[00:01] <Zelut> ok, I'm PXE booting ubuntu-server and I have: "append initrd=ubuntu/initrd.gz preseed/url=http://server/path/to/preseed.txt" in the pxeconfig.  Is that not valid?
[00:26] <xivulon> evand #226622 is all yours I uploaded the patches, feel free to edit the error msg
[06:50] <cjwatson> Zelut: fine as far as it goes. Is something going wrong?
[09:38] <GeorgesLeYeti> hello i'm looking for a preseed file for ubiquity
[09:38] <GeorgesLeYeti> where can i get one for 8.04
[12:57] <CIA-1> clock-setup: cjwatson * r194 clock-setup/debian/changelog: releasing version 0.95ubuntu1
[12:57] <CIA-1> partman-basicfilesystems: cjwatson * r562 ubuntu/ (82 files in 13 dirs): merge from Debian 59
[13:00] <CIA-1> partman-basicfilesystems: cjwatson * r563 ubuntu/init.d/auto_mountpoints: adjust indentation to match Debian
[13:02] <CIA-1> partman-basicfilesystems: cjwatson * r564 ubuntu/debian/changelog: releasing version 59ubuntu1
[13:14] <CIA-1> partman-ext3: cjwatson * r734 ubuntu/ (63 files in 3 dirs): merge from Debian 51
[13:19] <CIA-1> partman-ext3: cjwatson * r735 ubuntu/commit.d/format_ext3: adjust indentation to match Debian
[13:20] <CIA-1> partman-ext3: cjwatson * r736 ubuntu/debian/changelog: releasing version 51ubuntu1
[13:31] <CIA-1> partman-jfs: cjwatson * r723 ubuntu/ (63 files in 3 dirs): merge from Debian 26
[13:39] <CIA-1> partman-jfs: cjwatson * r724 ubuntu/debian/ (changelog control): Set Vcs-Bzr for Ubuntu.
[13:40] <CIA-1> partman-jfs: cjwatson * r725 ubuntu/debian/changelog: releasing version 26ubuntu1
[13:42] <Zelut> cjwatson: I noticed last nite that I have to put the language, keyboard and basic network settings in the PXE config to get that automated.
[13:43] <Zelut> cjwatson: however, if I use a kickstart syntax file (or even a hybrid) it all automates.
[13:56] <xivulon> evand, please review patches for #224697 and #226622 (initramfs-tools + lupin.proposed)
[13:57] <xivulon> also see yesterday night discussion re #207137 (in short: I cannot find anything wrong with the code, and am out of ideas :)
[14:03] <evand> on it
[14:03] <xivulon> thx
[14:04] <xivulon> for #217593 is it possible to add a reboot after starting ubiquity in /etc/init.d/ubiquity if automatic=1
[14:04] <xivulon> ?
[14:06] <xivulon> or to rephrase, any reason to do anything else other than rebooting when we start with ubiquity in unattended mode and ubiquity has completed/died?
[14:13] <evand> hrm
[14:14] <evand> yes, if ubiquity dies in say only-ubiquity mode or automatic-ubiquity mode (wubi excluded) we want the user to end up at the live cd desktop
[14:14] <evand> otherwise, imho, the result is even more confusing.
[14:15] <xivulon> hmm not sure I agree, wouldn't many people end up thinking that the installation was successful when they are in fatc in a live CD env?
[14:16] <evand> while that's possible, yes, they could also believe the install finished if the computer reboots
[14:17] <xivulon> maybe the best option is to get the ubiquity exit value and in case show a dialog saying installation failed, do you want to reboot or go in a Live CD desktop (r/o)?
[14:18] <cjwatson> or stick something in /var/lib/update-notifier/user.d/ so that they get a popup saying "installation failed; this desktop is so that you can investigate" or words to that effect
[14:18] <cjwatson> definitely a good reason not to immediately reboot - that loses us any possibility of investigating the failure
[14:19] <evand> indeed, I like the update-notifier idea
[14:19] <xivulon> A live desktop + notification would be perfect methinks
[14:19] <evand> ok, I'll add that to my todo list for the day
[14:20] <xivulon> can I reassign 218593 to ubiquity then?
[14:20] <xivulon> 217593
[14:22] <evand> done
[14:22] <xivulon> thanks
[14:35] <CIA-1> partman-reiserfs: cjwatson * r802 ubuntu/ (65 files in 5 dirs): merge from Debian 41
[14:37] <CIA-1> partman-reiserfs: cjwatson * r803 ubuntu/debian/changelog: releasing version 41ubuntu1
[14:48] <xivulon> evand for #225593 I think that the easiest way is to add a link /etc/skel/Host -> /host within lupin-support.postinst
[15:05] <evand> imho, I'd much rather fix it properly
[15:13] <cjwatson> please don't add things to /etc/skel
[15:13] <cjwatson> it's deprecated in Debian policy for a good reason
[15:14] <cjwatson> (adding things to it, not the directory itself)
[16:06] <xivulon> evand, I do think that fixing it might be a bit too extensive at this stage. I will give a look at nautilus/gnome-volume-manager/hal code though to confirm
[16:07] <xivulon> cjwatson if /etc/skel is not an option is there equivalent to it that could be used (assuming above assumption)?
[16:57] <cjwatson> xivulon: fixing it in gio/gvfs/whatever it is should be a one-liner or close to it. Talk with seb128
[16:57] <xivulon> hehe was navigating through gnome not familiar with that :)
[16:58] <evand> xivulon: I'm seeing quite a few reports in #ubuntu of being dropped into a busybox shell after updating.  Are you seeing these as well?
[16:58] <xivulon> I am at g_unix_mount_monitor_new
[16:58] <xivulon> evand, is this wubi?
[16:58] <evand> xivulon: ja
[16:59] <xivulon> no I haven't seen many reports
[16:59] <xivulon> so far...
[16:59] <xivulon> do you know if any particular update is relevant? is update-initramfs/update-grub triggered?
[17:00] <xivulon> most users that end up in busybox do so because of dirty flag in ntfs
[17:00] <xivulon> which is unrelated to updates
[17:00] <evand> I just noticed it in my #ubuntu backlog.  I'll follow up with them.
[17:00] <evand> Indeed, I've been telling people to chkdsk first.
[17:00] <xivulon> often simply booting into windows and then shutting down cleanly is enough
[17:00] <evand> And if it's still a problem to go the usual route of looking at casper.log and filing a bug.
[17:01] <xivulon> of course there will be no casper.log after installation
[17:01] <xivulon> if they boot with debug kernel option there will be a log in /tmp (IIRC)
[17:01] <xivulon> usually rescue mode will reveal the issue though
[17:03] <xivulon> this is wht 226622 is very important, not a real bug per se' but it generates LOTS of tickets...
[17:06] <evand> indeed
[17:12] <xivulon> evand, cjwatson, in my Mount Desktop icons traceback I ended up to mtab_file_changed
[17:12] <cjwatson> talk with seb128
[17:12] <cjwatson> I don't know that code well and am not prepared to review changes
[17:12] <xivulon> the issue being that /host is not in mtab and unlikely to be there
[17:12] <xivulon> not sure it is a gnome issue
[17:13] <cjwatson> seb128 will have a MUCH better starting point than me
[17:13] <cjwatson> failing that, try pitti
[17:13] <cjwatson> (hal)
[17:13] <xivulon> will ping seb128
[17:14] <xivulon> but as mentioned /host will only appear in /proc/mounts not in /mtab
[17:54] <xivulon> evand, in case you missed the discussion on #devel, for #225593 we go for bookmarks
[17:56] <evand> indeed, I've been following along
[18:01] <xivulon> evand, would be good to add an extra bookmark for My Documents in m-a
[18:04] <evand> it's already copied to the Documents folder.
[18:07] <xivulon> you mean the content? I think wubi skips that part :(
[18:08] <xivulon> even a bookmark might be complex though (since parent device might not be mounted)
[18:13] <evand> Oh, I see what you're saying.  I'm not keen on the idea of symlinking back to the Windows file system in migration-assistant.
[18:13] <evand> Part of the point of migration-assistant is copying documents and settings over so that you can erase Windows without fear of losing anything important.
[18:14] <evand> I'm afraid that if it started to symlink back to Windows, it would not be clear that the documents still resided on the NTFS partition.
[18:16] <xivulon> I agree :)
[18:18] <xivulon> http://www.brainofshawn.com/2008/05/07/wubi-what-a-fun-word/
[18:23] <evand> painful to watch, but fantastic.
[18:37] <xivulon> hahaha
[18:59] <tmmoyer> is it possible to disable virtual terminals in the installer?
[21:20] <Zelut> cjwatson: have a minute to give me some preseed tips?
[21:53] <Zelut> cjwatson: is it known that kickstart will cause DHCP to happen twice?
[22:15] <cjwatson> Zelut: I'll answer your questions from last night first
[22:16] <cjwatson> Zelut: yes, it's known that with preseeding you have to put certain things in the PXE config; eliminating that requirement was something I did intentionally when implementing Kickstart compatibility
[22:16] <cjwatson> Zelut: the requirements with preseeding are documented in the installation-guide
[22:17] <Zelut> which I found last nite.  I was confused as the preseed file appeared to have language/keyboard settings in it but they were also needed in the PXE config.
[22:17] <cjwatson> Zelut: Kickstart and DHCP: I *think* that's a bug; netcfg should end up configured once Kickstart is done, at least for a network installation. I'd appreciate a bug on kickseed with instructions on reproducing.
[22:18] <cjwatson> Zelut: you don't need the language/keyboard stuff in the preseed file if it's also in the PXE config
[22:18] <Zelut> ok, I'll head to LP and put something together
[22:18] <cjwatson> it's a little difficult because there are installation modes where it does make sense to put those items in the preseed file (e.g. initrd preseeding)
[22:19] <cjwatson> so the installation-guide just has to say "here are the things you can preseed, and here's how to express them on the kernel command line in the event that you need to do that"
[22:19] <Zelut> I'm assuming it doesn't hurt to have them in both places?
[22:19] <cjwatson> no, doesn't hurt
[22:19] <cjwatson> oh, hmm
[22:19] <cjwatson> sorry, it's been a while since I wrote kickseed :-) Just found this comment I wrote:
[22:19] <cjwatson>         # Run the postinst by hand so that main-menu will run netcfg again
[22:19] <cjwatson>         # later, possibly with different preseeded answers.
[22:19] <cjwatson>         /var/lib/dpkg/info/netcfg.postinst configure 2>&1 | logger -t netcfg
[22:20] <cjwatson> so that is intentional - I think that's because, conceptually, the first round of network configuration is very simple, just enough to retrieve the Kickstart file, and it's possible that the Kickstart file itself might then set up different network configuration
[22:20] <cjwatson> I *think* this duplicates Anaconda's behaviour given similar initial conditions
[22:20] <cjwatson> so probably not a bug, even though it is objectively a bit weird
[22:21] <Zelut> I use anaconda quite a bit with RHEL5 and that behaviour doesn't seem familiar..
[22:21] <Zelut> ..or at least not any way that I've used it.
[22:22] <cjwatson> hmm, admittedly this dates from RH9 or something like that
[22:22] <cjwatson> go ahead and file the bug anyway then, worst case I mark it invalid later ;-)
[22:22] <cjwatson> I can't really remember what was in my mind when I wrote that - I might have considered it an improvement
[22:22] <Zelut> what is kickseed written in?
[22:23] <cjwatson> shell
[22:24] <cjwatson> the only alternative at that point in the installer is C and shell was a *lot* faster to develop
[22:25] <cjwatson> it actually started out, briefly, as a Python program that ran outside the installer and did static translation, until I realised that the semantics of %pre and %include made that impossible
[22:26] <Zelut> that is odd.  I just added %packages to include LAMP, openssh-server, logwatch & denyhosts and told it to install from my mirror vs the http-exported CD contents and it added openoffice
[22:28] <Zelut> if its based from the http-exported ubuntu-server CD it chokes, of course, on denyhosts as that's not on the CD.
[22:37] <Zelut> cjwatson: would it be possible for kickseed to support the firewall option now that ufw is available?
[22:37] <tormod> xivulon: hi, I haven't got much user testing feedback on bug #136682, but I think the patch is finished now.
[22:42] <xivulon> tormod, I will test it in an hour or so
[22:43] <xivulon> doesn't seem anyone is having problems though, which is encouraging
[22:46] <cjwatson> Zelut: you probably need to add 'preseed pkgsel/install-language-support boolean false' to your Kickstart file. (Sorry.)
[22:47] <cjwatson> Zelut: firewall> possibly, though I haven't looked into ufw at all really
[22:47] <cjwatson> Zelut: you can grab the kickseed source package and use the ./test-kickseed program to see how it translates Kickstart files to preseeding; handlers/firewall.sh is the relevant bit
[22:48] <Zelut> well I'm doing a bunch of kickseed/preseed testing tonite.  If ya'll need any bugs verified let me know.
[23:45] <Zelut> cjwatson: I think I may be able to add some basic ufw support.  How would I test it?
[23:48] <cjwatson> create a sample kickstart file and feed it to ./test-kickseed
[23:49] <cjwatson> and check the output for sanity
[23:50] <Zelut> also I see in one of your examples you're using sed 's/foo/bar/' < input > input.new; mv input.new input.
[23:50] <cjwatson> for live testing, the most practical way is probably to add BOOT_DEBUG=3 to your kernel command line, use nano to edit your changes into /lib/kickseed/handlers/firewall.sh on the fly, and exit the shell a couple of times until it proceeds
[23:50] <Zelut> is sed -i not preferred for any reason?
[23:50] <cjwatson> I think at the time sed -i may not have been reliably available in busybox; it's fine to use it now
[23:51] <Zelut> ok, I'll use sed -i in my patch then.
[23:51] <cjwatson> I've made the corresponding change to handlers/auth.sh in my local tree
[23:52] <cjwatson> Zelut: ... and committed upstream
[23:52] <Zelut> cool
[23:53] <cjwatson> ufw unfortunately probably can't go upstream, being rather Ubuntu-specific
[23:53] <cjwatson> at least in its status as the default firewall
[23:53] <cjwatson> so I get to diverge kickseed again and merge it forever :-/