[12:27] pitti: Hi, yesterday's hardy kernel update has bad overrides. At least l-r-m landed in universe. [12:28] This makes linux-generic, linux-server uninstallable. [12:37] wgrant: I'm not so sure, since the install in question is done with universe enabled.. [12:38] lamont: What's the error? [12:38] 2.6.24-30 certainly does have bad overrides, but the specific one I quoted may not be directly relevant here. [12:38] that nvidia-kernel-common doesn't exist [12:39] because it's in restricted [12:40] though the top level error is simply that linux-restricted-modules-generic is uninstallable [12:42] lamont: Which is interesting, since linux-restricted-modules-2.6.24-29-generic also depended on nvidia-kernel-common. And nvidia-kernel-common has always been in restricted. [12:42] https://launchpad.net/ubuntu/hardy/i386/nvidia-kernel-common [12:44] Hmm [12:44] Interesting. [12:44] hardy's linux-generic was in restricted until March. [12:45] April, even. [12:53] wgrant: usually that already gets caught in -proposed; it was fixed yesterday, is any package still wrong? [12:54] wgrant: there's a bug in soyuz somewhere, it randomly puts some binaries into universe [12:54] we haven't yet written a script to fix overrides after copying from PPA [12:54] Well, it's not really a bug. [12:55] For non-queued copies of new packages it uses the default overrides [12:55] Which is universe. [12:55] well, it's not predictable [12:55] pitti: mdeslaur and jdstrand are also digging into this [12:55] as many of the kernels do land in main [12:56] wgrant@lamuella:/tmp/chroot-autobuild$ rmadison linux-restricted-modules-2.6.24-29-generic [12:56] linux-restricted-modules-2.6.24-29-generic | 2.6.24.18-29.9 | hardy-security | amd64, i386 [12:56] linux-restricted-modules-2.6.24-29-generic | 2.6.24.18-29.9 | hardy-updates | amd64, i386 [12:56] wgrant@lamuella:/tmp/chroot-autobuild$ rmadison linux-restricted-modules-2.6.24-30-generic [12:56] linux-restricted-modules-2.6.24-30-generic | 2.6.24.18-30.11 | hardy-security/universe | amd64, i386 [12:56] linux-restricted-modules-2.6.24-30-generic | 2.6.24.18-30.11 | hardy-updates/universe | amd64, i386 [12:56] That's the not whole problem here, but it shows that something is not right. [12:58] ok, overrides have been adjusted for yesterday's lrm [12:58] mdeslaur: ah, you did? I'm just running change-override [12:58] pitti: I just got jdstrand to do it [12:58] Hm, not showing up in the web UI yet. [12:58] right, so it needs a publisher [12:58] which runs in 5 min [12:59] I don't see !universe pubs on https://launchpad.net/ubuntu/hardy/amd64/linux-restricted-modules-2.6.24-30-generic [13:00] fyi, I got him to do this: http://paste.ubuntu.com/745874/ [13:01] Ah [13:01] That's only half the trouble :/ [13:02] mdeslaur: want me to run the same again for hardy-updates? [13:03] lamont: Do we have logs from previous installs to see what it actually did? [13:03] we do not [13:04] pitti, mdeslaur: re updates> I just did [13:05] ok, great [13:06] lamont: I don't see how you can install linux-generic without restricted :/ [13:06] wgrant: given what I've seen, I believe that main/restricted are what we install with [13:06] which would explain the universe thing breaking things [13:07] Oh! [13:07] That makes more sense. [13:07] yes, linux-image-generic -> main only [13:07] linux-generic -> includes l-r-m [13:07] for lucid onwards that's easier, as the kernel is purely main and everything else is DKMS [13:07] lamont: You said earlier that nvidia-kernel-common didn't exist. [13:08] main/restricted but not universe makes a lot more sense, but then it would be that linux-restricted-modules-2.6.24-30-generic didn't exist. [13:08] (because it's in universe, and won't be fixed this publisher unless someone is extremely quick) [13:08] nvidia-kernel-common was me walking down the uninstallable tree on an installed machine, which happened to be main/universe [13:08] Ahaa [13:09] damn red herrings [13:09] pitti, jdstrand: Could you bring linux-restricted-modules-2.6.24-30-generic back into main? [13:09] -29 is there, -30 is universe [13:09] it certainly should be in restricted? [13:09] this must not be in main [13:10] pitti: Well, it's been in main for a while. [13:10] m -s hardy -S linux-restricted-modules-2.6.24 [13:10] eg. linux-generic has been in main since April. [13:10] that's the reference [13:10] once we verify -30, we should maybe fix -29 [13:10] wgrant: that was also an error thehn [13:10] wgrant: that was also an error then [13:10] pitti: Certainly. [13:10] But I think we should make things installable first. [13:10] Then get a list of what's where and fix it up. [13:10] Because it is a hideous mess at the moment. [13:11] l-r-m and linux-generic were initially in restricted, but ended up in main a few months back. [13:11] wgrant: I wonder if maybe overrides shouldn't allow regexes [13:11] -28 of both was in restricted [13:11] Then -29 was in main [13:11] And -30 is split across main and universe [13:12] and kittens. I'm sure kittens were involved somehow [13:12] We could send them straight back to restricted now, but I'm more interested in making things work immediately. [13:15] wgrant: I thought we just did send them back to restricted? [13:15] how would that make things not work? [13:16] pitti: That didn't include everything. [13:16] linux-restricted-modules-common was there, but no other l-r-m bits. [13:27] * wgrant sleeps. === scott-work is now known as scott-sick-at-wo === scott-sick-at-wo is now known as scott-sick-work [15:51] hmm, found in new queue "Description: Unity configurator This program allows one to configure and tune Unity with a pleasant and comfortable graphical interface." [15:51] do we want this in the archive? [16:10] Riddell, good question. [16:11] it reminds me of ccsm [16:11] pitti, ^^ any thoughts on this? [16:11] what's ccsm? [16:12] skaet: compizconfig-settings-manager [16:12] this is at least friendlier than that. But does it expose things that it shouldn't? [16:12] thanks tumbleweed [16:14] tumbleweed: I'm not sure how you'd define things it shouldn't [16:14] * skaet thinks it needs some review by the desktop and dx team before letting it in. [16:15] are we talking about universe ? [16:15] ogra_: yes [16:15] that really shouldnt need a DX review [16:15] its universe after all, as long as the package fulfills the criteria for a universe package there shouldnt be any political reason to not have it [16:16] ogra_, I was thinking it might be useful for the DX team to make sure that the interfaces its exposing/using are stable and likely to persist. [16:16] eep, the code aint pretty (and is italian) [16:17] (the dev could as well have someone upload it to debian and wed would get it by sync ...) [16:17] ogra_ ack. [16:17] tumbleweed, well, that might be a reason why it doesnt fulfill the criteria ... [16:17] yeah, if it's not harmful, I don't have any problems with it, and don't think we should in general [16:17] but having a canonical entity block a universe package is a very bad idea imho [16:18] * skaet nods [16:18] I'd much rather things like this went into the archive, than we had people running crazy shell scripts or enabling a bazillion PPAs [16:18] true [16:19] though if the code is bad or harmful it definitely doesnt qualify ... [16:19] * tumbleweed doesn't think it's harmful, but I don't know what's safe to change in unity [16:19] tumbleweed: without reviewing the code I don't know if this is any different from a crazy shell script, it might change any random config option [16:19] it just changes gesettings options [16:19] that should be safe if it does it the right way [16:20] thats something every user can do using dconf-editor [16:20] (which is a horrid interface btw) [16:23] is there any screenshot of that utility? [16:23] or a list of the options it allows to tweak? [16:25] seb128: bzr branch lp:myunity; ack-grep -a gsettings [16:25] * tumbleweed would post screenshots, but about to run out for dinner [16:25] seb128: people.canonical.com/~jriddell/tmp/myunity.png [16:25] fyi, the overrides were still wrong for a bunch of stuff on hardy, so kernels were still uninstallable. this is hopefully fixed now [16:26] mdeslaur is investigating the other kernels, and I am applying overrides on his behalf [16:26] he is updating some documentation as well [16:27] this all seems to originate with the ppa to -proposed copy. so I have added another bullet item to ArchiveAdministration t verify the overrides, and add nag text to copy-proposed-kernel.py as well [16:28] tumbleweed, ogra_, Riddell: it doesn't seem to do anything "dangerous", it tweaks configs in gconf [16:28] we never banned stuff from the archive based on the fact that users could break their config with those [16:28] we wouldn't have ccsm if that was the case ;-) [16:28] I thought we didn't have ccsm in the archive [16:29] but at the same time I'm not sure if we should recommend those to go for extras rather, that seems the sort of hackish software that will not keep working long and stay broken in universe for years [16:29] Riddell, we do have ccsm in the archive for as long as we have compiz [16:30] so we do [16:41] seb128: what about ubuntu-tweak? [16:41] didn't that get rejected? [16:41] dunno but I would like to see it available [16:42] Riddell, the predecessor was rejected, because it was really harmful [16:42] it would be a better tools to recommends than gnome-tweak-tools (which depends on gnome-shell, so is not very nice to recommend to unity users who want to tweak gnome settings) [16:42] automatix ? [16:42] was that the name ? [16:42] it broke the package manager in various ways etc [16:42] ogra_: right [16:44] https://lists.ubuntu.com/archives/ubuntu-archive/2010-August/036627.html [16:45] wow, so nice and calm ... [16:46] i remember a blogpost from matthew garret about automatix where he literally commented on every line and its harmfulness [16:46] jdstrand is like that, I hear he doesn't even get flustered when linux updates go into universe [16:46] heh [16:46] which is a common issue :) [16:47] I think I'd be happier if this myunity package was clear in the description that it was unofficial [16:47] heh [16:48] Riddell, tell that to the packager then :) [16:53] done [17:05] dear archive admins... halp. [17:05] https://launchpad.net/ubuntu/hardy/amd64/linux-restricted-modules-generic [17:06] and yet cocoplum shows it in universe as of 1617 UTC [17:11] lamont: mdeslaur and I are working it out [17:12] and as we discovered, we are in a twisty maze of tiny passages, all different [17:13] yes, the grue is after us, but he will be denied! [17:52] skaet: I hear rumours about the release images being 750MB this cycle but I can't find any information on the topic, do you know where any UDS notes or WIs would be? [17:52] Riddell, just a sec and I'll hunt down the blueprint. [17:52] I thought it was https://blueprints.launchpad.net/ubuntu/+spec/desktop-p-dvd-image but there's nothing there [17:57] Riddell, the sabdfl acked the possibility to go 750mb if needed but consensus is that we will probably not need it this cycle [17:58] Riddell, in practice I don't think we will have any change this cycle [17:58] ok so we should worry about our images being 10MB oversized :) [17:58] thanks seb128 [17:58] well I guess kubuntu could decide to use the option to go over 700mb [17:59] that would be a decision in your team [17:59] I'm curious to know what the reaction would be [17:59] it was basically decided for Ubuntu that it's better if we can keep staying on the CD for the lts [18:45] seb128: if ubuntu one goes to pyqt you think you can fit that onto the ubuntu desktop CD? [18:46] Riddell, we need to have that discussion, the u1 team wanted to have another try at this ubuntuone-installer thing [18:47] I thought pitti vetoed the installer? :) [18:50] slangasek, yeah, in oneiric, that's why I say they want to have another try to it ;-) [18:51] heh, I didn't realize it was a temporal veto [18:52] anyone know why perl-modules 5.12.4-6 is still showing up in the precise Packages files? Seems like it should've been garbage collected already, but maybe it's sticking around for benefit of armhf? [18:52] slangasek: NBS rdepends [18:53] libperl5.12 specifically [18:53] ah [18:53] * slangasek bides his time then [18:53] slangasek, I'm not sure it got vetoed, it was on the CD until beta, we reverted because not having the music store on the CD was raised as an issue [18:53] slangasek: the LP people made it behave more like Debian so archs that are behind aren't uninstallable [18:54] slangasek, they were discussing having the music store showing on the CD this cycle but still having the installer triggering when you try to use it the first time [18:54] micahg: yes, I knew about that recent change, I just couldn't work out what was holding it here [18:54] though I was just about to check libperl5.12 in NBS :) [20:34] slangasek: yeah, it's due to perlkde not being buildable right now [20:34] which is due to a couple of bits of KDE being an upstream version or two back for some reason [20:35] libsmokekde-dev | 4:4.7.1-0ubuntu2 | precise/universe | amd64, armel, i386, powerpc [20:35] okular-dev | 4:4.7.2-0ubuntu1 | precise | amd64, armel, i386, powerpc [20:35] ^- specifically those two [20:35] it was quite handy that it was still there so that I could use it as a test case for debootstrap, though ;-) [21:41] cjwatson: heh, right :)