[12:09] bac benji frankban gmb hiya https://plus.google.com/hangouts/_/3c71b053e171414b2ca8546d398b7f1d82bb2838?authuser=1&hl=en-US [12:09] in 1 (or 2 if necessary) [12:29] bac`, Gah. I think we just swapped hangouts. I'll stay where I am, you come to me: https://plus.google.com/hangouts/_/443840122746ffd12ced54df205ab403d9ad6852?hcb=1&hii=117034961047606128906&hit=1340281662992&htp=0 [12:29] i'm sort of there but only see your still and no audio [12:29] ooh! at least for one brief shining moment, we are at 95% success rate for our tests! [12:29] bac`, Okay, weird; let me reconnect... [12:29] gmb: you may need to make a new one [12:30] given the usual rolling three day window approach, where the third day is today and I haven't run any tests yet :-P [12:30] but still, we are 19/20 for the last 20 tests [12:30] which is our goal statistic [12:30] bac`, https://plus.google.com/hangouts/_/2a9476c4e3d0d3bfe4c77790592e40065cdbedc9?authuser=0&hl=en-GB# has me in it. === bac` is now known as bac [12:33] bac, lp:~yellow/zope.testing/fix-tests [12:41] * gmb lunches [13:18] * bac was momentarily distracted by a swarm of pileated woodpeckers. i saw three large adults and at least two juveniles. very cool. http://www.allaboutbirds.org/guide/Pileated_Woodpecker/id [13:21] cool :-) [13:21] frankban, approved EuroPython [13:22] thanks gary_poster [13:33] frankban, no rush but https://plus.google.com/hangouts/_/f4daf72d067dd6bdd9be0aa8b2c5c805abbf025a?authuser=1&hl=en-US when you are ready [13:59] benji, I just realized that we missed something in our lpsetup description: we should be able to do something like rerun init and have the command update everything to the current "best" configuration. Your card does not need to implement this--I like keeping your card small--but it *might* influence your thinking or your approach, so I mention it now. (should have mentioned it earlier) [14:00] the spelling is up for grabs [14:00] it could be "rerun init" [14:00] or run some other command [14:00] hmm; ok, thanks [14:19] ugh. I don't know what we should do about this now. Another instance of lp.services.webapp.tests.test_error.TestDatabaseErrorViews.test_disconnectionerror_view_integration failing, with "TimeoutException: Launchpad did not come up after 60 attempts." :-( This is already with the 20 instances on a 32 core machine. [14:23] bac, How are we looking with that branch? [14:26] gmb: good, but running the full suite ends oddly: http://paste.ubuntu.com/1052716/ [14:26] note it just stops, no summary [14:27] also note that our new tests get run way at the beginning, so should not be influencing the summary output [14:27] gmb: thoughts on what might be going on? [14:29] gmb: changes pushed to lp:~yellow/zope.testing/fix-tests [14:29] grabbing... [14:30] gmb, we have our call now, but I see you are in the middle of looking at stuff, and no rush in any case. whenever you want, come over to https://plus.google.com/hangouts/_/cadff6c4e5227f872dd2392ff9ac5b51740d0d3e?authuser=1&hl=en-US [14:30] gmb: tried to interest murphy in my 70-200. no bite. [14:31] gary_poster, Okay, I'll be along when I've poked at this a bit. [14:31] cool [14:31] bac, Hah, unlucky :) [14:33] bac, So, I can confirm your results... it looks almost as though stdout is getting redirected somewhere [14:34] who would do such a thing? [14:34] :D [14:35] Ah, wait. [14:35] Maybe it isn't... [14:37] bac, Let's hang out. I'll start one. [14:40] bac, https://plus.google.com/hangouts/_/ae3a1c590e25a692ca618cd9b43312bf592778ac?authuser=0&hl=en-GB# [14:40] gmb: and i cannot tell if this is a reversion or not, since in trunk the tests are so messed up [14:51] lp.services.webapp.tests.test_error.TestDatabaseErrorViews.test_operationalerror_view_integration failure happened again, which is unusual. If it happens a third time, I'll say it is something new and different, and maybe repeatable... [14:59] gary_poster, I'll just grab a drink then I'll be right with you. [15:00] gmb, heh, ok, was just called away, but will be back in just a few myself [15:11] frankban: how does one get /usr/bin/lp-lxc-ip installed? [15:14] benji: python setup.py install [15:15] frankban: I'm confused. Since lxcinstall requires it, shouldn't it get installed by it? [15:16] benji: detup.py install is actually called by `apt-get install lpsetup`, so the subcommands can assume the script to be there [15:16] s/detup/setup [15:45] gary_poster: do I remember that we wanted to create bugs for the lpsetup cards as we got to them? if so, I'll create one for the one I'm working on so I can attach the MP to it. [15:45] benji, yes, sounds great, thank you [15:50] benji, I looked through the code and thought it looked good at first blush. Maybe we want frankban to do some starter reviews as we all get familiar with the code? [15:50] starter reviews == reviews that help us all get started, since he knows the code very well [15:51] That sounds like a good idea. We essentially did a pre-implementation call this morning because I treated my first branch as a prototype, this branch is the result of that call. [15:57] gmb: any progress? [15:57] bac, Yes. It's fun stuff this. hang on, let me fire up the hangout again... [15:58] rt [16:00] bac, https://plus.google.com/hangouts/_/cf98abf879ccb5bcaa52dc687ba22e9fe6fbabe3?authuser=0&hl=en-GB# [16:07] benji, ok, I'll just review it since you already have a preimp, I guess. Maybe we can ask frankban for a follow-up later. So, my belief/impression is that a lot of this stuff I can skim over because it is just moving code. In that vein, can you help me identify what I should look at more carefully? I'll go through and write what I see, and then you can look (live, or after I've gotten to the end and ping you) [16:07] and see if you think I missed anything. [16:08] tags: ok :-) [16:08] inthost registration: cool [16:08] inithost subcommand is almost all just moving stuff around, except you have a new pattern: [16:08] define the steps as class globals [16:08] so other subclass commands can use them easily [16:09] gary_poster: the inithost and initialize __init__s are the most important part; mostly code just moved from initialize to inithost; also "fetch" was broken out of one of the functions that moved and left behind [16:09] cool [16:12] benji, you say __init__ but I think you mean the "steps" attribute that the base class __init__ works with, yeah? [16:12] gary_poster: ah, right: initialize doesn't have an __init__ [16:13] cool [16:13] or the other way around [16:13] neither one does I think [16:13] they both use the base class I think [16:14] right, that was the prototype branch; it did some nasty things with slicing and dicing the install.steps to generate inithost.steps [16:15] bac, Changes pushed to the ~yellow branch. [16:15] thanks! [16:19] benji, I approved, mentioning I'd like to see a testing story in the next branch. Do you have thoughts on how to do that? [16:19] benji, gary_poster: I have some comments, do you prefer here or in the review? [16:20] gary_poster: frankban and I had a potentially fruitful idea of using ephemeral lxc instances to run whole-machine tests of the subcommands [16:20] we should think of why we want to test before deciding too firmly on what and how, though [16:20] frankban, in the review definitely, and here if you think it is valuable [16:24] benji, why: thoughtful and good point. balancing goals with test speed might or might not be interesting too [16:25] (that is, I expect the lxc instance tests will be much slower than basic "are we doing the right thing" tests, but the former is much more thorough, and I question the value of the latter beyond a Python syntax check) [16:26] but I think we need to put the workaround in for 1014916 next; and then make our testing story, before we do anything else. [16:28] The ephemeral tests, unless they install launchpad and verify that it works, which would take an hour or so, would actually not be much better than the mock-based "did we do what we expected to do" tests... [16:36] benji or gary_poster, would one of you have a moment to review the (excellent) work gmb and i have done to zope.testing? https://code.launchpad.net/~yellow/zope.testing/fix-tests/+merge/111442 [16:37] * bac notes with horror i'm the lead contributor now to the lp:zope.testing project. ugh. [16:38] I currently vote for the mock approach. [16:38] bac, reviewing [16:39] thx [16:50] ooh, look, we can fix 1013921 without changes to zope.testing! We just have to add -vv in our call to bin/test --subunit! [16:51] cool [16:52] hey bac, where does out zope.testing for take over __stdout__ for the subunit stream? [16:54] gary_poster: in options. when it instantiates the subunit formatter [16:55] ah! right, thanks [16:57] bac, __stderr__ is overridden there but not __stdout__ right? [16:59] (I'm looking at p14 while staring at the MP) [17:00] gary_poster, benji: I've added suggestions to the MP, and I hope they makes sense, because it's 7pm and hot as hell here in Italy... [17:01] frankban, :-) I skimmed them and they made since to me. Go relax. Thank you! [17:06] gary_poster: yes, sys.__stderr__ and sys.stdout are overridden there. and sys.__stdout__ is used by the subunit formatter as its output stream [17:07] bac, oh, I thought we had to override __stdout__ too but I was mistaken? [17:07] we temporarily override it in some doctests [17:08] ah, ok [17:09] gary_poster: :-) thanks and have a nice evening. [17:09] thanks frankban; stay cool [17:10] you too :-) [17:12] benji, I still see some stdout in subunit. :-/ try "lp.codehosting.sshserver.tests.test_session.Test_WaitForExit.test_dataReceived_bad_data" and you'll see "Got invalid exit information: 'bogus\\n'" [17:13] I'll make a card for it. no biggie, hopefully [17:13] goodness [17:16] (and in a similar "ugh" vein, "lp.services.webapp.tests.test_error.TestDatabaseErrorViews.test_operationalerror_view_integration" now fails consistently :-/ [17:16] ) [17:16] (it was intermittent; I'm assuming the cause is different) [17:42] bac, I approved https://code.launchpad.net/~yellow/zope.testing/fix-tests/+merge/111442 with some relatively small requests [17:42] gary_poster: cool [17:45] gary_poster: interesting suggestion. i'm happy to make the change. [17:48] cool [17:54] gary_poster: i'm also making this trivial change: http://pastebin.ubuntu.com/1053030/ [17:55] as you can see, options.module can be True even if -m is not specified so the warning gets emitted erroneously sometimes [17:55] bac, ah, thanks yeah. I've seen that behavior. +1 [18:00] hmm, new drobo with thunderbold/usb3. i'll bet that will be NotCheap [18:01] "thunderbold" -- i should trademark that [18:03] bac, naah, there's a domain squatter there already http://thunderbold.com/ [18:03] dang [18:03] benji, I haven't taken lunch yet, and I'm feeling peckish. And like I need a break. Do you mind if we start a bit late on our call today? [18:04] gary_poster: sure, that's fine [18:04] thank you [18:39] anyone know how to run the docests in lpsetup? nose doesn't run them by default and I get ImportError: No module named pkg_resources when I use --with-doctest [18:56] arg! the doctests were decoys! there are equivelent unit tests that /are/ actually run [19:19] yes, I think he left them as docs [19:19] benji, hey [19:19] hey [19:20] benji https://plus.google.com/hangouts/_/2f068ac30c507beee7b6527c6dd4e8e857f25c81?authuser=1&hl=en-US ? [19:29] gary_poster: when shall we talk? [19:29] bac, would 20 min from now be ok? [19:29] gary_poster: post review changes: http://pastebin.ubuntu.com/1053150/ [19:30] gary_poster: ok. 16:00 is better [19:30] cool bac, will talk then [20:02] * bac goes for beverage [20:06] yo yo gary [20:10] gary_poster: you about? [20:10] bac hey https://plus.google.com/hangouts/_/2f068ac30c507beee7b6527c6dd4e8e857f25c81?authuser=1&hl=en-US sorry [20:44] bac, lines 191-193 of http://pastebin.ubuntu.com/1053150/ maybe unnecessary? [20:46] gary_poster: yep. good catch [20:47] cool [20:48] gary_poster: actually it is called at line 213 [20:56] gary_poster: if you are so inclined you can review the changes I made in response to frankban's comments (https://code.launchpad.net/~benji/lpsetup/add-hostinit/+merge/111429) or we can wait for him to do so tomorrow [20:58] benji, looking [21:06] benji, I have an incident going on again in ops. It looks like you did what you asked but I'm pretty distracted. I'm happy to approve if you feel comfortable, but I can't focus. :-( [21:06] gary_poster: no worries, lets let it ride until tomorrow [21:06] cool, benji, ttyl