[11:21] morning [11:58] frankban: I reviewed and approved https://code.launchpad.net/~frankban/launchpad/bug-974585-unclean-reactor-error/+merge/102473 [12:02] frankban: I also reviewed the two MPs about adding to fstab [12:04] thank you benji [12:04] np [12:09] bac benji frankban gmb call in 1 or 2 [12:09] Okay [12:09] ok [12:19] gary_poster: quoth jml "soon. sorry for the delay" [12:47] gary_poster: I can't think of anything to add to the progress report. [12:47] ack thank you benji [12:48] gary_poster: I am going to change "974585: in progress (frankban)" to "fix committed" in the document, do you agree? [12:49] frankban, agreed, thank you [12:50] bac, fwiw I added comment 3 of https://bugs.launchpad.net/launchpad/+bug/981114 [12:50] <_mup_> Bug #981114: LP tests fail in lxc due to 'bzr whoami' failure

< https://launchpad.net/bugs/981114 > [12:53] thanks gary_poster, that looks good [12:53] cool bac, thanks for looking [13:06] hmm, that's odd: [13:06] % grep '^tags:' .testrepository/7 | cut -d' ' -f2 | sort | uniq -c [13:06] 1 worker-0 [13:06] 88 worker-1 [13:06] 88 worker-2 [13:06] 87 worker-3 [13:08] well, maybe not that odd, that one tests took 0.6 seconds which is about the length of the test run [13:41] bac gmb, have you reviewed the doc for the past two weeks? [13:42] gary_poster: The biweekly progress doc for today? [13:42] frankban, I replaced all the old lpsetup changes in the doc (that were from an earlier copy) with "various lpsetup improvements". I reviewed the recent changes and they seemed relatively small, but I may have missed something. Did I? [13:42] yes gmb, https://docs.google.com/a/canonical.com/document/d/1JsnEKxxID8GijjZcUOTQ671w9zeKNwoA4dbIhoSbGjA/edit [13:42] gary_poster: looking [13:43] ty [13:43] gary_poster: I'm looking now, too. [13:44] ty [13:49] gary_poster: you are right, I think the only changes for these two weeks are just those you can see in done-done [13:49] cool frankban, ty [14:32] benji: in charm tools we have this: [14:32] juju_status = lambda: command('juju')('status') [14:32] any reason that is better than the more straightforward [14:32] juju_status = command('juju', 'status') [14:32] bac: Testing. [14:32] bac: The charm-tools tests monkeypatch juju_status so that it doesn't call out to juju, IIRC. [14:32] gmb: er? how so, for the slow? [14:33] gmb: ok, i'll look for that [14:33] just curious, no biggie [14:33] bac: Yeah, it was a speed thing. However, it of course means that we need a test to make sure our mocker is actually accurate, which, also IIRC, it actually isn't. [14:33] (anymore) [14:34] bac: So feel free to actually make it, yanno, do what it should :) [14:34] bac: I think the second should work, the first is probably from the time before the command function grew the second parameter [14:34] gmb: i may have already [14:35] it seems a juju_status command can be monkeypatched regardless of how it is originally defined. benji i think you're right it may be a leftover [14:36] gmb: the tests all passed in my latest branch after i updated them to account for 'agent-state' returned by 'juju status' [14:36] bac: Ah, cool. [14:38] * gmb -> late lunch / changing locations [14:57] gary_poster: as a slack time task i'm gathering information about the juju SSL failures so i can file a bug. spurious juju failures were the doom of our scripting. [14:57] bac, sounds great, ty [15:28] could someone please request a build here? https://code.launchpad.net/~frankban/+recipe/python-shelltoolbox-daily-lucid (I have reached the quota) [15:32] frankban, on it [15:32] ty gary_poster [15:33] frankban, destination should be "Yellow Squad PPA" right? and Lucid, to state the seemingly obvious. [15:33] yes gary_poster [15:33] cool [15:33] pending frankban [15:34] thanks gary_poster [15:34] welcome [15:36] benji, am I right that you are working on "Add worker ID tagging to testrepository" but not on the buildbot changes necessary for 974622? If I am right, bac and/or I could look at that. If I am wrong, woohoo! more slack time! [15:37] gary_poster: you're right, specifically testing at this stage [15:37] benji, cool. bac, I'm going to try futzing around with the buildbot stuff. If you want to join me at some point, I strongly suspect that you would make things move much faster. [15:38] (but please feel free to finish the valuable juju bits you are doing) [15:39] gary_poster: here is the error i see doing repeated calls to 'juju status': http://paste.ubuntu.com/935599/ [15:39] gary_poster: i'll be glad to help with that in a bit. i'm about to leave to go biking [15:40] bac, benji, I only see testtools and testrepository in yellow PPA. we need the newest subunit, or at least the revision I identified in one of my emails, also. Is that somewhere, like in the LP PPA, or do we still need to add that too? [15:40] bac, cool, have a good time [15:41] gary_poster: I don't know of any changes we have made to subunit. [15:41] benji, we didn't make a change to subunit, but as I outlined in the "steps we need to take" email, we need a newer revision than is released. lemme find that... [15:42] yay, jml merged! [15:42] gary_poster: oh, ok [15:43] > - add test in subunit for TestProtocolClient.tags method [15:43] > - add .tags method to TestProtocolClient (see patch below) [15:43] These were added in revno 158 of subunit (unreleased). We just need to make our own release (after the work on subunit-filter, below). [15:57] back in a moment [16:38] * gary_poster going to lunch. subunit tests are failing, which means it won't build as is, which means we have work to do. [17:53] (I was wrong, confused by old builds) [18:06] but python-testtools does fail because of def _merge_tags(existing, (new_tags, gone_tags)): [18:08] benji: do you know what needs to be done wrt subunit for ppa? [18:09] a ppa already exists here: https://code.launchpad.net/~testing-cabal/+archive/archive that jelmer looks to be working on to get r162 built [18:09] bac: we need the head of subunit trunk packaged [18:09] bac, I was just going to copy that over when it was done [18:10] a slightly bigger problem is that testtools does not build [18:10] because of that code I gave above [18:10] SyntaxError in py3 [18:10] so build fails [18:10] so unavailable [18:23] gary_poster: it looks like support for py3 was only added recently to the packaging branch, but testtools looks to have lots of issue wrt py3 [18:23] not just the paramater format problem [18:23] bac, agreed [18:24] bac, I just said the next thing I found to jml [18:24] I think we should mae a branch with py3 build stuff removed, as I said to jml on -dev [18:24] gary_poster: oh, ok [18:26] testrepository is failing too? according to mail to yellow [18:26] gary_poster: and also stuff like : AttributeError: 'range_iterator' object has no attribute 'next' [18:26] heh [18:26] about 15 test failures for py3 [18:27] * bac looks at testr upload log [18:27] bac, so what's the most efficient way to rip out by3 build bits and get something working in the yellow ppa [18:28] gary_poster: branch the vcs packaging, fix it and push to ~yellow. they recreate build recipe referencing the new packaging [18:28] gary_poster: i'll do that now [18:28] we are using http://pastebin.ubuntu.com/935822/ [18:28] s/they/then [18:28] argh [18:29] bac, we are using "merge packaging lp:~vcs-imports/testtools/unstable" [18:29] perhaps there is a "stable"? [18:29] benji: it looks like testr recipe failed b/c the 'ppan' was not incremented. correct? [18:29] gary_poster: oh, maybe [18:30] bac, not that I see :-/ https://code.launchpad.net/~vcs-imports/testtools [18:30] bac: it did once yesterday, but I fixed it that time; did it fail again? [18:30] benji: yes, recently [18:30] hmm [18:30] and also http://anonscm.debian.org/bzr/collab-maint/python-testtools/ [18:30] gary_poster: i wonder if the build recipe lets you specify a revision number for the packaging branch [18:31] good question [18:31] would be nice [18:31] just roll back one or two rev [18:31] I just looked at the error in my email, it looks like incrementing the ppa# is all it needs. [18:32] r35 of testtools build won't be good enough bac [18:32] we could fork it though [18:32] and up the revision number [18:32] but r35 is 0.9.14-1 and we have 0.9.14-2 I think [18:33] gary_poster: no i meant a new recipe that uses an old version of the packaging branch but trunk of the source tree [18:33] bac I understood [18:33] bac, my point is that their is not an old version of the packaging branch that is simply usable [18:33] do we need to talk, gary_poster? i think we may be missing each other. [18:33] there [18:33] ok [18:34] I am in https://talkgadget.google.com/hangouts/_/extras/canonical.com/goldenhorde [18:57] it shouldn't matter but just in case it becomes important: I just merged the testrepository branch that I'm proposing to Robert into our testrepository "latest" branch [18:58] gary_poster: I /think/ all the tasks for Tasks for bug 974622 [18:58] <_mup_> Bug #974622: cannot determine order of test execution in a parallel worker < https://launchpad.net/bugs/974622 > [18:58] gary_poster: I /think/ all the task for bug 974622 are done now. [18:58] <_mup_> Bug #974622: cannot determine order of test execution in a parallel worker < https://launchpad.net/bugs/974622 > [19:08] gary_poster, benji: i've changed the testtools recipe to use trunk and created a packaging branch to get rid of the py3 build. [19:09] the recipe is at https://code.launchpad.net/~yellow/+recipe/testtools-daily [19:09] unfortunately, we have the same versioning problem now. trunk is at r253 where ~yellow/testtools/latest is at 265, so the latest ppa has that bzr revno [19:10] ... [19:12] benji, we still need to hook it up to buildbot...but we can make a separate card for that. This adds value, I agree. I'd like to sat that once the PPA is set up and works (we can show that all the tools we added can be used), we are done. [19:12] with this card I mean [19:12] then a separate card for the buildbot integration [19:13] bac, :-/ [19:13] suggestions, bac? [19:13] gary_poster: sounds good [19:13] bac: you are working on the PPA, right? [19:14] AIUI...bac is working on getting testtools in the PPA, and benji is working on getting the newest testrepository in the PPA (the complaint that we all got from the build system was for a problem we actually care about) [19:14] (because we want --full-results) [19:15] gary_poster: ok, I can do that (put testrepository in the PPA). [19:15] benji, cool, thanks. AIUI, you have to manually increase some number or other, from 2 to 3? [19:16] yes, i am working on testtools ppa [19:16] I have the feeling that we should have decomposed that 974622 card a little more. [19:16] but i am stuck [19:16] gary_poster: yep, that should be it [19:16] benji, maybe. I thought about it for a bit. We had a bit of decomposition. [19:17] bac, if we simply merged trunk into our own branch would that be good enough? [19:17] then we would be building from our branch [19:17] which would have a high-enough revno [19:17] gary_poster: yeah [19:17] yeah, it wasn't too bad; I just think a tad more would have been nice [19:18] benji, cool. maybe see if you can come up with a suggestion for the future for our Friday call? (assuming we remember to have it :-P ) [19:18] if not, no biggie [19:18] k, I'll think about it [19:19] but now, when things are fresh, is the time to see if you can turn that hunch into something actionable :-) [19:19] thanks [19:20] oh good, another linked in sort of thing for people to connect with: http://www.naymz.com/ [19:21] gary_poster: ~yellow/testtools/latest merged and PPA building [19:21] got an email from someone wanting to "connect with me" [19:21] great bac thanks [19:22] (from a canonicalite I suspect I've never met, no less) [19:26] and to continue my cynicism, look, it's not buildbot, it's not jenkins, it's not hudson, it's...travis! http://travis-ci.org/ kills my chrome though [19:27] fine on midori though [19:41] gary_poster: testrepository has been updated in the PPA, build successful [19:42] great benji, thanks. I saw that it was bopping along [20:03] bac, how goes it? [20:03] gary_poster: had a mis-start but it is rebuilding now [20:04] bac, cool. this is in https://launchpad.net/~yellow/+archive/ppa/+packages ? I don't see the rebuild [20:04] gary_poster: this go looks better: https://code.launchpad.net/~yellow/+recipe/testtools-daily [20:05] gary_poster: not done yet [20:06] gary_poster: 0.9.14-bzr266~ppa39~oneiric1 just published. precise is finishing up. [20:11] gary_poster: 0.9.14-bzr266~ppa39~precise1 is available [20:12] yay bac, thanks [20:12] trying [20:16] gary_poster: false start. it looked like it built but then didn't publish [20:16] bac, yeah, was going to say I didn't get it [20:16] ok, i missed something in the rules file. sorry. [20:18] gary_poster: it claims 13 minutes [20:19] cool np [20:20] gary_poster and bac: who is going to test the software in the PPA once that build completes? I was assuming I was, but I should make that assumption explicit so if not I can start on something else. [20:22] benji, the next task (I just made a card and then deleted it) is "buildbot results show subunit streams filtered by workers" [20:22] that depends on this stuff working... [20:22] anyone who wants to work on that task should see if they have a working version of stuff that does what we want. [20:22] (sorry, I'm thinking this through) [20:23] so...I was planning on trying out what I expected to work for buildbot (via running testrepository tests, since they are parallel) [20:23] oh, wait, I thought "verify that the stuff in the PPA does what we want" was the last bit of 974622, not the first bit of the next thing [20:23] well... [20:23] :) [20:24] verification and usage are darn similar I guess is what I was trying to think through [20:24] I was not planning on verifying except to the degree that I was going to see if I could use [20:24] it [20:24] true, and if verification fails, then that will spawn new cards [20:25] someone actually verifying and coming at it from that perspective would probably be wise [20:25] yep [20:25] ok, in that case I don't have anything to do :) [20:25] benji, oh, I thought you were going to be the verifier! [20:25] :-) [20:26] and I was going to hang out as the user, or sneak past you and see if I could use it while you were verfying it and not looking [20:26] So, behold A Decision [20:26] gary_poster: oh, I can be... even then, I'm blocked, but I suppose not for long. Although, I've been thinking that it wouldn't be much longer for a while now. [20:26] heh, yeah me too [20:27] benji, have you already watched inventing on principle, http://vimeo.com/36579366 ? [20:27] I have. Good stuff. [20:28] then I have no suggestions :-P [20:30] gary_poster: lifeless' clarification of the whoami situation doesn't help me. [20:30] bac, we can move card for bug 949950 right? [20:30] <_mup_> Bug #949950: Feature request: an option to show full subunit stream of running tests < https://launchpad.net/bugs/949950 > [20:30] then we have a spot free [20:30] gary_poster: indeed [20:30] bac, remind me that bug please? [20:30] the one with lifeless' comment [20:31] bug 981114 [20:31] <_mup_> Bug #981114: LP tests fail in lxc due to 'bzr whoami' failure

< https://launchpad.net/bugs/981114 > [20:31] he's still thinking it is a test isolation problem. [20:34] gary_poster: I'm going to move the 974622 card to... somewhere, because it's "done" [20:35] benji, you verified? [20:35] ... [20:35] heh [20:35] I guess we talked past one another [20:35] no, that is part of the next thing; isn't that what we just talked about [20:36] not exactly. we can talk about it to clarify things later. Please move the card once you have been able to verify. I apologize for the confusion largely caused or at least exacerbated by my earlier indecision. [20:36] gary_poster: ok, for real, the testtools are in the ppa: https://code.launchpad.net/~yellow/+archive/ppa/+packages [20:36] bac, great [20:36] bac, I think I understand what Robert is saying now [20:36] * benji is saved by the HVAC guy knocking at the door. [20:36] but it is a pretty odd way of saying it [20:37] bac, I *think* Robert is making the point that we can set whoami explicitly in that test [20:37] please translate, slowly [20:37] oh [20:37] and then it will give a nice value, irrespective of whether /etc/mailname is set [20:37] which makes sense [20:38] and Robert's ideas often make sense [20:38] so I choose to interpret his message in that way. :-) [20:38] ok [20:38] so what do you want to do? unwind the /etc/mailname change and do it this way? [20:38] if that's the case i'd make a new card for it and, uh, get to it when we get to it. [20:40] i need to run to town for a bit. bbl. [20:40] bac bye [20:40] and yes, let's unwind and add [20:41] I'll make a card [20:45] I made a card [20:45] testtools is not actualy available yet [20:56] benji, AFAICT the worker tags are not included, but maybe my test sucks... [20:56] oh [20:56] wait [20:56] my test sucks [20:56] hope springs eternal [20:56] no it doesn't :-P [20:56] hope dies instantly [20:56] lol [20:57] no, hope springs eternal, but I don't think my test sucks [20:57] pfft [20:57] so I have newest versions of three packages from ppa [20:57] hope VERBS ADVERBly [20:57] --subunit works [20:57] k [20:57] --full-results works [20:57] k [20:57] but --parallel does not give me worker tags [20:58] are worker tags added in your most recent .testrepository/## file? [20:58] good q looking [20:58] (worker tags are always added, not just when --parallel is used) [20:59] no they are not [20:59] I've checked testr and subunit version [20:59] checking testtools... [21:00] 0.9.14-bzr266 [21:00] yeah that's right [21:11] gary_poster: I'm going afk now. The next thing I would check is to see if the code is actually in testrepository. You want to see a _wrap_result function in testrepository/commands/load.py that is fed to ConcurrentTestSuite. [21:11] benji, ack, on call. ttyl [21:11] k