hmm. Would be nice if ArrayList had a function to spit out a plain list [14:28] kadams54: well you can map based on Y.Array.each [14:28] kadams54: but yea [14:28] kadams54: http://yuilibrary.com/yui/docs/api/classes/Array.html maybe? I don't recall [14:29] kadams54 typeof myArrayList._items === 'array' [14:29] Yeah, but accessing private members es no bueno. [14:29] yea, frowed upon. It'll be a smallish list [14:29] you're safe looping it out into a new strcuture [14:29] kadams54: as you'll probably want to turn the models into objects and such anyway [14:30] tweak things in process [14:30] Thinking about similar situations in python where there's usually a method on the iterator to have it spit out a list object [14:30] the _items property will -never- change, it's a YUI convention [14:30] kadams54: yea, I'd just add a prepForTemplate() handler [14:30] it's ok to access it [14:30] I promise :) [14:30] rick_h_: will do [14:31] grep _items in our codebase you'll se we use it already [14:33] jujugui: charmworld review please https://codereview.appspot.com/83620043 [14:33] kadams54: http://yuilibrary.com/yui/docs/api/files/app_js_model-list.js.html#l888 [14:34] Nice [14:34] http://yuilibrary.com/yui/docs/api/classes/ModelList.html#method_toArray for the prettier view [14:34] bac: can look in a bit [14:35] rick_h_: thanks [14:38] * rick_h_ heads back to the house from the coffee shop [14:50] jujugui call in 10 please kanban [14:52] hatch: looking at your commment [14:52] hatch: the code reads that this.environmentHeader is gone [14:52] hatch: replaced by var header [14:52] oh, it's added back in line 57 [14:53] hatch: any reason for the change/indirection? [14:53] minification [14:53] :) [15:00] jujugui call in 1 [15:01] rick_h_, hey what's the tldr on showing colocated services in the gui? [15:01] I feel like for 14.04 we should have a owncloud all-in-one bundle [15:02] jcastro: http://comingsoon.jujucharms.com/:flags:/il/mv/ in progress see you in vegas [15:02] whoa! that looks nice! [15:27] rick_h_: one thing we missed: the 1.18 change which no longer includes the unit public address in the mega-watcher breaks quickstart :-/ [15:28] frankban: ok, well that'll have to be the reason to force an upload update for quickstart [15:28] rick_h_: quickstart uses the watcher to retrieve the GUI address [15:29] rick_h_: also, without the change to the MachineInfo I mentioned in the email, there is no way currently for quickstart to get the unit address when using the local env [15:29] rick_h_: so 1) make the change in core and 2) update quickstart [15:29] rick_h_: 2) is tricky because we will need to support both new and old behavior [15:30] rick_h_: I suspect we should escalate those bugs [15:30] frankban: +1 [15:32] rick_h_: and the core fix should be included in 1.18, otherwise we have a limbo in which there is no way to get containers' addresses for API clients [15:32] rick_h_: what's the timing for 1.18? [15:32] frankban: it's moved so much I'm not even sure any more [15:32] heh [15:32] frankban: let's talk to sinzui and get that marked as a blocker for 1.18 [15:33] and then we can get that going asap [15:33] +1 [15:55] Makyo just ran it again and it still failed locally [15:55] I'm baffled [15:55] hatch: bisect? does it work on older code? [15:56] go back a week and make clean-all [15:56] there could be a real bug, better to catch it now while bisect path is short [15:56] yeah going on that now [15:56] I had a conflict i had to resolve [15:57] other thing is a fresh checkout and see if there's something else that's off [15:58] I ran into a very odd chrome bug debugging those routes yesterday - it would cache the previous url even though the path didn't have /inspector for example [15:58] the solution was to open a new tab [16:01] nananananananana bisect man bisect man [16:06] hmm nope even old versions are broken [16:06] new repo time [16:10] wow our git repository is getting large [16:10] 90MB or so [16:13] sorry bac getting back to your review [16:13] everyone seems to think today is monday and wants to break stuff :) [16:14] haha [16:14] even a new fresh checkout doesn't fix it :/ [16:14] wth [16:14] hatch: then your machine is broken, time to buy a new one :P [16:15] mine AND Makyo's? [16:15] c'mon! [16:15] Hahaha [16:15] hatch, you working in vagrant? [16:15] yeah [16:15] That's the common factor, I think. [16:15] yeah but what changed? It worked two days ago [16:16] d3 upgrade involved regenerating npm-shrinkwrap, maybe that? [16:16] ohh [16:16] lemme check out before that [16:16] It should still snag the same version - none of the versions changed - but maybe something's going on with that. [16:17] ok trying a pre-D3 branch [16:17] maybe we need to update to the latest phantom binary [16:18] anywho I'll keep you posted [16:18] hatch, you'll have to make clean clean-all etc. Sorry. I'll do the same to check it out. [16:19] yep it's npming now [16:19] ugh I long for the day there is an npm rollup [16:20] Makyo nope still crashed [16:20] :/ [16:20] ExceptionHandler::GenerateDump waitpid failed:No child processes [16:20] Boo! [16:20] Yep, same thing here. [16:21] What would it take to get test-debug/prod to run locally? [16:22] it does? [16:23] I haven't tried yet, I've been getting that in the vagrant. [16:23] what do you mean to run locally? [16:23] In osx term [16:24] ohh it can't the way we are running it [16:24] Okay. Hmm. [16:24] Wonder if it's a saucy thing. Let me see what happens if I go back to Raring. [16:24] https://github.com/juju/juju-gui/blob/develop/test-server.sh [16:27] frankban: sorry, one more call. Just talked with james/robie [16:27] rick_h_: same url? [16:27] frankban: https://plus.google.com/hangouts/_/7ecpjnf6uun25l4dh5sus8edhg?hl=en [16:29] Makyo I'm updating phantom and mocha-phantom [16:29] * hatch crosses fingers [16:29] hatch, okay. I'm ducking back to Raring real quick, will compare notes. [16:31] jujugui need qa on https://github.com/juju/juju-gui/pull/218 real quick. Fast one. [16:31] frankban: http://paste.ubuntu.com/7194939/ is the other package they've got that is a lot like this [16:32] ugh I just wish I could run Ubuntu on metal [16:32] someone fix the kernel [16:32] frankban: they make --no-adjust-repos the default in the distro package [16:33] Makyo new versions didn't help... [16:33] although now it's just crashing without any error [16:33] rick_h_: sound good, so --distro-only and --ppa? [16:33] hatch, improvement! [16:33] Well, okay. Not really. [16:33] lol [16:33] frankban: sounds good to me [16:34] I wonder if we are runnign too many tests and it's overloading something [16:34] too much domness or something [16:34] there are 1040 issues.... [16:34] yikes [16:34] Yeah. Curious if it's an NFS problem. [16:34] Will check raring, almost done provisioning. [16:34] cool - wasn't there a reason why we upgraded though? [16:35] rick_h_: cool, both == --ppa, none == --ppa [16:35] Don't believe it worked for bac? Something. [16:35] hatch, if the downgrade works, though, we can rule out NFS. [16:36] yeah it's just very odd that it just happened [16:36] frankban: sounds good to me [16:36] and keep nfs on raring, maybe trusty [16:36] YEah. [16:36] with no changes... [16:36] hatch, other than added tests, like you said. [16:36] yeah but rolling back to an old version didn't make it pass [16:36] hatch: when you rolled back you make clean-all? [16:36] you bet [16:36] hatch: so that you got the old node/d3 deps? [16:36] cool, just checking [16:37] I even tried a fresh new repo [16:38] https://github.com/ariya/phantomjs/issues/11706 [16:38] ^ hmm [16:39] hatch, I juist think it might be a system dependency, not a node dependency. [16:40] Oh, hmm. [16:40] works now? [16:40] Oh, no, you're GH issue [16:41] ahh yeah... [16:42] hatch, try modifying the phantomjs command to output more? --debug=true I think [16:44] hmm not according to it's docs [16:44] but I can try [16:45] oh --setting debug=true [16:45] ok trying that [16:46] hatch, okay, sorry. Was going off the global install on my linux box. [16:46] well because we use mocha-phantomjs to run the tests [16:48] Oh, right, hah. [16:48] still no output [16:48] ugh [16:48] Gah [16:54] rick_h_ I don't think this is a valid route /inspector/apache2/service [16:55] /inspector/apache2 would show the details by default [16:55] and /inspector/apache2#tabName would show the appropriate tab [16:55] unless 'service details' means something else? [16:56] oh s/service/charm [16:56] makes sense [17:04] hatch: rgr [17:04] hatch: oh wait, so that was the url to have service details shown from the inspector [17:04] so /inspector/apache2 is inspector [17:04] and /service on it means details of that service popout [17:04] hatch: welcome suggestions [17:05] isn't that the charm details? what are the service details? [17:05] hatch: from the inspector you click the link and the details pops out to the left [17:05] that's not quite charm details, it's different data? [17:05] I was pretty sure it was the exact same thing [17:05] well the data comes froma different place [17:06] service data vs browser model data [17:06] tabs are different etc [17:06] hmm [17:06] lemme mull it over [17:06] k [17:06] but ok I understand now [17:06] thx [17:10] hatch, charm version differences are taken into account there. [17:10] Store will always show you the latest version. [17:11] yeah right but to me /inspector/apache2/charm means the charm details for that service [17:12] Oh, sorry. Don't have the doc in front of me. [17:12] where /inspector/apache2/service doesn't really say anything about the service, it's talking about the charm [17:12] Trying to debug CI failure. [17:12] I'm mulling....I'm mulling! [17:12] Mmm, mulled wine [17:13] hatch: it's the information in the service. It's readme/etc [17:14] hatch: it's more of a comparison to charm details (browser) service details (/service) unit details (/unit/4) [17:14] those are the 3 versions/distinctions [17:17] I just think that /inspector/apache2 IS the service and we are showing the charm details used to create that service [17:18] hatch, on a more interesting note, tests work in raring. [17:18] s/interesting/positive [17:18] Didn't mean to imply, sorry. [17:18] hmm so everything works on raring? [17:19] That was test-debug, let me do test-prod [17:19] hatch: we've just got to port what already exists. Something has to say "this inspector is open on X" and something else needs to say "inspector is open on this with the details open" [17:20] yep for sure [17:20] arm the charm details don't have any url [17:20] it just minimizes the sidebar [17:20] just fyi [17:20] huh? [17:21] /precise/mysql-10 is a charm details url. [17:21] plus /sidebar/precise/mysql-10, and /sidebar/search/precise/mysql-10?text=mysql [17:21] not when you click it from the inspector [17:21] it minimizes the browser and opens the left panel [17:21] without a url change [17:21] right, that's what we do now [17:21] which is why we're inventing a new url [17:22] there's not space to show the sidebar + inspeector + details. In the new world we'll show inspector + details [17:22] right [17:22] hatch: so what's the issue/argument point here? [17:22] you don't like the word service and prefer the word charm? [17:22] hey rick_h_, regarding an XXX bug for removing the deprecated field, we could do that, but it is kind of out of our hands. i don't see it happening before the death of charmworld. [17:23] bac: k, nvm then [17:23] rick_h_ yeah - /service seems weird [17:24] hatch: ok, then happy to entertain a better word [17:26] details was ditched because there's two types of details off the inspector. I didn't pick charm because things would get called charmDetails and such and it's an overloaded term already. [17:26] you're right that technically the inspector is the details of the service [17:26] so service isn't great either [17:27] yeah....hmm [17:31] hmm apparently freenode didn't like it either lol === hatch__ is now known as hatch [17:38] hatch, test-prod fails with the same error :( [17:38] ugh [17:38] what in the heck is going on [17:40] No clue :( [17:40] Will keep poking. [17:40] Meanwhile, think I figured out CI problems, created discussion card for what I found. [17:42] ugh devops issues infuriate me [17:42] Also meanwhile, if chiptunes are your thing, the ChipWIN April Fools album is actually pretty good, despite being mostly jokes. http://chiptuneswin.com/album/chiptunes-srsbsns [17:43] I've tried to do the chiptune thing [17:43] can't quite get into it [17:44] Haha, that's fair :) [17:45] Took me a bit to get past the fact that it's not mixed down to the wall-of-sound that we've come to expect from music these days. Motown popularized that years ago, and it's just gotten more and more extreme since. [17:47] * Makyo rambles. [17:49] AGH I hate not having .lbox.check [17:49] Sorry for the spam. [17:50] Makyo: a pre-commit or pre-push hook added to the docs would be cool :) [17:53] rick_h_, Alright, will investigate over lunch! [17:56] yeah pre-push hook would be awesome [17:56] although also somewhat irritating [17:56] because you could never push incomplete code [17:56] heh [17:57] git pushncheck [17:57] :) [17:57] so it's an alias maybe [17:57] hatch, there's apparently a --no-verify option [17:57] oh interesting [17:57] At least, to pre-commit hooks, reading up on pre-push [17:58] ohhhh rick_h_ you changed our 1:1 ok lunching now [17:58] :) [17:58] bbiab [17:58] hatch: changed it? /me thought it was always here. [17:58] hatch: oh, this is the time change in effect maybe? [17:58] it's still 2pm for me [17:59] oh it says it was just changed [17:59] odd...ok whichever I thought it was right now [17:59] I'm hungry anyways [17:59] :D [17:59] it is right now [17:59] I'm in there, huh? [17:59] weird [17:59] ok joining [18:27] rick_h_: I have to go, I have a possible prototype to fix the issue, see http://pastebin.ubuntu.com/7195287/ (lp:~frankban/juju-core/machine-addresses). There might be still something to fix, you can take a look at the discussion in #juju-dev. Please feel free to email me if you get additional info. I'll check tomorrow morning if I can do something. Have a nice evening! [18:28] frankban: thanks so much, have fun! [19:18] Got an email of someone trying to recruit me through a dating app. Unfortunately, the recruitment company has debug mode on, and so now I have their DB server password. Classy. [19:19] Wonder if it's some ridiculous CTF. [19:22] lmao [19:25] Looks like it made it to HN, now their cache is failing. [19:26] Aaaand now it's just blank. That was fast. [19:28] Makyo what were they recruiting you for?... [19:28] :P [19:28] Never did find out. Their redirect thing is what threw the debug error. [19:29] It was a click-tracking thing on the recruiter's site. [19:29] ohh haha [19:29] the other day I was recruited for a mechanical engineering position [19:30] I think they read the 'software engineer' and was like....."close enough" [19:30] lol [19:38] kadams54: hey, meant to stab you and let you know today is you QA day [19:39] and welcome back to the interwebs :) [19:39] Yeah, I realized that around 1 today :-) [19:39] I'll need the pokey stick back for bac tomorrow [19:40] rick_h_: er? [19:40] lol [19:40] stabby stabby [19:40] oh, to do QA? [19:40] yeah, good luck with that [19:40] bac: friday chat was that we want to bring back the QA day. I'll use a pokey stick to help remind folks [19:40] i *always* forget [19:41] ok, i'll try to remember [19:41] thus the pokey stick, I've got to remember before EOD tomorrow unlike today [19:41] So… what's the best way to do this QA? [19:42] comingsoon.jujucharms.com? [19:42] kadams54: works for me [19:42] Production? [19:42] or do a live environemtn [19:42] test out your lxc or ec2 setup [19:42] Oh yeah [19:42] I was going to get EC2 setup [19:42] the goal is to test all kinds of things, whateer your fancy is and try to find a bug [19:42] yeah - do odd interactions too [19:42] try your non default browser of choice, non default env of choice [19:42] preted you think like hatch [19:42] pretend [19:43] lol [19:43] enter my mind..... [19:43] break outside the mold a bit and try to make sure things still work as expected [19:43] kadams54 my mind http://upload.wikimedia.org/wikipedia/en/thumb/a/a3/Escher's_Relativity.jpg/300px-Escher's_Relativity.jpg [19:44] :-) [19:45] hatch: that pic doesn't have enough promises in it :P [19:45] lol [19:46] kadams54, actually, if you're doing QA stuff, think you could do a quick check on my PR? [19:46] Sure [19:46] 218 [19:46] https://github.com/juju/juju-gui/pull/218 [19:46] Sorry [19:48] Question: how intelligent should bundles be? That is, if I deploy the hadoop:cluster bundle and then attempt to deploy mediawiki:single after, is it fair to expect the mysql service in mediawiki to be renamed? [19:48] kadams54 nope [19:48] it'll fail [19:48] there will be a bundle configuration step at some point [19:49] Yeah, I know it fails [19:49] That just seems sorta unfair to me as the user though :-) [19:49] Especially in complex environments [19:50] kadams54: because you might want to share that service [19:50] so let's say you have a beefy mysql server [19:51] and want to share it across wordpress and mediawiki bundles [19:51] kadams54: so it's a valid question, but we don't have the abliity to handle it right now [19:51] so we error [19:51] Except you can't deploy the bundle [19:51] Yeah [19:51] right, and you can't from the cli [19:51] OK [19:51] If the CLI can't do it :-) [19:51] if you've got scripts that rely on that service name (juju log mysql/0) [19:51] it'll break if you rename it automatically [19:51] renaming is a bad thing in some cases [19:52] Not the existing service [19:52] The new service [19:52] well a bundle is a repeatable deployment [19:52] so in one cloud it's named X and Y in another? [19:52] if you've got benchmark tools to automate testing you can't use it now [19:53] You just need to use a deterministic renaming algo [19:53] it's all imaginary for sure [19:53] mysql-1 in both clouds [19:53] bam [19:53] well then you're fine if you call it that in your bundle [19:53] just don't use the default service names in your bundle and you're fine :) [19:53] Now you sound like a Java guy [19:53] railsmysql, railsapp, railsmemcache :) [19:54] "Explicitly configure everything and you'll be fine!" [19:54] just playing devils advocate [19:54] ;-) [19:54] it's something we'll deal with as bundes mature I think [19:54] stacks ahoy! [19:54] crazy to think they're still young little things. [19:54]