[00:11] ah running ubiquity finally... virtualbox workaround for 2.6.27 really slows things down though... [00:19] partman/exception_handler DESCRIPTION Error opening /host/ubuntu/disks/swap.disk: invalid argument [00:19] was there any change in partman preseeding? [00:46] preseed: /host/ubuntu/disks/swap.disk 100 200 200 linux-swap method { swap } format{ } . === foka_ is now known as foka [03:54] Not sure if this is germain to this channel, but I did a bum install via Autoconfig... & I need to remove what did get installed. Any suggestions... directions? Thanks [03:56] HenryHeron: reformat the partition? The installer can do this automatically, if you select the "format the partition" option when selecting the partition. [03:56] Err. Missed "Autoconfig". I suspect the above advice is not correct. [03:57] persia: Yikes, no, not of the system... a PROGRAM. :) [03:58] Yeah. Probably not germane to this channel. :) I'd suggest looking at the Autoconfig docs, and seeing if it tracks the install somehow, or provides a log of the partial install. [03:59] persia: I can get the terminal to track the files... but when I do an apt-get remove, it "says" not installed. [04:00] HenryHeron: I suspect Autoconfig doesn't use apt-get, but I'm not actually familiar with Autoconfig. [04:01] persia: this is probably true, I too suspect. That's why I need someone more knowledgeable than I (a relative newbie to linux/ubuntu). [04:02] persia: Perhaps I'll chime in at the ubuntu irc. Thanks for the reply. [04:04] HenryHeron: Good luck. I'm sorry I don't know more about Autoconfig. [04:25] Someone say autoconf? [08:20] have an issue in partman: open_dialog OPEN /host/ubuntu/disks/swap.disk -> invalid argument [08:20] http://paste.ubuntu.com/46539/ [12:21] Goosemoose: just /etc/init.d/likewise-open start; I'd recommend that you stop it after joining, too [12:22] xivulon: method{ swap } not method { swap } [12:22] the whitespace matters [12:23] cjwatson: the d-i rebuild hung for a while from LP issues, but it's all built now. Thanks. [12:23] ah cjwatson thanks [12:23] Goosemoose: and stop putting /target in your late_command script. It shouldn't be there. [12:23] unfortunately I just wasted my xp vm :( [12:23] Goosemoose: sorry, that was confusing [12:23] Goosemoose: your late_command should have chroot /target /post_install_tasks, as I said [12:23] Goosemoose: post_install_tasks should NOT mention /target [12:24] Goosemoose: that's why you're getting an error about /target/etc/nsswitch.conf not existing [12:24] persia: oh good. er, I think I did fix the last bit of it and then forgetfully go to bed, sorry ... [12:25] cjwatson: No worries. I was a little surpised to get up today and discover my image was broken, but as the only user who claimed they wanted to test this weekend is in UTC-7, I'll be OK as long as I can get a 20080913.1 image built soon. [12:26] persia: it's on its way now [12:27] Goosemoose: from your pastebin, it actually looks like something already tried to run '/etc/init.d/likewise-open start' but it didn't work. You'll have to dig into that yourself to find out why that might be ... [12:28] cjwatson: Excellent. I didn't know you could also press that button. Thank you :) [12:28] * cjwatson has all the cdimage buttons [12:30] * persia scrounges for more to add to the collection [14:04] cjwatson: A couple oddities, for which I'm hoping you can give me guidance on what buttons need pressing. On the ubuntu-mid 20080913.1 image, the partman-partitioning changes to enable lpia are still missing. Also, for the ubuntu-lpia alternate CD, 20080913 failed to build because of what looks like d-i skew. [14:41] persia: ubuntu-lpia is my own stupid fault for not bumping the kernel-version field in the seeds. fixing now [14:42] persia: ubuntu-mid needs a ubiquity upload before the partman-partitioning changes are effective [14:42] Cool. That one's easy then. And the partman-partitioning changes? [14:42] OK. That gives me new incentive to fix the ubiquity bug I found :) [14:43] I can do the ubiquity upload now if it would make things easier [14:44] Well, it means I can make sure the install works on a non-hacked image, which is easier. On the other hand, it reduces my incentive to fix the problem with a null self.blacklist. [14:44] If you've nothing else keeping you busy, and feel like it, an upload would be nice, but it's not critical to me. [14:44] More important was understanding what needed to happen. [14:46] why would you want an empty self.blacklist? [14:47] You wouldn't, but it happens if you happen to either not have the filesystem.manifest files or have filesystem.manifest be identical to filesystem.manifest-desktop [14:48] ah [14:48] The point being that when this occurs, ubiquity ought not crash. [14:48] It's just a robustness fix, rather than a usability fix. [14:48] shall I just fix that now? it's trivial ... [14:48] obviously I can let you do it if you'd rather [14:49] If you like. I was actually enjoying the dig through ubiquity internals as an educational exercise, but there are other things I could be dong. [14:52] (and I'll admit that fixing this isn't at the top of my list) [14:52] well, the problem looks to be that there's one return path from generate_blacklist that doesn't set self.blacklist; unless I'm missing something it sounds like simply setting it on that return path is the answer? [14:53] is there a bug about this? [14:55] cjwatson, Hello! I have a little generic patch for bin/update-local-indices in cdimage. What is the best way to give this patch to you? ( [14:55] There's not a bug, although I promised evand one if I didn't get a solution this weekend [14:56] foka: in rough order of preference: bzr branch published on LP; patch in a bug report on ubuntu-cdimage; pastebin [14:56] I think generate blacklist returns the empty set, and if '/%s' % path in self.blacklist: gets confused. [14:56] cjwatson, Cool, thanks! [14:56] persia: do you get a KeyError on self.blacklist? [14:57] * persia rebuilds the environment that reproduces the error again [15:09] cjwatson: Actually, I get an AttributeError: Install instance has no attribute "blacklist". [15:11] ah, ok, I think that's what I meant. [15:12] so it looks like this should fix it? [15:12] I see I'm not the only one who hasn't memorised the exception table :) [15:12] not to spoil your fun but if it's not at the top of your list ... [15:13] ubiquity: cjwatson * r2810 ubiquity/ (debian/changelog scripts/install.py): Initialise self.blacklist even if it's going to be empty. [15:13] Heh. No worries. [15:13] I like trying to learn python, but I'll admit that I don't really know it. So far, my total useful python code is a two-line patch to dput, except I wrote it Ada style. [15:14] Thanks :) [15:14] basically attributes of classes don't exist at all until initialised, and trying to read from them raises an exception [15:14] generate_blacklist was only initialising self.blacklist on one of its return paths [15:16] Ah. I missed that call to return, which explains my confusion :) [15:19] Thanks for the trivial fix, and the explanantion. Please upload when you have a chance :) [15:19] on its way [15:19] ubiquity: cjwatson * r2811 ubiquity/ (d-i/manifest debian/changelog): [15:19] ubiquity: Automatic update of included source packages: base-installer [15:19] ubiquity: 1.86ubuntu5, hw-detect 1.63ubuntu2, partman-partitioning 59ubuntu5. [15:24] ubiquity: cjwatson * r2812 ubiquity/debian/changelog: releasing version 1.9.13 [16:00] cjwatson: I cannot seem to get the mdadm/boot_degraded value into /var/lib/preseed/log .... [16:00] cjwatson: but everything else seems to be working well [16:00] ubiquity: cjwatson * r2813 ubiquity/ (8 files in 6 dirs): bump to 1.9.14 [16:05] ubiquity: cjwatson * r2814 ubiquity/ (debian/changelog ubiquity/frontend/gtk_ui.py): [16:05] ubiquity: * GTK frontend: [16:05] ubiquity: - Give focus to the language treeview on startup (LP: #254635). [16:06] cjwatson: the last patch attached to https://bugs.launchpad.net/ubuntu/+source/partman-target/+bug/268580 i think is ready for review/sponsorship [16:06] cjwatson: i've tested it as much as i think I can [16:06] cjwatson: i do have a related question.... [16:06] cjwatson: when I build a new .deb (not a .udeb), and want to replace the one in the install media, how do I do that? [16:06] cjwatson: fedora has an "updates" directory, where you can dump your own, updated RPMs [16:07] kirkland: oh, meh, apparently debconf-set-selections needs an argument to read from stdin: use debconf-set-selections - [16:07] cjwatson: ahh..... [16:07] cjwatson: cool, i'll test that [16:07] if that doesn't work, stick set -x on /bin/debconf-set-selections and see what it's doing [16:07] cjwatson: i grep'd around, and most usages of debconf-set-selections used it in conjunction with a chroot and args [16:08] right, there's nothing to chroot to at this stage though [16:08] cjwatson: exactly [16:08] that's a different debconf-set-selections program, though its semantics are intended to be basically the same, just with a different debconf db [16:08] cjwatson: i seemed to be in somewhat uncharted territory [16:08] cjwatson: at least from what documentation/code i could find on the plane ;-) [16:09] ditch the parens round LP: #blah in the changelog, they're confusing here and not needed :) [16:09] yes, you are [16:09] my suggestion of debconf-set-selections was based on inspiration, not prior art :) [16:09] ;-) [16:10] replacing a deb in the media: err, awkward. I don't think we have any equivalent of "updates". You basically have to regenerate the Packages{,.gz} and Release files; probably just remove Release.gpg [16:10] cjwatson: what about my updated deb question? [16:10] cjwatson: hrm, okay... i may save that test for an updated iso [16:11] cjwatson: and be satisfied with my unit testing when i see the value in the preseed log [16:11] I think that's pragmatic [16:23] cjwatson: BINGO! that did it [16:23] woo [16:25] feed me an update and I'll sponsor that [16:26] cjwatson, I suspect that the white space in partman preseed was just me typing here, there might be something else [16:26] finished with the vm, running again [16:26] cjwatson: in that case, I think the last patch attached to https://bugs.launchpad.net/ubuntu/+source/partman-target/+bug/268580 should be ready to go [16:27] xivulon: ok, afraid I don't know what "invalid argument" there might indicate; that's a catch-all error returned by the kernel that can mean all sorts of things, from some kind of internal error to "you idiot, don't pass those arguments" [16:27] the relevant syslog is here: http://paste.ubuntu.com/46539/ [16:28] it happens in open_dialog OPEN /host/ubuntu/disks/swap.disk [16:28] yes, I know [16:29] would really need an strace to see what's happening; the syslog is not very informative [16:30] will try to get one [16:38] kirkland: can the partman-md task be marked Invalid? [16:38] cjwatson: yes, please [16:38] kirkland: uploaded, thanks! [16:41] cjwatson: you bet, thanks for your patience and assistance [16:42] not a problem [16:47] cjwatson: have you seen http://manpages.ubuntu.com/ lately? [16:47] cjwatson: as of today, anyway? :-) [16:47] cjwatson: i made a search engine pluggin for firefox, you can install from that page [16:49] cjwatson: i fly again tomorrow, i'll try and draft a patch for man that pulls .gz's from the web, if that's not too late for Intrepid... otherwise, i'll scrap it for now, and work on it for Jaunty (perhaps making the dman shell wrapper available for download as a stop gap) [16:49] I'll have to take your word for it for today, I should do something other than work [16:49] I'd prefer to target the man change to jaunty and get it done right upstream [16:49] I note idly that searching for "man" gives me man(1posix) rather than man(1) [16:50] cjwatson: thanks, i'll have a look [16:52] cjwatson: i'll take some advice from you (sometime later) as to how best to handle namespace conflicts in the repository, which is what you're seeing [17:10] weird kvm goes along with the installation, vbox gives the partaman error above... [17:29] kvm installation seems ok, will now try to get an strace in vbox (vbox though has to be run without paravirt and is sloooooow) [18:53] cjwatson, Hello! I just made my first bzr push: https://code.edge.launchpad.net/~foka/debian-cd/ubuntu-thiz [18:55] cjwatson, So, is that all I need to do, just push the branch and notify you on IRC? Or should I "Propose for merging into another branch" through Launchpad? Thanks! [19:04] foka_: you can do a merge proposal if you want to, certainly, but an IRC notification is good enough [19:05] cjwatson, Great, thanks! === foka_ is now known as foka [19:05] oh, ok, yeah, that makes perfect sense. Er, except that I think warty-hardy might just need to delete it later, rather than not at all [19:05] cjwatson, I've another question. Is it possible to publish a private branch on Launchpad? [19:06] hmm, no, I think I'm mistaken [19:06] merged, thanks [19:06] foka: only on a commercial basis, I think. Talk to #launchpad [19:07] private like.. use launchpad for internal work? [19:07] cjwatson, It was one of the thing I got stuck on when trying to install in a VirtualBox. (My memory is fuzzy) [19:09] no, it makes sense, I should have thought of that when making that change as generally I've been quite careful to preserve buildability of older releases [19:09] generally Mark is happy to fund free Launchpad services for public free software, but I believe he's said that hosting of non-public or non-free branches will need to be on a paid basis [19:15] I see. Many thanks! Yes, I saw the "This branch is public" note on the top right corner, and was wondering if there was any way to make a private branch which only subscribers to see. [19:17] You and Mark are right: Launchpad is supposed to serve the Free Software community, and thus private branches are naturally against that spirit, so yes, in that case, Mark's reasoning is very reasonable. [19:21] I think this is the case, anyway. As I say, check #launchpad to be sure [19:26] cjwatson, Gotcha, thanks! [19:26] cjwatson, BTW, were you able to see my private msgs? [19:31] cjwatson, http://paste.ubuntu.com/46682/ [19:36] xivulon: it's Saturday evening; I don't suppose this can wait? [19:36] also, that strace says nothing [19:36] sure [19:36] parted_server is the process that actually does the underlying open [19:37] you're tracing something else, looks like one of partman's shell scripts [19:37] so that strace is no more informative than the original syslog, I'm afraid :( [19:37] was hoping that strace -f -F would catch that [19:38] not if the process you're tracing isn't responsible for starting parted_server in the first place [19:39] best to catch the parted_server process itself; you don't need -f -F for that [19:39] (and in general you don't need -F on Linux) [19:40] got it, going to try again [20:29] http://paste.ubuntu.com/46698/ === hardwire is now known as CrankWidow === CrankWidow is now known as hardwire [21:02] cjwatson, tried opening the file from python with posix open call [21:02] posix.open(filename, os.O_RDWR|os.O_LARGEFILE) -> OK [21:02] posix.open(filename, os.O_RDWR|os.O_LARGEFILE|os.O_DIRECT) -> invalid argument [21:16] cjwatson, One more small patch for you at https://code.edge.launchpad.net/~foka/ubuntu-cdimage/thiz or lp:~foka/ubuntu-cdimage/thiz [21:23] yep it looks like the 2.6.27 + O_DIRECT + ntfs do not go well together [21:26] ouch [21:26] is that for the settings migration? [21:39] bug #269946 [21:50] Bye! [21:50] * foka falls asleep. :-) === hardwire is now known as CrankWido === CrankWido is now known as CrankWidow