[11:44] hi gmb [12:08] (bac is already here) benji frankban gmb: https://plus.google.com/hangouts/_/cc873feeffaf2d89f51f40706dd22be36570cb3d?authuser=1&hl=en-US in 2 [12:21] gmb, my last change to z.t was r43 "[r=bac] Cleaned up stdout/stderr data handling." [12:22] https://dev.launchpad.net/LEP/LaunchpadJujuCharmForDevs [12:22] https://dev.launchpad.net/LEP/ParallelEC2Command [12:23] http://codesinger.blogspot.com/2012/06/our-checklist-for-running-daily-meeting.html [12:32] gmb: i grabbed trunk, merged your changes, and ran the tests. all hell broke loose. [12:32] Boo. [12:32] can you confirm? [12:32] Wait, whut? [12:32] Hmm. [12:32] Yes... I merged trunk, and then everything re-broke. [12:34] gary_poster: I submitted the last blog post to hacker news: http://news.ycombinator.com/newest [12:35] benji, heh, cool, thanks. I guess I should follow along there too then. They seem to like to talk amongst themselves over there [12:35] bac, So, my branch at r43 works. [12:35] My branch with today's trunk merged into it, not so much. [12:36] bac, I need to go and grab some food - or else just eat a colleague - are you okay poking around at it for now? [12:36] gmb, sure. i just wanted to ensure we saw the same thing [12:36] well, given that my submissions almost never take off, I wouldn't worry too much about that ;) [12:36] Yep. It broke. [12:37] gmb: not completely unsurprising since we worked in parallel making changes to sys.stdout [13:24] hey benji, got a second? [13:24] bac: I will in about 2 minutes; be right back [13:24] ok [13:26] benji or bac, I want to start "lpsetup: add workaround for bug 1014916" but I'm not allowed without a partner. boo hoo. When either of you are available to pair, lemme know [13:26] <_mup_> Bug #1014916: simultaneously started lucid containers pause while starting after the first seven < https://launchpad.net/bugs/1014916 > [13:26] oh nm [13:26] we're full :-) [13:27] bac: 'sup? [13:28] gary_poster: you could pair with me on "lpsetup init" [13:28] benji: wanted to talk about stdout, __stdout__, etc in zope.testing. it looks like our changes are a bit incompatible [13:28] ok cool, benji lemme know when you are ready [13:28] gary_poster: will do [13:28] bac: do you want to discuss via hangout? [13:28] benji: https://plus.google.com/hangouts/_/52afa9329db48899dcaaf86f18e58df34ab497f3?authuser=0&hl=en === benji is now known as Guest14462 [13:32] bac: hangouts seem to crash my UI on occasion; will you give me that URL again? === benji___ is now known as benji [13:33] https://plus.google.com/hangouts/_/52afa9329db48899dcaaf86f18e58df34ab497f3?authuser=0&hl=en [13:46] benji: you've gone away [14:18] bac: you have been booted from the hangout [14:18] google hates me this i know [14:19] this is the same thing that has happened to me [14:19] I think termbeamer needs a new transport [14:19] i'll start a new one [14:19] jabber.benjiyork.com ? [14:19] termbeamer.com :) [14:19] I'll make it a pay service and get RICH! [14:21] google says "unable to reach chat servers. check your internet" [14:21] * bac gets a beverage [14:25] i have pissed off the GOOG [14:25] i cannot reconnect [14:28] benji: can you start a hangout and invite me? [14:29] is this thing on? [14:30] bac: trying [14:32] bac: https://plus.google.com/hangouts/_/053813e4a0cbd571b5704a06108437be2b6af73f?authuser=0&hl=en-US [14:33] first attempt failed [14:34] geez, do i need to reboot? get a new IP address? [14:34] benji: skype? [14:34] bac: ok [14:40] benji: http://paste.ubuntu.com/1051006/ [14:54] fun. stub introduced a branch that gives our ec2 parallel test three failures, but it does not fail on buildbot nor can I make them fail locally even with the full worker test list. [14:55] I mean, it does not fail on lpbuildbot [15:06] bac: lp:~benji/zope.testing/stdio-and-testing-debug-session [15:08] gary_poster: I'm going to get a refreshing beverage and then I'll be ready to work on lpsetup init [15:14] benji, ok cool. I think I need to figure out what is going on with those test failures I mentioned above [15:15] When I do that I'll ping you [15:15] k [15:16] gmb, bac, boo hoo (re the fix-all-the-tests branch) [15:17] bootstrap seems to be broken ob my machine: http://paste.ubuntu.com/1051061/ [15:17] please make sure that's handed off, and if the card is still around tomorrow let's escalate its importance (making sure someone is always available for pairing and all that jazz per our checklists) [15:18] benji, this is in lpsetup? [15:19] gmb, "please make sure that's handed off..." was to you, in case that's not clear [15:19] gary_poster: "this" being "lpsetup init command"? If so, yes. ;) [15:20] benji, um. OK, is that the lpsetup bootstrap, or the Launchpad bootstrap as driven by lpsetup? [15:21] ugh, I can't get the tests to fail locally that are failing in parallel tests! :-( :-( [15:21] oh! the error; yes, that error is from running the bootstrap with my system python [15:21] gary_poster, understood. [15:21] thanks [15:23] benji: lpsetup no longer uses buildout for tests, so, no need to bootstrap. that said, bootstrap was broken in my system too and I had to run it inside a virtualenv to work around some setuptools problem [15:26] bac, benji: So, what's the situation with the zope.testing branch (apologies for the delay in asking; I've had a spectacularly interrupty afternoon) [15:27] gmb: it looks like the use of doctests within testrunner is horribly confused by the stdout/stderr shenanigans [15:27] frankban: We should remove the [test] target from the buildout then. And add something to README or a new HACKING file to describe how to set up a dev environment and how to run the tests. [15:27] in fact, does it need buildout at all now? [15:28] gmb: our next approach is to rewrite those doctests as unittests. i have regretably rejected your MP as it appears DOA [15:28] gmb: though some parts of that branch may prove useful [15:28] Botheration. [15:28] bac, Okay. Is there anything that I can do at this point to help you? [15:29] gmb, not that i can think of [15:29] Okay. [15:29] * gary_poster performs small cheering routing about pairing. Everyone is frightened away and looks out the window, embarrassed. [15:30] bac, If, at your eod, you're not done, feel free to send me a summary of your progress so I can pick it up in the morning. [15:30] gmb: will do [15:31] benji, no buildout is not needed at all, and the buildout/bootstrap files can be deleted. The README(.rst) file is already there. [15:32] frankban: indeed! (I didn't expect the Spannish... .rst file) [18:35] whee [19:56] hey benji, this is the start to the conversion: http://pastebin.ubuntu.com/1051482/ [19:56] i've got most of the required structure. the last test is one of the troublesome ones from the doctest and it works fine. [19:56] lemme know if you see something boneheaded [19:56] bac: cool! [19:57] * benji looks. [19:57] now, just put on some loud Lynryd and plow ahead [19:57] or joshua bell [20:03] bac: the tests look good; hopefully it will be mostly mechanical [20:03] yep [20:03] bac: if you need some extra sonic inspiration I suggest http://www.youtube.com/watch?v=pmo1VW_wA_o [20:03] i can't type that fast [20:04] heh [20:04] * benji learns that bac types in rythm to the music. [20:17] benji, how goes lpsetup init? [20:23] gary_poster: (I was preparing to answer your question.) It goes well, I think. Because of the lack of test coverage I have taken a conservative tack, but I think I'm about done. [20:23] we should discuss testing strategies [20:29] great, benji. frankban had a thought on testing strategy: an object that had a minimal API that the commands could talk to. [20:29] then we substitute in a testing version [20:30] that approach captures at least some important aspects of testing [20:30] we talked about that for the host, and then also for the lxc commands [20:31] I've had some luck with an approach like that in termbeamer tests, I did it at the module level (I have a few fake GTK modules that I use in different circumstances) [20:31] gary_poster: i was just reminded i need to take an early lunch (11-12) tomorrow, lest i forget to mention it [20:32] writing the more unity of unit tests for this won't be too hard (and there are several already), the hard part will be deciding what we really want the higher level tests to demonstrate [20:35] bac, ack [20:35] agreed benji. I'm ok with aiming a bit low in that regard, but I'm also ok with be convinced otherwise :-) [21:35] hey benji, still around? [21:54] how does testrunner-subunit.txt ever pass, given that it has hard-coded expected output with a path to jml's home directory? [22:12] bac, it works because the test hookup normalizes the output [22:12] look at how that doctest is integrated into the python test suite [22:13] look for ReNormalizer or something like that [22:14] ah, ok