[08:00] jam, if your will to live becomes too strong today, I pulled together a massive leadership-backport branch: http://reviews.vapour.ws/r/2205/ [13:14] fwereade: is there more than just porting other code? [13:14] nope, I mentioned the couple of non-automated bits I hit [13:15] jam, ^^ [13:15] fwereade: why are you here on the weekend? [13:16] fwereade: backporting the state.Open() change seems a more serious change to the internal API in a stable release. We do depend on it, though? [13:17] that seems to be the bulk of the line-by-line changes [13:17] jam, I think we probably don't strictly *depend* -- but unpicking it felt riskier than including it [13:17] jam, yeah [13:18] jam, and state.Open isn't really API, is it? it's only ever called in-process [13:18] jam, so we can trust the compiler :) [13:18] hence "internal" API [13:18] fwereade: so I don't know that we've stated "we won't change our function signatures in a stable release" as much as our over-the-wire API [13:19] jam, I'd see the various, er, internal API methods as the "internal" API ;p [13:19] jam, why would we want to restrict internal function signatures? [13:19] fwereade: ABI better for you ? [13:19] jam, if I were worrying about dylibs, maybe [13:20] jam, but when it's all one big chunk of code compiled and linked together, I can't get too worked up about it [13:20] fwereade: well we do happen to know some people writing extensions for various parts that will need to be aware of something like this. [13:21] I think it falls more under if you look at "feature" vs "bugfix". I understand why we need to backport this stuff, but it does fall under the "questioning eye for a stable change" [13:22] fwereade: anyway LGTM [13:22] jam, I think I see your point, but I'm not really willing to commit to any stable package interface I don't absolutely have to ;) [13:23] jam, tyvm [20:52] hey whats going on everyone? [20:54] I'm hitting my head on some mongodb issues using local provider for juju [20:55] my /var/log/upstart/juju-db-bdx-local.log shows http://paste.ubuntu.com/11906031/ [21:27] bdx, shot in the dark based on that traceback -- is something using port 38017 ? [21:27] bdx, it's a bit surprising because I didn't think we ran mongod with --rest [21:43] o/ fwereade [21:49] thumper, o/ [21:49] fwereade: Y U make my life difficult? [21:50] thumper, mainly for the sheer devilment of it, I must admit [21:50] thumper, (what did I do in particular?) [21:50] system destroy [21:50] thumper, oh, yes [21:51] thumper, am I annoyingly correct or just annoying? or have you not decided yet? [21:52] there is no list blocks command for a system yet [21:52] thumper, (either way, it may please you to know that the mosquitos are killing me here) [21:52] and I'm not sure I agree about the general output formatter [21:52] not for this case [21:52] I like the formatters for general "list this" type commands [21:52] or about this [21:52] thumper, well, for this case I can put up with whatever makes sense to stderr [21:53] but this output was really to help inform the user [21:53] thumper, I think that if it's stdout it's a bit different? [21:53] why? [21:53] we output messages to std out [21:53] plain text stuff [21:53] this is just more [21:53] context.Infof etc [21:55] * thumper digs into the windows test failure [21:56] fwereade: it seems the intermittent uniter test failures are becoming more common on win 2012 [21:56] bug 1475724 [21:56] Bug #1475724: UniterSuite.TestUniterRelations

[21:58] thumper, I think context.Infof writes to stderr? -- and, yeah, I saw that earlier, I am a little bit suspicious of update-status but haven't looked into it [21:58] thumper, would you nudge wallyworld in that direction when he comes on please? [21:58] fwereade: hmm... maybe it does [21:59] fwereade: I'm looking into it now as it is blocking the landing of the jes-cli branch [21:59] fwereade: that and timeouts on ppc for uniter tests :( [21:59] they take almost 8 minutes on my machine [21:59] thumper, I dunno, I just have a vague but strong expectation that stdout contain results of what you asked for and stderr have more freeform explanatory stuff and other abrotrary bits that seemed liek a good idea [21:59] and ppc is slower [22:00] fwereade: re: output, I think that's ok [22:02] thumper, fwiw, I have someone on the internet who says it's true :) http://www.jstorimer.com/blogs/workingwithcode/7766119-when-to-use-stderr-instead-of-stdout [22:03] thumper, re Output, I'm not sure why people haven't been using it at all [22:04] thumper, did everyone just forget it existed? or is there something important it doesn't do and is hard to fix? [22:04] fwereade: what are you referring to exactly? [22:05] fwereade: ctx.Infof and Verbosef ? [22:05] thumper, juju/cmd/output.go [22:05] thumper, the source of all the --format args [22:05] fwereade: people do use it [22:05] for all the places it makes sense at least [22:05] I do think that 'juju block list' should be fixed to use it [22:07] thumper, cool -- and, ah, I see what bugged me [22:07] thumper, I saw the tabwriter, and saw no output formatter [22:07] fwereade: you mean this time? [22:07] thumper, and thought OMFG-have-all-these-new-output-formats-been-hacked-in-horribly [22:07] fwereade: in this case... personally, I'm ok with this [22:08] thumper, but no, I see formatTabular here and there as it shoudl be and am mollified [22:08] as this is not 'standard' output for the command [22:08] it is informational only [22:08] thumper, put it on stderr and I'll fold [22:08] ok [22:08] done [22:08] thumper, cool :) [22:08] but I agree in principle about the formatters [22:08] and I do use it, and expect others to [22:09] it is used for all the user list, env list, etc [22:09] but these default to --format=tabular [22:09] for niceness [22:09] thumper, yeah, I just saw those, should have been looking earlier [22:09] no more 'smart' bollocks [22:09] thumper, and +100 to default-tabular [22:11] thumper, and, yes, naming something "smart" is basically cursing it with an eternal suck magnet [22:17] * thumper tries to get this win 2012 VM running the juju unit tests again [22:32] windows just told me it finished the worker/uniter tests in -24672.05 seconds [22:32] I can tell you it took longer than that [22:33] thumper: it's kind of funny.. in a sad way :D [22:50] oh FFS [22:52] * thumper looks at perrito666 [22:52] thumper: ? [22:52] perrito666: the source of the windows critical blocker is uniter_test.go line 272 [22:53] startUpgradeError{}, [22:53] that doesn't work on windows [22:53] as it does chmod 555 $CHARMDIR [22:53] so getting start-failed [22:53] instead of started [22:55] bogdanteleaga: ping [22:55] menn0: do you know much windows? [22:55] thumper: apologies, I hadn't noticed that [22:56] I'm wondering how to rework this without disabling [22:58] thumper: you can dissable that particular one by running it in a separate runUniterTests with the windows condition before and add a todo with my name in it [22:58] so you are unblocked [22:58] I would but it is a bit sunday dinner time fo rme [22:58] understood [22:58] I'm just wondering if I can just use 'exit 2' at the end of the script [22:58] instead of a chmod [22:58] thumper: I know it a bit [22:59] thumper: but i'm no expert [22:59] thumper: I can help with this issue now though [22:59] thumper: https://bugs.launchpad.net/juju-core/1.24/+bug/1474606/comments/5 [22:59] Bug #1474606: $set updates may clear out the env-uuid field

[22:59] thumper: it turns out what I was looking at is not quite as critical (but still needs fixing) [23:00] quick hangout? [23:00] yep [23:02] thumper: we seem to require a different failure for windows we might need to do a fully different charm failure for that [23:02] anyway, dinner, cheers [23:07] thumper: I'm around for 3-5 mins [23:09] bogdanteleaga: don't worry, I'm working through it with menn0 [23:13] thumper: fwiw, you're probably going to have to find another way of making it fail, it *should* be possible with permissions but it might get hairy [23:13] thumper: and you could use check.f to only run that test so you don't have to wait for everything [23:13] bogdanteleaga: looking at 'cacls $CHARMDIR /p everyone:r' [23:14] bogdanteleaga: think that'll work? [23:15] thumper: not sure if it'll disable other permissions you'd have to try it out; also check whether that works from both ps and cmd as I'm not sure under which that particular script gets ran [23:15] bogdanteleaga: I'm trying it now [23:36] thumper: this seems to be the correct way: http://paste.ubuntu.com/11906714/ [23:39] menn0: does it still ask for verification? [23:40] thumper: what do you mean? I didn't see any prompts. [23:45] menn0: good, tested and copied in [23:45] running the test now [23:45] if this doesn't work, I'll be ripping the test out and use test.skip for windows [23:45] until someone can fix it [23:47] it fails [23:47] thumper: why? [23:47] start-failed [23:47] the test doesn't give a lot of output [23:47] no [23:47] I see it [23:48] * thumper goes to try some again [23:52] for my sanity, I'm going to the gym