=== gary_poster is now known as gary_poster|away [01:15] thumper, wallyworld_: I think the authorisation worker should be enabled for null [01:15] the ubuntu user is there, after all [06:40] morning all! [08:44] dimitern: yo! [08:44] dimitern: did you have a good one? [08:44] rogpeppe, hey [08:44] rogpeppe, oh yeah :) and much needed at that [08:44] dimitern: cool [08:45] rogpeppe, yourself? [08:45] dimitern: yeah, good thanks [08:45] rogpeppe, so what's cookin' ? i've been catching up on emails [08:46] dimitern: same old same old, really [08:46] dimitern: HA is coming along [08:46] rogpeppe, sweet! [08:46] dimitern: there was an interesting discussion about charms vs default series on the juju mailing list yesterday [08:46] dimitern: you might want to contribute [08:46] rogpeppe, yeah I read through it [09:04] rogpeppe, weekly meeting is in 1 hour, right? [09:04] dimitern: i think so [09:04] dimitern: yeah [09:07] rogpeppe, cheers [09:11] morning everybody [09:11] fwereade, morning! [09:12] * fwereade is having a happy fun water-main-leaking-into-electrics day [09:12] dimitern, happy new year :) [09:12] fwereade, to you too :) [09:13] fwereade, that sounds dangerous [09:13] dimitern, just got dominion hinterlands, looks interesting, haven't played it yet [09:13] dimitern, indeed [09:13] fwereade, sweet! [09:13] dimitern, well the circuit breaker works at least [09:13] fwereade, so in addition to leaky drains now the mains are causing issues [09:13] dimitern, and we're ok for the moment because the water is off [09:14] dimitern, well, we fixed the drains and they seem ok now [09:14] dimitern, and hopefully it's just the one pipe [09:14] dimitern, but, well, we shall see [09:14] fwereade, *fingers crossed* [09:14] cheers :) === _thumper_ is now known as thumper [09:59] standup time [09:59] https://plus.google.com/hangouts/_/calendar/bWFyay5yYW1tLWNocmlzdGVuc2VuQGNhbm9uaWNhbC5jb20.sbtpoheo4q7i7atbvk9gtnb3cc [09:59] welcome back dimitern [09:59] thanks jam [10:00] mgz: poke ^^ [10:14] I can't get back onto the hangout, Internet access is very flaky [10:15] axw: no problem, we'll just assign all the work to you :) [10:15] hah [10:15] I got on :) [10:17] axw: did you get the !dead stuff sorted out? [10:17] I managed to land the other changes for add-machine compatibility, but that is now the blocking point [10:18] jam: yes, there's a CL up [10:18] jam: it was pretty trivial, and there's a test now [10:19] jam: https://codereview.appspot.com/48670045/ [10:23] axw: why are you passing "Dying" to the $unset? [10:24] jam: the value is arbitrary, and ignored by $unset (but required, AFAIK) [10:24] axw: its a JSON object so it does need a value in the key: value context. [10:24] right, that makes sense [10:25] I suppose it could've been nil [10:29] axw: I'd prefer it be not something that looks like it might be a value, so "" or nil is fine [10:29] Dying made me wonder if it was only unsetting it if it was Dying, or etc [10:29] jam: no worries, will update before landing [10:29] yep, fair point [10:56] gah, dammit [10:57] mgz: well there you are [10:59] rogpeppe: btw, the pastebin in the meeting - most of those are git failures. I was going to fix them this morning, since they obscure real failures for me [11:00] natefinch: cool, thanks [12:58] anyone fancy helping me with an odd juju/gccgo build issue? - https://launchpadlibrarian.net/162007788/buildlog_ubuntu-trusty-i386.juju-core_1.17.0-0ubuntu1~gccgo4_FAILEDTOBUILD.txt.gz [12:58] can't reproduce this outside of a launchpad builder which is awkward [13:00] I shall look at it, and not be helpful am sure === gary_poster|away is now known as gary_poster [13:03] src/launchpad.net/juju-core/charm/config.go:14:32: error: $WORK/launchpad.net/juju-core/schema exists but does not contain any Go export data [13:03] jamespage: src/launchpad.net/goose/http/client.go:12:21: error: $WORK/launchpad.net/goose exists but does not contain any Go export data "launchpad.net/goose" [13:03] sounds like you have a stale build file [13:03] ? [13:03] several modules [13:03] jam: maybe - but the errors jump around [13:03] so I suspect something toolish [13:05] does feel ording related or something, if you diff with a local build log that worked, are the modules coming up in the same sequence? [13:07] mgz, broadley yes - I've tried a few things such as tweaking the -p flag - I think in a launchpad builder it default to -p 1 (single cpu) [13:08] you might be at the email-dave point then [13:08] mgz, yeah - that's what I guess - I already pinged him [13:09] mgz, trying to pull together a package that supplies both gccgo and golang-go built binaries [13:09] well I have it locally === BradCrittenden is now known as bac [13:18] rogpeppe: have you gotten a chance to look at the additions I made to replicaset? https://codereview.appspot.com/49290043 [13:19] natefinch: oops, perhaps i haven't published my comments [13:19] natefinch: one mo [13:19] rogpeppe: cool [13:23] natefinch: reviewed [14:37] * rogpeppe goes for lunch [16:00] arg, this kills me: // The Attempt and AttemptStrategy types are copied from those in launchpad.net/goamz/aws [16:43] mgz: there's a bit of confusion between hostname:port and hostname only in our code. i wonder if we should consistently name the former "Address" (fitting in with the "address" as passed to net.Dial) and the latter "HostName" [16:45] mgz: as it is, we've got net.Dial and now replicaset.Member.Address that hold a hostname:port combi, but instance.Address and state.Machine.PrivateAddress and others which hold host names only [16:45] rogpeppe: that sounds sensible [16:46] well, isn't there a better term... [16:46] * mgz goes to look at rfcs [16:46] mgz: quite possibly [16:47] mgz: net.Dial calls it "host" [16:47] 'authority' would be correct for the combo [16:47] [ userinfo "@" ] host [ ":" port ] [16:47] mgz: maybe somehow "instance.HostName" seems a better fit than "instance.Host" to me though [16:48] mgz: i think we should go with the Go convention of "address" for host:port [16:49] mgz: it's used all over the place, and we can't change that in the Go tree [16:49] host is ipvX / ipv6 / ipv4 / reg-name [16:49] mgz: alternatively we could use "HostPort" [16:49] mgz: yeah [16:50] so, "registered name" where we explicitly mean not a ip address [16:50] I quite like hostport [16:51] mgz: hostport is also used in go stdlib too, e.g. http://golang.org/pkg/net/#JoinHostPort [16:51] (looking at rfc 3986 btw, there have been several url ones that somewhat update each time) [16:51] mgz: but it's not so widespread [16:52] mgz: maybe we *should* consistently use Host and HostPort throughout [16:52] mgz: it mirrors the string concatenation nicely [16:53] that sounds like a plan [16:53] mgz: and instance.Host could grow on me, i think [16:53] mgz: thanks [16:53] mgz: i'll file a bug, and conform to that convention in new code [16:56] https://bugs.launchpad.net/juju-core/+bug/1267541 [16:56] <_mup_> Bug #1267541: The name "Address" is used inconsistently [16:58] mgz: the only problem with "host" is that it sounds like you're talking about the signified object, not the signifier. so "SelectInternalHost" sounds like it's looking for a particular host that happens to be internal, rather than looking for a host address that can be used internally. [17:00] right, works less well compounded, hostport likewise [17:02] perhaps hostName (hostname?) and hostPort might be better [17:48] dstroppa: you have review [17:49] er... if codereview is actually up atm [17:50] okay, now it's back and the post went through [17:52] mgz: got that, thanks [19:32] arosales, cmars, mgz, mramm2, I apologize for my absence this afternoon, the whole island had a power cut :/ [19:32] fwereade: no problem! [19:33] randomly, friend of mine just messaged me a malta related video: https://www.youtube.com/watch?v=LXfVlCuWG6g [19:34] fwereade, no worries, thanks for the fyi [19:34] mramm2, haha, I recognise that one :) [19:35] fwereade, no prob, we'll resched [19:36] cmars, most of tomorrow is free for me I think [19:37] (a 132kV something-or-other tripped at the delimara power station... 132kV seems like quite enough voltage for me to forget the other details) [19:48] 'ello [19:48] * thumper looks at the modest proposal email thread and gives it a miss [19:59] * thumper looks at the juju run work and tries to split it up into three chunks for review [19:59] also realised I missed some tests... [20:00] although in the last pipe [20:00] * thumper gets to work [20:00] * thumper things of music, perhaps dubstep [20:13] thumper, ok, I am just going to use the vagrant image for everything [20:14] :) [20:14] I don't understand why juju needs to put stuff all over the disk [20:14] the minute something goes wrong the entire installation is worthless [20:18] jcastro: there are reasons... [20:18] perhaps we can discuss over a beer sometime [20:18] sure [20:19] maybe we should just tell people to always use vagrant? [20:19] I've had to reinstall twice this cycle already. :-/ [20:19] ?! [20:19] really? [20:19] that seems real screwy [20:19] well, you couldn't fix it the first time [20:48] thumper: what does git --version return for you? I'm trying to figure out some of the tests that fail due to a newer version of git [20:48] natefinch: $ git --version [20:48] git version 1.8.3.2 [20:49] thumper: and I presume uniter tests pass for you? specifically the tests in worker/uniter/charm ? [20:49] natefinch: yep, all passing [20:49] just ran them :) [20:49] thumper: cool [20:49] busy breaking things up for review [20:49] thumper: they're broken with git 1.8.5 for whatever reason... one fix was easy, the other... well, I'm not sure yet [20:50] ack [21:23] thumper: ahh, I see it. "git add -A ." used to be a NOOP if there were no files to add.... now in 1.8.5 it errors out saying "no files match filespec '.'" [21:23] ha [21:28] it bugs me that "apt-get upgrade foo" does not work the way "apt-get install foo" or "apt-get remove foo" works. Inconsistent UI is aggravating [21:34] :) [21:35] you have seen git koans right? [21:46] thumper: sounds familiar [21:49] * thumper needs another coffee, and some food [22:04] thumper: how are you on the networking stuff? [22:05] looks like MaaS folks have scheduled an emergency huddle in 25 min [22:05] mramm2: not too hot, but I'm good at bluffing [22:05] o/ mramm2 [22:05] ;) [22:05] bugger, games up [22:05] I talked to thumper before xmas [22:05] you were useless :) [22:06] kapil is the best to talk to [22:06] he wrote most of the existing network sketch [22:06] I'll come along to throw comments from the sideline [22:06] but... he is in london at a sprint this week [22:06] not sure how much help I'll be [22:06] yeah he was a "tentative" on the invite [22:06] * bigjools in another call until :30 [22:12] thumper: bigjools: I have a candidate who would like to talk to me in 30 min [22:12] bwahaha [22:12] do you think I can duck out after the first few min of our call? [22:45] thanks mramm2