=== uru_ is now known as urulama [14:11] anyone ever run into https://bugs.launchpad.net/juju-core/+bug/1380337? [14:11] Bug #1380337: adding machine after destroying another fails

[17:08] minor update to the juju errors package to use the new location for errgo: https://github.com/juju/errors/pull/8 [17:08] reviews appreciated [17:27] rogpeppe2: LGTM [17:28] TheMue: thanks! [17:41] rogpeppe2: yw [17:46] anyone know anything about the bitbucket.org/kardianos/osext and bitbucket.org/kardianos/service dependencies? [17:46] they don't appear to be used by anything [17:46] rogpeppe2: aren't they windows dependencies? [17:46] jcw4: possibly, but i haven't found anything that uses them [17:47] rogpeppe2: I see... it's only in cmd/jujud/main_windows.go that it's even imported [17:47] osext is a dependency of service [17:48] jcw4: ah, thanks [17:48] jcw4: i don't know why i didn't find that [17:48] rogpeppe2: I just remember the second level dependency issue when I bumped into the same question recently [17:49] jcw4: it's a pity that the dependencies aren't findable by godeps. i should fix that. [17:49] yeah. since it's only in the _windows.go files go needs a little prodding to find them I guess [17:49] (when you're not on windows) [17:50] jcw4: yeah. godeps should try with different build flags. [18:22] did someone ping me or is my IRC client highlighting this channel for no reason? sigh [20:44] rogpeppe2 I'm seeing this error in builds: [20:44] Extant directories unknown: [20:44] gopkg.in/errgo.v1 [20:45] for example: http://juju-ci.vapour.ws:8080/job/github-merge-juju/914/console [20:46] doesnt make sense to me because nothing in juju master refers to that yet, so it must be a transitive dependency [20:50] Also, I can't find any transitive dependencies on that yet either. [20:50] mgz: I think it's a build script issue... do you know why it's trying to pull in gopkg.in/errgo.v1 since that isn't in the dependencies.tsv, and doesn't seem to be a transitive dependency either [20:52] jcw4: is it from the juju/charms package? [20:53] jcw4: that was a branch that landed today [20:53] jcw4: make sure to update, I saw it go by in the email [20:54] rick_h_: I am fully updated to juju master [20:54] thumper: I'm trying to decipher what "it" is :) [20:54] jcw4: looking [20:54] this error is when the build script is initially pulling down dependencies [20:54] * thumper sighs [20:54] found an intermittent test failure in worker/provisioner [20:56] thumper: I've seen worker/provisioner fail a couple times intermittently, but I see worker/peergrouper fail fairly often locally [20:56] jcw4: https://github.com/juju/errors/pull/8 is the pr I was thinking of [20:56] * thumper sighs heavily [20:56] rick_h_: yeah, that was why I pinged rogpeppe initially [20:56] * thumper is going to *fix* juju/errors [20:57] YAY [20:57] what baffles me is why the build script is even pulling it at all [20:57] juju-core doesn't reference it yet [20:57] and I couldn't find any dependencies of juju-core that referenced it either [20:58] like thumper said, probably juju/charm [20:58] I was told that errgo wasn't going to sneak into dependent packages [20:58] thumper, rick_h_ I don't see how it could have [20:58] and it was only going to be used outside of juju/errors in the charmstore [20:59] thumper: jcw4 ok, justcccc ccbgjgvcveghhillduiklnblheujreenhtsnkubdrcjf [20:59] bah gueccccsccbgjgvscuneviintrdikfgbvegt [20:59] idhvdfvudkdkjb [20:59] double bah [20:59] wat? [20:59] called rick sitting in the hotel bumping his yubikey [20:59] rick_h_: you ok? [20:59] haha [20:59] no, I'm still in brussels :P [20:59] rick_h_: you lucky fella [21:00] rick_h_: did you go see the Atomium? [21:00] rick_h_: I was looking at the schedule for the first few days [21:00] $ git diff master | wc -l [21:00] 1731 [21:00] hmm [21:00] this branch isn't going to break up well [21:01] thumper: errgo should stay within charmstore limits, yes. [21:01] * thumper puts hacking hat back on and ignores all y'all for a while [21:03] ah crap... [21:04] new struct from a merge has broken my work... [21:04] * thumper works around it [21:04] chase them down and make them eat pie [21:05] wait, before committing to that course of action... which struct? [21:05] rick_h_: isn't it LATE in Brussels? [21:05] 11pm or so? [21:06] jcw4: yes [21:06] jcw4: not you [21:06] * jcw4 wonders if the initial pull for imports uses the HEAD revision of juju/errors to find child imports [21:07] thumper: btw, let me know if errgo somehow sneaks into the code ... [21:07] thumper: whew [21:07] jcw4: was cmars's branch for login v2 [21:07] urulama: ack [21:07] urulama: you home? [21:07] thumper: was having fun trying to QA all the "landed" branches from last week ... could have missed some though [21:07] thumper: i am [21:08] thumper: my flight was about 1h 20min ;) [21:08] urulama: I have a branch for juju/errors that needs updating based on our conversations in nuremberg [21:08] urulama: I got home yesterday just before lunch [21:08] urulama: working today/tomorrow taking thu/fri off in lieu [21:08] but today is finishing off what I started pre-sprint [21:09] thumper: enjoy :) [21:09] \o/ [21:09] thumper: seen that errors branch, looks nice [21:10] I'll see if I can get to it tomorrow [21:10] or may this afternoon if I can fix all the test failures I caused [21:10] by changing a bucketload of API calls [21:10] that's always a good thing! :D [21:10] * thumper is full steam ahead on finishing identity bits [21:11] OOPS: 268 passed, 6 FAILED, 5 PANICKED [21:11] need to fix the failed and panicked [21:11] yikes [21:11] :) [21:11] :) [21:11] where is the code that actually runs the CI build? I dont' think its the jenkins-github-lander repo [21:11] don't panic [21:12] urulama: thanks for that :) [21:12] jcw4: not entirely sure sorry [21:12] thanks thumper you can ignore me while you're dealing with your panic attacks [21:12] :) [21:13] cheers buddy [21:13] * thumper goes back to it [21:13] I bet its https://launchpad.net/juju-ci-tools [21:17] jcw4: rog borked it [21:17] mgz: yay :) [21:18] with this change: https://github.com/juju/errors/pull/8 [21:18] mgz: it seems like the build script might be getting imports from the HEAD revisions of first level imports? [21:19] mgz: right, but that's just on the errors repo, and juju-core wasn't updated to use that revision in the dependencies.tsv [21:19] yeah, something's up [21:19] godeps likely doesn't do exactly the right thing [21:20] mgz: yeah, interesting [21:21] I'll have a poke around [21:23] mgz: just revert the merge for now [21:23] mgz: poke around later [21:23] jcw4: the repo is just getting lp:juju-release-tools and running `./make-release-tarball.bash master` [21:24] mgz: this is how we deal with broken builds [21:24] dvcs FTW === marcoceppi_ is now known as marcoceppi [21:24] thumper: fair enough [21:24] thumper: thanks! [21:24] mgz: thanks! [21:28] mgz: I see the problem I think [21:28] compare_dependencies in check_dependencies.py doesn't seem to like any packages that are not accounted for in the dependencies.tsv file [21:29] I bet godeps is just pulling down second level imports before updating first level imports to the specified revision [21:30] which would bring gopkg.in/errgo.v1 into the pristine GOPATH, and then compare_dependencies complains [21:58] mgz: thumper, rogpeppe2 -- I reverted https://github.com/juju/errors/pull/8 so that juju-core will build again. [21:58] rogpeppe2: sorry! Hopefully we can figure out the build process so that we can land your change again [21:58] jcw4: kk === jcw4 is now known as jcw4|afk === mup_ is now known as mup