[10:04] <CIA-4> ubiquity: evand * r4500 trunk/ (debian/changelog ubiquity/plugins/ubi-timezone.py): Append to the timezone model using the correct types (LP: #702898).
[10:06] <ev> preparing another upload
[10:06] <ev> speak now if you want anything in it
[10:07] <CIA-4> ubiquity: evand * r4501 trunk/debian/real-po/ (63 files): debconf-updatepo
[10:22] <CIA-4> ubiquity: evand * r4502 trunk/ (123 files in 3 dirs): Update translations from Launchpad.
[10:35] <CIA-4> ubiquity: evand * r4503 trunk/debian/changelog: releasing version 2.5.14
[11:06] <ev> cjwatson: Cody is seemingly accurately pointing out in bug 707749 that debian_support.Version (lp:usb-creator r320) is not applicable for comparing releases.  I just want to get an idea of your motivation behind using it before addressing the bug.
[11:06] <ubot2> Launchpad bug 707749 in usb-creator "ValueError: Invalid version string" [High,New] https://launchpad.net/bugs/707749
[11:06] <ev> distribution releases, that is
[11:09] <cjwatson> ew
[11:09] <cjwatson> yeah, ok, it was a bit of a cheat
[11:09] <cjwatson> if he's managed to construct something that actually breaks it then I suppose we need to roll our own :-/
[11:10] <cjwatson> (though looks like usb-creator needs a bit of work to understand that .disk/info anyway, if I'm understanding correctly?)
[11:12] <ev> indeed
[12:31] <ev> so much for the "ripping to bootloader out of the CD" idea.  I don't think it's going to work unless we set -hard-disk-boot in genisoimage, which we haven't.
[12:32] <ev> assuming I understand the El Tortio specification correctly
[12:32] <cjwatson> sorry, which idea?
[12:33] <ev> pulling the bootloader off the ISO to avoid syslinux/grub version incompatibilities
[12:33] <ev> my thought was some combination of geteltorito + syslinux and memdisk might work
[12:33] <cjwatson> isohybrid does some mangling which I think is roughly equivalent to ending up with -hard-disk-boot
[12:33] <cjwatson> you might find dumpet helpful if you're playing with this
[12:33] <ev> neat, thanks
[12:34] <cjwatson> I think that the -hard-disk-boot stuff only governs header bits, and doesn't necessarily require different contents
[12:34] <cjwatson> so in principle it ought to be possible to twiddle it after the fact
[12:35] <ev> oh? Maybe I'm misunderstanding the problem entirely then, as surely the value of that doesn't pass through to the boot code that geteltorito extracts, so perhaps the boot is failing for a different reason entirely
[12:36] <cjwatson> I think the way I would approach this would be to compare what you need to do with what isohybrid is doing (probably using dumpet), and try to work out which of the differences are relevant
[12:37] <ev> indeed, will do.  Thanks
[16:06] <superm1> ev, i thought the worry about grub version incompatibilities went away when the CDs switched to grub by usb-creator just dd'ing boot.img and core.img?
[16:06] <cjwatson> but the CDs haven't switched to grub and it continues to be really hard
[16:07] <cjwatson> I'm still working on it, but emulating enough of gfxboot to be acceptable is a pain in the ass
[16:09] <superm1> ah, i wasn't sure how far along you were with it.
[18:17] <ev> hm, some combination of the following may work: 1) write the small bootloader used to find isolinux to the MBR, 2) find out the LBA of isolinux.bin and write it to offset 432, 3) profit?
[18:18] <ev> something to play with tomorrow
[21:58] <CIA-4> grub-installer: cjwatson * r1195 ubuntu/ (73 files in 3 dirs): merge from Debian 1.60