=== wedgwood is now known as wedgwood_away [00:44] davecheney: are you available for a hangout later? [00:46] thumper: i certainly am [00:47] you are 2 ? 3 ? hours ahead of me ? === thumper is now known as thumper-afk === jam1 is now known as jam [04:43] thumper: if you're still around, are you free to chat ? [04:43] davecheney: hey [04:43] davecheney: let me just check on the family... [04:43] if you want, we can try agian tomorrow [04:43] maybe it would make more sense after le' meeting [04:44] probably a better time for your anyway [04:46] davecheney: let me just check on the family...\ [04:47] ugh [04:47] stupid up arrow [04:54] davecheney: yo [04:54] What's up folks [04:54] thumper: Heya, welcome to juju [04:55] hi niemeyer [04:55] thanks [04:55] niemeyer: I have been thinking serioursly about it now since we kinda chatted in miami about it [04:55] then it was more of a joke [04:55] happy to be on board now [04:56] * thumper needs to go and make dinner [04:56] niemeyer: isn't it late for you? [04:56] thumper: Yeah, I figured by then, but when we talked recently I got the impression you were actually jumping in [04:56] thumper: Depends on the point of view [04:56] thumper: It's about to be early too :) [04:56] :) [04:56] I suppose it matters if you have slept yet or not [04:57] niemeyer: I was going to just rotate in for a short while [04:57] but in the end, it became more of a dive [04:57] longer term [04:57] thumper: Very happy to hear that [04:57] it'll be fun [04:57] I've been keeping a file of questions :) [04:57] it is getting quite long now [04:57] some I have found out myself by reading more [04:58] others are still open [04:58] but I have some calls scheduled with jcastro and davecheney now [04:58] thumper: It's a brilliant team to be part of, and it's even better with you in [04:58] thanks for that :-) [04:58] I'm love working with smart people [04:59] makes for interesting times [04:59] s/I'm/I/ [04:59] * thumper steps away from the keyboard to go make dinner === thumper is now known as thumper-cooking [07:46] morning all! [07:46] morning rogpeppe [07:49] morning! [07:49] * TheMue just listens to The Dark Side of the Moon, 40 years of great music ... [07:50] ready for an early meeting? ;) [07:50] TheMue: +1, yeah, oldies-goldies [07:50] dimitern: Early? What shall Mark say? :D [07:51] :) [07:51] it's actually late there [07:51] too late to be early [07:52] dimitern: Hehe, indeed also a way to handle it. [07:52] morning [07:53] davecheney: evening :) [07:53] touche [07:55] davecheney: hiya === thumper-cooking is now known as thumper [07:56] hmm... [07:56] I hope this doesn't get too confusing [07:56] in another channel there is TheMuso [07:59] thumper: Never had any conflicts. [08:00] thumper: In 1997 mue has been enough, but later often other people claimed it. [08:03] https://plus.google.com/hangouts/_/3054ca9e05632bb677f4a136cc956486d9aac90f [08:03] for anyone not here yet [08:04] ta [08:04] TheMue: ^ [08:04] mgz: w7z: https://plus.google.com/hangouts/_/3054ca9e05632bb677f4a136cc956486d9aac90f [08:04] fwereade_: ^ [08:04] hold up, installing pluginz [08:06] rogpeppe: ah, thx, link in calendar doesn't work [08:14] and the hangout plugin segfualted [08:14] joy... [08:14] hmm, can't talk to any google properties at the moment [08:14] anyone else dropped off ? [08:14] davecheney: haha [08:15] davecheney: no, just you [08:15] nope, sorry [08:15] nope, no route to google properties at the moment [08:16] mark's talking about making a kanban board for the command line stuff [08:16] +1 [08:16] 'cos there's a big scrum on that [08:16] any it's still in the design phase, as I understand it [08:17] "if we get more need for coordination, we can try having a scrum-of-scrums meeting" [08:19] davecheney: is it working for you now? [08:20] davecheney: apparently not :-) [08:20] got a few mins [08:20] then dropped out again [08:20] mark was very choppy [08:23] all my google services have eaten a goose egg [08:23] gmail is slow [08:24] can't get to plus anymore [08:25] we're talking about the floating ip problem [08:26] please take notes [08:29] essentially: floating ips are in limited supply and we need some way to work around that [08:34] Is there any kind of framework for writing new Charms? At least with ability to handle templates? [08:35] It could be great to have some kind of dsl for common cases. Something like Chef have. [08:52] Another thing I thought about is custom events, not only joined, changed, broken and departed but any user defined. For now this can be done via relation_ids and relation_set but it seems like workaround for the case. [08:52] Do you have such thing in your plans? [08:56] mgz: do you remember the rt # + [08:56] ? [09:23] thanks everybody for a productive meeting [10:44] Pavel_: a framework for writing charms is a reasonable idea. i think various people have python helper scripts, which could be a start. lots of places to explore here. [12:36] so, lunchtime, biab === wedgwood_away is now known as wedgwood [15:49] interesting, live tests against the "enable password checking" branch fail because the tests try to access an api.Machine, which is denied by our policy rules. === wedgwood is now known as wedgwood_away [16:00] rogpeppe: did you guys know there's a discussion on juju/go on reddit? http://www.reddit.com/r/programming/comments/18atce/juju_canonical_109k_lines_of_go_code/ [16:01] jcastro: nope, hadn't seen that, ta! [16:01] tell the team, the discussion looks interesting enough for you guys to participate in [16:05] is the 32-bit 2.2.3 MongoDb version on the website good for using with Juju? [16:05] everything on http://juju-dist.s3.amazonaws.com/ is 64-bit [16:49] fwereade_: i had to make a slightly more extensive change than i'd expected in response to the live test failure - i implemented the Client type and a skeleton Status method so that live tests could talk to it. would appreciate if you could have another look: https://codereview.appspot.com/7299066 [16:49] * fwereade_ looks [16:52] rogpeppe, LGTM if you write a bug for Client.Status, I don't imagine that's going to be a priority for a couple of days so it's good to have it tracked [16:52] rogpeppe, (just that it's very unfinished) [16:53] fwereade_: ok, i'll make a ticket === wedgwood_away is now known as wedgwood [16:53] fwereade_: done [16:56] fwereade_: there are a couple of unresolved comments i responded to too. https://codereview.appspot.com/7299066/diff/3007/state/api/apiserver.go#newcode117 and https://codereview.appspot.com/7299066/diff/3007/state/api/client.go#newcode31 in particular [16:59] rogpeppe, sorry, missed those; thanks; responded [16:59] fwereade_: thanks [17:07] fwereade_: i've just seen a slightly concerning live test failure: http://paste.ubuntu.com/1639959/ [17:07] fwereade_: the jujutest code looks plausible, but it seems the unit hasn't gone properly. is it doing something wrong? [17:08] fwereade_: (that was in my branch, which i don't *think* would impact that behaviour, but am testing in trunk now) [17:09] rogpeppe, heh, that's that totally crackful bit [17:09] fwereade_: ah, i wondered if that was the case [17:09] rogpeppe, niemeyer's comment is a bit out of date but correct in intent [17:09] fwereade_: how *should* it remove the unit so it can remove the machine? [17:09] fwereade_: the kitchen sink comment? [17:10] fwereade_: ah no, i see [17:10] rogpeppe, I think it should be: (1) destroy unit, wait removed (2) destroy machine, wait removed [17:10] rogpeppe, I can see that that is maybe inconvenient for you though [17:11] rogpeppe, and, hmm, I think there's a bug pointing out that the provisioner doesn't actually remove dead machines [17:11] fwereade_: hmm, there should be convenience methods in juju.Conn to remove units and machines, presumably? [17:12] rogpeppe, huh, can't see it obviously, I'm sure I saw it yesterday [17:12] rogpeppe, yeah, there's DestroyUnits and DestroyMachines IIRC [17:12] fwereade_: presumably it *did* work, otherwise that test would never have passed [17:13] rogpeppe, yeah, but it's also total nonsense -- it tells us nothing about whether anything is working as expected and if anything just causes trouble by doing things that deployed code is expecting to do itself [17:13] fwereade_: really? it seems to verify that something is responding to machine removals by stopping instances, doesn't it? [17:14] rogpeppe, yeah, must have been seeing things, provisioner looks like it should work [17:14] fwereade_: should we just lose that part of the test and assume that the deployer is adequately tested? or perhaps you've got a better way of testing that things are working live? [17:15] rogpeppe, I tend to test live stuff by hand, I have more opportunities to do surprising things and see what happens ;) [17:16] fwereade_: i do too, but i think it's important to have some automated tests too [17:16] rogpeppe, I'd really prefer it if we did have that part of the test doing something sane [17:17] rogpeppe, it's not exactly an edge case, seems perverse to leave it out [17:17] fwereade_: so it would be ok to do the above dance (destroy unit, wait destroyed, destroy machine, wait destroyed) ? [17:17] rogpeppe, I think it should be, yes [17:17] rogpeppe, that's the expected interaction [17:17] fwereade_: ok, i'll add a TODO ticket for that. [17:17] rogpeppe, ok, cool [17:18] need to go to the shops; might not get back to work tonight [17:18] fwereade_: i'm going soon too [17:18] fwereade_: see ya tomorrow, probs [17:39] Yeah! Putting files works fine too. Will only extend it to a table driven test tomorrow. [17:50] yay, all the outstanding api branches now submitted! [18:01] rogpeppe: grats [18:14] good place to stop. g'night all. === arosales1 is now known as arosales [20:18] hi jcastro [20:19] thumper: fire up a G+ [20:19] kk [20:20] jcastro: created [20:43] jcastro: http://docutils.sourceforge.net/docs/user/rst/quickref.html [22:06] thumper: any time you're ready [22:06] hi davecheney [22:07] I'm in the middle of writing an email about constraints, but I have a feeling it will be a while, so lets chat now :) [22:07] nah, it can wait [22:07] you're documenting stuff [22:07] that trumps everything [22:07] ok, after lunch then :) [22:07] kk [22:08] although not so much documenting stuff, but arguing :) [22:08] and trying to get a better understanding [22:08] I'm already getting plans and ideas [22:08] I just hope that then end up sounding rational and useful [22:09] if you could see your way clear to cc'ing juju-dev, that would be awesome [22:09] too much silo'ing atm [22:09] who's on juju-dev mailing list? [22:10] it seems that only the admin can see the subscriber list [22:11] thumper: well, it is all of us devs [22:11] it is public [22:11] gustavo or william is probably the admin [22:11] davecheney: my only concern about taking it public is that the chat started private, and I'm always careful about dragging conversations public without letting the others know first [22:11] however I'll do more of a scan [22:12] and remove anything that could be construed weird [22:12] thumper: understood, see previous head shaking about silo'ing [22:12] mramm: got any objections to me taking the constraints email public? [22:12] davecheney: it was partly for my benefit as I'm still grappling with many concepts [22:12] thumper: understood that email is a poor documentation medium, but it is better than what we have at the moment [22:34] thumper, fwiw I'm happy to have the discussion in public [22:35] fwereade_: awesome, I've cut out most of the initial email and your reply, just leaving relevant bits [22:35] just finishing off the reply now [22:35] thumper, cool, cheers [22:42] * thumper heads out for an early lunch [22:42] davecheney: I'll ping you when I'm back [22:45] thumper, I'm not sure I'll answer that tonight; when you return you might find that some of it has been covered in the recent thread starting at https://lists.ubuntu.com/archives/juju-dev/2013-February/000479.html