/srv/irclogs.ubuntu.com/2014/03/05/#juju-gui.txt

MakyoWhew, got it.  Man, ingest is a long path!00:11
rick_h_Makyo: :)00:12
frankbanhi dimitern: so, IIUC, rsyslog-gnutls is now required to be installed in the local machine to bootstrap a local environment, right?09:10
dimiternfrankban, hey, that's right09:11
frankbandimitern: and this will be a requirement for 1.1809:11
dimiternfrankban, yes, but it seems juju-local doesn't list rsyslog-gnutls as a dependency09:13
frankbandimitern: asking because, if that's the case, we'll need to update quickstart as well.09:13
=== rogpeppe is now known as 17SAAOELG
dimiternfrankban, isn't quickstart about installing juju-local?09:14
=== rogpeppe1 is now known as rogpeppe
frankbandimitern: quickstart installs the juju stable PPA, and then apt get installs juju-core, lxc and mongodb-server09:15
dimiternfrankban, from source?09:16
frankbandimitern: apt-get install09:16
dimiternfrankban, ah, sorry - from the ppa09:16
dimiternfrankban, ok, so does it install juju-local package?09:17
frankbandimitern: no09:17
dimiternfrankban, it should09:17
frankbandimitern: I guess juju-local is a meta-package for  juju-core, lxc and mongodb-server, correct?09:17
dimiternfrankban, http://paste.ubuntu.com/7037530/09:17
frankbandimitern: cool, and you'll add rsyslog-gnutls I suppose09:18
dimiternfrankban, I pinged jamespage about it09:18
frankbandimitern: so effectively installing juju-local means a complete installation, also supporting cloud providers09:20
dimiternfrankban, it should be enough09:20
frankbandimitern: I'll investigate why we decided to install single packages instead of juju-local. If there is no valid reason, I'll change quickstart to just install juju-local from the PPA.09:22
dimiternfrankban, tyvm09:22
frankbandimitern: thank you!09:23
* hazmat totally missed were quickstart went to using urwid09:30
hazmatwhe09:30
hazmatwhen09:30
hazmatfrankban, there's a change in jujuclient 0.17.3 that's worth noting.. occassionally juju core will disconnect all api clients09:31
hazmatfor watches specifically, i added a default parameterized behavior that will auto reconnect them09:32
frankbanhazmat: interesting, auto_reconnect=True09:34
frankbanhazmat: IIRC there was a similar bug some weeks ago, is this different?09:35
hazmatfrankban, its the same bug.. roger had re-worked timeouts in the api server which is why it had got marked closed last time.. 09:36
hazmatfrankban, but another manifestation of the same err, is that juju api-server goes awol occassionally09:36
hazmatsomething about a failure around mongodb socket, and then it disconnects all clients09:37
hazmatcore agents auto reconnect  and handle09:37
frankbanhazmat: ack, so this will be fixed in 1.18 I suppose09:37
hazmatfrankban, unclear.. 09:37
frankban:-/09:38
frankbanhazmat: are you planning an update for the jujuclient packages in the juju stable ppa? if so, we'll need to also update quickstart09:38
frankbanhazmat: I am also thinking about the jujuclient tgz in the gui charm, perhaps we need to use the last release there too09:39
frankbanhazmat: moreover, a mega-watcher random disconnection on 1.18 would also be a big problem for javascript clients (e.g. the GUI)09:43
frankbanrogpeppe: ^^^09:43
rogpeppefrankban: i'm planning to look at that today09:45
rogpeppefrankban: i haven't yet managed to reproduce the problem yet though09:46
hazmatfrankban, i had it pushed to trusty09:46
hazmatfrankban, i haven't really touched the stable ppa..09:46
* hazmat wonders if he can bribe someone to update deployer and client in there09:46
frankbanrogpeppe: cool, thanks09:46
hazmatrogpeppe, i've gotten about a dozen people reporting in the last two weeks.. i think canonistack environments help to reproduce ;-)09:47
rogpeppehazmat: i guess i will try to repro inside canonistack then09:49
frankbanhazmat: could you please ping us on PPA update? unrelated: are you still living on the wrong timezone?09:50
frankbanhazmat: re urwid, yeah, we use that for the environment management CLI09:51
frankbanhazmat: the only downside is that it's not easy to make it work on windows, so maybe we'll need another solution when porting quickstart there09:52
rogpeppehazmat: from the bug report (lp#1284183) it's not entirely clear how platform specific the repro script is09:52
_mup_Bug #1284183: jujuclient.EnvError: <Env Error - Details:  {   u'Error': u'watcher was stopped', u'RequestId': 9, u'Response': {   }} <api> <status> <juju-core:Triaged> <https://launchpad.net/bugs/1284183>09:52
hazmatfrankban, toddler09:53
hazmatfrankban, if you want to push it yourself your welcome to.. your pushing quickstart there?09:53
* frankban thinks about pyside09:54
hazmatfrankban, over pyqt?09:54
frankbanyeah09:54
frankbanhazmat: toddler?09:54
hazmatfrankban, baby not sleeping well09:55
frankbanhazmat: OIC09:55
hazmatfrankban, yeah.. never been clear to me pyside was going to keep up.. its already well behind on qt 5 support afaics09:57
hazmatfrankban, the only reason i've seen people choose pyside is licensing..09:57
frankbanhazmat: interesting, I did not look at that ecosystem for a while09:59
frankbanhazmat: so https://code.launchpad.net/~ahasenack/+archive/python-jujuclient and https://code.launchpad.net/~ahasenack/+archive/juju-deployer-daily , right?10:01
frankbanhazmat: I'll take care of copying binaries over to the juju stable ppa. I plan to do that on my next quickstart slot (next week probably, maybe earlier).10:05
hazmatfrankban, awesome thanks.. i think the distro packaging is different, it fetches from pypi10:06
frankbanhazmat: yeah I think we have the same situation with quickstart, where trusty uses pypi and our PPA builds use a launchpad recipe10:09
rick_h_yea, they've packaged the deps up from the pypi12:06
frankbanrick_h_: when you have time, could we please have a call?12:07
rick_h_frankban: sure thing12:11
rick_h_frankban: https://plus.google.com/hangouts/_/CONVERSATION/UgxGu-UJzXYihe-SGLV4AaABAQ?hl=en&hscid=1394021535421286643&hpe=11omarnn9ff7sl&hpn=Francesco%20Banconi&hisdn=Francesco%20Banconi&hnc=0&hs=3512:12
frankbanrogpeppe: in the near future we might want to add the following fields to the mega-watcher MachineInfo: Life, Series, ContainerType, Jobs SupportedContainers, Constraints, ParentId and maybe also Adresses. do you see any problems with that path?12:38
rogpeppefrankban:  mostly seems fine12:42
rogpeppefrankban: ContainerType and ParentId can be inferred from the machine id12:43
bacfrankban: i need a charm review if you have a moment: https://codereview.appspot.com/7123004312:44
frankbanrogpeppe: yeah, that's what we do now, and we can keep doing it12:44
rogpeppefrankban: machine Constraints are an implementation detail and should not be exposwed12:44
frankbanrogpeppe: so, should we use HardwareCharacteristics?12:44
rogpeppefrankban: SupportedContainers may not always be set (but if it is, it is valid)12:44
rogpeppefrankban: yes12:44
frankbanrogpeppe: cool, so,  what about: Life, Series, HardwareCharacteristics, SupportedContainers, SupportedContainersKnown, Jobs, Addresses12:46
frankbanbac: sure12:46
jcsackettbac: i forgot to ping you yesterday after putting the test up. it's now up at https://code.launchpad.net/~jcsackett/charmworld/bad-average/+merge/20927213:06
bacjcsackett: cool13:06
* bac looks13:06
frankbanbac: done13:09
bacjcsackett: nice, thanks.  approved.13:09
bacfrankban: thanks13:09
jcsackettbac: thanks!13:09
benjihas anyone seen errors like this when doing a "make devel"?13:14
benjinpm ERR! Error: ENOENT, open '/home/benji/work/juju-gui/github/node_modules/node-spritesheet/node_modules/q-fs/test/partial.js13:14
frankbanbenji: npm -v && nodejs -v13:15
rick_h_benji: can you paste the whole block there? I've seen ENOENT stuff before. 13:15
frankban??13:15
benji1.4.313:15
benjiv0.10.2613:15
frankbanuhm... same here13:15
rick_h_benji: you've got the sys deps, imagemagick ?13:16
benjirick_h_: I did what HACKING said, other than that I have no idea13:16
benjithanks for checking frankban 13:16
rick_h_benji: ok, it's in hacking but asking to verify13:16
rick_h_the only thing around spriting would be base node issues and missing imagemagick library used to do the spriting13:17
rick_h_that I can think of off the top of my head with the one liner there13:17
frankbanrick_h_: cards created13:25
rick_h_frankban: thanks13:25
frankbanrick_h_: what's the goal of "Update machine/unit models with extra data now available"?13:26
rick_h_frankban: well, there was existing data and then there's the ability to pull data via new calls? Or maybe there's not. 13:26
rick_h_frankban: so the first card was about fleshing out our models with existing data, and then updating them with what we need from the new addmachine calls13:26
rick_h_frankban: so does addmachine give us access to anything we didn't have? Or maybe it sounds like it's all the in delta and we've got that covered now?13:27
frankbanrick_h_: so, addMachines returns something like {13:28
frankban            err: 'only defined if a global error occurred'13:28
frankban            machines: [13:28
frankban              {name: '1', err: 'a machine error occurred'},13:28
frankban              {name: '2/lxc/1', err: null}13:28
frankban              // One entry for each machine/container added.13:28
frankban            ]13:28
frankban          }13:28
frankbanrick_h_: so I don;t see any new info for us13:28
rick_h_frankban: ok, then we can ditch that card13:28
rick_h_frankban: gone, thanks for checking up on it. 13:29
frankbanrick_h_: cool. Fortunately we can parse the parent/container relationships from the machine names13:29
rick_h_benji: any luck?13:30
rick_h_benji: want to look at doing a screenshare/call to try to see what's up? 13:31
frankbanrick_h_: so right now we have name, agent_state, agent_state_info from the watcher + parentId parsing the name. I guess it is everything we need for machine view 1.013:31
rick_h_frankban: and we have the machine spec info ?13:32
* bac afk...bbiab13:33
frankbanrick_h_: machine spec is used to deploy units to a specific machine: this is now enabled, and IIRC units have an attribute for the machine name they belongs to13:33
rick_h_frankban: so we need the machine constraint metadata. 13:34
rick_h_frankban: sharing the machine view pdf with you, should have an email13:34
frankbanrick_h_: ok, I'll take a look after lunch13:35
benjirick_h_: here are the first few lines of "make devel" and the first (of many, many) errors I get: http://paste.ubuntu.com/7038531/13:39
rick_h_benji: check the owner of ~/tmp and ~/.npm and make sure it's you. 13:41
rick_h_benji: and if you try a make clean-all and make devel again does it hit the same issue?13:41
benjirick_h_: yep, me in both cases13:45
benjiif I remove those two directories and run "make devel" again, I get a different first error:13:47
benjinpm ERR! EEXIST, mkdir '/home/benji/work/juju-gui/github/node_modules/yui'13:47
rick_h_a make clean-all should remove all of those directories and start again13:48
rick_h_clear the cache that is and local copies13:48
benjiI don't think clean-all touches ~/.npm13:48
rick_h_but node_modules isn't in ~/.npm13:49
rick_h_/home/benji/work/juju-gui/github/node_modules/yui13:49
rogpeppe128418313:50
rogpeppefrankban: seems reasonable, although i'm not entirely sure we need SupportedContainersKnown13:50
rogpeppefrankban: s/reasonable/good/13:50
rick_h_benji: it looks from this traceback that you had an npm install get part way and fail one time and it doesn't like to re-run. So make clean-all should remove the existing stuff and restart.13:53
rick_h_benji: the thing to see if the failing traceback on a clean first-run of the npm install if it fails again13:54
benjirick_h_: I had the same thought and am running several sequentially now13:54
rick_h_maybe you just hit a network snafu on a pass and got unlucky?13:55
benjiI don't think so.  It has never failed consistenly like this before.13:55
benjiI suspect either some setup setup that we forgot to document or a recent change that breaks new installs but works on old installs and we haven't noticed because we use git13:56
rick_h_benji: ok so I'm confused. Are you getting any successful runs after a make clean-all? You mentioned running sequentially now?13:57
frankbanrogpeppe: fo you mean SupportedContainers can be null when they're not determined and a map when they are?13:57
rick_h_not sure how using git effects the npm install?13:57
benjiI have not had a single successful run since re-installing my OS.13:57
rick_h_benji: the only recent change was the nodejs upgrade. Several of us have gone through the hacking doc on new installs of trusty. I'll setup a fresh lxc to see if I can dupe13:58
benjisince we install locally, and any amount of "make cleaning" will never be enough, there can be a hysteresis that causes existing builds to continue to work while new ones break13:59
rick_h_benji: right, so an lxc test is a good sanity check. But I know I've done three trusty fresh installs and gotten it running14:00
frankbanrick_h_: we already have support for adding machines/containers with a set of constraints. We currently don't receive HardwareCharacteristics from the meta-watcher. To be clear, we can set constraints for new machines but we cannot show hardware for existing machines. Looking at the slides, the current situation seems ok for now.14:01
rick_h_frankban: ok, good to know. There was some design around putting machine hardware characteristics in the UX but if that got pulled out even better14:01
* rick_h_ migrates to coffee shop to run from cleaners. back in a few min14:02
frankbanrick_h_: an idea could be to replace the 6-weighted card we removed with two 3-weighted cards to 1) update core MachineInfo and 2) reflect new info in our machine models. But I am ok if, as you mentioned, you want this work to be done later14:06
rogpeppefrankban: yes14:07
rogpeppefrankban: or an array14:07
frankbanrogpeppe: +1 then, yeah I mean, a sequence in general14:07
frankbanrogpeppe: oh, yes I used map I intended slice14:08
rick_h_frankban: yea, let's stick in in the on deck for now. We've added the existing new stuff in maint. 14:13
rick_h_frankban: and if we get time we can pull them from on deck to work on this 2wk cycle if we can14:14
frankbanrick_h_: ok sounds good14:15
hatchbenji hey what error are you getting when trying to build the gui?14:36
benjihatch: here's what I get from a fresh build http://paste.ubuntu.com/7038531/14:36
rick_h_benji: http://paste.ubuntu.com/7038757/ works from a clean machine to running gui14:37
rick_h_benji: I created an empty trusty lxc, set that as install-gui.sh and run it and end up with a gui running14:37
rick_h_benji: this is without any bind to my home directory or anything14:37
benjirick_h_: I'll try that as a script14:37
rick_h_benji: I'm running it a second time to verify on another clean lxc 14:38
benjiI don't think this machine is dirty, I've only installed about 10 packages (other than GUI things)14:38
rick_h_benji: understood, just trying to duplicate your issues on as clean as I can get. Don't even have the ppa install command 14:38
hatchbenji does npm have access to /usr/lib ?14:38
hatchdoes it work if you run it with sudo?14:39
benjihatch: I doubt it; it's not being run as root14:39
benjiI'm not about to run it as root14:40
rick_h_benji: ok, verified that script on a new empty lxc brings up the gui in one shot14:41
rick_h_benji: just watch the user:group in the chown 14:41
benjirick_h_: thanks; running it now14:41
rick_h_and <3 google wifi at starbucks. 6Mb/s downloads ftw 14:41
hatchbenji npm needs access to ~/.npm and /usr/local/lib/node_modules14:41
benjirick_h_: chown?  where did those come from?14:41
rick_h_benji: that's because when you first run npm install -g (global) it creates those two directories for your user owned by root14:42
benjihatch: the gui's npm install doesn't install to /usr/local/lib/node_modules14:42
rick_h_benji: so to be able to use them for the make devel local deps, you have to chown them to writ ethem14:42
rick_h_benji: this is why I was asking about if ~/.npm and ~/tmp was owned by you or root. 14:42
benjirick_h_: is that in HACKING?14:42
rick_h_benji: yes14:42
rick_h_benji: well the chown isn't14:42
hatchbenji ok then ~/tmp needs access too then14:42
hatchI had these in my hacking doc updates but SOMEBODY thought that it was too harsh :P14:43
rick_h_benji: that's why it was the first thing I asked about. If the directories already existed you're fine, only on a fresh intsall does it bring that up14:43
rick_h_hatch: no, you had it where root owned everything14:43
rick_h_or no, you chowned /usr/lib 14:44
rick_h_to your local user 14:44
hatchyeah14:44
hatch:)14:44
rick_h_which was not going to fly14:44
rick_h_:P14:44
hatchit is pretty stupid that the node/npm install doesn't fix the folder permissions for it to actually work14:45
rick_h_well it's stupid that a -g install touches your home directory at all14:45
rick_h_if I'm doing something globally nothing per user should be involved14:45
rick_h_there's a system /tmp for that14:45
rick_h_javascript jabber just put out a podcast with the npm guys. I'm tempted to listen to see if my rage at them will go down at all14:46
rick_h_or just go higher14:46
hatchhigher14:47
hatchdefinitely higher14:47
rick_h_that's a safe bet14:47
hatchlol14:47
rick_h_hatch: any reason for me not to use Y.isInstanceOf to check for a charm vs a bundle model?15:01
* hatch needs more backstory15:01
rick_h_hatch: I want to do X if charm model and Y if bundle model. I don't see a single instance of Y.Assert.isInstanceOf in our code and it makes me think I'm missing something15:02
hatchassert.equal(x instanceof Y.juju.models.Bundle, true)15:03
hatchlike that?15:03
rick_h_no, it's for real code not tests15:03
rick_h_I guess I can just check object.NAME15:03
luca__rick_h_: hatch I'm working on the UX for upgrading a charm in the inspector in regards to this bug: https://bugs.launchpad.net/juju-gui/+bug/1252586 if we remove it from the list of quick action notifications and maybe put a "Change version" link in the header which takes you to a dedicated space within the inspector that would be ok?15:03
_mup_Bug #1252586: Upgrade service is given too much real-estate <juju-gui:Triaged by lucapaulina> <https://launchpad.net/bugs/1252586>15:03
hatchif (x instance of Y.juju.....) ?15:04
hatchluca__ well imho the header might be the wrong place for it15:04
hatchit's a very limited use action15:04
luca__hatch: ok, sooooo where do you think it should live? Maybe in the configuration section?15:05
hatchso I'd even go as far as to put it on the configuration panel15:05
hatchyeah15:05
luca__hatch: thats the only other place I could think of15:05
luca__I don't really want to have a new pane for it15:05
hatchno I agree15:06
hatchjust taking a look one sec15:06
luca__should we alert users to new upgrades?15:06
hatchI'm thinking it should go in the configuration pane, maybe above the 'Import configuration file' button?15:06
hatchhmm15:07
luca__so populate a quick action notification for new upgrades, which disappears once they've looked at it?15:07
rick_h_luca__: hatch what about just making it something that either pops out a panel or that clears the inspector space and takes it over ?15:07
hatchthis would look like the 'service is being destroyed' notifications? 15:07
hatch^ luca__ 15:07
hatchrick_h_ I was thinking expands to push the settings down15:08
luca__hatch: no, a black notification like errors or running15:08
rick_h_hatch: luca__ I'm not sure a fan in the config because changing the version will change the config available15:08
rick_h_hatch: it's kind of two different things there15:08
luca__rick_h_: interesting15:08
hatchquick hangout?15:08
luca__sure15:08
hatchloading.....15:08
hatchhttps://plus.google.com/hangouts/_/7ecpjqdgivcbmgclulgpeuekf8?hl=en15:09
luca__hatch: rick_h_ cheers!15:23
hatch:)15:23
hatchrick_h_ ok back to your object compare issue....15:24
rick_h_hatch: ignore me, I've got something that works for now15:25
hatchhah ok cool15:25
Makyojujugui call in 1015:50
rick_h_jujugui small review please https://github.com/juju/juju-gui/pull/16015:53
jcastrorick_h_, can you ping me when the gui can show recommended bundles?15:53
rick_h_jcastro: that's the branch ^^15:53
jcastrojawesome15:53
rick_h_jcastro: will do, should be on comingsoon in a couple of hours. Won't be on jujucharms until next release15:53
hatchjujugui deploying local charms is broken on comingsoon until I land my current branch....sowwy15:54
rick_h_hatch: doh15:54
* hatch blames the qa'ers15:54
hatch:P15:54
rick_h_ruh roh15:54
rick_h_what's wrong?15:54
hatchit's trying to create an instance which doesn't exist15:55
hatchsimple one line fix15:55
Makyojujugui call in 115:59
hatchbenji ping16:00
benjihatch: pong16:00
hatchcall now16:00
benjiah16:00
benjiI waved goodbye to everyone on the hangout and then realized that I don't have a camera plugged in.16:07
rick_h_lol16:07
rick_h_we all waved back, mustbe magic16:08
hatchhaha16:10
hatchrick_h_ yeah they are talking about the $ now16:17
rick_h_hatch: heh good to know16:17
hatchbasically nodejitsu was running a single huge couchdb for the entire npm db16:21
hatch....16:21
hatchthat's why it kept crashing16:21
rick_h_yea16:27
hatchso the $ was so nodejitsu could afford to keep npm running16:27
hatchit had nothing to do with npm inc16:27
hatchI'm just surprised noone said 'hey maybe running a single huge instance is a bad idea'16:28
rick_h_yea, but I can't see $300K going to a couple month's bandwidth bill16:28
hatchwell no, I'm pretty sure they raised that expecting to fund the year16:29
hatchthen npm peaced out16:29
hatchlol16:29
hatchthey didn't cover that, but that's my guess16:29
rick_h_exactly, it sure seems like they're sitting aorund with $250K sitting in their back pockets16:30
rick_h_:/16:30
hatchnodejitsu is, yeah....16:30
hatchnot entirely sure what they should do with that16:30
hatchI mean, the project they were hoping to support just kinda left 16:30
hatchlol16:30
hatchand I didn't know this....but issac wrote npm before he started at joyent as a personal project16:31
hatchrick_h_ your CI failed - in case you missed that16:33
rick_h_right16:33
rick_h_hatch: I got it, updated branch is running16:33
rick_h_make lint fml16:33
hatchlol16:33
hatchgota get into the habit of pre-lint and testing :P16:33
hatchI am thinking of putting it as my pre-push hook :)16:33
hatch*sigh* fixing tests16:38
hatchfixing tests is demoralizing lol16:38
hatch"awesome this works so great.......*32 test failures*.....son of a...."16:39
rick_h_hatch: looks like it's the theme of the day https://twitter.com/git_clone_makyo/status/44125466288521216016:51
rick_h_lol16:51
hatchhaha damnit16:52
MakyoSigh~16:52
MakyoI hate to ruin its excitement.  It just can't help itself!  EEEEEE16:54
hatchhaha16:57
MakyoGot the comment that it looks like a Roguelike.  "It is very dark. You are likely to be eaten by an out-of-date mock."16:59
rick_h_hah16:59
hatchhahaha16:59
benjijujugui: I have a very small documentation branch that would have prevented me wasting a day getting the GUI set up: https://github.com/juju/juju-gui/pull/16117:12
hatchlooking17:12
rick_h_benji: done17:12
benjithanks17:13
benjihatch: you missed your chance17:13
hatch32 failures > 3 failures > 16 failures > 5 failures AHHHHHHHHH17:14
rick_h_come on, you know that the failure count in the tests is never worth a dran17:15
rick_h_darn17:15
rick_h_once a test fails all kinds of things fail for no good reason17:15
rick_h_one test at a time17:15
hatchlol17:16
hatchaswing-and-a-miss17:17
hatchI have found that running suites in the browser is way faster than running them in the console17:18
MakyoYes.17:18
MakyoDon't even need to worry about .skip then17:19
MakyoEr... .only17:19
hatch2!17:23
Makyo102 -> 97  here.  So many mocks to fix.17:23
rick_h_Makyo: what changed so bad? the charm url?17:27
Makyorick_h_, We get the service info back from that method in the JSON, but previously, we ignored everything except errors in that object.17:28
Makyorick_h_, So half of those are "expecting 3 values to unpack, only got two" and half are missing arguments in mocked methods expecting to receive that info.17:32
rick_h_Makyo: ah, gotcha17:38
hatchoh I love yui bugs17:41
rick_h_what did you break now? :P17:41
hatchfor whatever reason, a view is being destroyed but it's leaving it's container in the dom17:42
rick_h_that's the way it works17:42
rick_h_isn't it? if you give it a container?17:42
hatchnope http://yuilibrary.com/yui/docs/api/files/app_js_view.js.html#l31817:43
hatchohhh17:43
rick_h_http://yuilibrary.com/yui/docs/api/files/app_js_view.js.html#l160 you have to pass the option to delete17:43
hatchwaitasecond17:43
hatchyeah17:43
hatchwhat the farquad 17:43
rick_h_if (options && (options.remove || options['delete'])) {17:43
rick_h_so it's the way it's supposed to work :)17:44
hatchwhy the heck would you ever want to leave the container in the dom?17:44
rick_h_because you often render a View into an existing container17:44
rick_h_or swap views in that container17:44
rick_h_taking the container out would be a pita17:44
hatchbut a view renders it's own container if one isn't provided17:44
hatchso it should know that and remove it17:44
rick_h_it's not a widget with a boundingBox to keep 17:44
rick_h_k, so maybe it should destroy ones it creates, but it doesn't17:45
rick_h_and hasn't worked that way17:45
hatchcrazy, maybe I've always passed in a container so I never noticed it17:46
rick_h_yea, not sure. 17:46
rick_h_learn something new every day17:46
rick_h_but yea, we often in tests do a makeCotnainer and then destroy that container17:46
rick_h_so it tends to get cleaned up that route17:46
rick_h_and maybe didn't notice17:46
hatchyeah, well good thing we had this test :)17:47
rick_h_:) 17:47
* rick_h_ pats the test on the head...good test, good17:47
hatchhaha17:47
benjifrankban: would "Support ToMachineSpec in fake backend  ServiceDeploy and  AddServiceUnits calls" be a good card for me to take?17:49
frankbanbenji: sure, the arg is already in the real go env17:50
rick_h_benji: https://github.com/juju/juju-gui/pull/152 is the recent branch to the go env17:51
rick_h_oh, you reviewed it, heh17:51
benji:)17:52
benjifrankban: I'm a little surprised it is marked as size 3.  Are there complications I'm not seeing?17:53
rick_h_benji: well, when we scored it we didn't know there was a 'crib this branch for how it's done' in the time allocated. And that score is time to get through review and land as well.17:55
benjiok17:55
frankbanbenji: as Rick said, and also I guess some time can be spent for some investigations on the real juju-core side. e.g. what happens if the specified machine/container does not exist in the env?17:57
benjimakes sense17:57
Makyo97 -> 7 :P18:00
hatchnice18:02
hatchI'm doing a final run now18:02
hatchplanning on a few18:02
hatchlol18:02
hatchw000t 018:03
MakyoWoo! OK (SKIP=1)18:26
rick_h_and there was much rejoincing18:26
rick_h_err joicing18:26
rick_h_or something18:26
MakyoMan, I can't hold both git and bzr in my head at the same time :(18:32
rick_h_heh, welcome to the rest of us :)18:33
rick_h_every day I do something backwards18:33
hatchhmm18:52
MakyoDid... did _mup_ just go out for coffee?19:16
hatchoo boy laggy19:19
hatchMakyo that was the NSA installing new logging software19:19
hatchlol19:19
MakyoRewriting _mup_ in coffeescript, brb19:19
hatchlol19:19
hatchMakyo the other day I was reading a blog post that said "Atom plugins can be written in JavaScript" lol19:20
hatchso I am clearly not the only person haha19:20
rick_h_lol19:20
MakyoPff :)19:20
hatchactually I saw a review on Atom which showed it and sublime side-by-side and Atom had a pretty big lag comparatively  - I hope that's not indicative of webkit based apps on the desktop 19:23
MakyoIt seems okay to me.19:23
MakyoGranted I only briefly played around with it with the CS game thing.19:23
hatchyeah - I'm not sure one would notice if they aren't side by side19:23
hatchbut it was probably 500ms difference19:24
hatchchanging tabs 19:24
MakyoWhich I mostly wrote in Vim because Vim had good literate CS support.19:24
hatchjujugui lf a review/qa on https://github.com/juju/juju-gui/pull/162 plz and thanks19:26
hatchrick_h_ ok on to reviewing your branch19:26
rick_h_hatch: ty19:26
hatchholy crud we need to index the bundle table19:31
hatchlol19:31
hatchwow that's slow19:31
hatchrick_h_ review done19:32
hatchanyone able to review my branch atm? 19:36
hatchI'm going to continue on with it so I'd like to try and catch any issues up front :)19:36
rick_h_hatch: can in a sec19:40
hatchcool thanks19:40
hatchI think someone is DOS'ing someone in my internets 19:41
hatchis so slow19:41
rick_h_suuuure, blame it on the DDOS19:41
MakyoOn today's Not-The-0nion: I'm glad to hear that Macaulay Culkin is in a pizza-themed Velvet Underground cover band playing percussion and kazoo. http://thepizzaunderground.bandcamp.com/19:48
hatchlol wut20:04
MakyoThat's 9 minutes of my life I'll never get back.20:05
MakyoStrange things happen when you leave people home alone.20:05
hatchrofl20:05
rick_h_Makyo: I know we talked about grabbing a inspector card, but if you could peek at that bug I mentioned yesterday next and walk me through that I'd appreciate it. 20:42
rick_h_after all this fun, but just to setup the next up circle20:42
Makyorick_h_, Sure, though leave for vet in 15 or so.  If it's after EOD for you, I can probably get it done  if you want to hand it off.20:43
rick_h_hatch: how many views does this make?20:43
rick_h_Makyo: yea, I returned it to the ready to code yesterday to grab this one for the recommended issue20:43
hatchumm20:43
hatchI need more context20:43
rick_h_hatch: well there's two cards for "update viewlets to be views"20:44
Makyorick_h_, okay.  That's the d3 error?20:44
rick_h_hatch: so curious if this is a good time to move the card or if we're not half way ther eyet?20:44
rick_h_Makyo: yep20:44
rick_h_damn, using my kenisis, the air keyboard, and my thinkpad keyboard I think I'm getting worse at typing on all three of them20:44
hatchrick_h_ oh I'm leaving it there until I finish my current branch20:46
hatchthen I'll move it20:46
hatchwe will be 1/2 way then20:46
rick_h_hatch: ok, put it in review then?20:46
hatchnot yet20:46
hatchnot till my new new branch lands20:47
hatch:)20:47
rick_h_hatch: or by 'current' you mean next?20:47
hatchyeah sorry20:47
rick_h_k cool20:48
MakyoAlright, vet now, shouldn't be long.20:49
rick_h_have fnu Makyo 20:49
hatchthe microwave just turned on20:54
hatchand there goes my wifi20:54
hatchlol20:54
rick_h_hah20:55
hatchI think I'm going to go buy me a cable today too20:56
rick_h_I'm out all, night 21:00
* rick_h_ runs away21:00
hatchcyaz21:01
hatchoh boy this databinding stuff really is complicated22:23
hatchmaybe we should move to react22:23
hatchrick_h_ https://bugs.launchpad.net/juju-gui/+bug/128845022:51
hatch1000+ line diff and the only lint warning is that I forgot to doc a file23:04
hatchw000t23:04
hatchlol23:04
rick_h_hatch: k23:07
hatchthat one should probably be critical?23:07
hatchmake test-debug23:07
hatchbah23:07
rick_h_hatch: yea23:08
hatchwell for the morning I'll have 16 test failures to fix23:08
hatchthen the service config viewlet will be done23:09
hatchthis one was a lot of work23:09
Makyojujugui charmworld reviews/QA: https://codereview.appspot.com/7107005123:23
Makyohatch, https://twitter.com/horse_js/status/44135517251031449623:33
hatchrofl23:33
MakyoThe one problem I ran into that I really didn't like about CS was the optional parens were irregular if you were dealing with passing the result of a function call as an argument in another function call.23:35
Makyobut you could get around that with newlines.23:35
MakyoAnd it screwed with order of operations.23:36
hatchoh, that is a little wako23:44

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!