[08:53] <infinity> cjwatson: The vexpress kernel change recently had something to do with d-i targets, right?  I was only vaguely paying attention.  If you turn on the d-i vexpress build, that should take care of keeping it in main, I assume, so Martin will stop wondering if it needs demotion? :)
[08:54] <cjwatson> infinity: right, was planning on it
[08:54] <cjwatson> that was bug 826021
[08:54] <ubot2> Launchpad bug 826021 in linux-linaro-vexpress "Please re-enable crypto-modules udeb" [High,Fix released] https://launchpad.net/bugs/826021
[08:56] <cjwatson> let me test-build it first; I'll get back to you next week :-P
[08:56] <infinity> Hah.
[08:57] <infinity> I think we have a developer-accessible Panda in the DC...
[08:58] <cjwatson> Meh, I'm already on kakadu
[08:59] <infinity> Kay.
[08:59] <infinity> scheat.canonical.com should be the porter box.
[08:59] <infinity> And if I didn't have StrictHostKeyChecking on, I'd be logged in by now...
[08:59] <cjwatson> (is it just me who always thinks "kaaaakaaaaaaadoo doo doo push pineapple shake a tree"?)
[09:00] <infinity> adconrad@scheat:~$ dchroot -l
[09:00] <infinity> Available chroots: lucid, maverick, natty, oneiric [default]
[09:00] <infinity> Look at that.
[09:00] <infinity> Shiny.
[09:00] <cjwatson> awesome.
[09:00] <infinity> You can -j2!  Tempting!
[09:00] <cjwatson> if I die of boredom waiting for kakadu then I'll try that.  still on the first cup of coffee though so I might just fall asleep again.
[09:14] <CIA-45> debian-installer: cjwatson * r1538 ubuntu/ (3 files in 3 dirs): Re-enable armel/linaro-vexpress and move it to 3.0.0-1007 kernels.
[09:16] <CIA-45> debian-installer: cjwatson * r1539 ubuntu/ (build/config/armel/linaro-omap.cfg debian/changelog): Move armel/linaro-omap to 3.0.0-1007 kernels.
[09:18] <CIA-45> debian-installer: cjwatson * r1540 ubuntu/debian/changelog: releasing version 20101020ubuntu67
[09:53] <CIA-45> ubiquity: evand * r4989 trunk/scripts/install.py: Make the update process termination more consistent with the rest of ubiquity, and provide some logging to help further diagnose the bug.
[09:54] <CIA-45> ubiquity: evand * r4990 trunk/ (d-i/manifest debian/changelog debian/control):
[09:54] <CIA-45> ubiquity: Automatic update of included source packages: console-setup
[09:54] <CIA-45> ubiquity: 1.57ubuntu26, grub-installer 1.68ubuntu2, hw-detect 1.81ubuntu3,
[09:54] <CIA-45> ubiquity: netcfg 1.68ubuntu5, partman-base 151ubuntu2.
[10:33] <CIA-45> ubiquity: evand * r4991 trunk/scripts/install.py: Kill the entire process group, and with it, apt-get.
[10:37] <CIA-45> ubiquity: evand * r4992 trunk/scripts/install.py: Remove duplicate errno import. Thanks pyflakes
[10:41] <cjwatson> I wish unity-2d worked properly in kvm
[10:41] <ogra_> it doesnt ?
[10:41] <cjwatson> it sometimes sort of works but randomly breaks a lot for me; I can't really use kvm for ubiquity development any more
[10:41] <ogra_> weird
[10:42]  * ogra_ hasnt seen much breakage since A1 in unity-2d 
[10:43] <ogra_> cjwatson, might be the composite extension thats on by default, it doesnt do any harm on arm frambuffer but i could imagine it behaves different on vesa
[10:43] <ogra_> iirc you can disable it in d-conf somewhere
[10:44] <ogra_> (a metacity key)
[10:45] <ev> virtualbox has 3d driver support, for what it's worth
[10:45] <ev> the non-ose one, anyway
[10:45] <cjwatson> ogra_: that's always lots of fun to arrange on a live CD
[10:45] <ogra_> hmm, if thats the case you might end up with compiz under unity-2d
[10:45] <cjwatson> ev: yeah, but I had a virtualisation tool already that was working fine :)
[10:46] <ev> and unrelated to testing live CDs, but http://www.vagrantup.com makes working with virtualbox quite nice
[10:46] <ev> cjwatson: indeed, I realize it didn't directly address your question
[10:46] <cjwatson> and lots of finger macros
[10:46] <ev> ah yes
[10:46] <ev> I think we should just go back to the days when we were being given vmware licenses like they were candy
[10:47] <cjwatson> it was a great day when I switched from vmware to kvm
[10:47] <ev> haha, oh yes, I'm forgetting the constant module breakage and workarounds
[10:47] <cjwatson> yep, looking up whoever it was's site somewhere in .cz for the latest runes
[10:48] <cjwatson> TBH I don't know that 3d support would improve things noticeably here, given that unity is sufficiently much more sluggish than unity-2d on my laptop that I switched to unity-2d
[10:49] <cjwatson> although to be fair I hear that they reckon they've fixed a lot of the leaks
[10:49] <ogra_> no, but unity-2d (if you select the session in lightdm) forcefully uses metacity ... if the auto-fallback thing in the normal unity session kicks in it might use compiz with -2d
[10:50] <cjwatson> I want it to use metacity :)
[10:50] <ogra_> right
[10:50] <cjwatson> oh I see what you mean
[10:50] <cjwatson> possibly
[10:55] <cjwatson> at least my networking is slowly getting less appalling; twice the bandwidth I had last week
[10:55] <cjwatson> (twice of not a lot is still not a lot, but even so)
[11:13] <CIA-45> ubiquity: evand * r4993 trunk/ubiquity/frontend/gtk_ui.py: Fix typo around setting the ATK name of widgets. Thanks unit tests.
[11:15] <CIA-45> ubiquity: evand * r4994 trunk/ (151 files in 3 dirs): Update translations from Launchpad.
[11:20] <CIA-45> ubiquity: evand * r4995 trunk/debian/changelog: releasing version 2.7.37
[12:01] <cjwatson> hmm.  my libsoup-based callback fetcher for the timezone page *seems* to be working, but some of the completions randomly don't seem to be filled in
[12:03] <cjwatson> maybe I'm calling set_model at the wrong time
[12:46] <cjwatson> ev: I'd appreciate your review of https://code.launchpad.net/~cjwatson/ubiquity/libsoup-timezone/+merge/77324
[12:46] <cjwatson> I think I've tested it about as well as I can now
[12:47] <cjwatson> I used 'sudo iptables -A OUTPUT -p tcp --dport 80 -j DROP' to simulate the appropriate kind of lossy network
[13:03] <ev> cjwatson: will do!
[13:09] <cjwatson> (done a full test install with that on the nobbled network, succeeded)
[13:17] <CIA-45> debian-installer: cjwatson * r1541 ubuntu/ (build/config/amd64/netboot.cfg debian/changelog): Bump amd64 netboot image size by another 256KiB.
[13:18] <CIA-45> debian-installer: cjwatson * r1542 ubuntu/debian/changelog: releasing version 20101020ubuntu68
[13:35] <CIA-45> ubiquity-slideshow-ubuntu: evand * r386 ubiquity-slideshow-ubuntu/ (8 files in 2 dirs):
[13:35] <CIA-45> ubiquity-slideshow-ubuntu: Update screenshots to reflect the latest desktop changes
[13:35] <CIA-45> ubiquity-slideshow-ubuntu: (LP: #861410).
[13:45] <stgraber> cjwatson: two questions for you
[13:45] <stgraber> got RAOF to do some tests last night for bug 854967
[13:45] <ubot2> Launchpad bug 854967 in friendly-recovery "boot to rescue mode in Oneiric" [Undecided,Incomplete] https://launchpad.net/bugs/854967
[13:46] <stgraber> apparently nouveau isn't affected by the problem but nvidia is, testers confirm setting gfxpayload=text works for them
[13:46] <stgraber> is that something you'd be fine with me changing in grub (only for the recovery entry)?
[13:46] <cjwatson> not in grub; that's what the gfxpayload blacklist is for surely
[13:47] <cjwatson> whatever the nvidia package is should probably blacklist nvidia cards
[13:47] <cjwatson> that way it won't be in place if the nvidia package isn't installed
[13:48] <stgraber> oh, that'd make sense indeed, I'm not really familiar with how that blacklist works. Will comment in the bug, re-assign to the nvidia package and then have a look at that blacklist
[13:48] <stgraber> cjwatson: second one is bug 848072
[13:48] <ubot2> Launchpad bug 848072 in netcfg "[oneiric] net-installer dhcp client fails with a DHCPDECLINE" [Medium,Confirmed] https://launchpad.net/bugs/848072
[13:48] <cjwatson> right, still in my review queue, your patch looked reasonable at first glance though
[13:49] <stgraber> perfect
[13:49] <stgraber> getting back to playing with nvidia hardware then, thanks!
[13:49] <cjwatson> might want to check with slangasek whether he minds nvidia being gfxpayload-blacklisted across the board; it would likely affect his flicker-free boot work
[13:51]  * cjwatson laboriously works on duplicating bug 851704
[13:51] <ubot2> Launchpad bug 851704 in grub2 "GRUB reports error after successful installation (Alternate)" [High,Confirmed] https://launchpad.net/bugs/851704
[13:53] <stgraber> ok, I'll poke slangasek about it, though testing on my nvidia system here currently gives me "BIOS => blank => cursor => plymouth => blank => lightdm" so not exactly flicker free :)
[13:56] <cjwatson> yeah, that's probably a good sign for blacklisting
[13:57] <cjwatson> right.  excessively complex partitioning layout created, installing base system.  time for coffee
[14:07] <ev> cjwatson: I've replied to the merge proposal with a question
[14:08] <ev> ooo, coffee is a good idea
[14:08] <ev> shame we're down to the grade 4 stuff (out of 10)
[14:09] <jibel> ev, I tried r4995 and 'download while installing' still fails.
[14:09] <ev> ARGH
[14:09] <ev> how on earth..
[14:09] <jibel> ev, killpg should be called with the pgid not the pid but then it will also kill ubiquity which is a member of the same process group than update-apt-cache sudo being the parent process.
[14:12] <cjwatson> killing the pg sounds wrong.  shouldn't the descendant processes die due to SIGPIPE or similar?
[14:13] <cjwatson> you aren't using the usual SIGPIPE fix when spawning update-apt-cache; I wonder if that matters
[14:15] <cjwatson> ev: I'll give http://paste.ubuntu.com/698518/ a try in a bit and see how that behaves; my previous attempts didn't seem to go well but I might have made a mistake
[14:16] <superm1> cjwatson, i think there are two parts to that firewall bug.  the geoname piece that you have that soup merge request, but there is also clock-setup's postinst which is calling rdate
[14:16] <cjwatson> superm1: clock-setup has its own timeout handling though
[14:16] <cjwatson> I'm a lot less concerned about that because AFAICS it doesn't break installation
[14:17] <cjwatson> if there are two parts, it should be two bugs
[14:19] <superm1> oh clock-setup calls into tzsetup which is using wget with a 15 second timeout
[14:19] <jibel> superm1, clock-setup doesn't freezes the installer.
[14:20] <cjwatson> slowdown => not great; total freeze => baaaaaad
[14:20] <superm1> yeah
[14:55] <charlie-tca> ev, cjwatson : Can one of you push the fix to bug 781385 out today?
[14:55] <ubot2> Launchpad bug 781385 in ubiquity "Ubiquity GTK should have useful accessible names set in the Glade .ui files instead of using the variable names" [Medium,Fix released] https://launchpad.net/bugs/781385
[14:56] <ev> charlie-tca: it's been uploaded, but there was a build failure
[14:57] <ev> in the process of sorting that out
[14:57] <charlie-tca> Thank you
[14:57] <ev> sure thing
[14:58] <CIA-45> console-setup: cjwatson * r423 ubuntu/debian/ (changelog po/el.po po/fa.po po/he.po po/pa.po po/zh_TW.po):
[14:58] <CIA-45> console-setup: Update Greek, Persian, Hebrew, Punjabi, and Traditional Chinese
[14:58] <CIA-45> console-setup: translations from console-setup 1.68, fixing encoding problems
[14:58] <CIA-45> console-setup: (LP: #860562).
[15:01] <CIA-45> console-setup: cjwatson * r424 ubuntu/debian/changelog: releasing version 1.57ubuntu27
[15:18] <CIA-45> ubiquity-slideshow-ubuntu: evand * r387 ubiquity-slideshow-ubuntu/ (381 files in 7 dirs): Update translations from Launchpad.
[15:21] <CIA-45> ubiquity-slideshow-ubuntu: evand * r388 ubiquity-slideshow-ubuntu/debian/changelog: releasing version 48
[15:46] <cjwatson> ev: so we call jockey-text both in and out of the chroot?  *scratches head*
[15:47] <ev> cjwatson: yes, because jockey-text -C (for the composite driver), run in the live system apparently ends badly
[15:47] <cjwatson> wait, simple-plugins shouldn't matter because that calls jockey-text without -C
[15:47] <ev> so both calls are intended to do different things
[15:47] <ev> the first call is for the broadcom driver
[15:47] <ev> the second call is for the nvidia driver
[15:47] <cjwatson> so I'm still confused
[15:48] <cjwatson> unless jockey is doing the wrong thing with locales
[15:48] <ev> but it's happening in the first call, no?
[15:49] <cjwatson> I don't think so - the log seems to show it being quite late in installation
[15:50] <cjwatson> it's after a bunch of packages being installed in the target, so it can't be prepare
[15:50] <ev> ah, I misread
[15:50] <cjwatson> I suppose we could try setting LC_MESSAGES too but WTF
[15:52] <cjwatson> very confused
[15:56] <ev> hm
[15:58] <cjwatson> and I can't see gettext.gettext (i.e. _) being broken with respect to LC_ALL handling
[19:13] <dmarkey> cjwatson: when is ubuntu-installer change freeze for Ocelot?
[19:17] <bdmurray> I neglected to mention bug 645449 this morning in our meeting
[19:17] <ubot2> Launchpad bug 645449 in ubiquity "Ubiquity hangs at Keyboard layout if you use keyboard to navigate / select" [High,Confirmed] https://launchpad.net/bugs/645449
[20:04] <cjwatson> dmarkey: there isn't a particular installer freeze, we go with the freezes for the rest of the distribution; see the release schedule
[20:09] <dmarkey> cjwatson: so.. that would be tomorrow? :)
[20:11] <cjwatson> yes, although in practice we normally seek freeze exceptions for a number of late fixes
[20:12] <dmarkey> are you aware of anyone testing installing on Xen?
[20:37] <cjwatson> dmarkey: I did a set of tests at the end of last week and the start of this one, and fixed several bugs affecting Xen in all of lucid, maverick, natty, and oneiric
[20:37] <cjwatson> dmarkey: as far as I know it should now work fine in Xen
[20:37] <cjwatson> this was largely due to a friend of mine who runs a hosting company (www.jump.net.uk) who cornered me in the pub about it
[20:38] <cjwatson> of course I don't know what sort of variety there is in boot loader setups on Xen, so independent tests wouldn't hurt
[20:38] <bdmurray> bug 861609 is interesting since bug 442941 was fixed ...
[20:38] <ubot2> Launchpad bug 861609 in ubiquity "Installer failed - debconf passthrough issue" [Undecided,New] https://launchpad.net/bugs/861609
[20:39] <ubot2> Launchpad bug 442941 in ubiquity "debconf failed to upgrade from 1.5.27ubuntu1 to 1.5.27ubuntu2: exit status 128 - Use of uninitialized value $reply in scalar chomp at /usr/share/perl5/Debconf/FrontEnd/Passthrough.pm line 66" [Medium,Fix committed] https://launchpad.net/bugs/442941
[20:39] <dmarkey> cjwatson: do you know if they were using pygrub or not?
[20:39] <cjwatson> but it now (a) autoloads xen-{blk,net}front properly (b) actually honours grub-installer/grub2_instead_of_grub_legacy=false (c) knows how to install grub on /dev/xvd*
[20:39] <cjwatson> dmarkey: no, pv-grub
[20:39] <cjwatson> so maybe only (a) will be relevant to your setup
[20:39] <dmarkey> i see.. i'll test tomorrow
[20:39] <cjwatson> bdmurray: sort of thing where looking at the topmost symptoms may not be very helpful
[20:40] <dmarkey> if the grub.cfg hasnt changed, we should be file
[20:40] <bdmurray> cjwatson: where should I be looking?
[20:41] <cjwatson> the passthrough bits sound like a red herring
[20:41] <cjwatson> apt-setup failing with exit code 141: that's signal number (141-128) == 13
[20:41] <cjwatson> 'kill -l' tells you that's SIGPIPE
[20:42] <cjwatson> which happens when a process tries to write to a pipe and all the processes reading from it have gone away
[20:42] <cjwatson> (mind you there is an *awful lot* of that passthrough noise isn't there)
[20:43] <cjwatson> what's all that "Reverting lockdown the desktop environment." stuff?
[20:43] <bdmurray> is this an upgrade using ubiquity?
[20:43] <cjwatson> hm, irrelevant I think
[20:43] <cjwatson> Sep 28 17:14:30 ubuntu kernel: [ 1384.574306] apt-cdrom[29439]: segfault at 4 ip 00a5de9e sp bfbb6b84 error 6 in libc-2.13.so[9c7000+176000]
[20:44] <cjwatson> doesn't look brilliant
[20:44] <CIA-45> ubiquity: cjwatson * r4996 trunk/ubiquity/frontend/gtk_ui.py: grammar
[20:44] <cjwatson> Sep 28 22:43:14 ubuntu ubiquity: /usr/lib/ubiquity/migration-assistant/ma-ask: 40: : Permission denied
[20:44] <cjwatson> gnome-control-center segfault earlier
[20:44] <cjwatson> I'm really not sure I can tell what's going on here
[20:46] <bdmurray> what's this? Kernel command line: BOOT_IMAGE=/boot/ubnkern file=/cdrom/preseed/ubuntu.seed boot=casper quiet splash --
[20:47] <cjwatson> somebody's homerolled boot-from-hard-disk system?  not sure
[20:47] <cjwatson> don't know if it matters
[20:47] <cjwatson> the passthrough noise is basically a verbose way of saying "the thing I want to read replies from gave end-of-file instead"
[20:48] <bdmurray> okay that helps clarify things for me
[20:49] <cjwatson> where in this case that's probably coming from debconf-apt-progress wrapping apt-get update so that we get a progress bar around downloads
[20:49] <cjwatson> and the thing it's trying to read replies from is probably ubiquity
[20:50] <cjwatson> so it's very unclear what actually happened; the next step is to ask, if possible, for that to be reproduced with 'ubiquity -d'
[20:50] <cjwatson> maybe something has screwed up the debconf protocol pipe
[20:51] <cjwatson> all this kind of thing is basically bad response to an assertion failure - we could probably make it more graceful, but it wouldn't make the actual error go away :-(
[21:37] <CIA-45> ubiquity: superm1 * r4997 ubiquity/ (debian/changelog ubiquity/plugins/ubi-partman.py):
[21:37] <CIA-45> ubiquity: Preseed a bootloader target if one hasn't been set but running in
[21:37] <CIA-45> ubiquity: automatic mode. (LP: #861465)
[22:08] <infinity> ev: What was the rationale for splitting out oem-config-slideshow?
[22:09] <infinity> ev: Other than one different stanza (which could, arguably, be a bug), it looks identical to ubiquity-slideshow.
[22:26] <infinity> cjwatson: Or you?  Any idea on the above?
[22:27] <infinity> cjwatson: Not only does the split seem to not make a bunch of sense, but it also means that we don't get oem-config-slideshows for anything other than Ubuntu.  (Yes, fixable, but I'm failing to see the point in the first place)
[22:34] <cjwatson> I'm not sure, it'd be an IRC-log-trawling exercise ...
[22:34] <cjwatson> I thought at one point they were intended to be qualitatively different slideshows
[22:34] <infinity> Possibly.
[22:34] <infinity> How would you feel about this, for graceful fallback?
[22:34] <infinity> [22:34] <infinity> --- ubiquity/frontend/gtk_ui.py	2011-09-28 20:43:51 +0000
[22:34] <infinity> +++ ubiquity/frontend/gtk_ui.py	2011-09-28 22:33:27 +0000
[22:34] <infinity> @@ -782,7 +782,7 @@
[22:34] <infinity>              self.shutdown_button.hide()
[22:34] <infinity>  
[22:34] <infinity>          # Parse the slideshow size early to prevent the window from growing
[22:34] <infinity> -        if self.oem_user_config:
[22:34] <infinity> +        if self.oem_user_config and os.path.exists('/usr/share/oem-config-slideshow'):
[22:34] <infinity>              self.slideshow = '/usr/share/oem-config-slideshow'
[22:34] <infinity>          else:
[22:35] <infinity>              self.slideshow = '/usr/share/ubiquity-slideshow'
[22:35] <infinity> Thus allowing people to still use oem-config-slideshow-* if/when they want to, but falling back to ubiquity-slideshow-* when not present?
[22:35] <infinity> Since the latter exists on all flavours.
[22:38] <infinity> cjwatson: Pasted, for marginally less ugliness: http://paste.ubuntu.com/698778/
[22:38] <superm1> by default the ubiquity slideshow is uninstalled after a run though before you get to oem config I thought
[22:39] <infinity> superm1: Not in every abusive case of oem-config.  But fair point, that could explain the intent behind the split (though it seems poorly executed)
[22:40] <infinity> Given that the kde frontend doesn't know about oem-config-slideshow, and no flavours other than ubuntu have a package for it.
[22:43] <infinity> I'm tempted to make the above change just to fix things on preinstalled images anyway.  I see no way it can cause regressions, just catches a weird use-case.
[22:45] <cjwatson> infinity: http://irclogs.ubuntu.com/2011/02/11/%23ubuntu-installer.html#t14:35 seems to be the origin
[22:46] <cjwatson> I have no problem with that graceful fallback; I don't see how it can make anything any worse
[22:47] <infinity> That IRC conversation is pretty light on details.  Except, I guess it might have removed a dirty hack where you had to exclude ubiquity-slideshow-* from the live-manifest removal filter?
[22:47] <infinity> The hack seems better than packages with duplicate contents that seem to be slightly out of sync ;)
[22:48] <cjwatson> yeah, afraid I don't really remember
[22:50] <infinity> Well, not my pet project.  I'm just going to commit the fallback.
[22:51] <infinity> And by "commit", I mean send you a patch.  I seem to not be in that team anymore. ;)
[22:56] <infinity> cjwatson: I guess you could just remove Fabio from ~ubuntu-installer and add me.  I'm sure I'd be 87% more useful.
[23:25] <infinity> cjwatson: Can you merge http://bazaar.launchpad.net/~adconrad/+junk/ubiquity/revision/4998 for me?