[00:00] which means that the version of openoffice.org-hyphenation in hardy (and incidentally also feisty and gutsy) shares a file with the current version in hardy [00:00] more usually the second named package will be from an older release, and you can use that as a hint for the << you need to use in Replaces [00:00] ok [00:00] it looks like that particular one is out of date, mind you [00:01] pochu, thanks [00:01] ah, it's because it was in hardy, and therefore you need a Replaces to account for that [00:01] not necessarily that it is currently in hardy - it keeps historical information [00:02] note that those lists will have false positives where dpkg-divert is involved and appropriate manual overrides haven't been added [00:03] oh ok [00:03] yea that list looks really nice :) [00:08] g'night all, and thanks again for the info [00:17] is the release still on for tomorrow? [00:22] morning sabdfl [00:25] #ubuntu-motu [00:26] #ubuntu-devel [00:26] I have a question about wubi? [00:26] #ubuntu [00:27] rats. [00:27] Just missed him. [00:28] he is in #ubuntu i think [00:37] What's the difference between a delete_event and a destroy event in GTK? [01:12] is there any way to debug a postinst without building the package? [01:17] you can edit them in place in /var/lib/dpkg/info/ [01:18] most are shell, so adding 'set -x' near the top will give you an execution trace [01:23] cjwatson: the problem is that i need it to be on fresh install [01:23] mm [01:23] cjwatson: i will try to check what to do, thanks :D [01:23] nxvl: 'dpkg --unpack foo.deb', edit /var/lib/dpkg/info/foo.postinst, 'dpkg --configure foo' [01:24] debugging the preinst is harder; for that you need to unpick the deb with ar and tar and then put it back together again the way you want it [01:24] doable but tedious [01:24] cjwatson: oh! thanks i will unpack it :D [01:27] mm [01:27] but unpack is running postinst [01:27] incorrect [01:27] no [01:28] postinst is the --configure stage, which happens after unpack [01:29] i'm doing dpkg --unpack and it is running things i have write on .postinstt [01:30] then you have made a mistake somewhere. dpkg --unpack does not run the postinst. [01:32] can it be running the .config? [01:32] (except in cases where it needs to roll back from errors) [01:32] it might run the .config if you have '. /usr/share/debconf/confmodule' in your .preinst script [01:32] swapped postinst with preinst ? [01:32] http://www.debian.org/doc/debian-policy/ch-maintainerscripts.html#s-unpackphase [01:33] dpkg doesn't call .config directly, only when you source the confmodule [01:33] cjwatson: thanks [01:33] (because debconf is only integrated into dpkg with spit-and-sawdust) [01:33] cjwatson: thats the why [01:33] wooohooo [01:33] * nxvl HUGS cjwatson [01:33] the difference between config and postinst is rather important ... :-) [01:37] if i run "dpkg --reconfigure foo" it will run the scripts on /var/lib/dpkg/info/ doesn't it? [01:38] err [01:38] dpkg-configure [01:41] there is no such program. Do you mean dpkg-reconfigure? [01:42] read its code and look around line 96 for what it does [01:42] it's quite a short program [01:48] is there a way to set upstart to boot up a system without starting gdm? [01:49] essentially, boot into root level 2? [01:49] er run level 2 [01:50] the default runlevel with upstart is 2; though I understand what you mean, that runlevel layout is RH-specific [01:51] hi all - i asked in #ubuntu-motu earlier, but got no response. the xubuntu-docs are finished. (you can view them at: http://doc.ubuntu.com/xubuntu/ ) would someone be willing to build them for me, or to help me build them? [01:51] slangasek: Ok, _MMA_ and I are happy with UbuntuStudio images. [01:51] https://code.launchpad.net/~ubuntu-core-doc/ubuntu-doc/xubuntu-hardy [01:52] TheMuso: http://iso.qa.ubuntu.com/qatracker/build/all/all doesn't show any test results, is that validation in progress? [01:52] ugh hang on [01:52] _MMA_: ^^^ [01:54] * TheMuso only tested a manual partition install with creating new partitions... I have no time to do other tests, so I supposed I'll have to leave it up to the others to give the final word then. _MMA_ stated that he was happy, dispite having only done one form of install, which one I am not sure... [01:56] hrm, well guess I can just have them twiddle /etc/X11/default-display-manager [01:56] bryce: so in theory, you could get what you wanted by mv /etc/rc3.d/{S30,K30}gdm, and then, if upstart's sysvinit emulation is feature-complete, append '3' to the end of the boot line in grub [01:57] ok [02:01] brb (rebooting fresh upgrade from gutsy to hardy, wish me luck) [02:04] yay [02:10] <_MMA_> slangasek: Sorry. Had alot going on here. I did 2 tests. Added the results. === jescis_ is now known as jescis [02:29] _MMA_: ok, cheers === superm1 is now known as Daviey_ === Daviey_ is now known as superm1 === thegodfather is now known as fabbione [04:03] are there any negatives to using wubi? [04:04] You have to have Windows? [04:05] heh [04:06] you'll be running two OSs at once, so things will be slow [04:06] at least Windows-slow :) [04:08] ash211: But you won't, right? You boot into linux on the metal; it's not virtualised. [04:09] I'm pretty sure it is, actually. [04:09] It's homepage says you don't need to modify partitions or bootloaders, so I'm thinking it must run as an app within windows [04:09] everything I'm reading seems to point that way [04:09] http://wubi-installer.org/ [04:11] It doesn't change the partitions because it makes a filesystem in a file on your existing NTFS partition, and it doesn't add a *new* bootloader because it uses Windows' existing bootloader. [04:11] ash211: it doesn't run on top of windows [04:11] hmm, didn't know that [04:12] looks like I had a fundamental misunderstanding in how it works [04:12] evand: any ideas? [04:13] i need to blow away my laptop anyway so i was considering just reinstalling ubuntu in wubi form since it appears ubuntu supports ntfs write properly now [04:13] i need windows occasionally on bare metal (not via vm) to update stuff [04:15] stuff sounds mysterious [04:18] hmm sounds like hibernation/suspend may not work on it [04:18] which isn't good for a laptop [04:19] Wonder if wubi runs in Wine... [04:19] now that's a headache! [04:21] Yeah :P [04:21] Recently, we got Cygwin running in Wine :) [04:21] Like in the last couple weeks, iirc. [04:21] There are still some problems, like GCC doesn't work properly, etc. [04:21] But the basic shell runs. :) [04:22] Wubi edits the bootloader? [04:22] wubi is just an installer i think [04:22] that causes the windows bootloader to load something to bootstrap a loopback linux boot [04:23] so no it wouldn't run under wine [04:23] or at least would be very unlikely to do so properly [04:24] Yeah [04:28] there is some info at https://wiki.ubuntu.com/Testing/Cases/Wubi === kitterma is now known as ScottK2 [04:47] looks like hibernation doesn't work according to the wiki beta page [04:48] but sleep should [05:12] anyone familiar with virt-manager? [05:14] I know it can be used to administer guests on a remote server, because that's what the release notes in front of me say ;) [05:14] heh yea [05:15] well i am trying to get virt-manager/kvm setup so i can get rid of vmware [05:15] but it won't let me select the "enable kernel/hardware acceleration" option [05:15] 64-bit system? [05:15] yea [05:15] laptop? [05:16] slangasek: I'm having issues using my wubi install of ubuntu should that be a wubi bug? [05:16] 64bit desktop that has the vmx extension [05:16] bdmurray: as I starting point, I suppose so? [05:16] calc: then you're already a step ahead of me in the analysis :) [05:16] thats all i know to do [05:17] * calc thinks there should a setup page on the wiki about it [05:17] I agree [05:17] keescook: you awake? [05:17] iirc keescook was mentioning he was using it earlier today [05:18] i was going to host all my images on my desktop to get rid of the space issue on my laptop [05:18] and then found out it is supposed to run it faster with hardware accel, but its still grayed out [05:18] my laptop is one of the cheapo chips with it disabled [05:19] apparently the chips they use in store laptops don't have it (or didn't last june) [05:19] my laptop was not cheap, but it's still disabled :P [05:19] I'm told that it's largely a BIOS issue [05:19] slangasek: oh [05:19] at least for mine [05:20] i think for some its disabled in cpu itself, and iirc mine is like that [05:20] its a C2D T5300 [05:20] right, mine's a T5500; I hadn't heard about the possibility of disablement within the CPU [05:20] I just know "no worky" [05:21] ah ok [05:21] as i understand it it's just your BIOS locking it out because you didn't buy the business laptop [05:21] not that it helps [05:22] Amaranth: so its not something that the kernel can just override like it can in most other cases? [05:22] I guess not [05:23] what laptop is this? [05:23] oh, desktop [05:23] hrm [05:24] well on my desktop it shows that it supports it in cpuinfo [05:24] but virt-manager won't let me select the box [05:24] on my laptop it doesnt show that it supports it in cpuinfo [05:25] ah, so what laptop? [05:25] toshiba satellite A205-4577 [05:25] of course the hardware extensions don't really speed things up [05:26] vmware will still blow it away [05:26] Amaranth: oh :-\ [05:27] the vmware guys found that when they made vmware use the hardware extensions things got slower [05:27] http://www.intel.com/products/processor_number/chart/core2duo.htm [05:27] according to intel, whether its in bios or not, my and slangasek's cpus don't do VT [05:27] ah, sucks [05:28] * calc wonders what Intel Trusted Execution Technology is [05:28] i guess they had to do something with the chips that failed testing [05:28] its not the execute disable bit, so wonder what tha tis [05:29] all of the T7XXX series does it [06:07] good morning [06:19] Hi all [06:22] good morning dholbach [06:22] hi highvoltage [06:38] calc: yeah, check with soren just in case. [06:38] * dholbach hugs keescook [06:38] hi dholbach! [06:39] lamont: say, I'm tracking down an issue, and I've traced it back to swapon in feisty. === jescis_ is now known as jescis [06:39] lamont: between feisty and gutsy 30swsusp-resume.dpatch vanished without reference in the changelogs. [06:40] lamont: the patch seems to be the cause of the "zomg my swap UUID changed" bug, and lacking is the cause of the "ack, I have no swap anymore" bug. [06:40] lamont: seems the correct fix is to teach mkswap how to re-use and existing UUID, and then update 30swsusp-resume accordingly -- that should fix both bugs. [06:55] so, I see http://brainstorm.ubuntu.com/idea/94/ (#1 idea, suspend), and have read some of the comments, such as this is extremely difficult for many factors. Is there any momentum on this front, in spite of the many factors? [07:02] lamont: oh, hey, hardy mkswap supports UUID. :) [07:29] Good morning [07:34] slangasek, verified the i386 works for us. [07:34] (on the mythbuntu alt disk that is) [07:36] superm1: great, thanks [07:36] pitti: morning [07:38] hunteke: it's something that will take more than ubuntu to fix. it's not just attention to a few details and patches away from "just working" [07:39] hunteke: I think mjg59's been doing a decent job cataloging the problems with the current "architecture" [07:39] unfortunately too, issues lie in the videobios sometimes [07:39] pwnguin: alright, I'll buy that. But is there a discussion somewhere about it? The only thing I see from Canonical/Ubuntu are borked scripts/hibernate,sleep buttons, and details through brainstorm. [07:39] pwnguin: cool where do I find that? [07:39] mjg59's stuff that is [07:40] linuxconf. lemme grab a link real quick [07:40] which on the windows side of things, that's part of the reason why they will only qualify a particular driver release to that machine. they have a special code path to work around a bad vbios [07:40] k, thx [07:40] http://mirror.linux.org.au/pub/linux.conf.au/2008/Fri/mel8-139.ogg [07:40] http://linux.conf.au/programme/presentations [07:41] superm1: yeah, I sorta figured, part of the buying-through-a-vendor-service and MS tax, I expect [07:41] pwnguin: oh excellent! it's a movie presentation. :-) [07:41] well not even that. it's exclusively that vendors don't always stick to the "spec" [07:41] much easier to digest [07:42] well, broken video bioses etc can probably be avoided by intelligent linux laptop vendors [07:42] but even in such limited and advantageous, I dont think the current suspend / hibernate regime is helping [07:43] pwnguin, superm1: so perhaps a better question, how can I help tilt against this windmill? [07:43] hunteke: read LKML? [07:43] the problem with them is that they typically freeze a month or two before the laptop is about to be released, so if linux is an afterthought by the vendor, they won't have much leeway in getting vbios fixed. [07:43] pwnguin: no, just the digest I get through LJ [07:44] hunteke: which is that? [07:44] Linux Journal [07:44] sort of more oriented towards beginners [07:44] oh. not liveJournal. [07:44] that say, linux magazine [07:44] makes more sense [07:44] heh [07:44] morning dholbach, I added a dialog to choose teams to https://code.edge.launchpad.net/~thekorn/five-a-day/teams.applet [07:45] hunteke: well, aside from patience and learning more, you could perhaps found a kernel forum on the subject? [07:46] or fund one ;) === thekorn_ is now known as thekorn [07:48] keescook: Are you around? [07:52] abogani: he might have gone to bed (0:52 at his place) [07:52] thekorn: you ROCK - looking at it in a bit [07:52] dholbach: Ops Sorry! [07:53] pwnguin: sorry had to run away for a min. hehe, seeing as how I'm basically a poah-ass-student, funding isn't an option. [07:53] I would like it to be though. so perhaps after I make my first million or something . . . [07:54] are there other things I, as a peon could do? [07:54] read lkml I suppose [07:55] the options basically involve "practice patience", or "become an expert and get irreversibly sucked in to the development process" ;) [07:55] hehe, alright. I guess I was looking for more social things [07:56] perhaps analogous to "writing my congress/senator" [07:56] abogani: while dholbach would normally be right, I'm addicted to fixing a swap/resume issue at the moment. what's up? [07:56] keescook: wow [07:56] hunteke: in reference to fixes for laptop support? you have far less leverage with your laptop vendor than you do with your senator :-P [07:56] hah [07:56] keescook: pff, who are you kidding, I see you keeping late hours all the time :) [07:57] say, where're you all? [07:57] kansas [07:57] slangasek: I want to deny! ... but it's true. [07:57] it's almost 4a here, well past my bedtime [07:57] (yes, it is way late here) [07:57] Cascadia! [07:57] and I thought I was mad getting up early because my girlfriend had to (after a very short night) :) [07:57] keescook: Sorry for bother you but #107209 is already fixed. Please set it accordingly for Gutsy and Hardy. Dapper seems not affected. [07:59] dholbach: oh, don't let the presence of either kees or myself leave you with any doubts about your own madness [07:59] I assure you, I'm also quite mad [07:59] slangasek: I wouldn't dare to excuse my own madness with that of other people :) [08:00] abogani: dapper is affected (it lacks the "fix" that was attempted in 2.6.17). Yes, Hardy is fixed (which is why there isn't a "linux" task) [08:01] I knew I was in trouble when I'd say thing like "Am I crazy, or $blah", and people would say "yes". [08:02] Why bug is marked "In Progess" fro Hardy and Gutsy? :-? [08:03] abogani: not sure where you're looking ... linux-* tasks are a bit oddball (there is no "hardy" version of linux-source-2.6.15 e.g.) [08:06] Opss i' looking wrong bug report! [08:07] * abogani really need a coffe! === \sh_away is now known as \sh [08:07] abogani: oh! haha. okay, sanity returns [08:38] lamont: debdiff attached to bug #66637 for your examination. [08:38] Launchpad bug 66637 in util-linux "After running mkswap, swap space is discarded, system fails to hibernate (invalid swap signature)" [High,Fix committed] https://launchpad.net/bugs/66637 [08:38] * keescook goes to bed [08:48] good night keescook [09:14] pitti, when will langpacks be released? it's been over a week I think [09:15] TomaszD: right after beta freeze is over [09:15] pitti, thanks [09:24] calc: "Executable Disable" bit is the ability to mark certain pages in RAM non-executable. It foils buffer overflow attacks. [09:24] calc: You may have heard about it as the NX bit. [09:26] pitti: I commited a small fix for dapper->hardy upgrades into the hal bzr, please review and if you are happy, I'm going to upload [09:26] calc: It's true that certain C2D's don't have the IVT extensions, namely the T52x0, T5300, T54x0, T5500 with stepping "B2", E2xx0, E4x00 and E8190 models. (reference: http://en.wikipedia.org/wiki/VT-x#Intel_VT_.28IVT.29) [09:27] soren: so if I see stepping: 6, does that mean the chips support it and my bios is ornery? [09:28] <\sh> if somebody is so kind, and can explain how to setup a sane and clean chroot for building lpia apps on archs like i386 or amd64...reading this maemo-sdk-setup-doing-manually-stuff on nokia pages scares me...are the any possibilities to do it in a debian way? :) [09:28] slangasek: If you're looking at the cpuinfo anyway, just look for "vmx" in the cpuflags. [09:28] \sh: debootstrap --arch lpia? [09:28] slangasek: (Notice how I avoided exposing my lack of ability to actually answer your question) [09:28] soren: yeah, no vmx in the cpuflags; I was vaguely hoping that was because the BIOS had hidden it from me :/ [09:29] slangasek: No. [09:29] slangasek: The cpuflags will show it regardless. [09:29] slangasek: When you try to load the modules, they'll tell you that the BIOS doesn't like you. [09:29] heh, ok [09:30] <\sh> slangasek: if this is so easy...is it compiling to i386 or ARM arch? [09:30] lpia is i386 compatible. [09:31] <\sh> soren: thx [09:32] * soren wanders off again (seeing as he's supposed to be on holiday) === asa2 is now known as asac__ [09:46] how can I reduce the 'wait for root fs' delay again? [09:49] pitti, isnt that just waiting until the device shows up (HW dependant) [09:50] right, but in my special case I need to fix the initramfs first, and I'm sick of waiting one minute :) [09:52] heya people [09:58] is the language selector menu item not working on the CD a known issue? [10:02] ummm [10:02] which one? [10:02] the one that opens right at boot? [10:02] slangasek: system, administration, language support [10:02] Ubuntu? Kubuntu? [10:02] seb128: what happens if you click on it? [10:03] seb128: and what is printed on the terminal if you run: gnome-language-selector? [10:03] mvo: the desktop has "/usr/bin/gnome-language-selector -n" as a command [10:03] mvo: the "" are breaking it [10:03] there was a bug reported against language-selector-qt only [10:03] so I guess the gnome one wasn't tested [10:04] clicking on it gives a "no such file or directory" [10:04] seb128: that is strange, my bzr looks fine [10:05] * mvo investigates [10:05] withou the "" around the command name it works correctly [10:06] hum, I don't get it, the version installed on my desktop doesn't have the bug [10:07] ah, 0.3.1 on the CD [10:07] I think its because the desktop file is modified for the language-selector .. [10:07] how come? [10:07] in livecd build (or whatever the script is called :) [10:07] its run with -n there to supress the message about missing language-pack packages [10:08] ah ok [10:08] so whatever does this change is buggy [10:08] cjwatson: ^? [10:08] it probably happend when we swichted it from root to use and removed the gksu [10:09] \o/ I managed to do an encrypted LVM installation with the desktop CD [10:12] ./scripts/casper-bottom/35fix_language_selector:24: sed -i '/^Exec/ s|/usr/bin/gnome-language-selector|"& -n"|' /root/usr/share/applications/language-selector.desktop [10:13] seb128: that code dates from October 2006 [10:13] seb128: have the semantics of Exec changed? [10:13] cjwatson: not that I know, but could be a side effect of the gnome-desktop switch to gio [10:14] I'm not sure it's correct to use "" there or if the fact it was working before was a bug [10:14] let me look at the spec [10:15] http://standards.freedesktop.org/desktop-entry-spec/latest/ar01s06.html seems to suggest it was a bug that it worked before [10:15] ogra ping [10:15] is there a bug number for this? [10:16] cjwatson: not that I know, I was asking what would be the right component [10:16] casper [10:18] calc, bdmurray: does bug #203984 really refer to suspending *from* the CD, or from a system installed using it? [10:18] Launchpad bug 203984 in linux "[hardy beta] [amd64] squashfs errors on resume from desktop cd" [Medium,Triaged] https://launchpad.net/bugs/203984 [10:19] cjwatson: bug #204185 [10:19] Launchpad bug 204185 in casper "language-selector menu entry broken due to extra quote use" [Undecided,New] https://launchpad.net/bugs/204185 [10:20] the history suggests that in fact it was necessary at one point [10:20] "Make sure to quote arguments to properly" is the (slightly mangled, apparently) log message [10:20] davmor2, pong ? [10:21] ogra: edubuntu failed to install from netboot (mini.iso) [10:21] oh, crap, we need to stop building that [10:21] seb128: thanks, fixed in bzr [10:21] cjwatson: thank you [10:21] ogra: no we don't, mini.iso is common to everything [10:22] you just need to stop testing it ;-) [10:22] hmm [10:22] well, it shouldnt be in the task list [10:22] davmor2: Edubuntu is only installable as an add-on CD now; discard all test cases that involve testing it directly [10:22] think it is to do with italc-client [10:22] oh, netboot [10:22] that actually should work [10:22] ogra: no reason why not, if installing over the network [10:22] oh, right [10:22] that should j8ust install edubuntu-desktop [10:23] davmor2: post syslog [10:23] ogra: I'm letting stgraber know the error [10:24] cjwatson: How do get the syslog? [10:24] italc-client only depends on libitalc libqt4-core libqt4-gui [10:24] so i wonder what can go wrong ther [10:24] e [10:24] davmor2: back up to the installer main menu and select "save debug logs" [10:24] * ogra tries to apt-get it [10:25] I tried edubuntu addon yesterday on i386 and it worked well, italc was installed and working after rebooting [10:25] and I'm running the devel package for iTalc on my lappy (amd64) [10:25] works fine from apt-get as well [10:27] and runs with no problems ... [10:27] yes, speculation is moderately pointless without the log file :) [10:27] yup [10:29] hum [10:29] actually i dont understand why it gets pulled in at all [10:29] its only a dep in the addon seed, not of edubuntu-desktop itselrf [10:30] * ogra wonders if tasksel exposes edubuntu-desktop-addon instead of edubuntu-desktop [10:30] mvo: ugh, good catch; please upload [10:31] guys goto www.davmor2.co.uk [10:32] cjwatson: ogra: stgraber: [10:32] thanks pitti, will do [10:33] davmor2, stgraber: use of deprecated chown syntax in postinst [10:33] slangasek: no idea it just stopped installing :) [10:34] stgraber: but I guess the more significant problem is that the package assumes group 'admin' exists when it doesn't? [10:34] indeed [10:34] when is the admin group created ? [10:34] i would be more intrested why its pulled in at all [10:35] dunno; it's not part of base-passwd, I would suggest that you simply try to create the group yourself before using it [10:35] slangasek: hmm, wouldn't that cause a fail for the package that actually creates it ? [10:35] shouldn't [10:36] "addgroup admin" when "admin" exists returns 1 [10:36] all such packages *should* know how to handle a group that already exists; I'm looking for the responsible package to verify, though [10:36] guys anyone who needs the log if you can copy it that would be great then I can carry on testing [10:37] stgraber: well, the only package I see creating the group in postinst on my system is jockey-common [10:37] davmor2: for me that's ok, I know what to change [10:37] I don't think you want to depend on that :) [10:37] indeed :) [10:37] so it's probably created by the installer when adding the user to the system [10:37] but, jockey-common does have a good example of how to do this idempotently [10:38] yes, probably [10:38] anyone else need the log? [10:38] stgraber: the same thing happen on 64bit too [10:39] yes, it's not architecture-dependent [10:39] slangasek: how can I quickly check if the group exists ? (!= parsing /etc/group) [10:40] stgraber: see the jockey-common postinst -- "getent group admin" [10:41] ok, will be fixed after beta (I need to upload a fix anyway) [10:42] ogra@ceron:~/devel/seeds/edubuntu.hardy$ apt-cache show edubuntu-desktop|grep italc [10:42] ogra@ceron:~/devel/seeds/edubuntu.hardy$ [10:42] * ogra scratches head [10:43] what's exactly the problem ? the edubuntu desktop task installs the content of edubuntu desktop addon task ? [10:43] ogra: education-thin-client-server suggests it, though it's a bit odd if that causes it to be pulled in [10:44] i wonder if -meta is out of sync [10:44] * ogra tests an update run [10:44] the Packages file lists italc-client as part of edubuntu-desktop-addon task, yes [10:44] right [10:45] that's right [10:45] but its not in -desktop [10:45] so netinstall installs edubuntu-desktop-addon instead of edubuntu-desktop ??? [10:45] -addon creates -desktop [10:46] or: is the base for * [10:49] Added italc-client to desktop-addon-i386, desktop-addon-amd64, desktop-addon-powerpc, desktop-addon-ia64, desktop-addon-sparc, desktop-addon-hppa, desktop-addon-lpia [10:49] Removed thin-client-manager-gnome from desktop-kde-i386, desktop-kde-amd64, desktop-kde-powerpc, desktop-kde-ia64, desktop-kde-sparc, desktop-kde-hppa, desktop-kde-lpia [10:49] * ogra cries [10:51] slangasek, fine with you to fix that after beta ? i dont want to delay anything [10:51] does that only affect the netinstall ? [10:52] ogra: yes, after beta should be fine [10:52] there is no other method i know of that would allow you to use edubuntu-desktop dirctly on install [10:52] my fonts in mutt are borked after last week's updates; it can't display line drawing characters or international characters [10:52] so likely only the netboot is affected [10:52] has anyone else seen something similar? [10:52] not here, mutt displays fine for me [10:53] mdz, checked your locale ? [10:53] if that's only netboot, it'll be fixed as soon as the fix is uploaded so will be fixed hours after Hardy's beta. === heno__ is now known as heno [11:10] _MMA_: Ubuntu Studio Alternate i386 (20080319.1) needs testing [11:18] mvo: what would you think to adding a Kubuntu KDE 4 Desktop file in app-install-data? [11:18] * Hobbsee wonders when jono will appear. [11:19] Hobbsee: do you need him for something? [11:19] Hobbsee: after the hard work is done ;) [11:20] Riddell: got some questions on dealing with abuse of power, and how to do that. [11:20] Hobbsee: he's on #kubuntu [11:21] Riddell: oh, he's in -women too. should have checked for that, thanks. [11:25] good to know he knows his place in life :) [11:28] Riddell: you'd think the beard was a give away though wouldn't you :)# [11:32] Riddell: heh :) [11:34] davmor2, or the duck :) [11:38] Riddell: that sounds good, that will install kubuntu-kde4-desktop ? [11:47] mvo: yes [11:48] mvo: can I add that somehow or do you have to do it? [11:52] Riddell: I added it to my bzr tree [11:54] mvo: groovy [11:55] mdz: mutt WFM [12:01] argh, just got that g-settings-daemon start failure on the live system [12:05] Hi. Can someone point me to the correct channel to ask questions relating to PPAs [12:05] timlinux: #launchpad [12:05] Fujitsu: many thanks [12:07] launchpad seems to be missing from the wiki: https://help.ubuntu.com/community/InternetRelayChat [12:07] * timlinux will see if he has perms to add it [12:07] It's not part of Ubuntu. [12:07] cheers [12:07] ah [12:07] what is it part of? [12:08] stgraber: the admin group is indeed created by user-setup while adding the user to the system; please use addgroup --system admin to create it, not just addgroup [12:08] It's not part of anything. [12:08] Fujitsu: :-) [12:08] Fujitsu: everything is part of something [12:08] * timlinux is going meta [12:09] ok thanks for the info, cheers [12:09] Fujitsu, well, ts part of the public conspiracy to lower microsofts market share :) [12:09] ogra: Heh. True. [12:10] even thought i'm not actually sure there is such a thing like a public conspiracy :) [12:11] One can conspire publically. Conspiracy is a matter of intent :) [12:12] ah [12:12] i thought secrecy was mandatory [12:12] ogra: nah, the secrecy just makes it more sensational once it gets out. it's a marketing trick :) [12:13] Also, it's typically easier to conspire secretly. When your opponent can see your plans, they are more able to counteract. [12:13] <_MMA_> stgraber: Sorry. I had. Just forgot to add the results there. I just added 64bit results. Added now. [12:34] cjwatson: ok, thanks [12:34] _MMA_: thanks [12:35] _MMA_: btw, during release time you may want to join #ubuntu-testing for ISO testing related stuff [12:42] hi [12:43] ogra__: haven't changed my locale [12:44] still en_US.UTF-8 [12:44] changing to .ISO-8859-1 gets it displaying properly again [12:45] but I don't know why [12:45] <_MMA_> stgraber: Sure. I try to. Just been hectic around here. [12:48] davmor2: Do you have a bugnumber for the iTalc problem you found earlier ? (I have the fix but would like to link to LP) === \sh is now known as \sh_away [12:59] stgraber: not bugged it yet was trying to find out what to report it against when you said you could fix it [13:00] stgraber: I can throw one to gether now for you if you want [13:00] davmor2: please report against /+source/italc/ [13:01] https://launchpad.net/ubuntu/+source/italc/+filebug [13:05] stgraber: Bug #204235 [13:05] Launchpad bug 204235 in italc "Edubuntu fails to install from netboot" [Undecided,New] https://launchpad.net/bugs/204235 [13:05] davmor2: perfect, thanks [13:11] calc: sorry to be so late in reply, but it doesn't use virtualization and suspend should work but hibernate does not === \sh_away is now known as \sh === DktrKran12 is now known as DktrKranz2 [13:49] Is there a developer that can take a look at bug 192845 ? A patch is attached so this bug might be easy to fix before Hardy and avoid regressions [13:49] Launchpad bug 192845 in udev "[Hardy] Intel 3945ABG Long boot delays while "Loading hardware drivers"" [Medium,Triaged] https://launchpad.net/bugs/192845 [13:51] Keybuk: would be the one [13:51] the patch is almost certainly invalid [13:51] (without looking at it) [13:52] yes [13:52] the patch is invalid [13:53] though it's a nice hack ... :) [14:14] Heybuk: invalid? Does it gives some hint on what caused the regression? [14:18] Heybuk: No matter, I just seen your comment on bug, thanks for your work :) === Mithrand1r is now known as Mithrandir === cjwatson_ is now known as cjwatson [14:26] slangasek: can you please push wxwidgets2.8 (universe) through the queue, it fixes a postinst-failing bug (bug #203526) [14:26] Launchpad bug 203526 in wxwidgets2.8 "python-wxgtk2.8 uninstallable due to post-installation script error" [Medium,In progress] https://launchpad.net/bugs/203526 [14:46] slangasek: yes suspending while booted off the cd [14:46] slangasek: works fine on i386, on amd64 kernel it gives those weird errors === wasabi` is now known as wasabi === dholbach_ is now known as dholbach [15:01] Is it possible for me to get rid of openoffice.org or is Xubuntu forced to ship with it because of the lang-pack dependencies? [15:09] cody-somerville, its only lang-support that should pull in oo.o stuff [15:09] use the lang-pack packages [15:11] Okay [15:15] Would someone be kind enough to merge ~xubuntu-devel/ubuntu-seeds/xubuntu.hardy/, regenerate xubuntu-meta, and upload for me? :) [15:16] cody-somerville, I wasn't aware those language packs actually translated xfce anyhow. do they ? [15:17] Probably not anymore [15:18] in the mythbuntu alternate disk at least, the part of d-i that picks languages has been taken out [15:18] cody-somerville: merged; ship seed changes don't need a xubuntu-meta upload [15:18] cjwatson, okay, thanks [15:23] Also, could someone upload the xubuntu-docs for me? [15:23] The bzr branch is located at http://bazaar.launchpad.net/~ubuntu-core-doc/ubuntu-doc/xubuntu-hardy/ [15:25] pitti | Mithrandir: can you please give back haskell-regex-base on sparc? thanks! [15:26] sistpoty|work: done [15:26] :) === amitk_ is now known as amitk === vorian is now known as help === help is now known as vorian [15:42] pitti: will soyuz automatically retry builds that are in dep-wait? [15:43] sistpoty|work: yes [15:43] ogra_cmpc: have you encountered bug 203844 during testing? [15:43] Launchpad bug 203844 in moodle "Hangs when installed in chroot in hardy" [High,New] https://launchpad.net/bugs/203844 [15:44] pitti: excellent, then I hope that I won't need to bother you with too many give backs to complete the haskell transition on sparc :) [15:45] sistpoty|work: don't worry; a give-back is that much --><--- effort for me [15:45] heh :) [15:45] cjwatson, intrestingly not, the package also didnt change since gutsy afaik [15:45] sistpoty|work: as long as your list does not use commas, giving back one or 500 packages is the same amount of work [15:46] but i never tried it in a chroot though (even though i dont see why that should matter here) [15:46] * pitti pats his script to drive the LP UI [15:46] :) [15:47] pitti or cjwatson: can you please approve wxwidgets2.8 currently in the queue? (see above, slangasek doesn't seem to be here) [15:48] Adri2000: will do soon (I currently have ssh hanging again) [15:48] I'll do it [15:48] ogra_cmpc: I'll push it to post-beta [15:49] mvo reported it just yesterday, we talked about it alredy [15:49] crap [15:49] seems to be a fix in Debian [15:49] and io know why i didnt see it [15:49] my ucf on the server is outdated gah [15:49] or maybe not [15:50] does that mean it *should* be beta-critical? [15:50] moodle ? [15:50] nah [15:50] iots just the order that needs to change as i understand it [15:50] ah, it's only shipped not installed by default [15:51] right [15:51] there are no server bits that are installed by default in edubuntu anymore [15:53] ogra_cmpc: anything new on moodle? [15:54] mvo, just discussing [15:55] ogra_cmpc: I think the fix is just moving the dh_stop further down, but I only glanced at it [15:55] yeah, thats how i understood the linked report [15:55] just change the ordering [15:59] libpam-foreground should probably be pulled in as a dependency on ubuntu systems. Otherwise, errors show up in auth.log when using sudo (https://bugs.launchpad.net/ubuntu/+source/ubuntu-meta/+bug/195880). I was directed here to ask what package I should recommend depend on libpam-foreground [15:59] Launchpad bug 195880 in ubuntu-meta "ubuntu-standard doesn't depend on libpam-foreground" [Undecided,New] [16:02] revno: 1138 [16:02] committer: Martin Pitt [16:02] branch nick: ubuntu.hardy [16:02] timestamp: Thu 2007-11-22 17:07:54 +0000 [16:02] message: [16:02] remove libpam-foreground; packages which need it need to depend on it explicitly [16:03] sudo is what causes the error itself... [16:03] I'll reassign the bug to the proper place [16:03] Assigning it as under sudo? Or another package? [16:04] see the bug [16:04] /etc/pam.d/sudo doesn't refer to pam_foreground; there is no reason why it should have the responsibility of delivering the dependency/recommendation [16:06] cjwatson: Right. Looks good. [16:06] speaking of bzr, do we plan on importing bzr 1.3 into Hardy? It seems imminent for release and the guys produce quality releases. [16:07] We were uploading new tool chain bits up to the beta freeze, so I don't see why that would be a problem. [16:07] :) cool [16:08] btw, how's the beta going? [16:08] largely OK except for screwed DVDs [16:08] we'll probably have to just not ship those [16:08] meh that's probably not a big problem for the beta phase [16:09] cjwatson, what happened with the dvds? oversized or failling alltogether? [16:09] mario_limonciell: bizarre gfxboot explosion [16:10] I clearly did something very bad in a previous life, because I'm sitting tracing through its memory allocator at the moment [16:10] ooh yuck. don't they use the same gfxboot code though? [16:10] they do, but it depends on exact layout of stuff in memory, which depends on other things in the /isolinux directory on the image [16:11] whew, didn't realize things were that hardcoded.. [16:11] but then again I do remember you poking around at some of the code at UDS [16:11] and it sure didn't look pretty [16:11] I like the old SYSLINUX put-up-a-splash-image days... [16:12] mario_limonciell: not hardcoding as such - I think it's a bug in its memory allocator which crashes on certain sizes of allocations, or something like that [16:12] the debug output shows a chunk with negative size [16:12] which I think is making it oops [16:28] tkamppeter: hi [16:28] tkamppeter: thanks for the cups patches [16:29] tkamppeter: BTW, I recently committed all our remaining patches to Debian, except the ones we just need for daper->hardy upgrade (which don't fit to Debian) [16:29] tkamppeter: so, in the future we should try to avoid adding a patch to the ubuntu branch only [16:29] tkamppeter: (you did that correctly, I just want to point it out explicitly) [16:29] so after hardy we can just sync cupsys [16:32] pitti, on today's patch I need to do a small correction, so wait a little with the upload. I will inform you. [16:34] pitti, tkamppeter: do we need to default printer menu item in the preferences menu? isn't that something that system-config-printer can also do? [16:36] seb128: hm, s-c-p usually needs 'lpadmin' privileges; I'm not 100% sure, though, I currently don't have a local printer [16:38] ok [16:38] * pitti tries, I forged a printer now [16:38] because I was testing on the CD and having a list with only pdf is not really useful ;-) [16:40] seb128, the default printer menu item in the preferences menu is for every user being able to set his personal default printer. s-c-p can set a system default printer (for all users without personal default printer), but it needs lpadmin privileges. [16:40] pitti: please give back haskell-regex-posix on sparc [16:40] seb128: I tried it, doesn't work; it asks you for the password [16:40] seb128: since it wants to set the global default, not the per-user one [16:40] ok [16:49] pitti, I have committed the corrected patch to the cupsys SVN (both Debian and Ubuntu) now, now I am doing a build test [16:50] doko: I've gotten the PIE patches working for attach and core, but not "run". should I upload that, since it's at least progress? [16:50] doko: (context is "gdb") [16:51] doko: also, I enabled the testsuite. was it intentionally disabled? it looked like a typo. [16:52] not intentionally. I'd upload for hardy [17:00] pitti, I have tested the new cupsys patch now. Can you upload the current SVN snapshot into Ubuntu? Thanks. [17:01] tkamppeter: we are still in freeze [17:03] pitti, can you then upload cupsys after the beta release? The patch is not so urgent. [17:10] doko: could it be that the gutsy-update of libc6 (2.6.1-1ubuntu10) has somehow broken lbreakout2? (bug #196334)? [17:10] Launchpad bug 196334 in lbreakout2 "segmentation fault - lbreakout2" [Medium,Incomplete] https://launchpad.net/bugs/196334 [17:27] tkamppeter: yes === \sh is now known as \sh_away === macd_ is now known as macd [17:38] tjaalton: hi, what kind of "progress" is bug #189343 "in" currently? :) [17:38] Launchpad bug 189343 in linux-restricted-modules-2.6.24 "DRI doesn't work with fglrx 8.01" [Critical,In progress] https://launchpad.net/bugs/189343 [17:40] slangasek: good point, I think it can be closed since 8.3 is in. at least it fixed the issue for the reporter [17:41] ok, I wasn't sure how to check that 8.3 was in - thanks [17:42] 1:7.1.0-8-3+2.6.24.11-12.31 [17:42] see, it's clearly fixed :) [17:44] OBVIOUSLY [17:44] :P [17:45] oh, sorry, I didn't know that "8.3" was supposed to be read as a regexp ;P [17:45] Can any of the buildd admins give back nautilus-share? [17:45] AMD can't decide whether to use . or - :) [17:45] heh [17:47] tkamppeter: where are the specs for the common print dialogue? [17:51] pitti: Can you please give back nautilus-share and mysql-gui-tools? === thekorn_ is now known as thekorn === jw2328_ is now known as james_w === mathiaz_ is now known as mathiaz [17:58] Adri2000: done === lamont` is now known as lamont [17:59] this is odd. I have a DVD which I cannot seem to read on Linux [18:00] plain ol' data dvd. [18:00] cjwatson: thanks [18:00] Looks like it's attempting to access beyond the end of waht it thinkgs is the device. [18:00] Wonder if this is a bug in the iso9660 fs driver or something wacky about this DVD [18:00] or just a screwed up device driver. [18:02] Yeah... weird. blockdev shows it as only being 1.1 GB. Disk is much longer than that. [18:02] the mounted FS says it's 3.8G [18:04] superm1: so has mythbuntu alt amd64 gotten any testing? [18:05] Daviey, ^ [18:05] cjwatson: Are you one of the buildd admins? [18:05] wasabi: did you burn the DVD in XP? [18:06] It's not mine. Came from MS though. It's a MSDN DVD. [18:06] I'm thinking it might be the two SATA DVD drives reading the disk length wrong. [18:06] "or something" [18:07] slytherin: no - ~launchpad-buildd-admins [18:07] It's been pressed. No idea where the original FS was built. Also I don't know how disk length stuff works with a DVD> [18:07] I'd assume it is some disk metadata at the beginning which only the drive cares about [18:08] wasabi: or may be there is problem with UDF driver in Ubuntu [18:08] The UDF driver would not lie about block device size. Only about it once it's been mounted. No? [18:09] Seems to be iso9660 anyways. [18:09] cjwatson: thanks. But it is hard to find who is awake at this time [18:16] slangasek: AFAIK amd64 alt has recieved _no_ testing [18:18] Daviey: will it, in time for us to be able to confidently include mythbuntu in the beta today? [18:18] Daviey: doing your own beta later is always still an option if you need more time... [18:18] hmm [18:18] <_MMA_> Daviey: Well, I as well as TheMuso did 64bit installs of Ubuntu Studio Alt. Pretty much the same thing as the Ubuntu disk. Mostly. [18:19] I don't have an amd64 dev machine.. i can only test it in vm :/ [18:19] one mo. [18:19] _MMA_: that's not the same thing. I'm not willing to announce, and take up archive space, for completely untested images that may be broken for unforeseen reasons [18:20] each image needs to get tested before it goes out in this mail... [18:20] slangasek: yep, just trying to pull someone to test it physically [18:21] If anybody else in here wants to help, that would be MUCH appreciated! [18:27] pitti: does bug 202848 seem valid? Jockey writes "Defaultdepth" which is wrong [18:27] Launchpad bug 202848 in linux-restricted-modules-2.6.24 "GeForce FX5200 drivers xorg bug" [Undecided,New] https://launchpad.net/bugs/202848 [18:27] <_MMA_> slangasek: I never suggested otherwise. :) [18:27] should be DefaultDepth [18:31] Riddell, the specs for the Common Printing Dialog are in the works by Peter Sikking and his current student. They will be ready end of May when we will have the coding kick-off for the dialog end of May (directly before LinuxTag) in Berlin. [18:33] If we are lucky, we will have two GSoC students working on the implementation of the dialog. The coding session of the GSoC begins end of May. See also https://www.linux-foundation.org/en/OpenPrinting and https://www.linux-foundation.org/en/Google_Summer_of_Code. I am GSoC admin for the Linux Foundation. === hunger_t is now known as hunger [18:39] cjwatson: are there known problems with non-working keyboard at the boot screen (new MacBook Pro)? [18:42] doko: is that with DVD ? [18:45] doko: macbooks don't always bring up legacy keyboard support [18:45] doko: I have to use rEFIt to reliably have keyboard on bootloader [18:49] doko: If you are not too busy, can you please give back nautilus-share and mysql-gui-tools? [18:56] doko: if it's DVD, that's known [18:56] doko: if it's not, could be USB issues [18:57] holding down shift at boot might help [18:57] cjwatson, that's the only known issue on the dvd right? We don't use gfxboot for our factory install, so it wouldn't affect us then. [18:57] pitti: I had some fsck related issues on boot yesterday; would you like to talk about them? [18:57] mario_limonciell: we can't tell because we can't boot it ;-) [18:57] hehe [18:58] cjwatson, well you can probably boot a usb stick with grub (or similar) installed, and call upon the initrd and vmlinuz files in casper/, I suppose, but fixing gfxboot is more of priority [18:59] I've found the gfxboot problem, so we shouldn't need workarounds [18:59] too late for beta though [18:59] going to re-generate one shortly after beta then at least? [19:05] stgraber: yes, dvd [19:06] can someone explain me how the XubuntuY.Z versions work [19:06] doko: ok, so it's a know gfxboot bug [19:06] y know that on debian a package versioned by -X is versiones as -X.Y if it is a NM upload [19:06] known [19:06] but on ubuntu why to we use ubuntuX.Y? [19:08] nxvl: X.Y is used for security and stable release updates, where we need something between X and X+1 [19:08] cody-somerville: I see quite a few outstanding tests on the xubuntu images, according to http://iso.qa.ubuntu.com/qatracker/build/xubuntu/all; do you have anyone available to help round out this testing? [19:08] jdong: that's what i thougt, thanks for it! [19:09] slangasek: given back [19:09] nxvl: XubuntuY specifies that it is a X version from Debian with added changes made in Ubuntu [19:09] doko: what that for me? [19:09] mario_limonciell: right, if I get it sorted before going on holiday [19:10] ahh, yes [19:10] doko: Thanks. :-) [19:10] doko: I'm sure I'm also grateful for it being given back, whatever it is ;) [19:10] (I know where the problem is - just a matter of finding an efficient fix) [19:10] slytherin: yes, that i know [19:10] slytherin: i wanted to know why does ubuntuX.Y exist [19:11] nxvl: ok [19:11] they're updates after the distribution has been released [19:11] ie, security or critical fixes [19:13] slangasek, I'll see about trying to round some people up [19:13] thom: thnx [19:13] cody-somerville: cheers === heno__ is now known as heno [19:33] slytherin: given-back [19:34] mdz: still there? sure [19:34] pitti: doko did that already. And the builds have succeeded. :-) [19:34] tjaalton: hm, that's done by guidance-backends [19:35] tjaalton: so, entirely possible that it is a bug, but so far it didn't matter [19:35] xorg.conf has been pretty much case insensitive so far? === kylem_ is now known as kylem [19:41] pitti: right, maybe some drivers are more picky about it, intel at least works with both cases [19:41] tjaalton: it's not that trivial to fix in guidance-backends unfortunately, I already tried it a while ago [19:42] but if it's actively breaking stuff instead of just being cosmetical, we should do something about it [19:42] tjaalton: I've subscribed to that bug. If there are pointers towards guidance-backends that I can do something with, I'll take a shot at it. [19:42] pitti: I've got some more Guidance work planned after beta, so if we can narrow it down, maybe I can hack something out that'll work. [19:43] cool, maybe I'll ask for the log with the broken setting, that should confirm it [19:44] I know exactly where to patch to look for the downcased letter when guidance reads xorg.conf in and change it if that would be useful. [20:09] doko: you know some bits about wxwidgets, right? Are there any caveats allowing an update of wxformbuilder (bug #203781)? [20:09] Launchpad bug 203781 in wxformbuilder "new upstream release 3.0.56 (RC8)" [Undecided,New] https://launchpad.net/bugs/203781 [20:11] sistpoty: I didn't follow upstream too close ... [20:11] doko: ah, ok... whom could I asked then? [20:13] doko: ok, package is from bug reporter, so nevermind ;() [20:22] doko: is there an up-to-date "here's how to make a python package from scratch" document anywhere? From my perspective things keep changing (python-central, -support, -foo) -- I can't keep up. :) [20:23] keescook: ok, I should write such a doc ... [20:25] doko: I'd really like that. :) even a "use pkg $BLAH as an example" in a wiki page would rock. :) [20:29] bah, seems that hardy started to make some pling and plong noises on suspend/resume. WTH??? [20:29] keescook: I don't know how up to date it is, but I've been using http://wiki.debian.org/DebianPython/NewPolicy [20:29] is that pulseaudio? [20:29] keescook: noone has complained with auth-client-config or ufw :) [20:30] keescook: and by up to date, I mean relevant [20:30] doko: is http://wiki.debian.org/DebianPython/NewPolicy [20:30] what we should be working off of? [20:31] jdstrand: cool, yeah. I just wanted to get some kind of officially blessed cookbook. [20:36] jdstrand: yes, this still looks fine, just add that we prefer python-central for new ubuntu packaging, because it handles upgrades better and lets dpkg/apt doing the file conflict/replacing stuff [20:39] doko: cool, thanks! [20:41] pitti: gnome-power-manager, fix is in the queue [20:41] er, or maybe it's in the queue; I talked to ted about it, not sure if it's been sponsored just yet [20:42] doko: Speaking of Python - I'm not sure pushing python-xml removal late in the cycle is turning out to be such a great idea. I've seen more than one package inadvertently broken. [20:43] ScottK: unfortunate, but ... which packages are broken? [20:44] doko: The most recent was wxwidgets2.8. The guy that broke it was looking after it, but I don't know if he's fixed it yet. [20:44] I'm more worried about breaking things and then not finding out until after release. [20:46] That and most of the remaining ones that I've looked at need code copied into them. [20:47] doko: Did you consider dropping python-xml to suggests for python-reportlab solve the transition or should it be removed entirely? [20:47] ScottK: can be completely removed for reportlab [20:47] doko: OK. I can do that one. [21:32] slangasek: g-p-m> what fix? [21:39] pitti: the pling/plong noises on suspend/resume [21:39] aah [21:39] pitti: it's not in unapproved though, so someone needs to sponsor it for ted still [21:39] great that it's not considered a bling feature *phew* [21:44] hi kids [21:45] Hi mom. === heno__ is now known as heno [22:29] slangasek: Mythbuntu beta amd64 alt seems fine [22:33] Daviey: lovely, thanks :) [22:35] have done, hit the wrong channel first :( [22:38] Is there a rebuild of the cds in progress? [22:39] cody-somerville: no, is there some reason you would expect rebuilds? [22:39] I thought we were planning to release the beta this evening [22:39] yes [22:40] cody-somerville: Which is precisely where there are no rebuilds. :) [22:40] Well, could you please rebuild Xubuntu alternative for i386? [22:41] It is over sized and I committed the fix earlier today [22:42] it can be rebuilt, but it won't be included in the beta [22:42] that fix should've been committed weeks ago, when I was trying to get someone to pay attention to it for alpha6 [22:43] Unfortunately Xubuntu has been going through a leadership crisis [22:44] if the Xubuntu developers want to release Xubuntu beta alternative CDs after the fact, that's fine with me and I'm happy to assist, but I can't hold back the beta for all the other flavors [22:45] Can you start the build and we can just slip in the new iso when it is ready? [22:46] not at this precise moment, my hands are busy with beta publishing [22:46] slangasek: Were you planning on doing an acpi-support upload at some point? [22:46] mjg59: to drop the suspend/resume code? [22:46] slangasek: Because the toshiba_acpi.modprobe file that someone removed needs to go back, or Toshibas are broken [22:46] oh [22:46] mjg59: Beta is already on its way out... [22:47] infinity: Then it'll have to be post-beta [22:47] mjg59: is there a bug report about this? [22:47] slangasek: I doubt it, I just noticed it in the changelog [22:48] ok - if I'm to upload it I'll need to understand why the change was wrong, so I can properly document it against further oopsery [22:48] I pointed out that this wasn't a good idea on ubuntu-devel, but getting anyone to listen to me nowadays is effort :) [22:48] so if you're around in a few hours we can talk about it, or if you file a bug and milestone it I can take it from there [22:48] Yeah, I'll do that [22:48] hrm, I don't recall seeing anything about this on ubuntu-devel recently :/ [22:48] Might have been devel-discuss === pedro__ is now known as pedro_ [22:58] slangasek: You have a 108x88 DPI monitor? [22:59] ion_: 1400x1050, 15", so sayeth xdpyinfo [23:00] slangasek: 108x88 DPI would mean non-square pixels. [23:00] ion_: er, sorry, 1280x800, 15" [23:00] Most laptop screens are non-square, ish. [23:00] yes, I'm aware that implies non-squareness; you're welcome to deduce xdpyinfo's math error if you like :) [23:00] slangasek: Try measuring the display's size and adding a DisplaySize setting to xorg.conf and see if it the X and Y DPIs are still different. [23:01] ion_: Mine is 122x127 DPI. [23:01] Although, 108x88 seems a bit more extreme. [23:05] slangasek: 199888 [23:06] blueyed: Please don't delete files from packages without trying to find out why they were added in the first place. In this case, you broke hotkeys for all Toshiba owners. [23:07] blueyed: Also, there was no point adding any suspend/resume related code to acpi-support in hardy. It's entirely unused. [23:07] blueyed: In summary, please try to work out what a package is meant to do before uploading new versions :) [23:10] mjg59: I know, it was a bunch of collected patches/supposed fixes. [23:10] blueyed: Yes, but it, well, wasn't [23:11] It was a bunch of fixes and breakage [23:11] slangasek: I can't milestone any more [23:12] mjg59: ok, done [23:12] slangasek: Argh. You'll need to delete events/lenovo-eject as well. [23:13] It's a dock eject button, not a CD eject button [23:13] mjg59: is there something that makes it work for that? [23:13] The firmware ought to basically handle it [23:13] But I don't have a dock to test [23:13] hmm [23:13] Using it as a CD eject button is definitely wrong, though [23:14] then isn't it also a problem that I'm able to see an acpi event for it at all? [23:14] mjg59: say, my crazy disasming friend of mine is coming over shortly, got a tarball of the usplash+libdisasm goo handy? I don't want to reinvent the wheel... :) [23:15] keescook: I've been trying to think through the possible ways of doing it [23:15] keescook: I'm beginning to think it's not practical without an nx machine [23:15] Oh, no, wait [23:15] mjg59: toshiba hotkeys problem because of bug 180678 ? [23:15] Launchpad bug 180678 in acpi-support "dmesg: toshiba_acpi: Unknown parameter `hotkeys_over_acpi'" [Medium,Fix released] https://launchpad.net/bugs/180678 [23:15] nx in this context is? [23:15] blueyed: Yes [23:16] blueyed: Don't delete that file. It's there for a reason. [23:16] blueyed: And #194609 even has someone commenting that the proposed patch is wrong! [23:16] mjg59: but it said "toshiba_acpi: Unknown parameter `hotkeys_over_acpi'"?! [23:16] blueyed: Yes, but that was a *kernel* bug [23:16] Not an acpi-support bug [23:16] The kernel's been fixed [23:18] slangasek: #194609 could do with milestoning as well [23:19] mjg59: oh, I see. I'll fix it [23:19] how's about I just talk someone into sticking a milestoner bit back on you... :) [23:20] slangasek: I think that's it for now :) [23:24] keescook: Ok, so here's roughly how I think it should work: [23:24] keescook: Map the BIOS [23:24] keescook: Register a SIGILL handler [23:24] keescook: Disassemble the instruction at the IP. Place an illegal instruction immediately after it [23:25] is fixing that why my power button works again? [23:25] keescook: If the instruction is a jump or a loop, also place an illegal instruction at the destination [23:25] (toshiba laptop) [23:25] keescook: If it's a call, place an illegal instruction at the destination [23:25] pwnguin: Unlikely [23:25] keescook: Enter vm86 mode [23:26] mjg59: but the thinkpad one really should go away, right? [23:26] slangasek: Yeah, I think that's ok [23:26] (I mean, as soon as I killed it locally, my hotkeys started working again, so) [23:27] keescook: In the sigill handler, invalidate the mapping and repeat [23:27] keescook: Oh! And print the instruction and registers, obviously :) [23:27] mjg59: heh. yeah, that part. :) [23:27] that mapping/unmapping is not currently happening? [23:28] keescook: None of this is currently happening [23:28] I was trying to find some way of avoiding some of this awkwardness, but I think this is the cleanest way [23:28] mjg59: right, -- i'm just trying to understand how vm86 works -- I haven't spent too much looking at it. [23:28] keescook: I found the gfxboot bug! [23:29] cjwatson: ooooh! what was it?? [23:29] (sorry, not to distract you) [23:29] keescook: vm86 is given a structure with original register state, and then executes stuff [23:29] yes, let's not distract kees from his bios sigill handling with discussions of forth code... :-) [23:29] * keescook cries [23:29] keescook: So, in theory, when you hit the sigill handler, the registers are now what they would have been after you executed that instruction [23:29] keescook: files read by gfxboot aren't zero-terminated (it gives you back a pointer to the middle of a cpio archive in memory), but we were using strstr on them and strstr expects zero-termination [23:30] heh [23:30] asac: Is there still no way to add https exceptions in epiphany? [23:30] and then we plonked a zero byte after the end of the string which happened to sit on top of the metadata for the next malloc chunk [23:30] mjg59: right, so, the BIOS mapping part -- do you mean duplicate the bios memory elsewhere so we can modify to with the illegal intrs? [23:30] keescook: mmap it from /dev/mem with PROT_FIXED [23:30] cjwatson: oooh snap. nice catch. how much longer did you spend on that madness? [23:30] keescook: That's done already [23:31] keescook: The issue is that you need to modify that mapping to include the illegal instructions [23:31] keescook: So you need to munmap() it and then mmap() it again afterwards to get the original contents back [23:31] mjg59: err.. and it'll let me _change_ the memory? right, this is where I'm a bit confused [23:31] keescook: It's the difference between physical and virtual address space [23:31] keescook: about two or three hours; I discovered a gfxboot primitive that dumps out all the malloc metadata, and discovered that it printed "oops: " when stuff went wrong, so given that I was able to binary-chop and figure out what was corrupting memory [23:31] ah, it'll let me mod it, but it's only for my mapping [23:31] keescook: You're mmapping 0xc0000 of physical address space into 0xc0000 of your virtual address space [23:32] cjwatson: nice -- did you submit that upstream? [23:32] And yeah, you can make it COW [23:32] keescook: not relevant [23:32] cjwatson: ah, dang [23:32] keescook: the code that was broken was entirely written by me [23:32] cjwatson: erf. heh. [23:34] mjg59: (not to interrupt you), do you have any opinion if we want eeepc-acpi-scripts as FFe (bug 195724) [23:34] mjg59: this'll be ... not fast. ;) [23:34] Launchpad bug 195724 in ubuntu "[FFe] Please sync eeepc-acpi-scripts 1.0 from debian unstable" [Undecided,Incomplete] https://launchpad.net/bugs/195724 [23:34] sistpoty: Erm. No clue. [23:34] mjg59: ok, thanks [23:35] I'd need to take a look at the package first [23:35] keescook: Yeah, but nor is x86emu, so... [23:35] keescook: The trick is then to try to get the debug output to look like the x86emu debug output [23:35] mjg59: you had said something the other day about bios memory and how it wasn't going to work? [23:35] mjg59: if you could do so, and comment on the bug, that would be great :) [23:35] keescook: Yeah, that was based on my original idea [23:35] But I think the current one avoids that [23:36] mjg59: ah, okay. yeah, this is just single-stepping. should work fine. ;) [23:36] why does it need to look like x86emu debug output? [23:36] keescook: So we can diff them [23:36] And figure out where it's going wrong [23:36] ah! heh. [23:36] so, remind me, we're just trying to do a simple POST, right? [23:37] or is it deeper? I'm realizing I need to actually get a "reproducer" of the calls that are failing now that I've got my hands on hardware [23:38] better yet -- what do I need to tweak to get the x86emu debug output? [23:40] keescook: We want to know why usplash dies on nvidia with x86emu, but not with vm86 [23:40] Hang on... [23:41] keescook: http://www.codon.org.uk/~mjg59/tmp/x86_debug/ [23:42] mjg59: eexcellent. so that for libx86 -- how do I run usplash with x86emu? [23:43] I believe it's nobbled so it always runs with x86emu [23:43] keescook: Oh, and remember that this all has to be on a 32-bit kernel [23:44] sistpoty: No, that package shouldn't exist [23:45] mjg59: ok, thanks, then I'll reject it [23:45] sistpoty: acpi-support should already do all of it [23:45] mjg59: thanks again for the info :) [23:49] mjg59: okay, I think I know where my confusion lies. vm86 is a syscall. x86emu is a library, libx86 is a library. usplash is linked against libx86. where is x86emu, and who is currently using vm86 calls? :P [23:51] libx86 ( by default) uses x86emu on x86_64 and vm86 on x86 [23:51] The magic code needs to be added to lrmi.c in libx86 [23:51] lrmi is the layer that does the vm86 handling [23:53] ah, gotcha. (currently reading http://www.codon.org.uk/~mjg59/libx86/) [23:54] okay, so, then why do I want to run a 32bit kernel? just to get the "it works" debug log? [23:54] Yes [23:55] okay. same page achieved. :) [23:55] vm86 doesn't work on 64-bit kernels [23:55] right -- that part i got. you didn't mean "all work will be done on 32bit kernels" [23:55] Oh, right [23:55] Once we've got a vm86 log, the bugs can be fixed on 64-bit [23:55] you meant, get a working dump on 32bit, then attack 64 [23:55] Yeah [23:55] groovy [23:55] But x86emu runs on 32-bit as well [23:55] And fails in the same way [23:56] and to force x86emu when running 32bit, I'll want the libx86 package from your tmp dir you just sent me? [23:57] Yup [23:57] Which also does the debug output [23:57] * keescook claps stupidly [23:58] heya people [23:58] hi emgent [23:59] keescook: about pentest team, next Security meeting (26th) we can start the project. [23:59] :) [23:59] cool [23:59] report-tool it's ready, i will update bzr branch to it