[12:09] bac benji frankban gmb call in 1 or 2, depending on how soon people get there [12:13] bac please land your cards in qa landing [12:29] gmb: when you can, please let's talk about the scriptactivity_pkey bug [12:30] frankban, Okay, give me 10 minutes or so and I'll be free. [12:30] gmb: cool thanks [12:31] gmb: https://juju.ubuntu.com/docs/constraints.html [12:31] instance-id is not listed [12:32] er, image-id [12:32] it's not listed too [12:41] bac, gmb, that irc conversation I found was of someone asking for the feature, sorry. It describes why we don't have it yet, and that there's some interest in getting it in the future. :-/ [12:41] Of course, that's from March 12 [12:42] But...not a good sign [12:42] Bottom. [12:43] gmb, default-image-id: in environments just breaks now? [12:43] gary_poster, It's just ignored. [12:43] gary_poster, per the doc bac pointed out [12:43] oic [12:44] gary_poster, Well, actually, not that version of the doc. There's one in the juju tree, though, that says... [12:45] I also see on that doc on the web, "All commands valid before the client upgrade should continue to work as before, as will the EC2 default-image-id and default-instance-type environment settings." [12:45] Also removed "default-instance-type" and "default-ami" from ec 2 environment config, and set bootstrap to provision a machine from juju defaults. Note: appears to actually work as expected.) [12:45] Ah, hang on. [12:45] Maybe default-image-id is not what I'm using... [12:47] Hah. Putting default-image-id in environments.yaml makes juju yell at me. [12:48] yeah, it started yelling a few weeks ago [12:48] yell and then do something, or just yell? [12:48] Just yell. [12:48] * gary_poster thinks talking on #juju sooner rather than later would be wise [12:48] Agreed. [12:49] I'll talk to frankban first though. [12:49] frankban, Shall we hang out? [12:49] gmb, ok [12:49] frankban, I'm in goldenhorde [12:51] oh right, we have calls with me today too [12:52] * gary_poster is slow on the weekday uptake this week. [12:55] bac, you declined our 3:30-4:30 Thursday meeting from now till kingdom come, I think. Was that your intent? [13:16] * gmb -> lunch [13:20] gary_poster: lp:~yellow/subunit/real-time and lp:~yellow/subunit/test-count are ready [13:20] benji cool on it [13:20] * benji start the slack-time task of "get coffee" [13:20] :-) [13:21] benji: join me in goldenhorde after being caffeinated [13:21] bac: will do [13:21] bac, did you mean to decline our Thursday 3:30 calls? [13:22] nope [13:22] cool [13:22] via gcal? [13:22] yeah [13:22] no, it's been on my calendar [13:22] did something just change? [13:22] yeah [13:22] you declined it about an hour ago [13:26] benji, for ~yellow/subunit/real-time, does testrepository then need a change to not look into _buffered_calls? [13:30] frankban, hi [13:30] https://talkgadget.google.com/hangouts/_/extras/canonical.com/goldenhordeoneonone ? [13:31] gary_poster: joining [13:31] cool [13:32] gary_poster: correct; we changed ~/yellow/testrepository/latest so that instead of it's subclass of FilteredTestResult that had these two properties, it just uses subunit.FilteredTestResult [13:33] frankban, #ubuntu-server [13:50] benji, cool. I'll assume I should make a branch/MP for that too unless you stop me. [13:51] gary_poster: +1 it should just be the last revision (or two) from ~yellow/testrepository/latest [13:51] cool [14:21] benji, bac, I have MPs for the real-time subunit branch and the associated testrepository branch. Please doublecheck what I wrote. https://code.launchpad.net/~yellow/subunit/real-time/+merge/103701 & https://code.launchpad.net/~yellow/testrepository/bug988481/+merge/103704 . [14:22] benji, is the other branch really associated with bug 988481, or is it a separate issue? [14:22] * bac looks [14:22] <_mup_> Bug #988481: Testr tests generate an AttributeError when run with latest testtools < https://launchpad.net/bugs/988481 > [14:22] gary_poster: cool, will do [14:22] thanks [14:23] gary_poster: if "other branch" the last revision(s) ~yellow/testrepository/latest, then yes; if fixes that error by depending on the two branches you just made [14:23] benji, no sorry, I meant ~yellow/subunit/test-count [14:23] is that for a separate issue? [14:23] it looks like it [14:24] if so, I think the MP should describe the problem, and I don't know what it is [14:24] other than the fact that it fixes testrepository somehow [14:24] it /might/ be possible to seperate the two [14:25] if you don't include this change, you still get that AttributeError on _buffered_calls? [14:26] right, it moves the testr behavior into subunit, but as you say it might be possible to just move the time (_buffered_calls) fix into subunit and keep the testr FilteredTestResult class with just the test count code and removing the time bits [14:26] * gary_poster should figure out how to set up a venv of the various dependencies [14:26] oh! [14:26] I think it should be possible to keep the time bits and still avoid the AttributeError [14:29] so, this test-count branch *plus* the real-time branch together make it possible to dump the testrepository version of TestResultFilter. If you only want to do one or the other, you'll have to add testrepository's TestResultFilter subclass back and do surgery of one nature or the other. The branches you are submitting seem like the cleanest approach, which is why you prefer it. [14:29] benji, is that on the right track? ^^ [14:29] they say they don't want lock step releases [14:29] gary_poster: exactly [14:29] ok cool [14:29] thanks, will clarify testrepository MP [14:30] * benji contemplates how to make this non-lockstep [14:30] gmb, https://talkgadget.google.com/hangouts/_/extras/canonical.com/goldenhordeoneonone [14:30] ? [14:30] * benji realizes that the breakage was lockstep [14:31] gary_poster, Yep. [14:31] cool [14:41] gary_poster, Okay, I'm free now :) [14:41] ok gmb, rejoining :-) [14:49] gary_poster: those MPs look good to me [14:57] benji: is the diff in the MP for subunit correct? i thought more changes were made to subunit. [14:57] bac: I think it's correct, he's doing two MPs, one for the time fowarding bits and one for the test count bit [14:58] ok. 2 for subunit and 1 for testrepository then? [15:01] gary_poster: it seems that serge didn't know about the missing sanity check in get_init_pid, see discussion in #ubuntu-server [15:01] frankban, serge is awesome :-) [15:21] benji, thanks for review of MPs. One more simple one: https://code.launchpad.net/~yellow/subunit/test-count/+merge/103717 [15:21] k, looking [15:22] gary_poster: it looks good, but the "We could instead continue to subclass in testrepository" might be better phrased as "We don't know how to continue to subclass in testrepository" [15:23] benji, oh? ok, don't understand why exactly but will go on faith. will adjust. [15:23] frankban approved https://code.launchpad.net/~frankban/launchpad/bug-987442-testSimpleScriptRun/+merge/103706 [15:23] gary_poster: thanks [15:23] I can explain in a hangout in a few minutes if you want. [15:24] cool benji sounds good thanks [15:24] not that important but would be nice to know [15:35] ok, gary_poster, want to chat? [15:36] frankban, am I correct in assuming that the proposed lxc-ip --search-interface option would ignore -s? [15:36] benji yes thanks! one more IRC message to write... [15:36] gary_poster: I was thinking about that... [15:37] gary_poster: it would ignore -i too... [15:37] bac, #juju comment was actually about canonical-tech list [15:38] gary_poster: found it [15:38] cool [15:38] gary_poster: the horde awaits [16:56] bac: building: http://ec2-23-22-36-49.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/0 [17:27] benji, I don't know if you saw, but somewhat as predicted, real-time was fine and has been merged, but jml has comments on https://code.launchpad.net/~yellow/subunit/test-count/+merge/103717 . I'll be interested to see how you think you would like to reply [17:28] gary_poster: yeah, I'm thinking about it now. Right now I'm leaning toward your idea of how to keep the behavior isolated to testr [17:29] benji, cool. Re http://ec2-23-22-36-49.compute-1.amazonaws.com:8010/waterfall , I expected to see more output--more logs, yes? [17:29] I will probably begin working on a branch momentarily. [17:29] and status changes [17:29] is this a pre-run? [17:29] branch: cool [17:30] gary_poster: hmm, yeah I would have too [17:30] the only change I see is that the output is now fully subunit [17:30] grr, firefox has frozen up on me [17:30] :-/ [17:30] it looks like your changes are not actually hooked up or something [17:31] the subunit output suggests that at least the master.cfg changes are being used; let me log into the ec2 machines and poke around [17:48] Did everybody else already realize that --load-list apparently discards the requested ordering? [17:48] So, we get the exact order of tests run now with our recent work [17:49] but --load-list ignores that ordering [17:49] kinda sucks [18:02] gary_poster: that's not good [18:02] yeah :-/ [18:02] so, that calls into doubt all of my recent "not an isolation error" evaluations for cards on our board [18:03] gary_poster: van hoof submitted his lp2kanban patch yesterday which i reviewed and merged a little while ago. nice addition of 'addCard' to a lane object. go community. [18:04] yay bac. I wonder if we should make a package for the generic bits (which are in one file IIRC) [18:04] actually a couple of file [18:05] gary_poster: seems premature until there is more uptake, i think [18:05] or maybe that's backwards thinking [18:05] dunno [18:05] it is advertised by lp2kanban [18:05] I mean [18:05] advertised by the kanban people [18:05] oh really? did i know that? [18:05] gary_poster: might be a fun test for pkgme [18:07] bac, http://blog.leankitkanban.com/2011/12/leankit-kanban-api-wrappers/ and http://support.leankitkanban.com/entries/20807108-leankit-api-wrappers-and-examples [18:08] nice [18:10] bac, this was from benji's most recent run. Didn't we have some sort of answer for this? [18:10] http://pastebin.ubuntu.com/947847/ [18:10] * gary_poster has forgotten [18:10] i ran into that problem earlier in the week and it mysteriously went away [18:10] :-/ [18:11] benji landed a branch a few weeks ago to address it [18:11] the bzrplugin throws away all of these messages until the correct sentinel is seen [18:11] that branch addressed somethign similar, I don't recall that particular error though [18:11] so, i remain confused as to why this is cropping up now [18:11] :-/ [18:12] benji: that specific error message was in the bug your branch fixed i recall [18:12] sounds like a bug to file...though I may wait till we have a pretty output. I could make one from the subunit output...we'll see how excited I get. [18:12] * gary_poster decides that it is a great time for a water break. [18:13] bug 972456 [18:13] bac: right, but I seem to remember the test failures were of a different flavor; they were all concerned with starting the bzr server; I don't think that fix will help this test [18:13] <_mup_> Bug #972456: Tests can fail when bzr emits an unexpected "unsupported locale setting" warning

< https://launchpad.net/bugs/972456 > [18:13] benji: o [18:14] * bac is glad the shop owner downstairs is playing something good today -- arcade fire, atm [18:27] so benji, those logs don't show up automagically? (i'm watching you edit) [18:28] bac: at this point I'm just trying to get any log to show up (other than stdio) [18:28] I can't for the life of me figure out why our logs aren't being displayed. [18:30] gary_poster: one-on-one? [18:30] jawohl [18:31] benji, https://talkgadget.google.com/hangouts/_/extras/canonical.com/goldenhordeoneonone [18:31] hm... [18:31] my image is not moving in the hangout...oh there it goes [18:39] benji: so did it barf? [18:39] bac: it's almost there: http://ec2-23-22-36-49.compute-1.amazonaws.com:8010/builders/lucid_lp/builds/5 [19:28] bac, more water, shoudld be ready in 2 [19:32] gary_poster, ok [19:32] hangout url? [19:32] bac, sorry was 4 https://talkgadget.google.com/hangouts/_/extras/canonical.com/goldenhordeoneonone [19:53] benji, pair in horde? [19:53] bac: sure [19:53] wow, sometimes you can communicate with making no sense [20:58] gary_poster, benji : when i run update-manager i still get the warning about packages being held back / partial upgrade. i thought that would clear up when the release was out (wishful thinking). y'all know anything about how to resolve it? [20:58] nope [20:59] bac, well, I don't have that problem now, but when I did, I danced around between update manager, apt-get and aptitude. [20:59] The three of them have different algorithms to determine how to clear things up [20:59] gary_poster, apt-get shows the same set as being held back [20:59] at one point aptitude was the only one that could tell me what to do [20:59] i'll try aptitude [20:59] and I had to examine the options [20:59] oh, ok [21:00] i think that 'partial upgrade' button is demonic. [21:00] it turned out that one of the options was to remove a set of packages that had become obsolete and replaced with another [21:00] that was the right option [21:00] that was really the worst I've had [21:00] everything else has been fine [21:00] nice [21:00] so is your install clean now? [21:01] aptitude is generally my go-to manager for this stuff, because of the ability to examine solutions, but even it sometimes is bettered by one of the other two. I think it is bac; I'll check. It was yesterday morning, I think. [21:01] That was the last time I tried an update. [21:02] trying now [21:08] bac, I have 16 updates right now with a resolution problem; trying. [21:10] bac, aptitude seemed fine with it when I ran sudo aptitude; downloading updates now [21:11] great. i'm in the middle of 'apt-get install aptitude' so i'll know soon [21:11] k [21:17] bac, do you have a moment to look at https://code.launchpad.net/~gary/launchpad/bug988541/+merge/103778 and see if you think I'm right that ordering doesn't matter? I figure two people saying "huh, I, uh, guess so" is probably better than one. [21:17] if not no biggie [21:17] sure [21:17] thank you [21:18] gary_poster, i think you could have used assertContentEqual [21:19] bac, oh, nice, ok [21:19] iirc it is an ordering agnostic list compararator [21:19] I'm trying to verify [21:21] def assertContentEqual(self, iter1, iter2): [21:21] """Assert that 'iter1' has the same content as 'iter2'.""" [21:21] self.assertThat(iter1, MatchesSetwise(*(map(Equals, iter2)))) [21:21] Sounds like it, I think. Will switch. [21:25] gary_poster, comparing sets is heckuva lot more understandable! [21:25] bye [21:26] bac, I thought so too :-) but this is better [21:26] bye [21:26] thank you