=== med_out is now known as medberry === medberry is now known as med_out [05:11] cinerama, https://bugs.launchpad.net/bzr/+bug/255687 [05:11] Ubuntu bug 255687 in Bazaar "log and annotate should work on removed files" [Medium,Confirmed] [06:02] still dark outside... [06:02] its dark outside here too! [06:06] hi wgz [06:06] beautiful day here [06:13] * fullermd searches around to find a window, and discovers darkness there too.... [06:14] :) [06:14] It's like a giant plague, taking over! It probably covers half the world by now :( [06:37] okay, nearly time for bus [06:38] cow-orking today. [06:39] oh great [06:39] i'm out for a bit to play squash, probably back later [06:48] poolie: i think I mentioned earlier, but they put a 1hr inetd timeout on savannah, i believe. [06:49] I don't think it is from bzr, I think it is from their configuration. [06:49] morning all [07:45] Hello. I just wanted to find out what the best way of moving/copying a bzr checkout to another PC is? [07:45] (I'm on a slow Internet connection and don't want to have to check it out again) [08:07] good morning. [08:10] kirkland deserves some praise, fix up etckeeper in oneiric from yesterday [08:23] hi mgz, jam [08:23] jam, oh it kills it after an hour regardless? [08:23] poolie: I'd have to track down the original message, let me check [08:25] https://bugs.launchpad.net/bzr/+bug/824797 [08:25] Ubuntu bug 824797 in Bazaar "bzr serve doesn't drop idle/dead connections" [High,Fix released] [08:25] I may have just been confused [08:25] because he closed the connection himself after 1 hour [08:34] i think that's imapd, or maybe inetd, closing it [09:25] hey, I'd like to have a working Debian dpkg import today for Ubuntu precise opening, and apparently it's blocked on bug 714622 [09:25] Launchpad bug 714622 in Ubuntu Distributed Development "import fails when lp branch has been push --overwrite'n" [High,Confirmed] https://launchpad.net/bugs/714622 [09:26] I've deleted the 1.16.0~ubuntu8 tag from the natty branch, since AFAICS it doesn't correspond to anything that was ever accepted into the archive [09:26] could somebody please requeue it and see if that's enough to fix this? [09:30] hi cjwatson [09:30] i'll do that now [09:30] mgz: hi meet cjwatson [09:31] hi cjwatson! [09:31] heya [09:32] cjwatson: do you mean 'an import of dpkg itself'? [09:35] I mean package-import [09:35] http://package-import.ubuntu.com/status/dpkg.html [09:35] sorry for slow response, I'm doing Ubuntu 11.10 publishing [09:37] it's running now [09:38] btw multi-tarball inputs should be supported fairly soon [09:43] cjwatson: it seems to be failing on [09:43] AssertionError: ('steve.langasek@linaro.org-20110325000519-s1f74q0j9esmvs1s', '0d73da5a0af194cfc458d9ec624b6f42209f35f7') != (u'james.westby@ubuntu.com-20110324170447-1m0lwg8kvwt6yfli', u'c7a0e118cb0af317cf3e7af7874192fa4c405d0a') for dpkg 1.16.0~ubuntu6 in natty, something has changed [09:44] could that be a similar problem? [09:45] hmm [09:45] ok, I'll have another look in a minute, worst case is deleting a tag likely to solve the problem? [09:46] it seems possible [09:46] hmm [09:47] I don't see the latter revision id in either lp:ubuntu/dpkg or lp:ubuntu/natty/dpkg [09:47] where might it be coming from? [09:47] hm, let me see if i can work out more [09:49] cjwatson: i'll try a full reimport [09:53] poolie: um [09:53] poolie: will that trash existing revision ids? we are actually using lp:ubuntu/dpkg [09:53] definitely don't want to lose that [09:54] i killed it [09:57] cjwatson: the 'latter' revision_id is probably the one the package importer stores in its sql database tracking what has been imported. [09:57] It is the one that the importer had imported to match the given debian revision. [09:57] well, ubuntu revision, obviously :) [09:59] lifeless: you mention that nose is evil, do you have a recommended test runner? [09:59] is it possible to excise that, or would that cause other problems? [09:59] cjwatson: it is possible, though I don't specifically know a way that isn't hacking the sql database directly. [10:00] lifeless: I'm trying "py -m testtools.run" but it is a bit wonky [10:00] like '-v' for verbose doesn't do anything, etc. [10:01] i think we do need to excise them, and that would be a good fix for this [10:01] jam: I usually use testr [10:01] cjwatson: i need to go and get S from uni in a bit, but jelmer and mgz will be here soon and may be able to fix it [10:02] jam: which builds on py -m subunit.run [10:02] (which extends testtools.run) [10:02] or maybe maxb? [10:02] ok, thanks [10:03] I can look at it at lunchtime [10:03] (London lunchtime, that is) [10:04] thanks [10:04] cool, I'm not sure which bits to poke with package stuff yet [10:04] maybe max can tell you about it [10:18] night all [10:19] night poolie. [10:34] night poolie [10:34] mgz: so, after tons of bug filing I found the problem :) [10:34] (why I was running more tests) [10:34] the old, non-scenario code was using mixins, of the form class TestsForX(Mixin, TestCase) [10:35] and I accidentally had the same "TestsForX" class name in two places. [10:35] So scenarios really did give me better coverage, as it is a lot harder to make the mistakes you can make with mixins [10:35] figuring that out was very clumsy with the bugs I ran into. [10:37] jam: ha, that's actually really interesting [10:39] mgz: also, if I use the 'load_tests' hook, then 'discover --list' works correctly, since the suite of tests is determined at load time, rather than at run time. [10:39] but -v is still broken, etc. [10:39] (and I had to patch --list to make it work :) [10:42] mgz: also lots of other oddities. Like testscenarios.TestWithScenario inherits directly from unittest.TestCase rather than testtools.TestCase [10:43] which matters when you try to run with --verbose [10:43] because TextTestRunner uses test.shortDescription() to determine what name to print [10:43] and unittest.TestCase returns the docstring or nothing [10:43] (thus using str(Test) for the representation) [10:43] while testtools.TestCase returns self.id() as the short description [10:43] so TestWithScenarios was doing the multiplying correctly, but it wasn't showing the new test ids. [10:44] because str(test) wasn't changing. [10:45] I'm currently reviewing our internal code hosting stuff for bzr. Anyone know of any tutorials for a decent setup (pref with stuff like hookless emails etc) or a good way of working within a team? [10:45] What we've got a the moment is a kind of "mish-mash" repository (basically bzr+ssh) and was wondering if anyone knows a better way of doing it. [10:46] I'd use something like LP's code-hosting if it wasn't a proprietary development [10:48] canonical-bazaar / LOSA: Please stop the package importer for Oneiric [10:48] (per http://wiki.ubuntu.com/ReleaseProcess) [10:48] cjwatson, ok, I'll take a look [10:50] Mez: launchpad is open source [10:51] Riddell: indeed it is - however - it's not exactly the best solution for us as a whole. [10:52] Mez: thumper made a standalone lp based codehosting system [10:52] If it were easy enought to pull out the code-hosting stuff on it's own, that'd be fine - but the rest of it would just get in the way (and the setup/dependencies to get LP running locally is a bit of a pain) [10:52] lifeless: oh, cool - that's exaclty the kind of response I wass hoping to hear :) [10:54] its designed for school projects and soon, but should be close enough to consider [10:54] yeah - just looking over it now. I have a feeling that I might even be able to contribute to it. [10:56] (Mainly, I see the ability to add stuff like code review/merges/pqm-esque stuff) [10:56] (I do like LP's Pqm-esque stuff) [10:56] jam: do you know how to stop the package importer for Oneiric as cjwatson asked? [10:57] gnuoy and I are... not completely sure what to do [10:57] james_w should but probably isn't up yet [10:58] hi [10:58] aha [10:58] I wish I wasn't :-) [10:59] mgz, gnuoy, it's mass-import on jubany, it has an init script [11:03] cjwatson: done (with more than a little help :) [11:04] thanks! [11:08] Ran 23 tests in 4.600s [11:08] OK [11:08] bah [11:24] Mez: I plan to do something like codehosting system eventually, basically by integrating bzr, sphinx, roundup and buildbot [11:25] lightweight, componentized & extensible [11:25] but it will take some work, mainly on the sphinx front I think [11:26] ccxCZ: actually - sloecode looks promising - however, I think zope is broken in Oneiric. [11:27] http://paste.ubuntu.com/707279/ [11:32] not zope, just older zope.interface than paste requires it seems [11:32] either upgrade or use different wsgi launcher (I use uwsgi) [11:33] zope.interface is independent of the rest of zope and used by other projects aswell, eg. twisted [11:36] okay, looks like the basic idea of the morning works: [11:37] I shall now go and have lunch [12:08] qbzr diff viewer ; any way to make the view options sticky or preconfigure them? [12:09] Specifically I want to make "ignore whitespace" sticky because the project team I'm working with have annoyingly different whitespace settings in their IDEs (yes, I should just slap them, but that's not an option. Sadly.) === beuno is now known as beuno-lunch === med_out is now known as medberry === beuno-lunch is now known as beuno [13:06] awilkins: you can pass the option on the command line to qdiff, but I don't think there's a conf option for it [13:07] awilkins: could file a bug against qbzr for that, but really sounds like you need to do some slapping [13:08] * fullermd is occasionally irritated at the unconfigurability of qdiff too. [13:08] If I used it more often, I'd have local hacks sitting around in the code like there's no tomorrow. [13:12] I think there's some blame for bzrlib for various parts not quite being flexible enough for reuse in a gui [13:14] Alas, I can only limit my slapping to people I manage [13:16] These are the guys who thought they were OOOO so clever to sit around wasting hours discuss code formatting standards and then don't stick to a consistent IDE config (as predicted by Yours Truly - the only config you should use is the DEFAULT one) [13:16] Unless your IDE is clever enough to infer formatting rules from the files it opens (which Eclipse isn't, apart from line endings, it seems) [13:19] If you just converted to APL, you wouldn't have to worry about formatting rules... [13:24] hi [13:25] i need to install from source, but i don't find the right download [13:29] under downloads on launchpad, the last one should always be a .tar.gz source package [13:29] eg, https://launchpad.net/bzr/+downloads [13:30] *ie [13:32] thank you, i didn't see the right bar, thought it would have to be under 'code' [13:49] there i am again... the downloaded package says python 2.6 needed... on the pages all i've seen is 2.4, whioch i have to work with [13:49] you need 2.3 then jay567 [13:50] bzr-2.3.4.tar.gz same page, further down [13:51] mgz: ;-) thanks [13:52] :) [14:10] next obstacle: ImportError: No module named distutils [14:10] in setup.py [14:14] jay567: ...do you have a full python installed, or just a core system package? [14:15] distutils is part of the standard library and should be packed in "python" [14:15] i want to run bazaar on a server of my university... [14:16] maybe i can get them to install some modules for me... [14:16] and you have no ability to request packages? [14:17] if they can give you a working python (and ideally some build tools, pyrex and gcc), then you can do a user install of bazaar [14:17] or they might just be able to install the bazaar package for you. [14:19] jay567: as a final option, you may be able to get away with not having bazaar there as you can still work over sftp for instance [14:22] what do you mean by user install? isn't this what i tried? (python setup.py install --home $HOME) [14:23] Well, you can run it in place without installing. Though you may not even be able to build the C extensions without distutils, so you may be stuck with the slower pure-python variants... [14:24] jay567: yes, that kind of thing. [14:26] as fullermd says, you can get away without using install at all, but without the extensions you may not be better off then you would be just treating the server as a filesystem where you can put branches [14:27] Well, even without the extensions, I'm sure using it for bzr+ssh would beat the crap out of sftp, unless the server itself is mind-zonkingly slow and very [network-topologically-] close by. [14:27] Dumb servers are occasionally a necessary evil. But they're way more the short word than the long one :p [14:27] * mgz defers to the greater wisdom [14:27] ;) [14:27] ehm [14:28] (it's not greater wisdom; I just take any excuse to say "mind-zonkingly", 'cuz it's way fun. You should try it sometime.) [14:29] Try just running 'gmake' in the dir, see if it can do enough to build the C extensions. I'm betting 'no', sadly... [14:29] so if i commit to a 'server', my local bzr calls the server-bzr via ssh... yes? [14:30] gmake and make just return the distutils error [14:30] There are a lot of variations, depending on details of your setup. But in practice I'd say 'no'. [14:30] Just think of the server and your local as separate branches, that you sync up occasionally (probably only working on one side at a time) [14:30] So you commit locally, then you 'push' to the server, which calls that bzr over an ssh channel. [14:30] yes [14:31] Yeah. Figured. Sounds like a pretty wacked out python install to not have distutils though. [14:31] i think they are pretty afraid of the students, every student from the computer science faculty has a shell account there [14:32] As well they should be :p [14:32] jay567: try asking nicely for a proper python and build tools. [14:32] * fullermd remembers being a CS student once, and it would be a pretty dumb staff/prof who wasn't scared to death of him :p [14:32] :P [14:33] so, what do I need to do to get the 'push' working? tell my local bzr where to find the bzr binary on the server? [14:35] Well, there's an env var I'd have to look up to give the remote path. [14:35] Easier probably just to put it in your $PATH though. [14:35] yes [14:35] You could e.g. have a ~/bin/bzr symlink that points to ~/src/bzr-a.b.c/bzr [14:35] (assuming ~/bin/ is in your path of course. But naturally it is, surely) [14:36] i see [14:37] It should Just Work(tm) then. With the python fallbacks, so slower in various operations than would be possible otherwise, but it'll work. [14:38] i don't thnk performance is a big issue [14:49] where do i put the config files in this case? === mnepton is now known as mneptok [15:09] and there is precise :-) [15:09] the package importer has just seen it and created all the jobs to create the branches, it will get to them when it is started again [15:16] james_w: should be be started again? [15:16] *it be [15:17] we should get notification when they are ready for it === zyga_ is now known as zyga-afk [15:17] I expect it could be started now, but let's wait === mrevell_ is now known as mrevell [16:21] jam, poolie: as far as bug #714622 is concerned, I think the correct course of action for *all* the affected packages is to take the revid currently on the branch as authoritative... bad form as it may be, if someone has bzr push --overwritten (or otherwise changed the tag), they've done so for a reason. As long as someone's doing db surgery, could you do so for the other 48 affected packages on http://package-import.ubuntu.com/status/ ? :) [16:21] Launchpad bug 714622 in Ubuntu Distributed Development "import fails when lp branch has been push --overwrite'n" [High,Confirmed] https://launchpad.net/bugs/714622 [16:21] (this would help me for sysvinit in particular) === deryck is now known as deryck[lunch] [16:31] Hi guys. How can I 'merge' / squash 2 commits into one? [16:43] CaMason: "bzr uncommit; bzr uncommit; bzr commit -m "Commit message for the squashed commit" === zyga-afk is now known as zyga [17:32] wgz: hi [17:36] jelmer: hi, just back from cow-orking [17:37] I see from your summary page the builddeb test tweak worked for maverick and natty [17:37] wgz: did you perhaps request a rebuild for bzr-builddeb? [17:37] I'm a bit surprised by the upload error [17:37] have you got a bug for the other lucid failures? I guess it's some known issue with the debian helper script not existing in that ancient a release? [17:37] jelmer: nope, I shouldn't have touched any buttons [17:38] wgz: I've fixed those I think [17:38] http://people.canonical.com/~jelmer/recipe-status/bzr.html === deryck[lunch] is now known as deryck [17:39] I had to backport cython for hardy as well [17:39] ha, I see what you mean about the 'failed to upload' [17:40] is that not just something about today being release day? [17:40] wgz: that usually happens when somebody hits the "Request a build" button [17:40] and the same version gets built twice [17:40] (looking at the page) [17:41] right, food time for me. === yofel_ is now known as yofel === aj00200 is now known as aj11u10 [21:48] Anyone know why the UDD importer is not running? [21:49] I stopped it. [21:50] and james_w said "let's wait" when I asked about restarting after the release was done. [21:50] it may well now be time to start it again. [21:52] if you have any idea, please say maxb :) [21:52] I'm not clear why it needed to be stopped at all? [21:53] cjwatson was asking about getting the dpkg import fixed today, but I'm confused, as it appears to have last run successfully [21:54] after that, another item on his checklist for the release was to stop the Bazaar importer [21:54] ...I don't have any more information than that, I'm afraid [21:56] Oh, wait, it'll be so it's not trying to do stuff whilst LP's branch-distro.py process is executing [21:57] exactly [21:59] we don't want to be pushing up branches for the new series before that has run, otherwise it will lead to a huge increase in storage used [22:00] what's the run time on that? :) [22:00] 18397 outstanding jobs [22:00] it [22:00] it's around 18 hours IIRC === medberry is now known as med_out [22:01] Hmm - the web UI currently says precise has no packages [22:01] jam, thanks for the package import status notifications on branch, it's very nice to have [22:02] Does that mean the initialize-from-parent / branch-distro stuff is not under way yet? [22:03] maybe there was a bug in i-f-p, let me check my scrollback [22:05] I suspect a cache of some kind, looks like only the counts on https://launchpad.net/ubuntu/precise are zero [22:07] there are definitely publications for precise - which explains the UDD queued jobs, actually [22:08] maxb, yeah [22:09] seems like there were some problems with publishing, but I can't tell if they were resolved [22:09] the people in question left for the day though, so it may be completed tomorrow [22:09] branch-distro has been started though [22:25] hi all [22:27] hey poolie [22:30] hey poolie, wgz [23:52] hi there [23:53] morning poolie [23:54] hi there [23:54] can anyone update me on the udd importer? [23:55] i guess there is scrollback, perhaps i'll just read that