[00:35] infinity: ^^^ Prety please .... [00:59] ScottK: Bribery would help. [01:24] infinity: Thanks. Anything you want me to cargo cult in in exchange? [01:25] ScottK: I have a ubiquity upload coming later tonight, but not just yet. [01:25] If you're around, a review would be lovely. If not, I'll find someone else. ;) [01:26] OK. [02:33] Such a perfect time to have a dead powerpc buildd.... [02:39] Did we kill one? [02:39] Apparently so. [12:12] * infinity accepts grub2 and goes to bed. [14:06] ScottK, understood. infinity, thanks for getting them through. [18:37] cjwatson, infinity: thanks for the quick grub fix! Now we just need to revert those nvidia/fglrx blacklists... :) [19:02] slangasek: just saw your comments in the bug. So seems like the nvidia blob is actually doing KMS now? [19:07] stgraber: no, it isn't, that's not the issue [19:08] stgraber: gfxpayload=keep hands a VESA framebuffer off to the kernel, which gets picked up automatically; and that apparently coexists fine with nvidia... but it also leaves the console in graphics mode, with nothing to set it to text mode for us in recovery [19:08] exact same thing happens with intel [19:09] now a C program to forcibly switch the console to text mode before launching the friendly-recovery menu would be trivial, but it's still a little late to go that route [19:09] so we should just use =text for recovery mode for this cycle [19:09] oh fun, that explains what komputes reported yesterday then [19:09] yes, indeed [19:09] if you remember his bug number, please dupe it :) [19:10] and then on the menu for today, I have bug #864149 :/ [19:10] Launchpad bug 864149 in nvidia-graphics-drivers (Ubuntu Oneiric) (and 3 other projects) "initrd doesn't set Fr keyboard layout anymore => couldn't open root luks at boot (affects: 1) (heat: 8)" [High,Triaged] https://launchpad.net/bugs/864149 [19:14] I guess fixing that one will make my fix for bug 524605 obsolete (starting console-setup to have the keyboard layout properly set) [19:14] Launchpad bug 524605 in friendly-recovery (Ubuntu) "Keyboard layout isn't configured by the time friendly-recovery starts (affects: 1) (heat: 6)" [Medium,Fix released] https://launchpad.net/bugs/524605 [19:16] stgraber: not exactly [19:18] not unless we decide to always do console-setup in the initramfs... which we currently do not, for reasons of boot speed [19:34] we do if it panics; it would be fairly straightforward to do that on entry to recovery mode too [19:35] though I don't know whether it'd be better to just do it from f-r instsead [19:35] *instead [19:35] cjwatson: for recovery mode, it all currently works [19:35] it's for regular boots, with luks root filesystem and binary video drivers where things are going wrong [19:35] because the binary video drivers are forcing FRAMEBUFFER=n for initramfs-tools, and console-setup keys on that [19:36] ah, right [19:36] well, again, the bits are all in the initramfs, it would be fairly easy to make c-s check the cryptsetup case too [19:36] (but probably not for 11.10) [19:37] * slangasek nods [19:37] maybe. I think it keyed off FRAMEBUFFER=n for a reason mind you [19:37] boot performance [19:38] yes, but I don't think it was just that [19:38] I have a feeling that it didn't work reliably unless the framebuffer was there [19:38] perhaps because we couldn't set the font yet unless it was [19:38] ah [19:38] well anyway, setting FRAMEBUFFER=n is the wrong way for the binary drivers to handle this [19:38] (that makes sense now I say it) [19:38] the question is whether we'll be able to sort out the right way in a timely manner [19:39] * slangasek goes back to fiddling with his test nvidia box :) [19:40] hmm, infinity got stalled on the flash-kernel-installer related bug he was working on and didn't upload ubiquity overnight; I think I'll do that now, it has several bug fixes which could use maximal testing time [19:40] ack [20:15] cjwatson: Your ubiquity upload adds gir1.2-soup-2.4 as a ubiquity-frontend-gtk depends. Was that intentional (not mentioned in changelog)? [20:15] Actually I can see it's used in the code now... [20:26] cjwatson: is en_US.UTF-8 locale guaranteed to be present? [20:28] I think so. [20:28] Don't we seed it on everything? [20:28] BTW, I accepted it, but don't let that stop you asking questions ... [20:28] I'm not sure === tkamppeter_ is now known as tkamppeter [20:42] hmm, promising lead on the nvidia-initramfs-framebuffer issue [20:45] ScottK: yes, needed for the async geoname fetching code; it was already on all the relevant CDs as far as I could see so I didn't scruple [20:45] OK. Thaks. [20:45] slangasek: language-pack-en-base creates it and currently that's always there [20:45] Thanks even. [20:45] cjwatson: ok [20:46] slangasek: ultimately it ought to be C.UTF-8 instead but I didn't want to risk that given that we were already using en_US.UTF-8 for a similar purpose elsewhere [20:46] (thanks for asking questions though!) [20:47] sure :) [20:48] hummm, did we have another regression in pkgbinarymangler? libxcb-render0 has changelog.Debian.gz listed in md5sums, but it's a symlink [20:48] same for a bunch of other desktop library packages [20:50] ah, and some of these builds are from June... I guess I should run debsums -s more often :/ [20:53] please don't tell me we have to rebuild the world [20:55] I don't think it warrants rebuilding the world for a wrong .md5sums [20:55] but there are a lot of wrong .md5sums on my system [21:02] argh, ubiquity failed to build, not funny [21:02] oh, missing build-dep I think :-/ [21:27] ok, got a solution for bug #864149 I think; just need to test it on a non-nvidia system now too [21:27] Launchpad bug 864149 in nvidia-graphics-drivers (Ubuntu Oneiric) (and 3 other projects) "initrd doesn't set keyboard layout when nvidia drivers are installed (FRAMEBUFFER=n override) (affects: 1) (heat: 8)" [High,Triaged] https://launchpad.net/bugs/864149 [22:52] slangasek: did you notice bug 862129? [22:52] Launchpad bug 862129 in samba (Ubuntu) "package samba 2:3.5.8~dfsg-1ubuntu2.3 failed to install/upgrade: subprocess new post-removal script returned error exit status 2 (affects: 1) (heat: 6)" [High,New] https://launchpad.net/bugs/862129 [23:05] cjwatson: no, I hadn't. what do you suggest? ignoring errors from update-inetd? [23:05] not a new issue in any case; I don't see a reason to treat this as a blocker for 11.10 [23:10] the general problem isn't easy to solve; perhaps an explicit test that File::Temp is loadable, but that violates layers ... [23:10] * slangasek nods [23:11] it probably isn't, though I'd like to confirm that it doesn't show up in default upgrades [23:11] or default upgrades after samba has been installed anyway [23:11] didn't doc-base run into something similar? [23:12] I don't recall [23:15] maybe that was more direct [23:17] * slangasek accepts ubiquity [23:18] ta [23:18] hmm, looks like I'm ENOTIME to finish testing this initramfs-tools change now; will have to upload it later this evening [23:37] I suspect that the simplest fix for bug 770711 looks rather like http://paste.ubuntu.com/700779/ (though I'm not going to start testing that at this time of night) - does anyone object to the X-Python-Version bump there, for collections.OrderedDict? [23:37] Launchpad bug 770711 in ubiquity (Ubuntu Oneiric) (and 1 other project) "Prefers to install to USB rather than to HD (affects: 3) (heat: 16)" [High,Triaged] https://launchpad.net/bugs/770711 [23:37] A backport does exist, but it's a fair chunk of code, and given that 2.7 is our default anyway ... [23:39] Seems easy enough to roll back if it does cause issues.