[01:01] <CIA-19> partman-target: cjwatson * r718 ubuntu/debian/changelog: releasing version 54ubuntu5
[01:03] <CIA-19> partman-target: cjwatson * r719 ubuntu/debian/ (changelog partman-target.templates): * Remove some mistakenly-placed po-debconf flags.
[01:04] <CIA-19> partman-target: cjwatson * r720 ubuntu/debian/po/templates.pot: debconf-updatepo --skip-merge
[01:11] <CIA-19> partman-target: cjwatson * r721 ubuntu/debian/ (59 files in 2 dirs): * Update translations of Ubuntu-specific strings from Launchpad.
[01:12] <bdmurray> superm1: can the other task for bug 215036 be closed now too?
[01:12] <ubotu> Launchpad bug 215036 in ubiquity "mythfrontend.real crashed with SIGSEGV in MythThemedMenuPrivate::keyHandler()" [Undecided,Fix committed] https://launchpad.net/bugs/215036
[01:12] <superm1> bdmurray, it will get closed with the upload of ubiquity
[01:12] <superm1> the fix is in ubiquity trunk right now
[01:12] <bdmurray> Okay, I didn't know both tasks were vaild
[01:12] <superm1> yeah
[01:19] <CIA-19> partman-target: cjwatson * r722 ubuntu/debian/changelog: releasing version 54ubuntu6
[01:21] <cjwatson> evand: could you make sure to use partman-target 54ubuntu6 when uploading ubiquity, for the sake of the updated translations? It should be available from https://launchpad.net/ubuntu/hardy/+queue?queue_state=1&queue_text= even if it isn't in the archive yet (or from http://people.ubuntu.com/~ubuntu-archive/queue/hardy/unapproved/ if you can't see that)
[01:24] <evand> sure, I'll stick it in my local archive.
[01:33] <CIA-19> ubiquity: evand * r2660 ubiquity/debian/po/ (79 files): debconf-updatepo
[01:36] <CIA-19> ubiquity: evand * r2661 ubiquity/ (d-i/manifest debian/changelog):
[01:36] <CIA-19> ubiquity: * Automatic update of included source packages: console-setup
[01:36] <CIA-19> ubiquity:  1.21ubuntu7, partman-target 54ubuntu6.
[01:48] <CIA-19> ubiquity: evand * r2662 ubiquity/debian/changelog: releasing version 1.8.5
[01:53] <CIA-19> grub-installer: cjwatson * r732 ubuntu/ (3 files in 2 dirs):
[01:53] <CIA-19> grub-installer: * Confirm the GRUB password after entry (LP: #42019). Note that
[01:53] <CIA-19> grub-installer:  grub-installer/password-again must now be preseeded in addition to
[01:53] <CIA-19> grub-installer:  grub-installer/password, and that grub-installer/password-crypted now
[01:53] <CIA-19> grub-installer:  takes an MD5-crypted password rather than a boolean.
[01:56] <CIA-19> grub-installer: cjwatson * r733 ubuntu/debian/changelog: releasing version 1.27ubuntu7
[02:00] <CIA-19> installation-guide: cjwatson * r419 ubuntu/ (debian/changelog en/appendix/preseed.xml): * Document GRUB password preseeding (LP: #42019).
[02:02] <cjwatson> evand: thinking about it, you can't select existing filesystems for autopartitioning at the moment (bug 195608), so I don't see a need to document the clear_partitions question
[02:02] <ubotu> Launchpad bug 195608 in partman-auto "[Hardy FR] enhance recipe format to allow specifying an existing partition" [High,Confirmed] https://launchpad.net/bugs/195608
[02:04] <CIA-19> installation-guide: cjwatson * r420 ubuntu/debian/changelog: releasing version 20080211ubuntu4
[02:05] <evand> ah
[03:25] <CIA-19> oem-config: evand * r448 oem-config/ (d-i/manifest debian/changelog):
[03:25] <CIA-19> oem-config: * Automatic update of included source packages: console-setup
[03:25] <CIA-19> oem-config:  1.21ubuntu7, localechooser 1.42ubuntu5.
[03:28] <CIA-19> oem-config: evand * r449 oem-config/debian/changelog: releasing version 1.34
[10:47] <TheMuso> MUAHAHAHAHAHAHAHAHAHAHAHAHAHAAAAA!!!! Orca/A11y + only-ubiquity, *YOU ARE MINE!!!*
[10:47] <cjwatson> TheMuso: cool, what was needed?
[10:48] <TheMuso> cjwatson: Basically, to split the a11y code block up. What I've currently done, which is what is working, is to put the first piece, the bit that checks the gconf setting + loads at-spi-registryd, just before gtk is imported, background is set, and the window manager is loaded.
[10:48] <TheMuso> The rest, i.e loading either onboard or orca, is where it is already.
[10:48] <TheMuso> So basically, the a11y framework has to be up, before anything GTK related is loaded.
[10:49] <TheMuso> I got this idea from seeing how gnome-session does things.
[10:50] <TheMuso> Still want to check a few more things, like try it on other machines etc, but all looks good.
[10:51] <TheMuso> Oh, and I also removed all the preexec calls for dropping privileges as evand said we might have to do.
[10:51] <TheMuso> So yes everything is root, but things are working.
[10:54] <TheMuso> And this also means a change to the casper-bottom/30accessibility script to make sure gconf settings are set as root when only-ubiquity mode is active.
[10:57] <cjwatson> oh, of course, that makes a lot of sense
[10:58] <cjwatson> happy to review a diff if you like, the idea sounds right
[10:59] <TheMuso> Let me get some food, test one or two of the other profiles, and double heck things, and I'll give you a diff, thanks.
[11:41] <xivulon> If anyone has windows+fat partition, could you pls try installing wubi selecting a largish size (>5GB) and check that at the end you have virtual disk files in c:\ubuntu\disks?
[11:44] <TheMuso> xivulon: I will do tomorrow.
[11:44] <xivulon> TheMuso: thanks
[11:45] <xivulon> that is to debug #216161
[11:45] <TheMuso> Ok.
[11:56] <xivulon> cjwatson, evand, do you know what is the behaviour of db_input critical partman-auto-loop/unclean_host (autopartition-loop line #252)
[11:56] <xivulon> In reference to bug #217593
[11:56] <ubotu> Launchpad bug 217593 in wubi "If the installer fails or is aborted, users can end up in Live CD Desktop, which is confusing " [Medium,Confirmed] https://launchpad.net/bugs/217593
[12:00] <cjwatson> 217593 is a well-known ubiquity issue
[12:00] <cjwatson> not sure why you're asking specifically about that question
[12:00] <cjwatson> it's just another error template AFAICS
[12:00] <xivulon> I just opened that bug, must be a duplicate
[12:01] <xivulon> there are 2 issues though
[12:01] <xivulon> 1) when it fails/is aborted and you lend in the live CD
[12:01] <xivulon> 2) when you reboot after failuer, you will still lend in the live CD (I guess because of found_images=non-empty which triggers #1)
[12:02] <cjwatson> 1) is one of those cases where both alternatives are bad in some way
[12:02] <xivulon> 2 could be addressed by triggering failure-command then (I guess)
[13:08] <TheMuso> cjwatson: http://www.pastebin.ca/985904 for ubiquity, and http://www.pastebin.ca/985905 for casper when you have a minute.
[13:21] <cjwatson> TheMuso: casper change looks fine
[13:21] <cjwatson> TheMuso: should initialise extras outside the if
[13:22] <cjwatson> TheMuso: (otherwise you'll break the KDE case)
[13:22] <cjwatson> TheMuso: I'd just write '~root/.gconf' rather than '~%s/.gconf' % 'root'
[13:23] <cjwatson> TheMuso: in the background image code, comment says "not in v2" but code says "not in v1"
[13:25] <cjwatson> TheMuso: is it necessary to keep metacity running as root? I'd rather drop privileges for that if it's possible
[13:25] <cjwatson> TheMuso: likewise gnome-settings-daemon
[13:26] <cjwatson> if I'm going over old ground with that last bit, feel free to ignore me
[13:26] <cjwatson> ah, you explicitly have that in the changelog
[13:27] <cjwatson> TheMuso: would it be possible to drop privileges only if accessibility != 'true'? I'd be slightly more comfortable with that
[13:27] <cjwatson> so something like:
[13:27] <cjwatson> maybe_drop_privileges = {}
[13:27] <cjwatson> if accessibility != 'true':
[13:28] <cjwatson>     maybe_drop_privileges['preexec_fn'] = self.drop_privileges
[13:28] <cjwatson> ...
[13:28] <cjwatson> subprocess.Popen(['onboard'], stdin=null, stdout=logfile, stderr=logfile, **maybe_drop_privileges)
[13:28] <cjwatson> that kind of thing?
[13:29] <cjwatson> that way we'd be making fewer changes to the non-a11y case
[13:33] <TheMuso> cjwatson: I'd be fine with metacity running as a user, but I *think* that window titles wouldn't be read if one was switching between windows with alt + tab.
[13:33] <TheMuso> cjwatson: And re gnome-settings-daemon, it starts pulseaudio, and if pulse is running as a user, and speech is root, theres a good chance it won't be able to use pulse for audio output.
[13:34] <TheMuso> However, I can test this to be sure.
[13:34] <cjwatson> ah, gotcha
[13:35] <TheMuso> But I'm certainly happy to only run as root for the a11y case.
[13:57] <TheMuso> cjwatson: Ok taking the above into account, is something like this what you were thinking of? http://www.pastebin.ca/985953
[13:58] <cjwatson> TheMuso: I think maybe_drop_privileges needs to be a dict {} rather than a list [] - but otherwise, yes, that's dead on
[13:59] <TheMuso> Oh ok.
[14:00] <cjwatson> http://docs.python.org/ref/calls.html documents that **expression syntax, btw
[14:00] <cjwatson> ('"expression" must evaluate to a (subclass of) dictionary, the contents of which are treated as additional keyword arguments' etc.)
[14:01] <TheMuso> Right, I actually just skimmed over it so didn't pick up the finer details unless I would have looked visually. :)
[14:01] <TheMuso> Right, one final test before I commit.
[14:48]  * TheMuso groans. Orca zombifying again, even after those few changes... Long night ahead me thinks.
[15:08] <evand> TheMuso: does /var/log/installer/dm show anything interesting from orca?
[15:10] <TheMuso> evand: Hang on, rebuilding/reburning the custom disk to try something. Will let you know how that goes.
[15:10] <evand> ok, best of luck
[15:10] <TheMuso> Thanks.
[15:18] <TheMuso> evand: No orca still zombifies, and nothing in dm. I'll run with debug-ubiquity
[15:19] <evand> hrm
[15:19] <TheMuso> I'm wondering... Whether its because orca is starting too soon after at-spi-registryd. Could we add say a 1 second or so pause? If so, how does one go about that in py thon?
[15:20] <evand> import time; time.sleep(SECONDS)
[15:20] <TheMuso> Oh that easy, nice. Thanks.
[15:21] <evand> you're welcome
[15:23] <TheMuso> Hrm however... Do all the processes get executed in one big pile? i.e after the extras etc variables are populated, or are things exeucted as the code is parsed?
[15:24] <evand> they're executed as subprocess.Popen is called, so the latter case of as the code is parsed.
[15:25] <TheMuso> Ok so the timeout would have an effect then. Time to try it.
[15:28] <TheMuso> s/timeout/sleep/
[16:32] <TheMuso> Ok. I can say with 100% certainty now that it is because orca was being loaded far too soon, and the a11y framework just wasn't ready. Since this will vary between slow and fast machines, (which explains why it worked on my dual celeron 466, I'm thinking 10 seconds is a good compromize, but thoughts/suggestions welcome.
[16:33] <TheMuso> Actually, I tried 20, which worked, so I'll try 10 now
[16:47] <TheMuso> Right, 10 seconds works, I'll finalize everything tomorrow. Time for sleep.
[16:48] <TheMuso> err later today. Shows my sleep readiness. :p
[18:19] <xivulon> cdimage.ubuntu.com/daily-live/current/ is down at the moment, is that in preparation of rc release? Is there some sort of schedule for planned outages?
[18:21] <xivulon> I will have to update the metalink files accordingly
[18:24] <cjwatson> blink
[18:24] <cjwatson> not planned, let me check after I bring the publisher back up
[18:25] <xivulon> 403 Forbidden error to be precise
[18:26] <cjwatson> that's not what I'd call a planned outage, it's what I'd call a bug. :)
[18:26] <cjwatson> drwxrwsr-x 2 cdimage cdimage 4096 Apr 14 08:30 20080414
[18:26] <cjwatson> drwxrwsr-x 2 cdimage cdimage 4096 Apr 15 10:02 20080415
[18:26] <cjwatson> lrwxrwxrwx 1 cdimage cdimage   10 Apr 15 11:24 current -> 20080415.1
[18:26] <cjwatson> dangling symlink
[18:26] <cjwatson> I suspect an aborted build run; I'll put it back
[18:26] <evand> heh
[18:29] <xivulon> ...but it raised a good point. Will the daily ISOs be always available up to the final?
[18:29] <cjwatson> yes
[18:30] <cjwatson> this was a bug due to (planned) buildd downtime
[18:30] <cjwatson> the consequences were unplanned though
[18:31] <cjwatson> restored, thanks for reporting
[18:31] <xivulon> np
[18:35] <cjwatson> I've fixed the code that produced a dangling symlink in that case as well
[20:36] <greearb> Hello!  I'm trying to respin a live-cd with a custom kernel (2.6.23.17 + my hacks).  I've read lots of howtos, but still missing something.
[20:37] <greearb> I think I must need to rebuild the installer, and maybe the thing that creates the .udebs
[20:38] <greearb> the info for udebs that I found seems to be outdated since it talks about using linux-kernel-di-i386-2.6
[20:39] <greearb> If someone could point me to a better howto, I'd greatly appreciate it.
[20:57] <cjwatson> greearb: the regular kernel source package build process spits out udebs these days
[20:58] <cjwatson> you do have to be using debian/ from the Ubuntu kernel or something like it, though
[21:01]  * cjwatson -> out for a while
[21:27] <CIA-19> oem-config: evand * r450 oem-config/ (configure configure.ac): Bump to 1.35
[21:29] <CIA-19> oem-config: evand * r451 oem-config/ (debian/changelog lib/frontend/kde_ui.py):
[21:29] <CIA-19> oem-config: * Properly initialize the UI object and its parent widget (LP:
[21:29] <CIA-19> oem-config:  #217844).
[21:31] <CIA-19> oem-config: evand * r452 oem-config/debian/changelog: releasing version 1.35
[21:50] <greearb> cjwatson, thanks.  I don't see any udebs in the debian dir, but maybe I'm not using the make-kpkg script correctly.  I'll read through it's docs again.
[23:24] <greearb> well, no luck..can't figure out how to build udebs for a new kernel