[00:05] <hatch> huwshimi so do you have enough work for today without any of the scale up stuff?
[00:05] <huwshimi> hatch: Yep, I think so.
[00:05] <hatch> awesome
[04:57] <urulama> huwshimi: morning
[04:57] <rogpeppe> urulama, huwshimi: hiya
[04:58] <urulama> rogpeppe: hehe, early bird
[04:58] <rogpeppe> urulama: :-) happens sometimes
[04:58] <rogpeppe> urulama: :-) happens sometimes
[05:04] <huwshimi> rogpeppe, urulama: Morning :)
[05:41] <urulama> wow, hp helion registration process is, well, could be easier :D
=== benonsoftware is now known as Guest69253
=== urulama is now known as uru-afk
=== uru-afk is now known as urulama
[08:01] <urulama> morning frankban
[08:02] <frankban> hi urulama 
[08:43] <urulama> is it possible to get more info then: "ERROR couldn't read the environment"?
[08:44] <urulama> been editing env.yaml for hpcloud and now whatever i do, even remove all entries in hpcloud: section still gives me this error
[08:48] <urulama> found error
[08:49] <urulama> so, parsing of env.yaml fails if there is an empty line with a tab (\t) at the beginning. this is probably known, right
[09:15] <rogpeppe> urulama: was there no more descriptive error than "couldn't read the environment". That seems like a bug - it should at least say that it found an illegal character.
[09:16] <rogpeppe> urulama: ah yes, that does look wrong
[09:16] <rogpeppe> 	environments, err := environs.ReadEnvirons("")
[09:16] <rogpeppe> 	if err != nil {
[09:16] <rogpeppe> 		return errors.New("couldn't read the environment")
[09:16] <rogpeppe> 	}
[09:17] <rogpeppe> it should be: return fmt.Errorf("couldn't read the environment: %v", err)
[09:17] <urulama> rogpeppe: i'll file a bug, it should state that tabs are not allowed and where the error happened (as you've written)
[09:18] <rogpeppe> urulama: i suggest proposing a fix that does what i suggested
[09:19] <rogpeppe> urulama: with that change, the error becomes:
[09:19] <rogpeppe> ERROR couldn't read the environment: cannot parse "/home/rog/.juju/environments.yaml": YAML error: found character that cannot start any token
[09:20] <rogpeppe> which i think you'll agree is a little better, even if it doesn't mention the line number or actual character that triggered the problem
[09:20] <urulama> rogpeppe: that's a bit better, agree
[09:21] <rogpeppe> frankban: you up for continuing on the API spec?
[09:22] <frankban> rogpeppe: yes, I'll be ready in a few minutes
[09:34] <frankban> rogpeppe: I am in the hangout
=== urulama is now known as uru-afk
=== uru-afk is now known as urulama
[11:37] <bac> morning frankban and rogpeppe. i am unexpectedly here today.
[11:37] <rogpeppe> bac: unexpected welcome, then!
[11:37] <frankban> bac: morning
[11:37] <bac> thank you
[11:40] <bac> frankban: azure change to qs looks good.  i'll do qa now.
[11:41] <frankban> bac: thanks, if you updated to juju-core 1.20 you will likely find that the new juju breaks quickstart
[11:42] <bac> frankban: i have not yet, i don't think.  how does it break? is there a card for that?
[11:42] <bac> frankban: i have 1.18.1.  i can qa with that
[11:43] <frankban> bac: cool, please QA with 1.18, I'll create a card and look at it later
[11:43] <bac> frankban: i can look at the card if you like
[11:43] <frankban> bac: there is a backward incompatible change in the mega-watcher for machines
[11:59] <frankban> bac: I am going to grab some food now. After your QA, could you please try to dupe the bug using 1.20? You should be able to do that by just trying "juju quickstart" in a local env. You should see a KeyError: u'NetworkScope' while juju is provisioning a machine for the GUI. If you are able to confirm the problem, AFAICT we should involve rbasak because this regression affects the distro quickstart.
[11:59] <bac> frankban: my first QA attempt with your azure branch has failed.  will try again.  does not look to be related.
[11:59] <frankban> bac: ok thanks
[12:01] <urulama> frankban: 1.19.4 has the same NetworkScope error?
[12:01] <urulama> frankban: with juju-quickstart on local env
[12:03] <rogpeppe> bac: the API proposal has changed quite a bit since yesterday, if you want to take another look
[12:04] <bac> rogpeppe: thanks, i will in a bit
[12:22] <frankban> urulama: did not try juju dev, if you have 1.19 installed you can easily dupe
[12:22] <frankban> bac: what error are you encountering?
[12:24] <bac> frankban: second QA worked
[12:24] <frankban> bac: cool
[12:24] <bac> frankban: first time there was a timeout trying to ssh to machine 0
[12:24] <bac> frankban: more likely azure weirdness or my local connection
[12:25] <frankban> bac: ok, this seems unrelated to quickstart
[12:25] <frankban> bac: yeah
[12:25] <bac> yes
[12:25] <bac> frankban: i marked it qa-ok
[12:25] <urulama> frankban: i mean, i do get the same error on 1.19.4
[12:25] <bac> well, i mean to.  i've actually pressed the button now
[12:26] <frankban> urulama: cool thanks for confirming
[12:28] <frankban> bac: thanks for the review, merging it. Working on the new card. I am starting to think we should really prioritize quickstart CI with current stable and development versions of juju. This would also help preventing other accidental public API changes like this
[12:30] <bac> frankban: prob a good idea
[12:30] <bac> frankban: prevented the change or just given us early warning?
[12:31] <bac> frankban: i mean was their api change really accidental and we just spotted or was it intentional and we didn't know it was coming?
[12:36] <frankban> bac: IIRC we consider API v0 to be the one in 1.18. I assume a change to the public API to be accidental. A quickstart test in these cases would not prevent the change to land, but would surely prevent it to be released
[12:36] <frankban> bac: filed 1337831
[12:36] <bac> frankban: right
[12:36] <frankban> bug 1337831
[12:36] <_mup_> Bug #1337831: Quickstart crashes when used with juju 1.20 <juju-quickstart:In Progress by frankban> <https://launchpad.net/bugs/1337831>
[12:37] <frankban> rbasak: for when you are available: ^^^
[12:37] <bac> frankban: are we the only consumers of that API?  i mean, should we fix QS to adapt or will there be larger fallout meaning juju-core should revert?
[12:39] <frankban> bac: I am quite sure the GUI does not get unit addresses from MachineInfo yet, so it should not be affected. Not sure about other consumers. The problem is that we now have this released. Maybe we should raise the problem with juju devs
[12:40] <bac> frankban: yeah, i'll bet william would be interested.  curtis too, but he's prob not here today.
[12:43] <frankban> bac: pinged him in #juju-dev
[12:49] <rbasak> frankban: thanks. After it happened the first time I'm making sure to test both juju-core and juju-quickstart together before uploading, and this is an MRE requirement for Trusty now, too.
[12:49] <rbasak> frankban: so I'll make sure that juju-core doesn't hit the archive without juju-quickstart also working with it.
[12:50] <rbasak> frankban: I guess I need a fix for juju-quickstart first, then, though I don't know that I'll get to doing it today anyway, so no panic for you guys.
[12:51] <frankban> rbasak: cool thanks, I am working on the quickstart fix right now
[12:53] <bac> frankban: i can reproduce the bug with 1.20, fwiw
[12:54] <frankban> bac: thank you
[12:55] <bac> frankban: should we plan on a qs release today?
[12:56] <frankban> bac: it would be great, 1.4.1
[12:56] <bac> frankban: i'll make a card
[12:56] <frankban> thanks
[12:59] <rbasak> frankban: I'm interested to know whether the fix will be backwards compatible with older Juju. I'm not sure it matters to be either way, but will probably be useful for me to know.
[12:59] <frankban> rbasak: it will be backward compatible
[13:01] <rbasak> Cool!
[13:24] <frankban> bac: could you please take a look at https://codereview.appspot.com/111860043 ?
[13:26] <bac> frankban: sure
[13:26] <frankban> thanks
[13:29] <frankban> rogpeppe: ready when you are back
=== urulama is now known as uru-food
[13:47] <bac> frankban: done
[13:48] <frankban> bac: thanks
[13:55] <bac> frankban: if you merge your 1.20 fix i'll do the release
[13:55] <frankban> bac: merged
[13:56] <bac> ty
[13:58] <frankban> rbasak: I merged the fix for bug 1337831. Now we are working to make a new 1.4.1 release
[13:58] <_mup_> Bug #1337831: Quickstart crashes when used with juju 1.20 <juju-quickstart:Fix Committed by frankban> <https://launchpad.net/bugs/1337831>
[14:10] <rbasak> frankban: great, thanks!
=== uru-food is now known as urulama
[14:39] <hatch> man testing UI interactions is such a pita
[14:50] <hatch> jujugui call in 10
[14:57] <hatch> urulama jrwren friday's use a different room than the rest of the week so you'll want to click the link in the calendar for the call today
[14:58] <hatch> jujugui call in 2
[14:59] <urulama> hatch: this one? https://plus.google.com/hangouts/_/calendar/cmljay5oYXJkaW5nQGNhbm9uaWNhbC5jb20.t3m5giuddiv9epub48d9skdaso
[14:59] <bac> urulama: https://plus.google.com/hangouts/_/calendar/cmljay5oYXJkaW5nQGNhbm9uaWNhbC5jb20.t3m5giuddiv9epub48d9skdaso
[15:00] <hatch> rogpeppe you coming?
[15:21] <rogpeppe> frankban: i'm back in the hangout, BTW
[15:21] <hatch> jujugui https://gist.github.com/087d9e30e3eb92a46657 runWhenCalled
=== urulama is now known as uru-away
[15:23] <frankban> rogpeppe: joining
[15:23] <frankban> hatch: looks good
[15:23] <hatch> frankban so simple yet solves so many problems!! heh
[15:23] <frankban> cool
[15:25] <hatch> frankban I updated the gist with usage
[15:27] <hatch> hmm I think I have a better update
[16:28] <frankban> bac: would you like to take another look at the final apiv4 spec before we send it out for discussion?
[16:59] <bac> hi frankban, just back from lunch.  will look now.
[16:59] <frankban> bac: thanks
[17:00] <bac> frankban: i cannot edit nor make comments on that doc
[17:03] <bac> rogpeppe: ^^
[17:05] <frankban> bac: it's set to "People at Canonical who have the link can edit"
[17:05] <frankban> need to go now, have a good weekend all!
[17:07] <bac> sorry frankban, the card is linked to https://docs.google.com/a/canonical.com/document/d/1wkgLrSZRcpz7zv5Sfc-lh21OZaYui-PhPolpedpWlCI/edit
[17:07] <bac> but that's the wrong doc
[17:08] <bac> rogpeppe: is this the correct document for your API proposal? https://docs.google.com/a/canonical.com/document/d/1wkgLrSZRcpz7zv5Sfc-lh21OZaYui-PhPolpedpWlCI/edit
[17:09] <rogpeppe> bac: here: https://docs.google.com/a/canonical.com/document/d/1ILHRpOe-qDlmjxHBbLUea7InDpehx5_roJ1ynZmcZDc/edit
[17:09] <bac> rogpeppe: please fix the link in the kanban card
[17:09] <rogpeppe> ah
[17:24] <bac> rogpeppe: is this a typo: GET meta/color?id=djang
[17:24] <bac> rogpeppe: or are you doing partial match on charm names?
[17:24] <rogpeppe> bac: typo
[17:24] <bac> yay
[17:24] <bac> rogpeppe: fixed
[17:25] <bac> rogpeppe: is meta/color new?
[17:25] <rogpeppe> bac: yes
[17:28] <hatch> oh man this new addition to the test suite is so nice
[17:29] <hatch> updated the gist https://gist.github.com/hatched/087d9e30e3eb92a46657
[17:31] <bac> rogpeppe: the doc looks good.  the examples you added are nice and help clarify things.
[17:34] <rogpeppe> bac: cool, thanks
[17:41] <hatch> so I need to load the css into the test suite to make these tests work, but that breaks many other tests..hmmm
[18:12] <hatch> hmm new addition does not scale well
[18:12] <hatch> need to rethink this
[18:58] <bac> hey hatch
[18:58] <hatch> ahoy
[18:58] <bac> hatch: you have a precise machine available?
[18:59] <hatch> umm....
[18:59] <hatch> one sec
[19:00] <hatch> bac I do
[19:00] <bac> hatch: could you upgrade juju to 1.20 and see if you can bootstrap a local env?
[19:01] <hatch> ok firing it up
[19:02] <hatch> it'll be a bit
[19:02] <hatch> I'll report back
[19:21] <hatch> bac 1.17.7 is the highest I can get for some reason
[19:21] <hatch> you sure they released 1.2 for precise?
[19:22] <bac> hatch: it may be in ppa:juju/stable
[19:22] <bac> hatch: but i discovered i'm hitting this bug: https://bugs.launchpad.net/juju-core/+bug/1316174
[19:22] <_mup_> Bug #1316174: 'precise-updates/cloud-tools' is invalid for APT::Default-Release <juju-core:Triaged> <https://launchpad.net/bugs/1316174>
[19:22] <hatch> ahh
[19:23] <hatch> yeah I see it in the package list
[19:23] <hatch> but it's not installing it
[19:24] <hatch> yeah who knows....
[19:24] <hatch> shall I spend more time on it? We could also spin up an ec2 instance
[19:24] <hatch> I think that still has precise
[19:32] <hatch> bac so you ok, or would you like me to keep on it?
[19:36] <bac> hatch: nm
[19:36] <hatch> ok sounds good
[19:37] <hatch> btw I ended up getting the nvidia drivers working on my MBP - now to figure out if I can use them to turn off the discrete card and just use the iris gpu
[19:37] <bac> hatch: unless you can translate comment #2 from that bug report into something actionable.  i cannot.
[19:38] <hatch> bac, add more package repositories? :)
[19:38] <bac> shouldn't the metapackage juju-local do that?  if not wtf does it do?
[19:38] <hatch> that's what I thought
[19:39] <hatch> tbh I have no idea how their packaging is set up
[19:53] <hatch__> The power just went out here and it's going to be a while before it comes back, so have a good weekend everyone
[21:45] <hatch> yay power