[00:00] thumper: how -> where [00:01] btw, nite/morning [00:12] why does revbiewboard hide the diff by default [00:12] that seems to be at odds with its primary function in life [00:13] * perrito666 fails at making dinner for reading this channel [00:55] thumper: ping [01:03] * thumper is back [01:17] axw: morning, you feeling better today? [01:27] wallyworld: hey. much the same as yesterday actually [01:27] :-( [01:28] axw: i have a favour - i have 3 PRs for 1.20, https://github.com/juju/juju/pull/760/761/762 - can you take a look for me? this will then allow 1.20.8 to be finalised for release this week [01:28] wallyworld: certainly [01:28] 2 are trivial [01:28] ty [01:41] axw: you make a good point about the gotool dependency, i'll remove it [01:42] maybe i should retain bash script licence also; the bug implied we wanted consistency, so i wnet for that [01:42] but i agree it doesn't make much sense [01:42] wallyworld: it isn't used in the same way as juju proper, so I don't think consistency makes sense [01:43] AGPL is about running things as a service, getting the same guarantees as GPL does for shipped code [01:43] yep, agreed [01:45] wallyworld: are we going to 1.20.8 as an RC again? would be good to have the MAAS change tested more thoroughly [01:49] axw: i will see if we can get it tested [01:50] wallyworld: all reviewed [01:53] axw: tyvm === Ursinha-afk is now known as Ursinha [02:13] playing with the gui on maas on the orange box and was bummed that there was no hardware characteristics. Does anyone know if that's a limitation of maas, or juju doesn't load/query it? [02:13] rick_h_: those are in maas so I suspect the provider code doesn't query it [02:13] bigjools: gotcha, /me ponders a bug for that [02:16] bigjools: ok, looks like it should be there: https://bugs.launchpad.net/juju-core/+bug/1193998 [02:16] Bug #1193998: maas provider doesn't return hardware characteristics of started instances

[02:17] oh man, the orange box is still on 1.18 [02:21] rick_h_: it's literally just been backported to 1.20.8 today, about an hour ago ... doesn't help if you're stuck on 1.18 tho [02:21] axw: gotcha, yea thanks. [02:22] axw: just playing with screenshots and machine view work on here and wish it was a little bit prettier https://plus.google.com/116120911388966791792/posts/dR3zqM6E66B [02:22] good to know it's coming though [02:22] ah, shame [02:23] wallyworld: did you ever get my email? [02:23] axw: no :-( [02:24] weird [02:24] resend? [02:24] just sent [02:27] got it [02:30] axw: did you want to have a hangout to talk about it? [02:31] wallyworld: in about 5-10 please [02:31] sure, just ping henevr [02:40] wallyworld: just doing the survey, will let you know when I'm done... not sure how long it's going to take [02:40] no hurry [02:40] i need to do it too [03:02] wallyworld: all issues raised for http://reviews.vapour.ws/r/20/ have been addressed. PTAL. [03:02] menn0: will do, thanks [03:03] wallyworld: ready when you are [03:04] axw: ok, will just finish a review and push up the licensing fix to trunk [03:04] sure [03:09] menn0: did you really mean to leave "foo" in as a way of saying "no"? :) [03:10] axw: yes. basically anything that isn't "y" or "yes" is treated as no so I wanted to include a nonsense value. [03:10] fair enough [03:10] axw: I followed the approach used by destroy-environment [03:11] oh that's a texst [03:11] except there on a non-response is tested for the no case [03:11] test [03:11] yep :) [03:11] my mistake [03:12] * axw blames it on the flu [03:18] menn0: +1 with suggestions [03:19] wallyworld: cheers [03:20] wallyworld: I like the 2nd and 3rd suggestion but I don't understand what you mean by the first one. [03:21] axw: no problems [03:22] menn0: just add the word "status" to the text, but ymmv (your mileage may vary) [03:22] ie a suggestion [03:23] wallyworld: got it [03:23] wallyworld: I thought you wanted me to include "your mileage may vary" in the help text :) [03:23] ah, lol [03:23] wallyworld: now that I understand it, that sounds good [03:23] ok :-) [03:24] axw: standup hangout? [03:25] wallyworld: yup, brt [03:48] wallyworld: anecdotal evidence is showing me more stability in our tests [03:48] my ratio of good to bad test runs is better now than it has been [03:48] thumper: indeed, but still a few failures to fix [03:48] landed my branch first time [03:48] * thumper nods [03:48] hit one earlier [03:48] but happy days [03:48] me too [03:48] but four times out of five today, it succeeded [03:49] i hit the mongo mac error too :-( [03:50] github.com/juju/juju/worker/uniter - filter_test.go:467: Error: unexpected config event [03:50] and FAIL: replicaset_test.go:175: MongoSuite.TestAddRemoveSet in the same test run [03:53] thumper: well, those tests are supposed to be getting removed [03:53] * thumper nods [03:53] i hope it happens soon [03:54] wallyworld: can I remove {state,params}JobManageStateDeprecated [03:54] axw: ^ [03:54] davecheney: i *think* so [03:56] davecheney: I think that a machine doc's Jobs may still contain it if upgraded all the way from 1.16, so we'd have this unknown job in parts [03:57] (if we removed it) [03:57] that's okay as long as we don't start barfing on unknown jobs [03:57] there's a TODO in cmd/jujud/machine.go to start barfing though [03:57] / TODO(dimitern): Once all workers moved over to using [03:57] / the API, report "unknown job type" here. [04:03] right, well i won't touch that [04:03] not with 1.18 being around for forever [04:03] yay [04:03] i found an unused function in envrions/manual which relies on an unused function in state [04:04] \o/ [04:06] * thumper runs out to collect cat from cattery now it is open [04:06] bbs [04:52] thumper: menn0 https://github.com/juju/juju/pull/765 [04:52] here is a quick one [04:52] % rbt post [04:52] ERROR: There don't seem to be any diffs! [04:52] ¯\_(ツ)_/¯ [05:16] davecheney: so you're in the branch you want to post? [05:17] yup [05:17] try "rbt post --tracking-branch=upstream/master" [05:18] davecheney: ^^^ [05:18] thanks [05:18] that worked [05:18] but now it wants my credentials [05:18] which ones are those ? [05:18] github ? [05:18] canoinical ? [05:18] gmail [05:18] ? [05:18] that's in Eric's email... it's your GH username and "oauth:" for the password [05:19] as per my email earlier today, we need to get the tracking branch config set up in the .reviewboardrc that's checked in [05:19] davecheney: ^^^ [05:20] davecheney: remember to Publish your review request so that others can see it [05:21] nup [05:21] won't let me login [05:21] is it really oauth: [05:21] or is that a placeholder [05:21] placeholder [05:22] so for you it's "oauth:davecheney" I suspect [05:22] nup [05:22] no joy [05:22] davecheney: what happens? [05:22] ==> HTTP Authentication Required [05:22] Enter authorization information for "Web API" at reviews.vapour.ws [05:22] Username: davecheney [05:22] Password: [05:22] ERROR: Error creating review request: The username or password was not correct (HTTP 401, API Error 104) [05:23] davecheney: I think the account has to get created first... visit http://reviews.vapour.ws, go to the login page (top right) and click the oauth button [05:24] i have an account [05:24] i have logged in [05:24] ok [05:24] davecheney: and is your RB account "davecheney"? [05:24] can you just take a quick look at https://github.com/juju/juju/pull/765 [05:24] yes [05:25] it was crated from my gh oauth creds [05:25] davecheney: then IHNFI :) [05:25] davecheney: will review on GH [05:26] thanks [05:31] davecheney: I don't like the duplicated function as I imagine you don't either [05:31] davecheney: how temporary will it be? [05:32] menn0: i can remove it as soon as state stops depending on apiserver/params [05:32] given the shifting nature of our priorities it's hard to give an eta [05:32] but it is the goal of this piece of work [05:32] it is not tangental [05:32] davecheney: but Real Soon Now? :) [05:32] davecheney: understood [05:32] re of our priorities it's hard to give an eta [05:32] err [05:32] as soon as state stops depending on apiserver/params [05:33] davecheney: you have my LGTM [05:33] davecheney: but I'm a junior developer [05:33] well junior reviewer [05:34] i'm sure thumper would have given it the thumbs up if he were here [05:34] but this is good [05:34] this leaves the last problem [05:34] the watchers depend on apiserver/params [05:34] * davecheney insert jarring chord [05:37] * thumper is here [05:37] * thumper looks [05:40] * thumper is here [05:41] * davecheney insert dramatic crescendo [05:45] WTF does the "review" button on a review do ? [05:45] it sure doesn't show you the code [05:55] wallyworld: the least horrible thing I can think of right now is to add a method, Environ.SetPrivateMetadataSource(DataSource) [05:56] the machine agent will call that with a data source backed by environment storage [05:56] when bootstrapping with --metadata-source, I'll create a local filesystem datasource and call it with that [06:01] * thumper is down to about 150 unread emails [06:01] got through over a thousand today [06:01] * thumper is done for today [06:01] night all [06:01] night [06:02] menn0: I'll be out part of the morning tomorrow as I get an ultrasound and xray on my shoulder [06:02] menn0: should be back before the standup [06:02] * thumper crosses fingers hoping no surgery needed [06:30] axw: i think that could work === uru_ is now known as urulama === BjornT_ is now known as BjornT [09:33] rogpeppe1: hey was just replied to you on the mailing list :) [09:34] eagles0513875: ta! [09:34] fwereade: POKE :P [09:34] rogpeppe1: im asking in the git channel but i have a hunch you do not need to rebase if you squash multiple commits together [09:35] eagles0513875: i think rebase -i is the way to do that [09:35] let me do some further research though [09:35] eagles0513875: at least, that's the way i've been doing it (it lets you choose which commit messages to keep, etc) [09:37] fwereade: im in here as well hehe :) [09:38] fwereade: out of curiosity here is an appropriate question to ask as i would love to start testing and helping out with juju on a non ubuntu specific distro [09:38] eager to learn golang for sure :) [09:39] eagles0513875, gsamfira has been leading the windows workloads, and I think that the extension of that to other OSs is on his roadmap? [09:39] eagles0513875, he'd be best able to say how to support him [09:39] gsamfira: ^ ping :D [09:39] hey there :) [09:40] gsamfira: im very eager to help you with testing and development of juju for non specific linux distros [09:40] eagles0513875: sure thing. What did you have in mind? [09:40] gsamfira: im using gentoo these days and I am more then willing to help you out with testing debugging etc and coding [09:41] eagles0513875 : I have personally never used gentoo, but I can write up a document of things that need attention when adding a new platform if you would be interested [09:41] if you dont mind that would be awesome [09:41] gui devs, I need a review for https://github.com/juju/juju/pull/768 (megawatcher), anyone available? thanks [09:41] oops, juju devs ^^^ [09:41] eagles0513875: I have to wrap something up, and it would be my pleasure. [09:41] also what about cloud providers as i would love to add the cloud provider i uise [09:43] eagles0513875: that is a topic for someone else I believe. But if you have a look inside github.com/juju/juju/providers, you a few providers as example :) [09:43] eagles0513875: If you plan on starting development for Juju, best place to start is: https://github.com/juju/juju/tree/master/doc [09:44] :D ok [09:45] im starting to find the work flow a bit confusing [09:45] though from what im seeing via emails [09:47] fwereade: who would i need to talk to about adding another cloud providr [09:47] provider [09:48] eagles0513875, if you think you've got the time, come round to mine for a day with your provider credentials and I'll see if I can get you pointed in the right direction [09:48] i have them already [09:48] would you need access to their API? [09:49] eagles0513875, (the first step is to make sure you can get juju building and running the tests ofc) [09:49] agreed [09:49] eagles0513875, well, having a golang library for interacting with their API would be a great first step [09:49] ok would need to contact them about that [09:50] eagles0513875, I expect that we'd have to write that tbh [09:50] fwereade: ? what do you mean [09:50] an golang wrapper? [09:50] eagles0513875, yeah -- unless one already exists that you know of? [09:51] nope none that i know of :) [09:51] need to find out if golang is available on gentoo [09:51] hell would be great to get an ebuild setup for juju :) [09:51] eagles0513875, ah looks like there are a couple on github [09:51] couple of what? [09:51] eagles0513875, no guarantee they're any good [09:51] eagles0513875, linode api libs [09:51] i know they exist [09:51] eagles0513875, (that's what you're using iirc?) [09:52] yes [09:52] thing is [09:52] to use the api you have to have an account with them [09:52] especially if we want to target clients etc [09:52] to make it easier to deploy services etc [09:54] fwereade: ^ not sure if that would cause new problems [09:54] eagles0513875, sure, that applies pretty much across the board [09:55] eagles0513875, configure the provider with your API keys and go [09:55] exactly so we coudl ask the users to enter their api key etc [09:57] eagles0513875, yeah -- that'd just be defining a Config that accepted the api keys [09:57] dimitern, simple one for you http://reviews.vapour.ws/r/31/ [09:58] mattyw, sure, will have a look in a moment [09:58] dimitern, thanks [09:59] fwereade: out of curiosity canonical wouldnt host a git repository for you guys and gerrit code review site either? [09:59] mattyw, oh, boy! my first review at RB :) [09:59] dimitern: from what ive been reading its making things a hell of alot more complex [09:59] dimitern, I still find it a bit clunky [09:59] but so was rietveld when I first started using that [09:59] eagles0513875, the --amend option? [10:00] dimitern: RB [10:00] eagles0513875, we want to be on github because that's where the developers are -- and we have our own reviewboard up now [10:00] mattyw, it's a bit UX-unfriendly, but it takes some time to grow on you I guess [10:00] fwereade: gotcha i dunno as i dont develop for juju yet but i find its rather a complex setup [10:01] for instance fwereade with a gerrit git setup [10:01] when you push your changes you push to gerrit which adds them to the list of things to be reviewed [10:01] then you can lets say have x number of reviewers and last reviewer would push to the repo [10:01] for example libreoffice use it [10:01] eagles0513875, FWIW, if it sanely supports patch series (i.e. master -> branchA -> branchB -> branchC) it's totally worth it for better productivity [10:01] dimitern: gerrit does [10:02] for example libreoffice require 3 reviews for anything going into stable [10:02] and it keeps track of who reviewed what etc [10:02] stable branches need two reviews and master needs 1 or 2 reviews for them [10:02] eagles0513875, have you read the workflow for dependent branches in gerrit? it's not super-nice [10:03] fwereade: sadly havent looked into it in detail [10:03] eagles0513875, and... you seem to be under the impression that we haven't been doing code review? [10:03] but im sure though gerrit can be adapted for your needs [10:03] fwereade, dimitern: time for a quick review? https://github.com/juju/juju/pull/768 [10:03] mattyw, reviewed [10:03] frankban, looking [10:03] fwereade: i knwo you guys do code review i never said you didnt :) [10:03] dimitern: thanks! [10:03] eagles0513875, we looked into it and decided we preferred reviewboard :) [10:03] :) at least it was concidered [10:04] anyway sorry for derailing the channel [10:05] fwereade: gsamfira hopefully in a few days time ill finish off this website and can start with helping you guys out [10:06] ericsnow, ping? [10:07] dimitern: I used subordinate in unit info for symmetry, and we don't need the principal name [10:08] frankban, ah, sorry, I meant Principal bool, not Principal string [10:09] dimitern: so Principal == !Subordinate? [10:09] frankban, yeah, it kinda feels more appropriate [10:10] frankban, more like UnitInfo.Principal: u.Principal != "" [10:10] dimitern: on the other hand, from clients perspective, we have two opposite flags to parse for the same concept [10:11] dimitern: isn't it UnitInfo.Principal: u.Principal == ""? [10:11] frankban, it's not the same concept actually [10:12] frankban, right, sorry, it is [10:12] dimitern: so, sticking with Subordinate makes sense to me [10:12] frankban, being a subordinate comes from the unit's service's charm having subordinate: true [10:13] dimitern: yes [10:13] fwereade: before I complain to the list, can I run something past you? [10:13] frankban, it's not a property of the unit itself, units only know if they have a principal or not, but strictly speaking I think you should check the ServiceInfo to determine whether the unit is subordinate [10:15] dimitern: yes, it's a denormalization, so that we can immediately know if a unit belongs to a service deployed from a subordinate charm [10:15] dimitern: it's just like UnitInfo.CharmURL: in theory the charm URL is an attribute of the service [10:15] jam, ofc [10:16] frankban, fair point [10:16] frankban, ok, objection withdrawn, go with UnitInfo.Subordinate then [10:16] dimitern: cool [10:16] dimitern: just another question [10:16] dimitern: in what cases do we need to increase the API facade version? only backward incompatible changes? [10:16] fwereade: so, apiserver/params/params.go imports gopkg.in/juju/charms.v3 [10:17] because stuff like ResolveCharm takes a "charm.Reference" object [10:17] which *actually* takes a 'string' [10:17] because it has a custom UnmarshalJSON type [10:17] but when trying to call it from python code, I was thoroughly confused [10:17] fwereade: so for a starting point, shouldn't all of apiserver/params be defined in basic types? [10:17] fwereade: the next thing was trying to Deploy [10:17] jam, right, that feels like the same problem as storing charm.Meta in the database [10:18] but you can't deploy "cs:ubuntu" or "cs:trusty/ubuntu" [10:18] you have to deploy "cs:trusty/ubuntu-#" [10:18] frankban, yes, but I need to know more details on the change you have in mind [10:18] but you have to go talk to the charm store to figure out what # to put in there. [10:18] jam, yes -- didn't that come up on the list semi-recently? it's *not* meant to be a charm.Reference at all, it's meant to be a charm.URL that gets deployed [10:19] dimitern: I was thinking about the change you just reviewed: it's not backward incompatible, so I guess we are ok [10:19] dimitern: just double checking [10:19] fwereade: sure, but it is *really* inconvenient from a "lets do this from Python" POV that to talk to our API you have to debug and figure out yet-another API [10:19] fwereade: can i make a suggestion for reviews [10:20] fwereade: and going to a page like: http://manage.jujucharms.com/charms/precise/ubuntu I can't actually tell what revision it is [10:20] jam, and -- yes, it should probably really just be a string over the API [10:20] jam, with the requirement that that string be a valid stringification of a charm url [10:21] fwereade: sure, it turns out that it is just a string, but it is decoded magically for the API rather than the API server taking a string and doing the decoding. [10:21] jam, I agree that that is a Bad Thing [10:23] fwereade: just would like to suggest with RB is it possible to have multiple people review something lets say 3 before approving and pushing [10:23] eagles0513875, it would be [10:24] eagles0513875, our experience is that that helps less than you might imagine [10:24] frankban, adding an optional field to a result should be ok, as long nobody depends on it [10:24] dimitern: yeah, that's what I suspected. cool then, and thanks for the review! [10:25] frankban, backwards-incompatible change will be if you require Subordinate to be present and give errors if not [10:25] frankban, no worries [10:25] dimitern: I may be late for standup, taking dog out [10:26] jam, ok [10:30] jam, ok [10:30] ha.. so hitting enter on a blank line in xchat resends the last msg [10:41] dimitern: no it doesn't [10:41] dimitern: you must have hit the up arrow [10:46] fwereade: thats news to me. wouldnt that improve code quality and start discussion about fixes etc if its the right way etc [10:47] voidspace, I must have then :) [10:48] dimitern: I take it we're waiting for jam for the standup [10:48] dimitern: I have a question about dialling mongo in the meantime [10:48] dimitern: I don't think it's possible for us to establish connections to all mongos in a cluster [10:48] voidspace, it seems I'm the only one in the standup call [10:48] dimitern: from just a session [10:48] dimitern: I was there a minute ago [10:49] voidspace: dimiter and I are here, where are you and TheMue [10:49] ah, ok [10:49] * fwereade lunch [10:52] jam: dimitern: I don't think you guys can hear [10:52] jam: dimitern: *hear me [10:52] voidspace: we cannot [10:53] jam: dimitern: leaving and returning [10:53] voidspace, we're hearing you typing [10:54] TheMue: poke? === fabrice is now known as fabrice|lunch [11:19] morning [11:19] perrito666, morning [11:41] how do I get from the charm/number format to the tag name of a unit? === fabrice|lunch is now known as fabrice [11:42] uh, I think I found it [11:43] perrito666: unit-charm-number [11:44] jam: I was looking for a method, apparently NewUnitTag would return that [11:44] perrito666: though be careful, as NewUnitTag (IIRC) will Panic if you don't have a valid content [11:44] and IIRC davecheney said "this is for tests only, you shouldn't be using this" [11:44] oh [11:44] ok [11:44] perrito666: so *maybe* you can use it, but you *must* validate the input first [11:45] jam: its is odd we dont have soemthing like that that does not panic (I was wondering why the panic) [11:46] * perrito666 try to recover his head side movement by ibuprofen-ing it [11:54] perrito666: i am right there with you. apparently i slept funny. [11:56] katco: I went hiking on sunday and there was this strong cold wind (not feeling the face level) and I am still suffering random muscle failures [11:58] perrito666: well your reason sounds much more healthy. [11:58] perrito666: all i got is "i slept weird" [11:58] katco: sleeping is good [11:59] I am sure its much better than trying to be asymptotic to hypothermia [12:00] perrito666: at least with that you sleep well :) [12:01] yup, apparently hiking was good on my memory only because last time I did it I was like 20kg lighter and 5 years younger [12:06] perrito666: I went jogging today after about a 6 month hiatus.... yeah, things aren't what they used to be [12:08] it gets better I decided to take some form of morning exercise [12:08] since I have lost my walk to the office in the morning exercise [12:08] well I havent but 5mts is not very demanding [12:09] I wish I could go jogging [12:21] perrito666: well, we got a dog 6mo ago, and I was spending my jogging time walking her, but today I was reminded that walking the dog != jogging [12:21] heh if your dog is athletic enough you can jog together, takes some training for the dog to be able to though [12:23] perrito666: yeah, I was trying to get there, but it is hard to control a dog and get them to jog for a while. My dog would jog for about 30s-1min. But I'm trying to jog more like 20+min. [12:24] jam: give the dog a reward when properly jogging and a yell when not, that worked with mine [12:25] yet my dog is a border collie and I cannot have her running at a time that suits me or she will faint bc of the temperature [12:25] perrito666: yeah, it has been difficult here as well [12:26] it has just started to cool down, which is why I went jogging today [12:26] here winter has become more of a technical term, we get like a month of cold a year [12:27] with luck [12:27] which reminds me [12:27] * perrito666 shops for AC for his dig [12:27] dog* [12:27] any ocr in da house? http://reviews.vapour.ws/r/18/ [12:28] my poor little pr has been feeling very unloved [12:28] wallyworld: o/ [12:28] hey :-) [12:28] * perrito666 reviews [12:28] the other one is dimitern [12:28] ty :-) it's mainly mechanical [12:29] test data unchanged === Ursinha is now known as Ursinha-afk [12:47] wallyworld: have a bit of patience I am trying to figure how to add a MustBeValidatedByDave(LGTM) to reviewboard [12:48] perrito666: no problem :-) thank you for looking [12:48] maybe just add a comment and i can ping dave tomorrow [12:57] wallyworld: I believe I did it :p [12:57] I sent an email to dave just in case [12:57] \o/ thank you :-) [12:57]