[00:32] anyone know of a quick way to show all the recomended charm store charms @ jujucharms.com? [00:33] arosales: search for 'charms' and open up the top list [00:33] it'll take a while to load [00:33] but the top section should be all recommended and the bottom is the rest [00:33] rick_h_: ack thanks [00:33] rick_h_: is that new with the "bundle" search capability? [00:33] arosales: yep [00:34] very cool [00:34] two keywords charm, charms, or bundle, bundles [00:34] but it's so slow we're not screaming from the root tops [00:34] more for you guys to show/demo/look at vs a real user feature :/ [00:35] understood, I get a lot of requests from folks to where can I look at the current list of recommeded charms. [00:35] I want to send them to the gui over manage [00:35] arosales: yep, understand [00:35] rick_h_: thanks [00:35] keep rocking it gui team === _mup__ is now known as _mup_ === _mup__ is now known as _mup_ === jcsackett_ is now known as jcsackett [11:19] frankban: looks to be going better this time. Thanks for the lik to upgrade npm [11:19] and morning [11:20] rick_h_: morning, cool [11:21] rick_h_: npm management is a bit disappointing [11:21] yea, not a huge fan myself [11:21] better than none I guess. [11:21] rick_h_: I suspect this breaks our charm as well (e.g. juju-gui-source=develop) [11:22] rick_h_: so, an idea could be to "npm set ca "" in the make file [11:22] hmm, where do we get nodjs from in the charm? [11:22] I've just updated the nodejs from the ppa [11:22] if it's using the ppa it should be fine on next update/deploy [11:23] rick_h_: trying [11:25] * rick_h_ goes to get ready for the day and the boy off to day care now that landing is unblocked [11:27] rick_h_: thanks for updating CI [12:21] frankban: the charm test go ok? [12:24] rick_h_: no, we take deps from ppa:juju-gui-charmers/stable, we need to update the ppa. created a critical card [12:25] rick_h_: the charm installs both nodejs and npm: the latter seems to be included in the former now, so a charm update might also be required [12:26] rick_h_: https://launchpad.net/~chris-lea/+archive/node.js/+packages effectively only includes the nodejs package [12:26] rick_h_: if we need to remove npm from the charms deps, then we might want to release a new charm ASAP [12:26] rick_h_: if the above makes sense to you, I'll dig into that ASAP [12:27] gotcha, sounds like a plan [12:28] yea, npm is both packages now. surprised that hasn't caused us issues before now as apt-get install fails due to that these day [12:28] rick_h_: perhaps the charm installs npm from precise? [12:29] frankban: maybe. The hacking docs have been the ppa for a long long time. It'll be good to get it in sync I guess. [12:29] rick_h_: no, we currently have npm in our ppa [12:29] right [12:29] gothca [12:29] so we've hit this in dev but the charm has been fine using the old stuff [12:30] of course they considered Friday as a good day for a backward incompatible change [12:30] hah! can't have a quiet friday [12:33] frankban: just reading backwards. so you're seeing the charm ftests failing? === BradCrittenden is now known as bac [12:34] bac: no, the charm "build from source" fails [12:34] frankban: in the charm trunk, i'm seeing the tests fail on lxc and ec2. [12:37] bac: logs? [12:37] bac: it could be related [12:38] frankban: http://paste.ubuntu.com/7010106/ [12:38] frankban: the second error is a problem in tearDown that is masking the real error in that test [12:39] if the deploy doesn't work then tearDown fails because self.serviceName is unset. i'll fix that and re-run to see what the real problem is [12:42] frankban: is it true that you cannot run the charm ftests on lxc if your host is not precise? i was getting errors yesterday that seemed to indicate that when run on my trusty VM. [12:42] bac: you'll have to deploy the charm with make deploy and look at the unit log to see why it's in error state [12:42] it could be related [12:43] bac: but have to double check the unit log to see why it's failing [12:43] rick_h_: ok [12:44] bac: since the error is in test_branch_source I strongly suspect you are encountering the same error [12:48] bac: IIRC you cannot run ftests on a local env for two reasons: 1) the tests co-locate the GUI on the bootstrap node (for speed reasons) and this is not possible in local envs, where machine 0 is localhost, and 2) bootstrapping a local env requires sudo [12:50] thanks for sending the email frankban [12:50] frankban: yes, i was encountering 1). couldn't colocate since they were of different series [12:51] bac: well lxc can't colocate regardless of series [12:52] bac: as I said, I suspect you would not be able to colocate even if you were on precise. That's because machine 0 in local envs is not a container [12:52] frankban: ah, right. i'll update the HACKING doc to remind folks (future me) of that [12:52] bac: great thank you [13:24] marcoceppi: when you have time could you complete the charmworldlib review at https://codereview.appspot.com/69430043 ? [13:25] bac: thanks, off today so will likely get to it monday -ish [13:39] * bac relocates to coworking spot [13:53] frankban I saw your email, does the charm build with the new version of node? [13:54] I've been running the latest node locally without issue.... [13:54] hatch: I am trying [13:54] cool [13:54] I thought you may have already so I was just curious :) [13:54] hatch: could you please take a look at https://github.com/juju/juju-gui/pull/149 [13:54] on it [14:00] hatch: nodejs update solves the SELF_SIGNED_CERT_IN_CHAIN issue, but there is another error. make log: http://pastebin.ubuntu.com/7010492/ npm log: http://pastebin.ubuntu.com/7010496/ [14:01] looking [14:02] frankban do you have g++ installed? [14:02] node-gyp requires that the user's home directory is specified in either of the environmental variables HOME or USERPROFILE [14:02] didn't we have to add g++ months ago to get things working? [14:02] or this is another case where we added the dep for dev but not for prod/charm? [14:02] and now we're hitting all kinds of upgrade fun? [14:03] rick_h_ I THINK we only added it locally [14:03] hatch, rick_h_ yeah g++ is in the charm [14:03] ohh [14:03] hmm [14:03] that's what typically causes the contextify issue [14:03] frankban: can we try to add a HOME= to the make target and see if it'll build then? [14:03] rick_h_: that's what I was trying to do [14:04] Error: node-gyp requires that the user's home directory is specified in either of the environmental variables HOME or USERPROFILE [14:04] ok [14:04] rick_h_: so the HOME will be root's home in the charm context [14:05] frankban: yea, :/ and kind of stupid it can't work in tmp or something. [14:06] ENOENT typically means that the directory doesn't exist or it can't access it [14:06] hatch: right, there's no HOME dir and so it can't build in it, or whatever it wants to do there [14:06] blarg [14:07] https://github.com/TooTallNate/node-gyp/issues/270 [14:09] https://github.com/TooTallNate/node-gyp/blob/0cc243e4593bfca8fb5cbbc7413d89b8ba3b1a03/lib/node-gyp.js#L49 [14:10] so yea, we need to give it a place to extract to. We might be able to fake it out if we had to, but building in /root will have to do as a start [14:12] btw, this is why I love OSS. "WTF..." *go load source* [14:13] OSS also has "RTFS MAN" [14:13] :P [14:15] oh man this diff is hard to read, git split it up in a very funky way [14:15] hatch: try the whitespace trick? [14:16] https://github.com/juju/juju-gui/pull/149/files#diff-8d19084702f50f400641361a9cb78591L1069 [14:16] it picked stuff from within another closure and considered it part of another fn [14:16] which is 'technically' correct [14:16] it's just hard to read [14:17] yea, the side by side is a little easier [14:17] but not much [14:19] I'll call this "Friday's hot fixes vs charm development" hell :-/ [14:19] so hot http://i.kinja-img.com/gawker-media/image/upload/hrlzerlexsbdxj8u0r5h.jpg [14:20] frankban lol [14:20] why do you think we haven't updated node in 6 months :D [14:20] we did not update node, it's node that updated us [14:21] lol [14:23] we were happy with our old good node, they decided to betray our trust, and to make us haters [14:24] there is too much drama in the node world right now [14:24] heh, no kidding. You see the npmjs stuff? :P [14:24] ugh what a shitshow [14:24] wheeee [14:24] community building at its best [14:24] I feel like I'm back in highschool student body elections [14:28] all I can say is that I'm super glad I didn't "donate" to npm [14:28] yea, i almost did and very glad. I'm be fuming [14:29] I'm still trying to figure out how a product a guy made while working at another company was able to break off into another company [14:29] unless he did it in his own time... [14:29] but I was sure it was done during the normal work [14:29] heh, guess who's investing and getting a bit of the company :) [14:29] * rick_h_ says in a cynical voice [14:29] lol [14:30] I'm wondering what their revenue plans are [14:30] they are pretty unstable as-is using couchdb [14:30] so not sure if people would want to pay them for local instances [14:30] s/local/private [14:31] I'm just blaming couchdb because that always seems to be their reasoning when something goes down [14:33] jujugui: i need a prelimary review of this charmworldlib branch that we're going to include several places as a forked version until it can be landed upstream. any takers? https://codereview.appspot.com/69430043 [14:35] bac: I'll try to take a look later if nobody else already did, now in the middle of a critical task [14:35] frankban: no rush [14:36] bac: looking [14:40] jcastro will the app developer week videos be recorded? [14:44] I am assuming so [14:44] they're on G+ [14:54] jcastro ok cool, I really want to watch but I don't think I can convince rick_h_ to give me that much TO :D [14:54] :P [15:05] rick_h_ I've NEVER used middle click in the course of my daily activities :) [15:05] always ctrl+c/ctrl+shift+c [15:05] the latter being VERY irritating when coming from iterm2 [15:05] hatch: ugh, neanderthal :P [15:06] haha, I just wish someone would make a good linux terminal like iterm2 [15:06] oh you've got to be kidding me [15:06] urxvt is 10x the terminal iterm2 [15:06] do I have to use ctrl+shift+c/p to interact with the terminal? [15:06] no, you use middle click and all is well [15:07] there is no middle click on a touch pad :) [15:07] which is why I'm so irritated [15:07] hatch: I use ^C ^V and remapped ^C to ^K [15:07] no idea what you use keyboard wise to copy/paste in a linux terminal [15:08] interacting with the terminal in Ubuntu always feels like the terminal is a second class citizen [15:08] heh, I feel that way in osc [15:08] osx [15:08] it doesn't remember the size it was, home/etc is stupiud "fn arrow" crap [15:08] urls aren't highlighted and ready to go [15:09] yeah that's osx for ya [15:09] are you using iterm2? [15:09] it's highly customizable and you don't need to do any cryptic commands to change font colours :) [15:10] really though....it should remember all your settings [15:10] mine does [15:10] yes, I'm using iterm2 [15:10] and any terminal that isn't config file configurable sucks... [15:10] no, you've got to go into the UI and set x, and y, and then z. Then go over to tab 5 and enter blah blah [15:11] instead of git clone github:config_files.git && make install [15:11] frankban, hey I just noticed, is there no way to make a manual provider with quickstart? [15:11] jcastro: no, maas and manual are not yet supported [15:12] jcastro: of course quickstart does not refuse to start unknown envs if they already exist [15:12] rick_h_ I am pretty sure you can import/export iterm2 profiles [15:12] ah ok [15:13] hatch: yea, I got some colors doing that. [15:13] hatch: not sure how much config it'll support [15:14] my biggest issue with things like vim/terminal etc is the discoverability of features is 0 [15:14] oh you want to show 256 colours? sorry we only show you 16 [15:14] :) [15:15] hatch: it's why I love my git with my config [15:15] solve it once and never worry about it again [15:16] until new features come down and you realize you didn't know about them until a year later :P [15:16] :) [15:31] * hatch longs for a code-coverage tool for our tests [15:31] * hatch sits up late at night sobbing for it's coming [15:32] hatch: good sprint idea to hack together [15:33] tbh the real reason I want it is so that I don't have to keep going through the tests and my refactored code to make sure i hit every branch :) [15:33] it's very time consuming [15:33] hah [15:34] http://www.autoblog.com/2014/02/28/lexus-rc-350-f-sport-geneva-photos/ [15:34] I think I need a second job [15:35] not a fan of that front end [15:35] yeah it's a love it or hate it design [15:35] I'm in the former category [15:35] "hey, you guys forgot to finish the bumper" [15:36] it's like what would happen if under armor sponsored a car design [15:36] "new breathable mesh" [15:36] hahaha [15:36] typically that much air intake on the front of a car is an aerodynamic nightmare [15:37] I'm curious as to how they solved that [15:37] if they did... [15:37] rick_h_: could you take a quick look at https://codereview.appspot.com/67990044 ? [15:37] frankban: loading [15:37] cool thanks [15:37] frankban: need a qa? it looks good here. [15:38] rick_h_: good point. I think a QA might help. make deploy && juju set juju-gui juju-gui-source=develop && check veerything is ok [15:39] everything even [15:39] frankban: rgr, will do a ec2 test just to make sure we hit as close to live as possible. [15:39] rick_h_: sounds good thanks [15:41] frankban: seen https://pastebin.canonical.com/105721/ ? [15:42] rick_h_: it's a core error. are you using trunk? [15:42] frankban: testing something out. Looks like I might have had an old env laying around :/ [15:42] rick_h_: if not, do you have a stale jenv file? [15:43] :-/ [15:43] yea, think I had some jenv around. My bad. Running juju on too many machines and don't realize I leave things in a bad state [15:44] good to see that someone finally fixed that imagemagic issue that we had when upgrading before :) [15:50] jujugui call in 10 [15:57] jujugui call in 2 [16:05] rick_h_, I have questions for you after your call [16:05] or if you guys want me in there at the end to make fun of me collectively I can do that [16:05] jcastro: after call is cool, be a bit [16:06] no worries! [16:08] http://mitechie.planningpoker.com/participants/new?token=hm67co [16:44] everyone votes 5 on leaving [16:45] jcastro: got time when you want to chat [16:45] bac haha [16:47] I was going to work a bit on my ghost charm this weekend [16:47] and create a bundle which deploys everything to machine 0 [16:47] hatch: woot [16:47] but there is a bug which will block that :( [16:47] the gui binds to port 80 [16:47] where ghost needs to be [16:48] hazmat filed it a while ago actually https://bugs.launchpad.net/juju-gui/+bug/1256426 [16:48] <_mup_> Bug #1256426: charm, allow binding to just 443 so can colocate with webapp on port 80 [16:48] it would be awesome if we could squish the fix into the schedule somewhere [16:48] yea, it'd be good to get that in as a config param [16:49] hatch: put a card in the backlog maint pool [16:49] and we'll try to get it up for looking at as part of a 2wk cycle [16:49] ok cool can do, could the card be 'supply new http and https ports for the gui' ? [16:49] rick_h_, https://plus.google.com/hangouts/_/72cpi1ph534ssfjd37rq6d5png?hl=en [16:49] so that the gui could be on any ports (in case the webapp needs https as well) [16:50] hatch: currently the default is 443 and 80 redirects to 443. with secure=false you only have 80 [16:51] hatch: so I guess those are two separate tasks 1) an option to only bind to the real port without redirecting 80 (defaulting to false) and 2) and option to set the real port (defaulting to 443) [16:52] ok great thanks I'll update the bug and make appropriate cards [16:52] it would be super awesome to have a Ghost bundle all on a single machine [16:52] hatch: the only missing bit is what happens when you set secure=false [16:52] juju quickstart ghost-simple [16:52] :) [16:54] rick_h_: is charm deployment going well? [16:54] frankban: yes [16:54] frankban right - I'd be happy to say certain config options are invalid when others are set [16:55] frankban: looking and testing it out but otp [16:55] if secure=false then you MUST have provided a insecure port [16:55] for example [16:55] sounds reasonable [16:55] rick_h_: cool thanks [16:59] frankban: looks good here thanks! [17:00] rick_h_: great [17:01] jujugui: starting the process for a charm release [17:01] :+1: [17:01] * hatch wishes his irc client did emoticons [17:02] Just make it do a train whistle on :+1: [17:02] lol [17:03] :-) [17:03] perfect [17:03] jujugui I'm not getting canonical admin emails right now so if you need approvaly for anything please ping me [17:03] rick_h_, ping :) [17:03] Makyo: lol [17:07] anyone else having issues sending emails? [17:10] jcastro: I lied, that is me [17:11] rick_h_, ok, let Joey know [17:11] jcastro: yep, replied [17:22] UGH [17:23] hatch: loves life... [17:23] I am irrationally agitated about the RL-1 not being available in English [17:23] RL-1? [17:24] http://www.revenuquebec.ca/en/sepf/formulaires/rl/rl-1/default.aspx [18:06] * Makyo ducks out to laundromat, will be on once there under diff. nick. [18:11] yay! new yubikeys arrived. Time to yubikey all the machines! [18:14] yay [18:14] can stop swapping them across machines now [18:33] jujugui: cs:precise/juju-gui-85 released and ingested. EOD, have a great weekend! [18:33] frankban: thanks so much for that! [18:33] have a good weekend [18:34] ty [19:18] jujugui if you've got any holiday time in the next two weeks please make sure to submit it so I can setup the number of cards correctly please [19:18] hmmm [19:29] oh so DRAB [19:29] :) [19:29] DrabMakyo I got a funny story for you [19:29] ok it's not funny at all [19:31] ruh roh [19:31] I must prefer funny stories...especially when they're funny [19:33] Hah! Oh yeah? [19:33] this one time, at band camp, hatch wrote CS [19:33] there is this reality tv show where they follow customs agents and all the dumb/funny things that happen [19:33] SHH Don't get him started. [19:34] like people trying to get into the country with warrants and stuff like that [19:34] anyways [19:34] this one guy got pulled aside for whatever reason and they pulled out a fox suit and was like 'wtf?' [19:34] and he said "I'm a furry..." [19:34] I smiled [19:34] Pff. [19:34] Nice. [19:34] because I knew what we was talking about lol [19:34] umm, ok [19:35] hah, come to penguicon, you'll get a lesson or three [19:35] It's super important, okay? Haha [19:36] I had never heard of it until, what, 3 weeks ago? and now I've seen it on TV [19:36] crazy times [19:36] It's weird enough to be a media darling, yeah. [19:36] lol, you've never heard of it? [19:36] nope never [19:36] I'd think in that cold up there you'd all have nice warm furry costumes [19:36] HAhaha [19:37] rofl [19:37] hatch is sheltered. [19:37] hey! I'm sure you guys have never heard of a snowmobile poker rally! [19:37] You got me there :) [19:38] haha [19:41] jujugui lf a review on https://github.com/juju/juju-gui/pull/150 [19:41] * hatch is stick of looking at that darn branch [19:41] hatch: looking [19:42] thanks [19:42] jujugui I've gotten the cards pulled out for the 2 week cycle. Lots of maint hopefully we can burn that down. [19:42] let me know if anything looks off, but basically there's the plan for the next 2 weeks [19:43] rick_h_ heh, it looks like a lot of work for two weeks [19:43] I know it's "accurate" it just looks like a lot :) [19:43] 5 devs * 10 days * 2 points per day = 100 points [19:44] that's about 79 points on the board [19:44] rick_h_ did you make changes to the CI? [19:44] hatch: my bad, I forgot that jujugui was our bot [19:44] and removed him from the juju team...oops [19:45] added back, sorry [19:45] np, I just noticed a 'Merged build started' and 'Merged build triggered' and got concerned :) [19:46] yea [19:46] same here [19:46] can he be renamed to jujuguibot? [19:46] hatch: I suppose [19:47] whichever :) [19:47] 11 files?! [19:48] yeah see why code coverage would have been nice? lol [19:56] responding while the comments come in is pretty cool [20:07] rick_h_ I responded while you were making comments so you will want to go through them again :) [20:07] lol, now to go read all the replies that are build up [20:09] Haha, ouch: http://twitter.com/appcode/status/437896886649757696/photo/1 [20:09] hahaha [20:09] well played [20:09] well played [20:10] very retweet worthy [20:12] jcastro do you have someone fixing that weird busted sticky header thing on discourse.ubuntu.com? Would you like someone to look into fixing it? [20:12] hatch: replies to your replies done [20:13] rick_h_ thanks, I'll get to making the changes [20:13] I'll get a QA setup going here [20:23] rick_h_: now that I see what you mean by the JSON we get from cs, I think it might actually be easiest to just build the URLs from the GUI side. At that point, we know we're getting charms from teh store because we requested the bundle from the store. [20:24] DrabMakyo: right but the json on a bundle details includes the bundle info so we don't have to make 100 calls out to the api [20:24] DrabMakyo: so I'm not sure how we can move that logic gui side [20:27] It looks like it contains most of the info there... https://gist.github.com/makyo/9279138 Maybe I misunderstand, though. Let me poke around a bit more. [20:27] Just need to change these lines: https://gist.github.com/makyo/9279138#file-bundle-json-L1214 etc. [20:28] ugh I gota run to the bank real quick [20:28] I'll be on the cell [20:29] I'll keep prowling, though. Definitely want to be sure. If nothing else, at least we have most of that info already in guiland [20:29] DrabMakyo: hangout? [20:29] maybe I'm just confused [20:29] rick_h_: Uh, sure, can step outside. [20:30] DrabMakyo: ah, nvm [20:30] we can catch up monday [20:30] my thoguht is that charmworld can update the charm section for each service when it writes out that json [20:31] and then everything should 'just work' and it knows the url since it's building that charm_metadata [20:31] rick_h_: oh, yeah, that'd work perfectly. [20:31] having gui aware that there's some magic there seems the wrong place for the magic [20:31] Ah, yeah, that makes sense. [20:31] rick_h_: would that be on outputting the json or on ingest/migration? [20:32] I had mentioned ingest in the doc, but could do either. [20:32] well my thought was to do it on ingestion so the data in the db is clean [20:32] Okay. [20:32] +1 [20:32] vs havnig to remember to clean it any time we output it [20:32] YEp, sounds good. [20:33] Laundry's done. Going to run back home. [20:33] Back in a few. [20:52] jujugui I'm out. Have a good weekend. Thanks for the branch hatch [21:13] back [21:13] ugh Friday bank and road traffic