[09:11] <ev> stgraber: having a look nowish
[09:18] <ev> joy, launchpad is timing out on bug attachments
[09:50] <ev> cjwatson: with respect to bug 758658, is there any historical reason why we don't retranslate once we get past the keyboard step and thus have a more complete locale (en_GB, for example)?
[09:50] <ev> perhaps that would be horribly inconsistent
[10:24] <ev> mm, that patch for 759804 isn't working quite as expected.  Digging.
[11:05] <ev> ah, jockey and nvidia-glx rely on uname -r to work. Bum.
[11:45] <cjwatson> ev: the keyboard step shouldn't affect translations.  The timezone step does, and I thought we did retranslate after it
[11:45] <cjwatson> we certainly should
[11:48] <ev> we don't appear to as far as I can tell. We call reset_locale in cleanup but that's it.
[11:48] <ev> for ubi-timezone, that is
[11:50] <ev> so then are you happy keeping country-specific pofiles and translating to them after that step?
[11:52] <ev> rather than just limiting ourselves to what localechooser can support
[11:53] <cjwatson> what do you need beyond what localechooser can support?  I'd rather not have divergent treatments of locales
[11:53] <cjwatson> you can't properly select a locale that localechooser doesn't support anyway
[11:55] <ev> distinction between en_US and en_GB in the slideshow (as people are whining in bug 758658), which is possible without involving ubiquity, but it seems like a logical extension to support better translations there as well if we can.
[11:57] <cjwatson> but localechooser *does* support en_GB vs. en_US
[11:57] <cjwatson> d-i generally doesn't ship different translations for them, but localechooser is entirely capable of selecting either
[11:58] <ev> ahh, I've been going off of languagelist rather than following the code
[11:58] <cjwatson> it doesn't differentiate them at the language selection step, that's all ...
[11:58] <ev> apols
[12:00] <ev> okay, so roughly: tell ubiquity to retranslate with the complete locale after the timezone page, and stop removing country-specific pofiles from the source package.
[12:02] <cjwatson> or just add a couple of country-specific po files when necessary
[12:02] <cjwatson> it's often just noise, there's a certain amount of judgement involved
[12:02] <ev> sorting the slideshow should be as simple as adding the country-specific pofiles back, which I imagine is acceptable for natty. Do you think the ubi-timezone change is a step too far? (in which case I'll start the natty-beta branch divergence from there)
[12:03] <ev> indeed, I'll do it manually rather than modifying my existing remove-unused-languages script
[12:03] <cjwatson> no, I think it's fine; you could reduce the risk by only retranslating the slideshow
[12:03] <ev> okay, cool
[12:03] <cjwatson> ev: note BTW that your script is wrongly removing Norwegian
[12:03] <ev> oh yikes
[12:03] <ev> thanks for the heads up
[12:03] <ev> I'll fix it
[12:03] <cjwatson> I put it back recently and it looks like you've deleted it again
[12:04] <cjwatson> thanks
[12:04] <ev> sorry
[12:04] <cjwatson> np, I guessed the second time that it must be scripted ...
[12:05] <ev> heh, indeed
[14:08] <ev> stgraber: your patch doesn't appear to fix the bug.  If I manually set the device for bootloader installation to /dev/sda1 and then add another partition, it shouldn't jump back to /dev/sda for the bootloader.
[14:08] <ev> Mind you, if I delete /dev/sda1, it should.
[14:12] <ev> ugh, when constructing a test for functionality in a particular configuration, it's probably best to make sure it works first in the general case :-/
[14:13] <ev> nvidia-glx-173 being uninstallable, in this particular case
[15:06] <CIA-7> choose-mirror: cjwatson * r1157 ubuntu/ (debian/changelog mirrorlist):
[15:06] <CIA-7> choose-mirror: Record a dummy "GB" entry in templates files for non-country-specific
[15:06] <CIA-7> choose-mirror: mirrors (i.e. ports.ubuntu.com) so that ports installations have more
[15:06] <CIA-7> choose-mirror: choices than just "enter information manually" (LP: #756719).
[15:08] <CIA-7> choose-mirror: cjwatson * r1158 ubuntu/debian/changelog: releasing version 2.37ubuntu2
[15:35] <ev> how very odd. The man-db trigger is failing in the chroot when installing the nvidia driver via jockey, but jockey sets the frontend to noninteractive.
[15:42] <stgraber> ev: yeah, it definitely needs more testing and fixes ;) I'll work on that once I'm done testing Edubuntu
[15:42] <ev> cool, thanks
[17:49] <CIA-7> cdebconf: cjwatson * r2340 ubuntu/debian/ (palette palette.dark palette.original palette.ubuntu): add debian/palette*, present in 0.154ubuntu2 upload but missing from bzr
[18:06] <CIA-7> base-installer: cjwatson * r951 ubuntu/ (debian/changelog kernel/i386.sh): Support AMD CPU family 18 (LP: #760490).
[18:08] <CIA-7> base-installer: cjwatson * r952 ubuntu/debian/changelog: releasing version 1.116ubuntu2
[18:11] <ev> cracked that chestnut
[18:11] <ev> ish
[18:11] <ev> I've got hacks all over the place, but lets see what I can pear off
[18:12] <ev> final piece of the puzzle was that jockey sets the frontend to noninteractive, but ubiquity leaves DEBIAN_HAS_FRONTEND set
[18:30] <rbelem> ping cjwatson
[19:10] <cjwatson> rbelem: yes?
[19:35] <stgraber> ev: http://paste.ubuntu.com/594160/ seems to work relatively well with the few tests I did here
[19:46] <cjwatson> rbelem: in future, please just say what you want rather than pinging first
[20:00]  * highvoltage is considering adopting a policy of ignoring contentless pings
[21:52] <rbelem> cjwatson, ok sorry
[21:52] <rbelem> cjwatson, i have some questions about oem-config
[21:53] <rbelem> cjwatson, i would like run some tasks after oem-config finishes
[21:54] <rbelem> cjwatson, how to do that? adding a script in /usr/lib/ubiquity?
[22:58] <bdmurray> How is "Download updates while installing" supposed to work?  Should they be installed or just downloaded or something else?