[10:55] morning gmb [10:55] hi frankban. welcome back. hope you had fun. [10:56] hi bac, thanks, and yes, the EuroPython was cool [10:56] Hi bac. [10:57] Also: hi frankban; I forgot you were back this morning. [10:57] gmb: can we call you Graham Green(e) now? [10:57] Arf. [10:57] buh dump dump [10:57] though i do quite like his books [10:57] Ah. [10:58] So you mean the writer. good. [10:58] There's a few of them :) [10:58] ah, yes [10:58] oh, i don't know the others [10:58] bac, Native American actor for a start. [10:58] Also a British comedian [10:58] though i think i have tried to read 'brighton rock' about three times and never can finish [11:11] bac, frankban: Can one of you take a gander at https://code.launchpad.net/~gmb/lpsetup/death-to-doctests-in-subcommands/+merge/114393? [11:22] gmb: you have a conflict in the MP [11:22] Crap. [11:23] Forgot to merge trunk. [11:24] gmb: btw, i've claimed the review [11:25] gmb: can you do a very quick review for me? https://code.launchpad.net/~bac/lp-tarmac-configs/tarmac-approvers-list/+merge/114397 [11:25] bac, Sure. [11:25] gmb: please vote but do not approve the MP. that way it will be self-testing [11:25] bac, Fixed conflicts and pushed. [11:25] bac, Okay. [11:26] gmb: why not add the cleanup of the file in the setUp? [11:27] bac, Because not all the tests create the file. The cleanup will then error because the file doesn't exist. [11:27] gmb: also, i think if you make one big mask by ORing together all of the expected file modes and then XOR with the actual file mode you can do the test in one go. [11:28] bac, Ah, good point. I'll try that. [11:28] gmb: gotcha [11:28] bac, I've voted to Approve. [11:29] gmb: also, the text conflict you just fixed may break your tests. i removed the insertion of the trailing \n. even if your tests don't break, you could remove the strip() and ensure the contents are the same. [11:33] Noted, thanks. [12:08] frankban, welcome back! [12:08] gary_poster: thanks, it's good to be back [12:08] bac benji frankban gmb, http://tinyurl.com/yellowsquad [12:15] https://tbe.taleo.net/NA3/ats/careers/requisition.jsp?org=CANONICAL&cws=1&rid=501 [12:45] Przemysław Pająk [12:54] gary_poster, https://tbe.taleo.net/NA3/ats/requisitions/RequisitionView.jsp?act=view&id=502&et=REQU&actionId=1. [12:54] I don't think you're on it as a hiring manager, though, so you might need to ping Alice. [12:54] Oh ok thanks gmb [12:55] * gmb -> lunch [14:00] gary_poster: for testing purposes, would you vote 'Approve' on this MP that graham has already reviewed? https://code.launchpad.net/~bac/lp-tarmac-configs/tarmac-approvers-list/+merge/114397 [14:00] bac, done [14:01] thanks [14:25] bac, dunno if you saw frankban's blessing for removing public and private key args to inithost. I made a card. Probably easy to do (or you could fold it into your current work, if that makes sense) [14:25] great [14:55] gary_poster: have you found something about -s not returning a flat list? [14:55] frankban, no sorry, on calls [14:56] gary_poster: np [15:32] frankban, if, as a non-root user, you pass multiple steps to a command that must switch to root access, and it works, just move on. I looked for the email and it didn't help. What I just said is what I remember [15:32] gary_poster: ok cool [15:41] hi frankban, if i add 'ignore-files=disabled*' to the lpsetup setup.cfg file then nosetest does not report the test failures and complains about distribute. i've poked at it but cannot figure out why it is unhappy. thoughts? [15:46] were we all aware, where 'we' == 'us' - 'me', that the handler listing was order-dependent? [15:51] hi benji, in the README.rst file for pocket-lint you specify sinzui's ppa. it appears in the normal repos so is there any reason to require the ppa version, other than shininess? [15:52] bac: it may be that in flailing about to find the right package name I got confused and believed that it was only in sinzui's ppa. If it is available in the regular places, then that is fine too. [15:52] cool [15:52] We can delete the PPA bit or make that a sidebar ("If you want an ultra-fresh version of pocketlint, get it from...") [16:06] bac: same error here, if the goal is to ignore TestCase.disabled_* tests you can try to add to setup.cfg: exclude=handle_testing|disabled_ . Ignoring whole files is broken here too :-/ [16:06] frankban: i'd wanted to disable whole test files [16:07] frankban: chmod +x works, but it is very easy to overlook [16:16] bac: interesting, it seems to work if you comment the line with-doctest=1 (in setup.cfg). Could you confirm? [16:18] frankban: yes, it works for me the same way. i note that passing --with-doctest on the command line causes the same problem [16:19] in conjuction with ignore-files [16:34] gary_poster: can you sanity check this diff for my unbreaking lpsetup branch? http://pastebin.ubuntu.com/1086450/ [16:35] bac, on call; looking [16:41] bac: that distribute error is not print by nose, it's present in distribute_setup.py#153 [16:42] * not printed [16:42] frankban: yes, but it is odd it is triggered dependent on the nose options [16:47] bac: it works for me if I add "where=lpsetup" to setup.cfg (or -w arg) [16:51] frankban: that is odd [16:51] thanks for looking into it [16:53] np, and I agree it's odd [17:05] bac, why did you remove sinzui's ppa? [17:05] I think it is still necessary [17:05] (remove from README) [17:05] pocket-lint is in the precise repo, i'm pretty sure [17:06] bac, was not for me [17:06] maybe I installed it with wrong name or something [17:06] Package: python-pocket-lint [17:06] Versions: [17:06] 0.5.30-0ubuntu1 (/var/lib/apt/lists/us.archive.ubuntu.com_ubuntu_dists_precise_universe_binary-i386_Packages) (/var/lib/dpkg/status) [17:07] ok bac, dunno. I just had to do this in the past few days. no idea. [17:07] not worrying about it [17:07] * bac just trying to streamline [17:08] gary_poster: i'm reverting my vm to the clean snapshot and will then try a install-lxc. [17:08] ok [17:09] well, i have reverted but i'm not bringing my baseline vm up-to-date. i think it was pre-release precise so i have lots of packages to install [17:15] bac, I have only skimmed so far (looks good) but Karyn tels me I have to take lunch and have a break, and she is right [17:15] I'll review and ping when I return of that's ok [17:15] if [17:27] gary_poster: enjoy your lunch. is it really 1:30? [17:40] why did we move initialize_lxc into init-host? that seems like an odd thing to want [18:27] bac, yeah, it was. Been busy. BTW, I am skeptical that it will accomplish anything, but you might want to look in on #launchpad-dev [18:28] benji, the biggest driver for that move is the data center. [18:29] gary_poster: who was that masked man/woman? [18:29] benji, in the data center, the webops people will not have root in the lxc host [18:30] but they will (hopefully) have root in the container [18:31] hmm, but "normal" people won't want that, right? wouldn't that be better as a seperate subcommand? [18:31] Therefore, making init-host able to initialize the Launchpad "host" even when it is an LXC container gives them the maximum amount to share with us [18:31] beyond that, technically it is initializing the Launchpad host whenever it is an LXC container [18:32] benji, notice that it will only happen if it is running in LXC (call_initialize_lxc I think?) [18:32] ah! that's different [18:32] And everyone should want it when their Launchpad host is an LXC container [18:33] bac, I dunno. Laura mentioned that we were looking for juju help, on the basis of our "please help!" section in the blog [18:33] this person replied that they would help with juju [18:33] oh, so there is some mechanism in the framework that constructs function names from step names and calls "call_initialize_lxc" and uses that to determine if the step should be called or not [18:33] At the least we applied indirect pressure to SpamapS ;-) [18:34] benji: yep [18:34] if they exist [18:34] that's... irritating <0.5 wink> [18:35] * benji hasn't enguaged in frational winkery in quite a while, it's refreshing [18:37] bac, I missed Bulgaria in my search :-P [18:38] sorry bac. I meant gmb, ^^^ [18:38] gary_poster: i'm sure i should understand that [18:38] oh [18:38] maybe not [18:38] :-) it was re candidate search [18:38] bulgaria might be a new country for us [18:39] as will PR...sort of [19:18] gary_poster, Is the TL call still on the conference no., or is it a hangout these days? [19:19] gmb, hangout. I'm the whatever it is today, so I'll include you in the ping with the url [19:19] Thankee. [20:33] bac, finally, to http://pastebin.ubuntu.com/1086450/ : [20:34] docstring in line 202 maybe should include note that the command is desired to go away? Maybe? Wdyt [20:34] gary_poster: cool. my testing revealed some problems. re-doing [20:34] bac, oh ok [20:34] should I wait [20:34] gary_poster: proper MP is up, if you're interested [20:34] sure bac [20:34] no, they were trivial (but fatal) [20:34] ok cool [20:34] cannot test install-lxc due to lp-setup in container from package issue [20:35] i made a card for that [20:36] gary_poster: https://code.launchpad.net/~bac/lpsetup/fix-inithost-install-lxc/+merge/114486 [20:37] lp-setup in container from package: bac, right, thanks. (There is a temporary work around fwiw, which involves running individual steps and installing the code manually inbetween starting the container and running all the commands in the container) [20:38] (and I have a hack in mind to make this work in the future) [20:38] gary_poster: i'm doing the workaround [20:38] cool [20:40] gary_poster: my tarmac branch didn't land b/c rockstar's tarmac was broken. he's taking care of it now. [20:41] bac, heh, great [20:41] yummy tarmac chow [20:44] bac, make_launchpad has install arg. Does anyone ever not pass True to it? [20:44] AFAICT no [20:44] which suggests a cleanup :-) [20:45] Is there a compelling reason to separate setup_launchpad and make_launchpad? unclear. don't care a lot though. [20:47] As already mentioned, I think IWBNI finish-inithost had a docstring (and help text?) that indicated that it was slated for eventual removal [20:49] bac, install_lxc should not include inithost.SubCommand.initialize_lxc_step [20:49] that is run in the lxc container for this story, when we run setup_lxc (which calls inithost in the container) [20:50] gary_poster: IWBNI? [20:50] it would be nice if [20:50] oi [20:51] :-) [20:51] bac, otherwise looks good to me [20:51] gary_poster: ok. you noting that stuff in the MP? [20:51] bac, yes [20:52] thanks for the suggestions [20:52] thank you! Nice to have this working again [20:55] bac, review made [20:55] gary_poster: wait, i think finish-init-host is hosed [20:55] bac, ok [20:55] bac, could it be inithost.SubCommand.initialize_lxc_step ? [20:55] it is running locally, not in the container. no one calls it in the container as far as i can tell [20:55] As I said, that shoud not be run [20:55] oh [20:56] yes, that needs to be run with ssh in the container. Sorry for missing that. [20:59] gary_poster: well it needs to run in the 'host', whether that is the container or your 'real' machine [20:59] bac, right. for install-lxc, it needs to be run in the container, I mean [21:00] gary_poster: right, but if you are running the steps manually you can't do it unless we add a wrapper [21:00] why bac? I don't understand. Woud quick call help? [21:00] l [21:01] sure [21:01] bac, I'm in yellowsquad