[08:12] <macogw> i have had really annoyingly big issues with installing feisty from cd (i can upgrade over the internet from edgy just fine), and i *think* it's to do with OOo because it's an OOo file that can't be copied when ubiquity segfaults and it's OOo that won't install claiming dependency problems when i try to install ubuntu-desktop from the alternate cd (with apt-cdrom add) after doing a "text only install" from alternate cd (alternate cd fails if
[08:12] <macogw> i let it try to install the gui by default)
[08:13] <macogw> i burned only at 4X speed, i checksum'd the isos (i downloaded it 2x for live, once alternate), i checksum'd the burned disks, and i did "check disk for defects" before installing.
[08:14] <macogw> i used the alternate cd to successfully on two friends' computers, so i know it works fine in general and is a good burn
[08:16] <macogw> i used one of the live cds last night to install on my other friend's new computer which has unsupported nvidia graphics (had to get the .run from nvidia.com), and tried using that same (obviously working) disk today to install with "ubiquity --debug" on my laptop which is 99% supported (2 of the 4 slots on my 4-in-1 card reader lack working drivers but hey all 4 lacked drivers in dapper and that installs fine)
[08:19] <macogw> my cd drive is fine, my hard drive is fine.  i can install dapper or edgy with no problem. it's -just- feisty.  it is driving me insane.  as far as i can tell, feisty has no compatibility issues with my laptop.  i can upgrade to it from edgy without a problem.  and while yay i can do that, that's a workaround.  it should be installable from the cd.  after ubiquity crashes, everything in the live environment tends to go to hell, so the first t
[08:19] <macogw> ime with --debug i couldnt get the logs off (it said i didnt have permission to use vi or gedit or nano!), but this time i managed to get them.  the log suggests that it's a bad cd drive (but i can install dapper/edgy, so no) or bad cd (but i can install on other comps with this cd, so no), so that doesn't help.  all i know is it says it's an OOo file where it fails, and the fact that OOo is what refuses to install on ubuntu-desktop with a wo
[08:19] <macogw> rking text-mode-only install makes me think that its linked to that
[08:20] <macogw> does anybody have any suggestions of a way to get more info about whats wrong to pin it down?
[08:47] <evand> macogw: take a look at /var/log/syslog, look for device errors.  If you're confused as to what they mean, paste them here (without flooding, of course)
[08:47] <macogw> there's tons of squashfs errors
[08:48] <macogw> i watched the output with tail -f to see it as it broke
[08:52] <macogw> however, this looks like its the actual cause of the segfault (it had squashfs errors before the crash with no effect, but crashed at this part) http://paste.ubuntu-nl.org/28763/
[08:56] <macogw> which is python error stuff
[09:00] <cjwatson> macogw: if there are squashfs errors first, then those are the cause
[09:00] <cjwatson> and I'm afraid that it *is* due to a problem with your CD drive or with the disk
[09:00] <macogw> all 3 disks?
[09:00] <cjwatson> it is probably just on the edge of some tolerance or other
[09:00] <cjwatson> then chances are the CD drive
[09:00] <cjwatson> I'd recommend getting a drive lens cleaning kit and using it
[09:01] <superm1_> perhaps the brand of the blank disks that you had purchased
[09:01] <macogw> then how did i install edgy?
[09:01] <cjwatson> I have no idea, but if it were a genuine problem with the CD images it would affect everyone, not just you
[09:01] <cjwatson> there's a faint possibility that the kernel driver for your CD drive has regressed somehow
[09:02] <cjwatson> but some physical problem would definitely be the first thing I'd investigate
[09:02] <macogw> so i should ignore the python exception about invalid literals then?
[09:02] <cjwatson> we've had the squashfs maintainer working on making squashfs fail more obviously rather than the live environment just gradually going to hell in a handbasket, which should make it easier to report such errors properly
[09:03] <cjwatson> the python exception is almost certainly a knock-on effect - it's just the first thing that happened to visibly fall over
[09:03] <macogw> the problem with hell in a handbasket is that it makes it impossible to open a text editor and see the logs
[09:03] <cjwatson> you're screwed anyway in this case
[09:04] <cjwatson> the live environment isn't going to last long once read errors from the CD start zapping random bits of important stuff
[09:04] <macogw> one error that i noticed on the live cd after it goes to hell is that adding applets to the panel gets OAFIID errors popping up
[09:04] <cjwatson> what that python exception actually means is that some subsidiary program that ubiquity was running crashed, or possibly printed garbage
[09:05] <macogw> which is something i had happening on installed-feisty during the unstable bit
[09:05] <macogw> OAFIID is what gconf refers to things as though O_o
[09:05] <cjwatson> could easily be the same sort of thing, code for "program fell over"
[09:06] <cjwatson> "panel failed to communicate with applet" or some such
[09:06] <macogw> mmm where do you get one of those cleaning things?
[09:06] <cjwatson> in the UK, electronics stores stock them
[09:06] <macogw> i really dont wanna try sending my laptop back to the manufacturer.  those idiots think reinstalling windows fixes it
[09:06] <macogw> they reinstalled windows when the problem was overheating due to hardened thermal grease
[09:06] <cjwatson> no, I doubt that would help
[09:07] <superm1_> in the US, best buy, frys, or any shop that sells cds
[09:07] <cjwatson> the cleaning kits usually look like CDs with little brushes attached
[09:07] <macogw> mmmk
[09:07] <cjwatson> I'd be interested in the logs from the alternate CD install to ensure it's the same thing
[09:07] <macogw> how do i get logs from that one?
[09:07] <cjwatson> I suspect the reason OOo is affected both times is that it and its dependencies are just so damned big that statistically it's quite likely
[09:07] <macogw> alternate cd wont automount my flash drive for me to save them
[09:08] <macogw> well on alternate if i do a non-gui install, it works fine and installs
[09:08] <macogw> but its not very pleasant to not have a gui
[09:08] <cjwatson> if you have the machine connected to a network, you can go back to the main menu, select "save debug logs", start up a mini web server from there, and connect to the machine via HTTP
[09:08] <macogw> i can only have one computer online at a time and i would have no idea how to do what you just said
[09:09] <cjwatson> I mean a local network; do you have ethernet connections between multiple machines at home?
[09:09] <macogw> no
[09:09] <macogw> there's only one machine here usually
[09:09] <macogw> i'm just home from school with my laptop and havent convinced them to get wireless so i can use my computer
[09:10] <cjwatson> ah. if it has a floppy drive, then the "save debug logs" option can use that
[09:10] <macogw> laptops dont have floppy drives
[09:10] <macogw> at least not nowadays
[09:10] <cjwatson> otherwise you have to mount the flash drive by hand from tty2
[09:11] <macogw> used to be you could yank out the cd drive and put in the floppy drive, but that wont work running from a cd :p
[09:11] <cjwatson> USB floppy drives still exist
[09:11] <macogw> :-/ how do you mount a flash drive?
[09:11] <macogw> i mean, i can mount hard drives
[09:11] <macogw> because its not hard to figure out which hdb or sdb wasnt there before
[09:12] <cjwatson> if there's no hd* or sd* for it, you may just not be able to
[09:12] <cjwatson> you can look at the logs ('nano -v /var/log/syslog' on tty2, page down to near the end) and see if there are obvious input/output errors
[10:58] <tepsipakki> hum, netcfg could/should have a longer dhcp-timeout, or a configurable one
[10:58] <cjwatson> it is configurable
[10:58] <tepsipakki> really?
[10:59] <cjwatson> preseed netcfg/dhcp_timeout
[10:59] <cjwatson> it's an integral number of seconds, default 15
[10:59] <tepsipakki> not in feisty though :)
[10:59] <tepsipakki> but good to know
[10:59] <cjwatson> is too
[10:59] <cjwatson> that's been around for several releases
[11:00] <tepsipakki> hum, not in my debconf-selections.. I'll dig elsewhere
[11:00] <cjwatson> it was introduced in netcfg 1.13, pre-dapper
[11:01] <cjwatson> it may well not be in debconf-get-selections, but that doesn't really indicate anything since it's never asked and just fishes the default out of templates
[11:01] <cjwatson> it's documented in the installation guide in the preseeding appendix
[11:01] <tepsipakki> ok, excellent
[11:06] <tepsipakki> yeah, there it is.. shame on me
[01:02] <CIA-19> ubiquity: cjwatson * r2127 ubiquity/ubiquity/frontend/mythbuntu_ui.py: PEP-8 import spacing
[02:28] <CIA-19> ubiquity: cjwatson * r2128 ubiquity/ (debian/changelog ubiquity/frontend/gtk_ui.py): - Suppress some spurious uncaught exceptions on startup.
[02:38] <CIA-19> ubiquity: cjwatson * r2129 ubiquity/ubiquity/frontend/ (gtk_ui.py kde_ui.py): take oem_config_title out of language_questions and handle it separately; fixes crash in language_treeview signal handler
[02:39] <cjwatson> superm1_: that should address the problem that (IIRC) you reported
[02:39] <cjwatson> one of those things you have to see yourself before it's obvious ...
[03:15] <CIA-19> ubiquity: cjwatson * r2130 ubiquity/ubiquity/frontend/ (gtk_ui.py kde_ui.py): set window title correctly in OEM mode when changing language too
[03:18] <CIA-19> ubiquity: cjwatson * r2131 ubiquity/ubiquity/frontend/ (gtk_ui.py kde_ui.py): set username_edited when hardcoding user name to oem
[03:18] <cjwatson> gar, that didn't work either
[03:19] <evand> heh
[03:22] <CIA-19> ubiquity: cjwatson * r2132 ubiquity/gui/glade/stepLanguage.glade: oem id entry should activate default
[03:25] <CIA-19> ubiquity: cjwatson * r2133 ubiquity/ (debian/changelog ubiquity/frontend/gtk_ui.py):
[03:25] <CIA-19> ubiquity: - Don't reconfigure the username entry box as a combo box if
[03:25] <CIA-19> ubiquity:  migration-assistant didn't find any users.
[03:25] <cjwatson> that's more like it
[03:25] <cjwatson> evand: is the test I added there the right one?
[03:27] <evand> indeed it is
[03:27] <evand> thanks
[03:28] <cjwatson> that's actually been bugging me for ages but it was never serious enough to bother fixing until it interfered with oem work just now :)
[03:29] <cjwatson> commit early, commit often ...
[03:29] <evand> indeed
[03:29] <cjwatson> right, time to take the dog for a w-word
[03:29] <cjwatson> (she gets excited if you unwarily say that word without obfuscation)
[03:30] <evand> hahaha
[03:30] <evand> what breed?
[03:30] <cjwatson> labrador/rottweiler cross
[03:30] <evand> adorable
[03:30] <cjwatson> utterly placid
[03:30] <evand> sounds big though.  I've seen some quite large rotties in my day.
[03:31] <cjwatson> http://userpic.livejournal.com/57407285/106700
[03:31] <cjwatson> yes, she's a fair size
[03:32] <cjwatson> not like an Irish wolfhound or anything, but
[03:32] <cjwatson> weighs 40-odd kilos
[03:32] <evand> cute
[03:34] <thom> 40kg is pretty sizeable, yeah :-)
[03:53] <cjwatson> oh, I see why ubiquity isn't managing to install packages from the CD
[03:54] <cjwatson> it needs to emulate some stuff that base-installer does, to bind-mount /cdrom into /target and configure apt-cdrom a bit differently
[04:56] <CIA-19> ubiquity: cjwatson * r2134 ubiquity/ (debian/changelog scripts/install.py):
[04:56] <CIA-19> ubiquity: * Replicate the apt configuration done by base-installer (trust CD-ROMs,
[04:56] <CIA-19> ubiquity:  allow unauthenticated packages if debian-installer/allow_unauthenticated
[04:56] <CIA-19> ubiquity:  is true, bind-mount /cdrom into /target, and configure apt-cdrom/apt not
[04:56] <CIA-19> ubiquity:  to mount CD-ROMs).
[04:56] <cjwatson> that's not quite enough though - something also needs to actually install stuff
[05:04] <CIA-19> ubiquity: cjwatson * r2135 ubiquity/scripts/install.py: factor out reading of apt-installed file
[05:24] <superm1_> cjwatson, ooh looks interesting, installing packages that are on the cd in a apt-cdrom compatible repository :)
[05:26] <cjwatson> that was actually intended a long time back
[05:26] <cjwatson> looks like it's been broken though
[05:30] <superm1_> AH
[05:30] <superm1_> oops, ah
[06:26] <CIA-19> ubiquity: cjwatson * r2136 ubiquity/ubiquity/frontend/ (gtk_ui.py kde_ui.py): let's make fullname and username insensitive too if they're uneditable
[06:27] <evand> getting GNOME to ignore partitions seems to be hit or miss.  bug 110904
[06:27] <evand> ah, no ubotu
[06:28] <cjwatson> perhaps he's using Kubuntu; it's not clear
[06:28] <cjwatson> or indeed Xubuntu
[06:29] <evand> oh wow, that didn't even occur to me
[06:29] <cjwatson> both of those definitely had some breakage in that area in feisty
[06:29] <cjwatson> I've added a comment
[06:29] <evand> thanks
[06:57] <CIA-19> ubiquity: cjwatson * r2137 ubiquity/ (82 files in 3 dirs):
[06:57] <CIA-19> ubiquity: * Install packages passed to apt-install even if they aren't on the live
[06:57] <CIA-19> ubiquity:  filesystem, as long as they're available from the apt archive on the CD
[06:57] <CIA-19> ubiquity:  (LP: #114296).
[06:59] <cjwatson> phew, that was more work than I expected ...
[07:00] <CIA-19> ubiquity: cjwatson * r2138 ubiquity/scripts/mythbuntu/mythbuntu_install.py: mythbuntu duplication => more work for Colin
[07:29] <CIA-19> ubiquity: cjwatson * r2139 ubiquity/scripts/mythbuntu/mythbuntu_install.py: remove trailing whitespace
[07:39] <CIA-19> ubiquity: cjwatson * r2140 ubiquity/scripts/mythbuntu/mythbuntu_install.py: let install_extras deal with installing drivers and services
[07:40] <cjwatson> superm1_: I hope r2140 works; I haven't tested it ...
[07:42] <CIA-19> ubiquity: cjwatson * r2141 ubiquity/scripts/mythbuntu/mythbuntu_install.py: imports no longer needed
[07:42] <cjwatson> vim highlights trailing whitespace in python by default, I think, though not if on a line by itself
[07:42] <cjwatson> I use 'let python_highlight_all = 1' which may be implicated; not sure
[07:43] <evand> hrm, it might be that I don't have syntax highlighting enabled by default, so I wouldn't notice such things unless I turned it on
[07:43] <cjwatson> really? I couldn't live without it enabled by default
[07:43] <CIA-19> ubiquity: cjwatson * r2142 ubiquity/scripts/mythbuntu/mythbuntu_install.py: unnecessary constructor
[07:43] <cjwatson> it's nearly the first thing I do on new machines
[07:45] <evand> heh
[07:49] <CIA-19> ubiquity: cjwatson * r2143 ubiquity/ (9 files in 3 dirs): more trailing whitespace
[08:06] <superm1> cjwatson, i'll see if there is any breakage tonight when i get home
[08:06] <superm1> thanks
[09:07] <evand> cjwatson: Is LVM by default just blocked on changes to Ubiquity to support it, or are there technical reasons why it's not a path we intend on pursuing?
[09:09] <evand> LVM came up a few times in UDS discussions (backups for one), and mpt mentions it in his reply to your email on ubuntu-devel-discuss.
[09:25] <superm1> evand, just speculating: but i'd expect there to be a gui management tool for volumes too before it was by default?
[09:25] <superm1> similar to the tool that red hat ships on their boxes
[10:29] <cjwatson> that class of thing is certainly the reason we opted not to do it by default after investigating it
[10:29] <cjwatson> (that decision was actually before ubiquity existed)
[10:29] <cjwatson> but certainly nowadays ubiquity support would have to be added
[10:33] <superm1> have you looked into adopting red hat's existing tool?
[10:33] <superm1> i don't know off hand what the license is on it
[10:35] <cjwatson> we looked it a while back and concluded it really wasn't good enough to inflict on all users
[10:35] <cjwatson> it's a pretty high bar, unfortunately
[10:37] <evand> ah, indeed
[10:38] <cjwatson> it's been two years, though
[10:42] <superm1> last i looked at their tool was maybe a year ago, and i wasn't "happy" with it persay, but it did the trick for the purpose/.  i'm assuming this spec will be revisited in Boston this year though, at which point it can be reinvestigated
[10:44] <cjwatson> I'd be very happy if it were better now, and we should certainly get to the point of being able to do LVM in ubiquity ...
[10:44] <cjwatson> (and kickseed, etc. etc.)
[10:44] <superm1> what is kickseed?
[10:46] <evand> kickstart -> preseed file
[10:47] <superm1> i was going to ask about that: preseeding isn't active in ubiquity atm then - so on the cds, when file=..../ubuntu.seed and such, does it actually do anything?
[10:47] <evand> yes
[10:48] <evand> ubiquity noninteractive
[10:48] <evand> makes use of it
[10:48] <superm1> that was just added as a frontend though right?
[10:48] <evand> indeed
[10:49] <evand> but preseeding is active in ubiquity
[10:49] <superm1> well for the mythbuntu frontend however, it won't be relevent yet then
[10:49] <evand> the only problem is that pages that already have all their questions answered do not gracefully skip to the next page, which I'm working on ever so slowly
[10:49] <cjwatson> superm1: ubiquity has always had some level of preseeding, even if not complete
[10:49] <superm1> okay
[10:50] <cjwatson> at least ever since it was called ubiquity - not in the initial ubuntu-express implementation
[10:50] <cjwatson> superm1: we use it for language pack selection, in particular
[10:51] <cjwatson> for example the Kubuntu preseed file has:
[10:51] <cjwatson> d-i     pkgsel/language-pack-patterns   string language-pack-kde-$LL kde-i18n-$LL
[10:51] <cjwatson> and ubiquity honours that
[10:51] <superm1> Is that in the ubuntu preseed file though?
[10:51] <superm1> or just kubuntu
[10:51] <cjwatson> it's in everything except Ubuntu
[10:51] <cjwatson> the default for that question is set to make sense for Ubuntu
[10:52] <superm1> which would be how i missed it then - looking only at the ubuntu preseed
[10:52] <cjwatson> right, the Ubuntu preseed file is not very interesting
[10:52] <cjwatson> in fact it was empty for a long tim
[10:52] <cjwatson> e
[10:52] <cjwatson> (nonexistent)
[10:53] <mirkobuholzer> I did some tests with evan's branch with preseeding and works nicely ...
[10:53] <cjwatson> I haven't tried out the automation stuff yet, I must confess
[10:53] <cjwatson> hopefully I'll get Evan to give me a demo next week
[10:54] <evand> indeed!
[10:59] <cjwatson> the bandwidth in the London office is fantastic for mucking about with CD images
[10:59] <cjwatson> we basically have something like Ethernet to the datacentre
[10:59] <evand> now if only I had the space to store them all
[10:59] <cjwatson> that's what the big vmware-server box is for
[11:00] <evand> yeah, I'm definitely going to build one once I move
[11:00] <evand> I'm running it on here, but 250GB...not enough.
[11:01] <cjwatson> yeah, I need a chunk more disk too :-/
[11:03] <evand> do I have access to the LP wiki?  I'd like to learn more on this whole PPA thing.
[11:05] <evand> ah, nevermind.  I get the gist of it from the spec description.
[11:05] <cjwatson> evand: see /msg