[10:11] <seb128> cjwatson: I confirm that yesterday's daily CD fix the windows vista detection issue I had the day before
[10:29] <cjwatson> seb128: great!
[01:13] <xivulon> hmm "fallocate() is a new system call which will allow applications to preallocate space to any file(s) in a file system."
[01:13] <xivulon> that should allow to create images quickly
[02:04] <xivulon> I finally have a wubi build that might even work! If someone has a spare windows partition that can be wasted give it a go!
[02:04] <xivulon> http://wubi-installer.org/devel/minefields/Ubuntu-7.10-alpha.exe
[03:15] <cjwatson> seb128: could you remind me exactly what your language pack problem was the other day? Unfortunately my home client wasn't joined so I've lost the logs
[03:32] <evand> xivulon: I will when things settle down a lot
[03:33] <xivulon> I had a couple of reports of hangs while copying files
[03:33] <xivulon> Any idea what could it be? One reason might be that the images are not preallocated, but that should result in a slowdown and be progressive.
[03:34] <xivulon> It works fine for me using a daily image of a couple of days ago'
[03:35] <xivulon> evand, quick question: at the moment, upon reboot, there is a usplash message to pull out the CD and press enter. Where is the code responsbile for it (am on windows now)?
[03:35] <cjwatson> evand: do you have logs of our conversation with seb128 about language packs yesterday?
[03:35] <cjwatson> xivulon: /etc/init.d/casper
[03:35] <evand> yes, sorry
[03:35] <evand> one second
[03:35] <xivulon> cjwatson thanks
[03:37] <evand> cjwatson: http://evalicious.com/ubuntu-installer.log
[03:38] <cjwatson> wow, back to December
[03:38] <CIA-18> ubiquity: evand * r2339 ubiquity/ (configure configure.ac): bump to 1.6.6
[03:39] <cjwatson> hmm, right, so seb128 is getting a hang
[03:39] <evand> I tend to be somewhat of a pack rat
[03:39] <cjwatson> haven't managed to reproduce that
[03:39] <cjwatson> disk space is less expensive than trying to reconstitute old stuff you didn't mean to delete
[03:39] <evand> haha, indeed
[03:40] <evand> did you try kubuntu?  I believe slangsek had a hang using kde_ui.
[03:40] <evand> https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/145012
[03:40] <xivulon> I normally leech off fabbione logs (http://people.ubuntu.com/~fabbione/irclogs/), guess have another link...
[03:41] <cjwatson> evand: I suspect it's more dependent on local network setup than frontend
[03:42] <evand> ah, indeed
[03:42] <CIA-18> ubiquity: evand * r2340 ubiquity/ (debian/changelog ubiquity/components/usersetup.py):
[03:42] <CIA-18> ubiquity: * Back out recent test for the user existing in the m-a debconf
[03:42] <CIA-18> ubiquity:  section. It isn't necessary and it breaks the installer when m-a is
[03:42] <CIA-18> ubiquity:  used (LP: #151243).
[03:42] <cjwatson> xivulon: oh, yes, I did check and for some reason thought Fabio's log bot wasn't in here even though it blatantly is
[03:42] <evand> that very much confused me, as I knew there wasn't any difference in that code path between kde and gtk
[03:42] <evand> it's exited a few times
[03:42] <cjwatson> evand: hmm, maybe I should try French after all
[03:42] <cjwatson> evand: it'll be different live seeds between Ubuntu and Kubuntu, I bet
[03:43] <evand> hrm
[03:43] <evand> so to explain the above bug
[03:43] <cjwatson> oh, I see https://bugs.edge.launchpad.net/ubuntu/+source/ubiquity/+bug/145012/comments/8, interesting
[03:43] <xivulon> Can I bring back the attention to the reports above about freezes during file copying, I am not sure they are wubi specific.
[03:43] <xivulon> I can ask the chap who reported them to run some tests
[03:44] <evand> I was looking at this: http://launchpadlibrarian.net/9527551/debug , and for some reason that now escapes me, in my head I was treating the setting of questions that were not registered as errors.
[03:44] <cjwatson> evand: SET would be, but .preseed() registers it if it isn't there
[03:45] <evand> ah, indeed
[03:45] <evand> somehow I did not trigger this in my testing
[03:45] <evand> it only happens when you are going to import something or when partitions isn't empty somehow.
[03:46] <xivulon> evand at what stage are lang pack retrieved? The hang reported are at 31% and 44%
[03:46] <evand> I'm going to stress test ubiquity, specifically m-a, now, unless you want help tracking down the langpack issue.
[03:46] <evand> though I imagine I'll be much more useful in the former as the latter has confused me for quite some time now
[03:49] <evand> terribly sorry about screwing up the release twice in a row.  My commits should have occurred days earlier than they did.
[03:50] <cjwatson> xivulon: way later than that
[03:50] <cjwatson> xivulon: hangs at your stage can really only be filesystem breakage
[03:50] <xivulon> cjwatson, that is what I suspected...
[03:55] <cjwatson> evand: langpacks> I'm beginning to understand it, but it sure isn't easy
[03:56] <cjwatson> part of the problem is that if it can't download updated package lists from the network then what it has in /var/lib/apt/lists/ is archive.ubuntu.com_*, which may not be right
[03:56] <cjwatson> but I don't think we can do much about that for gutsy
[03:57] <cjwatson> I'm really just trying to figure out how not to regress bug 131294 in the process
[04:00] <evand> indeed, this would be a bug in apt_pkg, no?  Perhaps mvo would be helpful?
[04:01] <cjwatson> no
[04:01] <cjwatson> apt_pkg is just doing what it's told :)
[04:02] <cjwatson> archive.ubuntu.com_* is there because that's what's copied over from the live filesystem
[04:02] <evand> ah
[04:04] <cjwatson> I think it's an apt-setup bug
[04:17] <evand> heh
[04:33] <seb128> cjwatson: re
[04:34] <seb128> cjwatson: do you still need some informations?
[04:34] <seb128> cjwatson: basically french install without activate network = hang at 87% on language pack download and no way to cancel
[04:40] <cjwatson> seb128: I assume this is Ubuntu?
[04:40] <cjwatson> (not Kubuntu)
[04:48] <seb128> cjwatson: yes
[04:48] <seb128> cjwatson: that's yesterday daily ubuntu desktop cd
[04:51] <seb128> cjwatson: did you try a non english install on a computer without connected network? what happens then for you?
[04:51] <cjwatson> seb128: it varies by language and exact circumstance
[04:51] <cjwatson> of course, I did try that
[04:51] <cjwatson> I got error messages, but a completed install
[04:52] <cjwatson> however French is not the same as Catalan (which I tried) because the French language pack is on the live filesystem
[04:52] <cjwatson> there are more variables here than just "non-English" ;-)
[04:53] <cjwatson> I'm trying French now; I had initially forgotten to disconnect the network but have done so partway through the install
[04:53] <seb128> ok
[04:53] <cjwatson> I suspect disconnecting the network device in vmware is not quite the same as pulling out the network cable, though
[04:55] <seb128> probably not
[05:07] <cjwatson> seb128: here, I get an error message "Une erreur est survenue lors de l'installation des paquets"
[05:08] <cjwatson> seb128: the main progress bar is at 87%
[05:08] <seb128> hum
[05:09] <seb128> I don't get the error here, but I've only tried on real configurations (my laptop and a desktop, both have the issue), maybe vmware is different as you said
[05:09] <evand> cjwatson: do you want me to wait for possible changes regarding the langpack issue, or am I safe to release a new ubiquity now?
[05:09] <evand> pitti is asking for it
[05:09] <cjwatson> evand: I think we should wait
[05:09] <seb128> the error means "an error happened during the packages installation"
[05:09] <cjwatson> seb128: in this case apt is immediately failing with a resolver error
[05:09] <cjwatson> seb128: oh, yeah, I recognise the error, it's a ubiquity template ;)
[05:10] <seb128> :)
[05:10] <cjwatson> seb128: in your case I bet the resolver is just sitting there timing out instead
[05:10] <seb128> well, it doesn't even time out
[05:10] <seb128> or it takes over an hour for that
[05:10] <seb128> I let the computer running for quite some time yesterday
[05:11] <cjwatson> would be interesting to know what processes are running
[05:11] <cjwatson> although, it's python-apt
[05:11] <cjwatson> so it might not be all that informative
[05:11] <seb128> I'm about to start a new install
[05:11] <cjwatson> but if something like /usr/lib/apt/methods/http is sitting there, that would be useful info
[05:11] <seb128> let me know if I should run ubiquity with some debug options from a command line or something
[05:12] <cjwatson> I don't think that will help much in this case, but switching to a terminal and seeing what it's doing would be useful
[05:12] <cjwatson> with strace if necessary
[05:12] <cjwatson> my current working hypothesis is that it is trying to talk to the network
[05:13] <seb128> bah
[05:14] <seb128> today's ubiquity is crashing on my desktop
[05:14] <seb128> let try on my laptop
[05:14] <seb128> looks like the migration assistant bug pitti pointed this morning
[05:15] <seb128> weird that it tries to run it where I asked to format the current Ubuntu install partition
[05:15] <cjwatson> seb128: try 'ubiquity --no-migration-assistant', should work around that
[05:15] <cjwatson> it's fixed in bzr
[05:22] <cjwatson> seb128: somebody else claims there's a D-state /usr/lib/apt/methods/http process; is it the same for you?
[05:23] <seb128> cjwatson: installation in progress, I'll tell you that to a few minutes
[05:23] <cjwatson> thanks
[05:29] <cjwatson> evand: so was bug 135149 really fixed elsewhere already?
[05:29] <cjwatson> since you backed out the ubiquity change that was originally part of that fix
[05:30] <evand>  * 135149 - failed to unmount migrationassistant: In some cases the user
[05:30] <evand> would be presented with a dialog saying that a partition couldn't be
[05:30] <evand> unmounted and asked them to manually take care of it before the install
[05:30] <evand> continued.  In most cases, the install would continue without issue if
[05:30] <evand> the user just pressed continue.  While I believe I've fixed all the
[05:30] <evand> cases where m-a would cause the installer to crash, I think this bug is
[05:30] <evand> caused by ntfs-3g reporting the partition as busy when it was just
[05:30] <evand> waiting on a write to finish.  NTFS is used in all the cases where it
[05:31] <evand> bug occurs.  This may be fixed by the newer version of ntfs-3g on the
[05:31] <evand> most recent CDs as I have not seen the issue in my limited testing of
[05:31] <evand> these CD images.
[05:31] <evand> wow, whoops
[05:31] <evand> I had hoped that would paste better
[05:31] <evand> the short version, I *think* it's fixed, but at worst it's an extra dialog they have to click through.
[05:32] <evand> a bigger problem is that after seb128's comment above, I discovered that m-a chooses to use a partition and its users despite them not being selected.
[05:32] <evand> I'm trying to fix it now, but if I can't by the time you're done with the langpack issue, perhaps we should make --no-migration-assistant the default.
[05:33] <seb128> cjwatson: nothing matching "apt" in the ps list
[05:33] <cjwatson> seb128: 'ps axf'; anything below /usr/share/ubiquity/install.py?
[05:34] <cjwatson> evand: that would be pretty unfortunate, as the marketing people are rather keen on the migration assistanmt
[05:34] <seb128> install.py is stucked in a read
[05:34] <cjwatson> -m
[05:34] <seb128> no
[05:34] <cjwatson> seb128: can you tell what that file descriptor is attached to?
[05:34] <seb128> 0
[05:34] <cjwatson> it's fd 0?
[05:35] <cjwatson> odd
[05:35] <evand> oh wait
[05:35] <evand> I may be wrong
[05:36] <seb128> it's blocked on "read(0, "
[05:37] <seb128> (strace install.py)
[05:37] <cjwatson> seb128: is there just one install.py process, or more than one?
[05:38] <seb128> the is a log-ouput -t ubiquity ....
[05:38] <cjwatson> not that
[05:38] <seb128> \_ /usr/bin/python ... install.py
[05:38] <seb128> that's all
[05:38] <cjwatson> mind you - what's the full argument list of log-output?
[05:39] <seb128> -t ubiquity --pass-stdout /us/share/ubiquity/install.py
[05:39] <cjwatson> ok, that's expected
[05:40] <cjwatson> seb128: do you think you could try this again, applying a small patch by hand first?
[05:41] <seb128> sure
[05:41] <cjwatson> seb128: http://paste.ubuntu.com/756/ applied to /usr/share/ubiquity/install.py before starting the installer
[05:41] <cjwatson> my suspicion is that the fetchprogress and installprogress objects are racing with the rest of the installer to read from debconf; and furthermore we don't really need to display IO errors anyway
[05:42] <cjwatson> I might make it only do that for language packs
[05:43] <cjwatson> or I might not :)
[05:45] <evand> ok, so migration-assistant/partitions ends up always being set to something, which is why seb128 ran into that block of code despite not selecting any users to use.  However, because no users are selected, they do not get fed into debconf, which means that nothing will get imported.
[05:46] <evand> cjwatson: I can quickly write a patch to check if there are any users that are going to be imported and set migration-assistant/partitions to '' if not, if you think it's worth it to kill m-a ealier on in such cases
[05:47] <cjwatson> if you think that's the safest approach, it sounds reasonable
[05:49] <seb128> cjwatson: installation in progress I'll tell how it works in a few minutes
[05:50] <evand> ok
[06:00] <seb128> cjwatson: your patch fixes the bug
[06:00] <seb128> cjwatson: it passed on the language pack download without error nor blocking
[06:01] <seb128> installation successful
[06:01] <seb128> cjwatson: you rock ;)
[06:02] <cjwatson> thanks ;)
[06:03] <cjwatson> let me try to squeeze in this usplash fix too, and then I think we'll be golden
[06:04] <CIA-18> ubiquity: cjwatson * r2341 ubiquity/ (debian/changelog scripts/install.py):
[06:04] <CIA-18> ubiquity: * Ignore IOError exceptions while fetching packages; they happen when the
[06:04] <CIA-18> ubiquity:  network is disconnected, and aren't important enough to result in scary
[06:04] <CIA-18> ubiquity:  error messages or even hangs (LP: #145012).
[06:05] <cjwatson> vmware is rather caning my laptop so I may still be a little while
[07:29] (cjwatson/#ubuntu-installer) superm1: no, but it scares me far too much to apply pre-RC
[07:30] (superm1/#ubuntu-installer) understandable :)
[07:30] (cjwatson/#ubuntu-installer) superm1: it may well be correct, but I need more time to test it so it needs to be post-RC I think
[07:30] (superm1/#ubuntu-installer) yeah i'm going to do some tests myself with it right now.
[07:30] (superm1/#ubuntu-installer) if it works as expected, i'll do the mythbuntu rc with it in
[07:30] (superm1/#ubuntu-installer) just building ubiquity on the ppa
[07:30] <cjwatson> ok
[07:31] <CIA-18> ubiquity: cjwatson * r2343 ubiquity/ (d-i/manifest debian/changelog): * Automatic update of included source packages: apt-setup 1:0.21ubuntu4.
[07:35] <CIA-18> ubiquity: cjwatson * r2344 ubiquity/debian/changelog: releasing version 1.6.6
[08:19] <CIA-18> ubiquity: superm1 * r2345 ubiquity/ (4 files in 2 dirs): adjust diversions for mythbuntu frontend
[08:22] <CIA-18> ubiquity: superm1 * r2170 mythbuntu-ubiquity/ (11 files in 7 dirs): merge with trunk
[08:23] <CIA-18> ubiquity: superm1 * r2171 mythbuntu-ubiquity/scripts/install.py: apply mvo's patch to resolve hang ups with cdrom method
[08:26] <CIA-18> ubiquity: superm1 * r2172 mythbuntu-ubiquity/debian/changelog: release 1.6.6+mythbuntu1
[09:17] <heno> bug 151355
[09:17] <heno> hm, no bot :)
[09:17] <heno> https://bugs.launchpad.net/ubuntu/+source/debian-installer/+bug/151355
[09:18] <heno> looks like we have a problem with the crypted LVM install
[09:18] <heno> evand: do you work with the d-i version as well?
[09:26] <evand> heno: not enough to diagnose that.  I wasn't responsible for any of the work going into it.
[09:26] <evand> I can try if you cannot find cjwatson, though.
[09:27] <heno> evand: ok, thanks. He'll be back in a bit and is aware of the problem
[09:29] <evand> ok
[09:29] <xivulon> is wubi-liveboot going to be on the CD? I need to hardcode the version/subversion string
[09:31] <evand> not sure, check with cjwatson
[09:35] <xivulon> it's in the daily build
[09:36] <xivulon> the subversion now points to Beta
[09:40] <xivulon> should it be Release Candidate...
[10:02] <cjwatson>     export OFFICIAL="Release Candidate"
[10:02] <cjwatson> ^- that's what we have
[10:05] <cjwatson> heno: oh, it's a seed bug
[10:05] <xivulon> cjwatson I just disabled the subversion check
[10:05] <xivulon> I'd need to change more code because of the space
[10:05] <cjwatson> xivulon: good plan
[10:05] <xivulon> I just uploaded but I'd need someone to double check with me
[10:06] <xivulon> I am going to burn the latest daily ISO and test with that
[10:06] <xivulon> That obviously require a physical CD
[10:07] <xivulon> I apologize for that it completely slipped my mind
[10:07] <cjwatson> evand: what was wrong with gobuntu, btw?
[10:09] <evand> cjwatson: it doesn't install gobuntu-desktop.  I'm slightly confused though...
[10:09] <evand> pkgsel calls tasksel with in-target but doesn't include tasksel/first when running debconf-copydb.
[10:10] <evand> How does it know to pick up the task as specified on gobuntu.seed
[10:10] <cjwatson> I don't think that's needed for explicitly preseeded items
[10:10] <cjwatson> only for those set by other bits of d-i
[10:11] <cjwatson> have a look at preseed/post-base-installer.d/05preseed
[10:11] <cjwatson> that copies in anything with an owner other than d-i
[10:11] <evand> well, cutting out the debconf progress bits and adding a -t and >/somefile to tasksel at that point only shows standard and minimal
[10:11] <evand> ah, ok
[10:11] <cjwatson> check in /target/var/cache/debconf/config.dat though
[10:12] <evand> yeah, I need to stick a sleep in there somewhere
[10:13] <evand> I should point out, if you go back into main-menu after the install finishes and run the pkgsel option again, it works
[10:13] <cjwatson> now that's odd
[10:13] <evand> tell me about it
[10:13] <evand> :/
[10:13] <cjwatson> (I'm downloading the daily, ETA 1:12
[10:13] <cjwatson> )
[10:14] <cjwatson> evand: have you tried DEBCONF_DEBUG=developer?
[10:14] <evand> negative, where would I stick that in d-i?
[10:16] <cjwatson> kernel boot parameter
[10:16] <cjwatson> evand: however - the gobuntu-desktop package isn't on the CD
[10:16] <cjwatson> I bet that can't be good
[10:16] <cjwatson> oh, gar
[10:16] <cjwatson> my fault entirely
[10:16] <cjwatson>         gobuntu)
[10:16] <cjwatson>                 # Gobuntu uses Ubuntu seeds.
[10:16] <cjwatson>                 GERMINATE_SEED_DIST="ubuntu.$DIST"
[10:16] <cjwatson>                 ;;
[10:17] <cjwatson> I should have taken that out ... (cdimage/bin/run-germinate)
[10:17] <evand> wow, you know I knew that a few days ago and completely forgot
[10:17] <evand> ugh
[10:17] <evand> the former, that is
[10:19] <cjwatson> evand: fixed on antimony - rebuild gobuntu at your leisure
[10:19] <evand> thanks a ton!
[10:19] <evand> and will do
[10:19] <cjwatson> with any luck that's enough
[10:27] <xivulon> I am going to reboot to test wubi-cdboot
[10:46] <xivulon> My wubi-cd test went well
[10:46] <xivulon> but I'd urge someone else to give it a go
[10:47] <xivulon> I'll be awake till late in case there are issues
[10:49] <evand> I'll give it a go after I'm done testing gobuntu
[10:51] <xivulon> thanks evand
[10:52] <evand> anytime
[10:52] <xivulon> will have to reboot again I guess
[10:52] <evand> heh
[10:55] <CIA-18> ubiquity: superm1 * r2173 mythbuntu-ubiquity/ (3 files in 3 dirs): update .desktop file
[11:14] <xivulon> 1 more change for me: disable early command/success command in preseed
[11:14] <xivulon> do I need: ubiquity	ubiquity/summary	note
[11:14] <xivulon> ?
[11:14] <xivulon> for the live cd? guess not
[11:16] <evand> yes, otherwise the summary page will be shown
[11:16] <xivulon> ok
[11:17] <xivulon> that's for the standard live CD of course
[11:19] <cjwatson> not for just running the live CD though
[11:19] <cjwatson> for just running the live CD, I don't think you should preseed anything, except maybe locale/keyboard if you're doing that
[11:22] <xivulon> gonna change that
[11:22] <xivulon> one sec
[11:24] <xivulon> anything else that has to be on the preseed file?
[11:25] <xivulon> debian-installer/local
[11:25] <xivulon> console-setup/layoutcode
[11:25] <cjwatson> locale
[11:25] <cjwatson> (typo)
[11:26] <cjwatson> that's all, you shouldn't preseed anything else for plain booting
[11:26] <xivulon> sure I pasted n-1 chars
[11:26] <cjwatson> (console-setup/variantcode is fine too if you start doing that)
[11:28] <xivulon> that's in, but at the moment I only found the windows code for dvorak
[11:28] <xivulon> us
[11:31] <xivulon> up rev 327
[11:31] <xivulon> pushing now
[11:32] <xivulon> ok going to reboot again, ciao
[11:33] <xivulon> cjwatson have a look at preseed.cfg in rev 327
[11:46] <xivulon> wubi-cdboot looks good to me, but I only tested on Vista
[11:46] <xivulon> I am going to see if I can fetch XP somewhere
[11:47] <xivulon> again, pls test it out, particularly if you have XP
[11:55] <cjwatson> xivulon: looks fine to me
[11:56] <xivulon> thanks colin
[11:56] <xivulon> testing on xp now
[11:56] <xivulon> can boot...
[12:03] <evand> doesn't everyone?
[12:04] <evand> AWESOME!  It works, green usplash and all.  Thanks cjwatson!
[12:06] <cjwatson> rock on
[12:06] <xivulon> hmm wubi on xp uninstalls but does not delete registry key (hence remains in add/remove)
[12:06] <xivulon> code looks good though
[12:06] <xivulon> unless registry is case sensitive