[01:28] jujugui looking for a review and qa on https://codereview.appspot.com/29400043/ [01:29] Will review. [01:29] thanks [01:29] May not get to QA [01:29] Will try, though [01:29] I just did a very deep qa so it's probably ok.....but ya know should probably have one anyways :) [01:29] stepping away for supper now [01:31] Cool, yeah, same reason. Code LGTM, will see about Qa after dinner [02:30] gary_poster, re unit errors, that's a previous behavior, ie errors at any point are considered fatal, part of it goes back to the original dev of deployer around stack testing/dev. noted that its not good behavior for this use case [02:34] ack hazmat thanks. if we provide branch that changes, should we make it optional and non-default, or something else, or you're not sure yet? [02:38] gary_poster, so there's quite a few cleanups in that same vein.. deployer has poor error handling for a library [02:38] agreed, not designed for that [02:38] but now that's what we want [02:39] gary_poster, ack, so a flag usage for it is fine. [02:40] hazmat, cool, thanks. default to erroring, to keep current behavior? [02:40] yup [02:40] k [02:40] thanks [02:42] gary_poster, incidentally commited that 0 int fix that frankban was looking for. [02:43] awesome, thanks, will pass that on :-) [05:09] gary_poster, fwiw just put in a deployer merge proposal to update error/warning feedback handling, frankban as reviewer. [13:11] jujugui hi. I have to be daddy at home this morning for a bit. Today and tomorrow are vUDS. I have added five sessions today and tomorrow to the UI team Google calendar that I think are interesting for us (and some we are essential) [13:11] Please take a look and see what you can attend [13:11] ack [13:11] k [13:22] gary_poster: ack, available for the pre-imp when you want. guihelp: I need one review+qa for https://codereview.appspot.com/29540044 (quickstart). Anyone available? thanks [13:23] Makyo: ^^^ includes bootstrapping with sudo, so you might be interested [13:25] thank you [13:26] rick_h_, you back to the land of the living today? feeling better? [13:27] * gary_poster off on morer daddy duty [13:33] benji, rick_h_: i think i've figured out my mongo problem. if you bootstrap a local env in juju it creates /etc/init/juju-db- files to start mongo on port 37017 for its use. destroy-environment should remove those files. if, however, they are not removed, when you reboot that mongo instance will spin up and then the upstart for your real mongodb will see a mongo instance running and start-stop-daemon will refuse to launch another. [13:34] this seems like a bug in juju; running a local env shouldn't disable an already-running mongo [13:41] bac: yea [13:41] gary_poster: definitely [13:41] bac: I wonder if this will be fixed by the new mongodb package they're looking to use [13:42] frankban: I can peek at it since I've been wanting local quickstart [13:42] rick_h_: thanks [13:47] rick_h_: i don't know. i filed bug 1253084 and have stirred the pot in #juju-dev [13:48] <_mup_> Bug #1253084: local use of mongo prevents default mongodb from starting [13:54] if I ever have a mock question I'll be banging on frankban's door [14:14] rick_h_, great. You happen to be attending the charm quality vUDS right now? Was hoping you could be our representative [14:14] rick_h_, http://summit.ubuntu.com/uds-1311/meeting/22007/t-cloud-juju-charm-quality-points/ [14:14] gary_poster: I'm listening in, I didn't want to jump in the middle [14:14] gary_poster: I missed getting in on time as I was reviewing and didn't look at the calendar until a little bit ago [14:15] I can jump in [14:15] rick_h_, ack. could you please ping arosales and let him know that you are the GUI representative, if he'd like one participating? [14:15] thank you [14:15] yay, wasted afternoon trying to figure out juju/mongo fighting only to have bug acknowledged and marked 'low'. go team. [14:15] :-( [14:16] gary_poster: do you know how to join the hangout? [14:18] rick_h_, you ask arosales or other participants on channel for the link, I think. Are you on the irc channel? [14:18] gary_poster: got a link [14:18] gary_poster: yep, all good [14:18] awesome [14:23] Makyo: hey are you in yet? [14:24] jujugui anyone available for a qa? Review is already done https://codereview.appspot.com/29400043/ [14:24] frankban, ready for call any time soon? My two year old may distract us, but I need to try and move forward. [14:26] hatch I know you qa'd this a lot. if you get a sandbox qa I think that's probably fine, if that helps [14:26] I'll see if I can do it quickly [14:28] gary_poster: I am using the phone connection now :-/ but we can try [14:28] frankban, can do the qa first. would it help to wait a bit? [14:29] frankban: woot, boostrap to gui: .venv/bin/python juju-quickstart 7.14s user 0.48s system 6% cpu 2:01.61 total [14:30] gary_poster: yes it could, could you ping me later? [14:30] frankban, sure, or you ping me when you are ready [14:30] rick_h_: cool [14:30] gary_poster: ok [14:36] good to see rick_h_ got a link [14:36] arosales: all good, thanks [14:48] hatch, silly, but in sandbox, difference: Status Info: hook failed: db-relation-changed vs Status Info: hook failed: "config-changed" (notice quotes) [14:48] hatch, also Status Data does not seem to update properly [14:48] in sandbox [14:48] hmm [14:48] should disappear when no longer in error [14:48] yeah? [14:48] it should.... [14:48] checking on that [14:49] ahh right, I never clear the agent state data on non errors in sandbox [14:49] good catch :) [14:50] the quotes are in the response from core so I'll add them to the relation-changed text [14:50] cool. also, hatch, more concerning for larger system (not your branch, probably), but I can pretty easily (1 in 12 attempts, say) click on a unit in error and get no information [14:50] then if I click again it will show error info [14:51] yeah I noticed that too [14:51] just going to make sure it happens on comingsoon [14:52] good idea [14:52] yeah it's there [14:52] I'll file a bug [14:52] thanks. odd [14:54] yeah that doesn't make any sense - probably some type of race condition with the databinding [14:54] tried the "break on any error" setting and triggered the problem: nothing. so that won't help. [14:56] gary_poster: the network seems to work better now [14:57] cool frankban. hatch, sandbox QA OK with given notes. [14:57] brb [14:57] great thanks gary_poster they are fixed, just testing locally now [14:58] great [14:58] watching, waiting, of course I picked the only unit that the simulator loves to keep in error hah [14:59] :-) [15:01] frankban, https://plus.google.com/hangouts/_/76cpjkfegd2n1821q2jb6snlo4?hl=en no rush [15:04] bac: do you have any getting-started hints on charmtools dev? I'm library-ifying proof and both the readme and makefiles are crap [15:12] benji: don't forget the test runner! [15:12] i mean in your list [15:12] bac: definiately; maybe that should be the first thing I do (along with making the README and Makefile better) [15:12] but, no, i have no tips [15:13] k [15:13] benji: getting a sane test suite would be the first task i would think. [15:13] yep [15:13] benji: note there are lots of python tests that exist, and may even be good, but are not run. [15:14] I guess I'll use nose, since it seems to be the popular choice of the times. [15:14] I'll lay odds that tests that are not run will not be very good. [15:22] gary_poster: I don't hear you [15:32] -19C today....oooo winter hath arrived [15:33] hatch: beat me, 1.6C here [15:33] and you have winter tires....pssshht [15:33] heh, it's more for Jan [15:33] and the utter crap stock tires [15:34] hey jujugui dudes, is there anything y'all have done for charmworld that's not yet safe to release to production? [15:34] not me [15:34] jcsackett: not that I am aware of, let me skim the recent changes real quick [15:35] jcsackett: when you releasing? [15:35] bac: if you guys don't have anything dangerous, i'll put in the RT today. [15:36] jcsackett: ok, we may follow up with another shortly [15:37] jcsackett: it looks safe to me [15:38] benji: thanks. [15:39] np [15:40] jcsackett, btw re comments in bug 1253123, please make sure that orange squad is coordinating with UX and UI team on any big charmworld changes: there are some gigantic plans up in the air there and we should make sure that we don't work to cross purposes [15:40] <_mup_> Bug #1253123: No javascript test infrastructure [15:40] * gary_poster returns to call [15:40] gary_poster: when we get to it, you got it. [15:40] cool [15:42] jcsackett: kernel panic != crash? [15:42] :) [15:43] hatch: kernel panic is more descriptive, and i was punting it over to the ubuntu folks so i thought i would spare them interpreting the bug. :-) [15:43] oh haha [15:43] *really* not sure how that got filed against charmworld. [15:44] it makes sense when someone accidentally files against, say, launchpad, since that's the site you're on. can't fathom the pattern that led to that filing. [15:44] it was probably the first in the dropdown and they didn't change it [15:44] haha [15:46] man it feels so weird not having Twitter installed on my phone [15:46] I never realized how often I checked it haha [15:50] jujugui call in 10 [15:58] jujugui call in 2 [16:00] bac: http://www.voidspace.org.uk/python/mock/patch.html#where-to-patch [16:00] bac: the big thing is to patch the module using vs the original source [16:00] benji, starting [16:36] * benji takes an early lunch. [16:38] hey bac, could you give me the link to the bundle docs, so I can refer to them on the vUDS call? [16:38] benji, frankban is the juju core call over already? [16:38] gary_poster: yes [16:39] anything interesting? [16:40] gary_poster: for us, ha the most interesting. unit leader election could be something we'll had to represent too [16:40] gary_poster: yep [16:41] frankban, we don't need to represent ha though, right? except maybe as a switch? IIUC, their HA approach leaves us out in the cold for being able to transparently switch [16:41] yeah, it was things we already knew were coming [16:41] frankban, "unit leader"? [16:41] new idea, for master slave? [16:42] gary_poster: ha, isn't it something we need to support in the guiserver? [16:42] frankban, the way that mramm described it to me lately is that they plan to only deliver db level HA [16:42] gary_poster: oic [16:42] you have to do something manual to switch state servers IIRC [16:43] maybe that has changed yet again, dunno [16:43] yeah, forgot about that distinction [16:57] Makyo: ping [17:01] frankban, Hey [17:02] Makyo: I landed a branch which makes use of sudo in quickstart. I just tested it by ensuring (with mock) that utils.call is called with the expected args. Is it any different from what you are doing re sudo stuff? [17:04] frankban, I'm mocking subprocess.Popen: http://bazaar.launchpad.net/~makyo/juju-quickstart/ensure-juju-lxc/view/head:/quickstart/tests/test_utils.py [17:04] gary_poster: is the session going? i see nothing. [17:04] bac, trying to figure out where to go :-/ [17:06] gary_poster: you wanted this: https://docs.google.com/a/canonical.com/document/d/1paE70HEVgKA6LIjngzMa83OYr5jkZrOJ0hVNrYUuva0/edit [17:06] thank you bac [17:06] jujugui, jcastro points out I got the time wrong. sorry. starts in 1 hr [17:07] gah! [17:07] bac: gets extra lunch time! [17:07] gary_poster: where is the vuds links and whatnot? I can't seem to find them anywhere on the ubuntu website [17:07] rick_h_: i raced through my leftover steak and beans [17:07] Makyo: ack, quick hangout? [17:08] bac: desert time! [17:08] frankban, sure [17:09] the Chrome Dev Summit starts today as well :) [17:09] rick_h_: hmm, i could walk up to ben & jerry's [17:09] rick_h_: here is what i have, test-wise http://pastebin.ubuntu.com/6448832/ [17:09] doh! jealous [17:09] bac, sorry, I suck [17:09] Makyo: https://plus.google.com/hangouts/_/7ecpjmcijt0iomhsrqqr05lf00?hl=en [17:09] rick_h_: utils is where it is defined. views.bundles imports and uses it. i tried mocking both. [17:10] but, like i asked, can you be too far up the call chain for the mock to work? [17:10] bac: right, but how does views.bundles import it? I think you want from charmworld.views import bundles; patch(bundles.utils.get_bundle_downloads) [17:11] bac: assuming that in bundles it's from charmworld import utils [17:11] otherwise shrink it down until it directly matches [17:11] that might be patch(bundles.get_bundle_downloads) [17:11] rick_h_: OH. i don't think i imported it in the test b/c the test doesn't use it. [17:11] `juju run` sounds sweet! ME WANT!!!! [17:11] sorry, just listening to this http://www.youtube.com/watch?v=ZjzImjEinB8&feature=share&t=2m :) [17:12] hatch: so what's juju run [17:12] rick_h_: allows you to run any script as if it's a hook [17:12] bac: right, but you want to patch it in the view namespace so you'll want to import it and patch it there [17:13] so it has the hook api/env [17:13] hatch: ah, cool [17:15] rick_h_: something like http://pastebin.ubuntu.com/6449182/ [17:15] bac: yea, I think so [17:15] rick_h_: yeah, that doesn't work either [17:16] bac: push a branch up? [17:17] rick_h_: lp:~bac/charmworld/add-stats-to-api -- it is test_api.py that is failing [17:17] bac: k, second [17:18] I've always wanted one to call my own: http://www.hatchpet.com/ [17:20] like tamagochi [17:20] but not nearly as cool [17:20] oh THESE are the people who have @hatch [17:25] bac: you're patching the views/bundles but you're calling the api/__init__.py [17:25] bac: change that from views/bundles to views.api and it works [17:27] rick_h_: doh [17:27] rick_h_: thanks for the eyeballs [17:27] bac: np, been there done that [17:47] gary_poster: so the error titles are too long at 'hook error: db-relation-changed' so to allow longer hook names I tried 'error: webiste-relation-changed' which is also too long. so....I was thinking using ellipsis and then show the full title on hover? [17:48] hatch: I feel your pain, there's a guy who has @benjiyork who has tweeted about 5 times, over 3 years ago [17:48] so it would be 'hook error: db-relation-cha...' [17:48] benji: ugh - I wish Twitter allowed people to trade/give away names [17:49] but I just deleted the Twitter app from my phone because of their new permissions so my twitter use will likely drop dramatically [17:49] they have been hinting at a policy change for a couple of years now regarding "unused" handles, but nothing has come of it yet [17:52] yeah they seem very slow to move on things [17:52] hatch sorry was lunching. what about allowing the line to break? [17:53] gary_poster: it looks really bad [17:53] even with reverse indent? [17:53] the first thing it does is push the ^ arrow to the next line [17:54] gary_poster: https://www.evernote.com/shard/s219/sh/78b56078-9a9e-48ed-b935-734c3f78620b/93c2d0e6586dbf5a390a4669a481d93a [17:54] see the landscape line [17:54] darn, jcsackett; I was just about to start on bug 1252782 when you un-assigned it from charmworld [17:54] <_mup_> Bug #1252782: Ejecting external USB hard drive causes kernel panic [17:55] hatch, oh! I see. How about ellipsis, which simply line breaks when it is open? [17:55] lol benji [17:55] :) [17:55] gary_poster: hmm.... [17:56] gary_poster: https://www.evernote.com/shard/s219/sh/436d81d0-dfbd-44ad-af4d-1a51040e76ea/02aed2d352ef1bbed2132197a792cc62 [17:56] still looks kind of bad...no? [17:57] hatch, yes. ideally arrow would be on far right, and word woruld line break [17:57] so [17:57] 12 machines need to be ^ [17:57] restarted [17:57] ? [17:57] 12 machines need to be ^ [17:57] restarted [17:58] ? [17:58] heh [17:58] yeah that might work [17:58] so.. [17:58] hook failed: website-relation-.... ^ [17:58] then [17:58] hook failed: website-relation- ^ [17:58] changed [17:59] rick_h_, gary_poster were you guys going to join the Juju GUI Roadmap and Update session? [17:59] hatch yeah [17:59] that sounds good I'll do that [18:00] * hatch wishes there were an evernote and skitch Ubuntu app [18:00] * benji wonders why the "Starting Soon" screen on the hangouts looks so funny with the offset text and all. [18:06] that icon looks more like sliced bread. should [18:13] they say for gaming? I say for development! http://www.anandtech.com/show/7529/gaming-immersion-overclockersuk-launch-infinity-emperor-chair-and-workstation [18:19] benji &/| rick_h_: charmworld review at your leisure. https://codereview.appspot.com/29850043 [18:21] hatch, "powder coated Canadian steel." Y'all do something funny to your steel up there? :-) [18:21] (from your link) [18:21] bac: I'll take a look [18:23] gary_poster: lol maybe it's from the 'olden days' when people were using Chinese steel that would rust really fast [18:23] heh [18:24] "our steel may be from the 1970's but it was made in Canada!" [18:25] I can envision that in the FAQ [18:25] lol [18:25] bac, benji, pretty fascinating GUI update? :-) [18:26] and rick_h_ :-) [18:26] * benji just stopped applauding and sat back down. [18:26] lol [18:26] gary_poster: :) [18:26] "Is still not determined..." [18:26] Should have gone all secret agent "I could tell you, but then I'd have to kill you" [18:40] gary_poster, did you have some thoughts on how we should field bundle bugs? [18:41] gary_poster, right off hand I am thinking juju-core and we can triage from there, but wanted to get your thoughts [18:42] arosales, probably file against gui and/or gui + deployer. The interrelationship between GUI and deployer is somewhat subtle. [18:43] ya, and perhaps we triage the bug to that project [18:43] right [18:43] but for users may hit juju-core since it is all juju to them. [18:43] arosales, ack. then triage it again, I'd say. [18:44] I think nick was seeing some deploy errors around the wiki bundle. Still needs some further investigation. But if we can reproduce we will file against GUI for now. [18:44] cool thanks arosales [19:06] bac: got a sec to chat? [19:13] rick_h_: sure [19:21] bac: card added and assigned over to [19:22] bac: added in the notes what I found on the call/token issue [19:22] thanks rick_h_. i'll get right on it. [19:22] jujugui review please of my hacky addition of no op'ing all the things! https://codereview.appspot.com/29860044 please [19:22] rick_h_: ok [19:23] benji: did you grab bac's review then? If not I can look now [19:23] rick_h_: yeah, I just finished [19:23] benji: cool I've got perfect timing then [19:23] :) [19:24] gary_poster: can I get a couple of min when you get a sec please? I know you're backed up with some calls [19:29] rick_h_: "- Drive by fix for deploying a bundle via drag-n-drop." this is no longer true, right? if so,be sure to edit it out when you 'lbox submit' [19:29] bac: ah right sorry [19:30] rick_h_, sure. you need me now, or is in 30 min ok? [19:30] gary_poster: whenever is good [19:31] gary_poster: so 30min is cool [19:31] rick_h_, ack, thanks will ping [19:33] rick_h_: did [19:33] bac: ? [19:33] benji: the four character code review is the best. [19:34] rick_h_: i did your review. your review has been did. [19:34] I did something? [19:34] oh! lol [19:37] bac: commit message "fix all the engrish" [19:38] thanks for the review! [19:43] rick_h_, the hangouts, they are a-callin' : https://plus.google.com/hangouts/_/7ecpjmspmqkgcq1hpt7fl71lso?hl=en [20:43] jujugui two quickstart reviews + QA https://codereview.appspot.com/29980043 (down from +300 lines, thanks to frankban's branch this morning!) [20:44] I will take one in a few if there are still any spots [20:45] going to get boys from school; back soon [20:48] hey rick_h_, token.js shows a Token has a type (charm or bundle). but i don't see it getting set or used. am i missing something? (otherwise in that _addDragability i don't know how to tell them apart) [20:59] rick_h_: nm [21:02] * bac -> wtd [21:10] bac: cool, yea there's another method that uses it I was going to copy it from [21:41] Makyo, LGTM (but no QA) with extremely trivial [21:41] gary_poster, thanks! [21:41] welcome, thank you [21:45] rick_h_, if you are around, more than a little curious to see if you think I caught the gist of our CI conversation. Please correct/elaborate as necessary [21:48] gary_poster in the hangout notes? I can look at the coffee shop tonight. [21:49] gary_poster: oh, just noticed the email, all good. Will review [21:49] rick_h_, no emailed to peeps [21:49] cool thank you :-)