[14:17] <kadams54> rick_h_: Looking to take on another card… any suggestions?
[14:49] <hatch> bahaha http://www.cbc.ca/news/canada/saskatchewan/saskatchewan-winter-too-much-for-one-cbc-radio-listener-1.2840916
[14:49] <hatch> guy just snapped
[15:29] <hatch> hmmm, didn't we remove the pyjuju stuff? https://github.com/juju/juju-gui/blob/develop/app/store/env/sandbox.js#L244
[15:30] <rick_h_> probably the env but not the sandbox?
[15:30] <rick_h_> you did it if I recall :P
[15:30] <hatch> yeah....I must have failed hard
[15:30] <hatch> or I'll blame it on a bad merge
[15:30] <hatch> yeah, bad merge
[15:31] <hatch> created 
[15:31] <hatch> a
[15:31] <hatch> card
[15:50] <rick_h_> uiteam call in 10ish please kanban
[16:08] <kadams54_> hatch: for you: http://cl.ly/image/1h3z271n1a1e
[16:08] <hatch> I don't see a -a there :P
[16:37] <kadams54_> hatch: http://cl.ly/image/1w2e2H2u1u0D there, you happy now?
[16:45] <rick_h_> kadams54_: :P
[16:45] <rick_h_> kadams54_: sorry, tried to reply to you but you dropped off irc and such. Looks like you're good on cards?
[16:46] <kadams54_> I'll need another one soon, so if you have any suggestions I'll take them.
[16:48] <rick_h_> kadams54_: if you're up for it, the generating ecs entries would be one I'd love to see go through. The changing a service name validation issue is another one
[16:48] <kadams54_> k
[16:49] <rick_h_> kadams54_: for a lighter workout the removing the extra pyjuju code hatch found is a light weight task
[16:52] <hatch> kadams54_: yeah it should be as simple as remove the code and remove the tests - you'll just want to do a bunch of sandbox qa afterwards
[17:24] <TheMue> rick_h_: ping
[17:27] <rick_h_> TheMue: pong
[17:29] <TheMue> rick_h_: the idea I wrote about yesterday is to let users, regardless of CLI or GUI, create action arguments as yamls, e.g. when they are complex or the user want to reuse them, and let them pass then to the server
[17:29] <rick_h_> TheMue: k
[17:29] <TheMue> rick_h_: the CLI already has such a command, but converts the yaml a json structure the API expects for arguments
[17:30] <TheMue> rick_h_: so right now the GUI would have to to the same
[17:30] <rick_h_> TheMue: ok, so that command is not an api command, but purely on the client?
[17:30] <TheMue> rick_h_: but when parsing the yaml on server-side both front-ends could use the parsing and validation
[17:31] <TheMue> rick_h_: right now yes, the passed yaml is parsed and validated there
[17:31] <rick_h_> TheMue: so the rule is, if the user is sending yaml we won't read it or touch it but pass it along
[17:31] <rick_h_> JS doesn't have a good yaml support and json/jsonschema is all we have plans to interact with
[17:31] <TheMue> rick_h_: and here I thought it would make sense if bith front-ends could benefit from it
[17:31] <rick_h_> TheMue: fine by me
[17:31] <rick_h_> TheMue: an api call to send the yaml uploaded from the client is peachy
[17:32] <rick_h_> TheMue: as long as we're 100% sure the results of that, errors/etc come back to the client (GUI) in json form to be able to attach errors and the like to the form generated frmo the action's jsonschema
[17:32] <rick_h_> TheMue: just like we do with charm config currently
[17:32] <hatch> rick_h_: ok I believe I have resolved the issue - it's now sending null to juju, unfortunately I can't actually test that it's fixed because now postgres won't deploy at all because there are null values lol
[17:33] <rick_h_> hatch: ummm, huh?
[17:33] <TheMue> rick_h_: ok, will talk to bodie and john about it
[17:33] <hatch> I've contacted 'stub' on irc for input so now I wait I guess
[17:33] <rick_h_> TheMue: cool, let me know if you need to me to attend or help in any way
[17:33] <rick_h_> hatch: ok
[17:33] <TheMue> rick_h_: will do
[17:34] <hatch> I actually get two errors https://gist.github.com/hatched/944c67756fcc3bf29c91
[17:34] <hatch> first it can't find it, then it doens't like the config option
[17:34] <hatch> hehe
[17:34] <hatch> so confusing
[17:36] <hatch> oh he is far far away - i'll just reply in the bug
[17:36] <hatch> interestingly enough it deploys from the command line
[17:36] <hatch> :/
[17:36] <hatch> bah
[17:37] <rick_h_> hatch: right, we're doing something. 
[17:37] <rick_h_> hatch: what about null vs undefined?
[17:37] <rick_h_> hatch: those are two different things in the json?
[17:37] <hatch> yeah they are
[17:37] <rick_h_> maybe we need to really be doing a 'juju null is a JS undefined' thing?
[17:39] <hatch> I don't see how null is a necessary value in juju anyways :P
[17:40] <rick_h_> doh, want to call and get my snow tires put on but the tire places phone doesn't work. Google is telling me 'it's busy' 
[17:40] <rick_h_> guess there's a line up for those snow tires a bit early this year :)
[17:47] <hatch> uiteam any idea why my gui deploy won't switch to a remote branch? "ValueError: u'git@github.com:hatched/juju-gui.git null-config-values': release not found"    
[17:47] <hatch> it was JUST deployed via the exact same string
[17:47] <hatch> in fact, I only changed the console logging option
[17:48] <frankban> hatch: seems like a charm bug then
[17:48] <hatch> ahhh it's running here url, filename = get_launchpad_release(project, origin, version)
[17:48] <hatch> that's definitely not a lp release
[17:49] <kadams54_> uiteam: I've been having problems recently with cached SVG icons in Chrome and it's to the point where it just drive me crazy.
[17:49] <kadams54_> So I did this: http://cl.ly/image/0x0c1M0u2J3m
[17:49] <kadams54_> Adding a cachebuster to the end of the SVG URL worked.
[17:50] <kadams54_> Any reason we don't currently have cachebusters on those URLs?
[17:50] <hatch> I just turn caching off
[17:50] <hatch> don't ever have cache problems with the devtools open
[17:50] <hatch> now when they are closed....
[17:50] <hatch> well!!!
[17:52] <jcsackett> kadams54_: i like it.
[17:52] <kadams54_> I also have caching disabled when devtools are open, but I'd still get the wrong icons.
[17:52] <hatch> very odd
[17:52] <hatch> what if you use stable chrome?
[17:53] <hatch> wonder if it's an issue with canary
[17:54] <kadams54_> hatch: Don't know. The problem is sporadic, so I'd have to use stable Chrome for awhile until something in its SVG cache went stale.
[17:54] <hatch> ahh
[17:54] <kadams54_> Or maybe nothing would break and I'd just go on intentionally using less advanced devtools for a long long time
[17:54] <kadams54_> And that would make me sad.
[17:54] <hatch> not going to lie, I've never once needed the canary devtools
[17:55] <kadams54_> Says the developer who also does not use vim
[17:55] <kadams54_> I rest my case.
[17:55] <hatch> and yet somehow......I get by :)
[17:55] <hatch> unfortunately right now the gui charm has a bug which I need to investigate
[17:56] <rick_h_> kadams54_: because for the icon to change the id has to change (new rev) and they auto cache bust themselves?
[17:56] <rick_h_> kadams54_: and until the rev goes up there's no reason not to keep using the cache
[17:56] <kadams54_> Except that we've seen caching issues pop up in both Chrome and Safari with these…
[17:57] <kadams54_> Where all of a sudden the icon is off-center or (currently, for me) the wrong size
[17:57] <kadams54_> But yeah, I see what you're saying. I'm not sure a cachebuster would really help if the number didn't change.
[17:57] <rick_h_> kadams54_: well yea. I'm just explaining the " Any reason we don't currently have cachebusters on those URLs?"
[18:10] <hatch> darn frankban left
[18:10] <hatch> I'm a little confused as to how this ever worked in the gui (specifying a github url)
[18:11] <rick_h_> hatch: it has known issues on precise
[18:11] <rick_h_> make sure you're on trusty
[18:11] <rick_h_> hatch: and the tests say it works in trusty
[18:11] <rick_h_> the test is skipped in precise because it fails/isn't supported I think
[18:11] <hatch> hmm nope deployed on trusty
[18:11] <rick_h_> what are you entering? I just did this with kadam's branch on guimas the other day
[18:11] <hatch> investigating
[18:12] <hatch> sec
[18:12] <hatch> juju set juju-gui juju-gui-source="git@github.com:hatched/juju-gui.git null-config-values"
[18:12] <hatch> oh it needs to be the https url
[18:12] <hatch> bleh
[18:12] <rick_h_> hatch: have to use https url
[18:12] <rick_h_> it can't do git because it can't do ssh?
[18:12]  * hatch reading the charm source
[18:13] <hatch> I have no idea - but the parse_url method in the charm requires it to be https heh
[18:13] <rick_h_> hatch: try setting it to https://github.com/hatched/juju-gui.git null-config-values
[18:13] <rick_h_> :)
[18:13] <hatch> yeah wow....damn
[18:13] <rick_h_> because the charm can't talk ssh (git@) for you
[18:13] <rick_h_> not unless you want to upload your github ssh keys 
[18:14] <rick_h_> hopefully that works for you now
[18:14] <hatch> wow that was a huge mind blank
[18:14]  * rick_h_ goes back to lunch
[18:14] <hatch> I'm sure I knew that
[18:14] <rick_h_> :)
[18:27] <hatch> uiteam is there a way to destroy a service with units in error without spamming `juju resolved <service>/<unit>` ?
[18:28] <jrwren> hatch: not afaik :(
[18:28] <hatch> jrwren: is there reasoning behind this? 
[18:28] <hatch> or just a feature not yet implemented?
[18:28] <jrwren> hatch: why would any unit ever be in error? :)
[18:28] <hatch> lol
[18:29] <hatch> oh you're HILARIOUS 
[18:30] <jcsackett> hatch: juju destroy-service --force?
[18:30] <jcsackett> not sure if that has a force on it or not.
[18:30] <hatch> jcsackett: not an option according to the help
[18:31] <jcsackett> well damn.
[18:34] <jrwren> if you are using a reasonably modern version of juju you can for killer in $( status --format tabular | awk /magic/  ) ; do juju destroy $killer ; done
[18:34] <jrwren> or some such
[18:34] <hatch> haha 
[18:34] <kadams54> FYI, seems like comcast has a cable outage
[18:34] <hatch> I am surprised this isn't a real issue for larger environments
[18:35] <hatch> kadams54:  I thought that comcast was always out :P
[18:35] <kadams54> Going to see if I can find another location with an up connection.
[18:35] <jrwren> hatch: real larger deployments don't error.
[18:35] <hatch> the way you bounce on and offline
[18:35] <hatch> :)
[18:35] <hatch> jrwren: "real developers don't error"
[18:35] <kadams54> hatch: I've had good service from them in this area.
[18:36] <hatch> subprocess.CalledProcessError: Command '['make', '-C', '/var/lib/juju/agents/unit-juju-gui-0/charm/juju-gui-source', 'distfile', 'BRANCH_IS_GOOD=true', 'HOME=/root']' returned non-zero exit status 2
[18:36] <hatch> yup it's time for lunch
[18:36] <hatch> later
[18:44] <jcastro> rick_h_, you home or at the sprint?
[18:59] <rick_h_> jcastro: home
[20:11] <hatch> kadams54: does chrome canary auto update the websocket frames without having to click 'ws' again?
[20:11] <hatch> That feature alone would make switching worth it :)
[20:13] <hatch> teslanick: ahoy!
[20:13] <teslanick> hatch: ahoy-hoy
[20:13] <kadams54> hatch: don't know
[20:14] <hatch> after what felt like 100 failed deploys the null config issue appears to have been resolved
[20:15] <hatch> rick_h_: we now send `null` through the ws and it comes back as `undefined` lol....oh boy
[20:16] <hatch> BUT it appears to work
[20:17] <teslanick> "Null" types per language:  JS: 2.  Everyone Else: 1.  Haskell: 0.
[20:17] <hatch> lol
[20:18] <hatch> teslanick: so this one is extra special, it goes YAML > Python > JSON > JS > JSON > GO
[20:18] <hatch> who knows what it comes out as at the end :)
[20:18] <hatch> at least now the JSON > JS > JSON part is now fixed
[20:19] <hatch> look at all those failing tests *sniff* it feels so good when tests fail when they should
[20:19] <teslanick> hatch: So: http://i.imgur.com/jfwcVa5.jpg
[20:19] <hatch> lol exactly
[20:27] <hatch> teslanick: what editor are you using these days?
[20:35] <teslanick> Depending on the task: Intellij Idea, LightTable, TextMate2
[20:38] <hatch> yikes
[20:38] <hatch> why not just intellij all over?
[20:38] <teslanick> It's slow and has stupid opinions about thinks.
[20:38] <teslanick> *things
[20:38] <hatch> ahh I've been thinking of giving webstorm another try...but that was my experience last I tried too
[20:38] <teslanick> If I could use LightTable for everything, I would. But it's buggy and has very limited language support.
[20:39] <teslanick> Though I spent a few minutes yesterday figuring out how tough it would be to make an ES6 plugin for it.
[20:43] <hatch> I've been trying atom every now and again - it's still too slow compared to sublime
[20:43] <hatch> not sure it'll ever be able to match the speed being written in coffeescript
[20:44] <teslanick> Is CS really any slower than regular JS?
[20:45] <jrwren> no, but it is still js.
[20:45] <jrwren> which should be super fast given v8.
[20:46] <jrwren> does atom ship a modern v8 with it? or is it node?
[20:46] <hatch> I'm pretty sure it's using chrome's v8 as it's backend
[20:46] <hatch> it's a massive download so it better be haha
[20:46] <jrwren> hatch: REALLY!?! so its more like a chrome fork?  that is interesting.
[20:46] <hatch> it's just slow doing system things, like searches
[20:46]  * jrwren looks
[20:47] <hatch> it has devtools and everything
[20:47] <hatch> lol
[20:48] <hatch> which IS pretty cool
[20:51] <jrwren> hatch: its node: https://github.com/atom/atom/tree/master/build
[20:51] <jrwren> hatch: which uses an old an unsupported v8 :p
[20:53] <hatch> hmm
[20:53] <hatch> well where does it get the devtools from?
[20:53] <hatch> it must wrap it in chrome
[20:53] <teslanick> IIRC, the UI is written in Chrome which communicates with a nodejs "backend" that handles native system calls etc.
[20:54] <hatch> ohh that makes sense
[20:54] <teslanick> s/Chrome/chromium
[20:54] <jrwren> hatch: on node's use of unsupported v8 https://code.google.com/p/v8/issues/detail?id=3692#c6  
[20:55] <hatch> yeah I knew that node was using an unsupported v8
[20:55] <hatch> little scarry
[20:55] <teslanick> As a note, the "old" version isn't that old. And V8 hasn't gotten that much faster as of late.
[20:55] <teslanick> The unsupported thing is scary, I agree.
[20:55] <hatch> I'm not really concerned about the speed of v8 but it's the bug fixes that can't even be rolled back
[20:56] <jrwren> seems like using newer v8 would be worth it just for better harmony support
[20:57] <jrwren> The version in that issue is 2+yrs old
[20:57] <hatch> ahh just use Dart
[20:57] <jrwren> on the desktop/server?
[20:58] <jrwren> use go! :p
[21:00] <rick_h_> lazyPower: fyi we're going to use https://github.com/CanonicalLtd/jujucharms.com/issues to track jujucharms.com issues and get them to the right projects
[21:00] <hatch> haha yeah Go decimates Dart in serverside performance
[21:00] <teslanick> Go is a crazy-fast language by design, though, right?
[21:00] <hatch> Dart also creates masssssive js files when using dart2js 
[21:00] <hatch> yeah it is
[21:00] <lazyPower> rick_h_: thanks for the update i'll pass it on :)
[21:04] <hatch> uiteam lf two reviews and two qa's for this backend change https://github.com/juju/juju-gui/pull/664 requires real env (LXC works)
[21:05] <hatch> oops I might as well remove the unused method too
[21:08] <hatch> uiteam ok all done - review and qa away
[21:15] <hatch> teslanick: if I were to start a new project today I'd PROBABLY use Go for the server and js for the client...
[21:16] <hatch> there are just not as many modules for dart so one would have to write quite a few 
[21:16] <teslanick> s/(Go|js)/clojure/
[21:17]  * teslanick likes his webapps like he likes his women: full of parentheticals.
[21:17] <hatch> rofl
[21:18] <hatch> I really don't think that makes any sense
[21:18]  * teslanick likes his webapps like he likes his summer days: homoiconic.
[21:18]  * teslanick likes his webapps like he likes his politics: full of terse s-expressions.
[21:19] <hatch> you got a lot of these don't you?
[21:19] <teslanick>  </dadaist-humor>
[21:19] <hatch> you'd use closure for the client too eh?
[21:19] <jrwren> hatch: reminds me of : http://axioms.io/zen/2014-11-06-swift-go/
[21:20] <teslanick> Yeah, I have a few middling-sized experiments in clojurescript UIs.
[21:20] <jrwren> if I were to consider clojure, I'd skip it and use ocaml instead. :p
[21:21] <teslanick> Clojure can interact really nicely with the JVM, so you get a lot of great libraries out of the box. Also, I don't think there are any (decent?) OCaml->JS compilers.
[21:21] <hatch> jrwren: teslanick is a closure pusher
[21:21] <hatch> jrwren: haha yeah but swift is only for one OS :)
[21:22] <teslanick> The one I'm working on now has a clojure server, clojurescript front-end, and they communicate back-and-forth using a websocket.
[21:22] <jrwren> hatch: the only OS that matters :p
[21:22] <teslanick> (two OS's -- OS X and iOS)
[21:23] <teslanick> The really big advantage of using clojure is that you can repl into both systems and build new pieces it as it runs. A little semantics around safely restarting stateful components, and you can literally build the plane while it's flying.
[21:23] <jrwren> that sounds nice. I do miss a repl when using go.
[21:23] <jrwren> in C# I even had a repl.
[21:27] <hatch> jrwren: lol did you forget who you work for? :D
[21:31] <hatch> jrwren: from your link.... """Swift was released just this year in classic Apple style: A complete surprise, with massive fanfare, and no acknowledgement that many of the ideas had been pioneered elsewhere"""
[21:31] <hatch> lol!!!
[21:32] <teslanick> heh
[21:33] <hatch> from my limited experience with swift it does look like a pretty cool language
[21:33] <hatch> maybe at I/O this year Google will tell us that Dart is supported as Dart on Android and Chrome 
[21:34] <teslanick> It does look it. It's crazy that it has ARC baked in. So it's a non-garbage-collected language without explicit destructors.
[21:34] <teslanick> hatch: I half expect that Apple will announce Swift for Safari/MobileSafari at the next WWDC.
[21:34] <hatch> I kind of doubt it because noone uses safari :P
[21:35] <teslanick> A ton of people use mobilesafari
[21:35] <hatch> ok ok mobile safari
[21:35] <teslanick> And they kinda-sorta share a codebase
[21:37] <jrwren> hatch: yeah, that is obviously by someone who doesn't follow languages :)
[21:37] <jrwren> hatch: I also am not serious about swift. I'd not want to use a close source language.
[21:38] <hatch> this is true
[21:38] <hatch> Dart is open source with an ECMA standarsd
[21:38] <jrwren> I use safari, because everything syncs to my mobile safari. The apple convenience switched me :(
[21:38] <hatch> haha - chrome does the same :)
[21:38] <jrwren> hatch: so dart is an ecmascript? :)
[21:38] <hatch> and I think FF did that before then
[21:39] <jrwren> yup... i'd stick to FF, it is just too stupid on iOS
[21:39] <hatch> ahh right becuase it's just a wrapper around safari right?
[21:41] <teslanick> I just want a spec for a browser bytecode language with decent capabilities and a great GC.
[21:41] <hatch> I like that guys website - so many don't use serif fonts any longer
[21:41] <hatch> teslanick: asm? *snicker*
[21:42] <teslanick> You snicker, but it's close-ish. The problem is that it has manual memory management.
[21:42] <jrwren> yup, asm.
[21:42] <jrwren> nacl?!
[21:42] <teslanick> C and C++? No thanks.
[21:43] <teslanick> Also not memory-managed. And pointer math?
[21:43] <jrwren> go used to target nacl
[21:43] <jrwren> you just want the world eh? :p
[21:43] <hatch> I thought Google dropped nacl from Chrome
[21:44] <hatch> that's why Unity is now compiling C# > C++ > asm.js
[21:44] <hatch> for the web
[21:44] <teslanick> You have to enable nacl from chrome://flags
[21:45] <hatch> hmm I was sure they were saying their plugin will no longer work because it was removed - maybe it's essentially useless if random joes have to enable it manually
[21:46] <hatch> uh oh, something bad happened in the apple/unity world https://twitter.com/MattRix/status/535542224164225024
[21:50] <jrwren> huh, too bad, I thought nacl was cool.
[21:51] <hatch> http://forum.unity3d.com/threads/no-nacl-in-unity-4-3.210450/
[21:52] <jrwren> hatch: well that is waht they get, they should pay xamarin.
[21:53] <hatch> who?
[21:55] <jrwren> unity, if they want to be in app store in those months.
[21:56] <hatch> I thought it was Apple who needed to pay....I'm so confused 
[21:57] <jrwren> hatch: nah, its Unity.
[21:57] <hatch> ahhh 
[21:57] <jrwren> Unity uses mono.
[21:57] <hatch> they should just switch from C# to Go
[21:57] <hatch> ;)
[21:58] <jrwren> Mono is open source, much of it MIT, but runtime is GPL. GPL can't go into app store, and so requires a different license, this is how Xamarin makes $$$
[21:58] <hatch> wait...GPL can't go into the app store? 
[21:58] <jrwren> correct.
[21:58] <hatch> why not?
[21:58] <jrwren> everything in app store is statically linked.
[21:58] <hatch> oh haha
[21:58] <jrwren> you can't statically link to GPL libraries wtihout still being GPL.
[21:59] <jrwren> so... you can be GPL in app store but your app would need to be GPL too.
[21:59] <jrwren> I wrote that very poorly at first. sorry.
[21:59] <jrwren> GPL can be, but apps using GPL libraries cannot be, because, linking.
[22:00] <hatch> licensing is complicated 
[22:00] <huwshimi> Morning
[22:00] <hatch> morning huwshimi
[22:00] <teslanick> It seems like statically linking vs. dynamically linking is a technical distinction, not a legal one. But really, the GPL is awful.
[22:02] <hatch> I wonder what license I picked
[22:02] <hatch> GPL 3
[22:02] <hatch> is that any better?
[22:03] <hatch> :)
[22:03] <hatch> https://github.com/hatched/ghost-charm/blob/master/copyright 
[22:04] <teslanick> BSD always seemed like the "best" open source license. It should just be: "BSD License DGAF"
[22:05] <hatch> I just want one that says "here, take this, use this, no I don't offer warranty"
[22:05] <hatch> and preferably "pr's accepted" :P
[22:14] <hatch> hmm I'm not having any luck getting these spurious ci issues to stop
[22:15] <rick_h_> hatch: firewall ports aren't open I bet
[22:15]  * rick_h_ checks sauce videos
[22:16] <rick_h_> hatch: shoot an email to bac/jcsacket that we need the ports opened up for 8888 and 8889 for gui landings so that saucelabs can connect
[22:16] <rick_h_> hatch: note that in the sauce videos no html ever loads
[22:16] <hatch> ahh I didn't even think of that being the issue
[22:16] <hatch> will d9o
[22:31] <jrwren> hatch: CC0 license for you.
[22:33] <hatch> maybe I'll write a new license
[22:33] <hatch> the FTL license "F'it just Take'it License" 
[22:33] <hatch> lol
[22:34] <jrwren> hatch: I assume you know about the WTFL.
[22:34] <jrwren> http://wtflicense.com
[22:35] <hatch> nope but I do now
[22:35] <hatch> lol
[22:35] <hatch> WTFL 2.0 for me
[22:35] <hatch> :D
[22:36] <hatch> I like how nothing else on the page works