[07:59] <lifeless> 'lo
[08:00] <cjwatson> mpt: ubuntu-express was written by Guadalinex in response to our specifications; however it did not meet my expectations and making it do so was going to be a sufficiently painful change in direction that I needed to rename it
[08:00] <cjwatson> mpt: I do not know whether Guadalinex are still using their ubuntu-express code
[08:00] <cjwatson> this is an object lesson in why not to call specifications "Ubuntu Foo"
[08:04] <evand> hello lifeless
[08:06] <lifeless> I love DRY violations. not.
[08:06] <cjwatson> DRY?
[08:06] <lifeless> dont repeat yourself
[08:06] <lifeless>  /cdrom was changed to /live_media in multiple initramfs scripts
[08:07] <lifeless> rather than a single variable being sourced from casper-functions or some such, so I missed a place first time around setting it back
[08:08] <cjwatson> note that the one in debian/casper.init could not be sourced from casper-functions since that is not available when that script is run
[08:08] <cjwatson> so be careful if you attempt to implement that
[08:09] <lifeless> I dont have time right now to do more than whinge
[08:09] <lifeless> I might file a bug for reference
[08:10] <lifeless> seems to be much happier
[08:10] <lifeless> thanks!
[08:10] <cjwatson> np
[08:10] <lifeless> 34%, will know in 5 minutes whether its good
[08:11] <lifeless> how likely is it to have other timebombs for me?
[08:11] <cjwatson> if there's an obvious way in which ubiquity should be changed to do NFS for itself, let me know - it would have to be controlled by something passed in by/via casper though
[08:11] <cjwatson> medium likely
[08:11] <cjwatson> there are plenty of scripts that run after the copy stage (up to 75/80% or thereabouts) that use /cdrom
[08:11] <lifeless> root FS is know by the kernel
[08:12] <lifeless> as long as they dont try to actually mount or unmount the CD itself, just loopback stuff, it should be all good
[08:12] <cjwatson> yeah, unpicking it to get at the squashfs is fiddly though
[08:12] <lifeless> 72%
[08:12] <cjwatson> I wouldn't put it past apt-setup to *try* to unmount the CD, but hopefully it won't
[08:13] <lifeless> 78%
[08:13] <lifeless> configuring locales
[08:13] <cjwatson> note that once we do live-cd-stacked-filesystems, the set of stuff ubiquity needs to mount will be != the squashfses underlying the rootfs
[08:13] <lifeless> running 22gnome_panel_data
[08:14] <lifeless> cjwatson: wheres the spec for that stacked stuff?
[08:14] <cjwatson> live-cd-stacked-filesystems. that's the spec name.
[08:14] <lifeless> ok
[08:14] <lifeless> google-juice
[08:14] <cjwatson> (I nearly always refer to things by spec name to give people a hook
[08:14] <cjwatson> )
[08:14] <lifeless> ah, I see.
[08:15] <lifeless> but they will still live in a FS told to casper
[08:15] <cjwatson> yes
[08:15] <lifeless> i.e. by netboot=nfs nfsroot=blah
[08:15] <lifeless> which is what the debian casper patches do
[08:15] <lifeless> 96%
[08:16] <lifeless> 98%
[08:16] <lifeless> I'm getting excited
[08:16] <lifeless> woot
[08:16] <lifeless> rebooting.
[08:17] <lifeless> cjwatson: thank you! I owe you one cerveza
[08:34] <cjwatson> excellent
[09:06] <evand> Anyone have thoughts on whether it would make more sense to let the user change the details of the default user in migration-assistant, or to just gray those options out when the default user is selected?
[09:08] <mpt> evand, do you allow changes to details of any of the other user accounts?
[09:09] <evand> mpt, yes, if I understand your question correctly.  The interface is at https://wiki.ubuntu.com/MigrationAssistance under design.  They can currently change the details of any of the users they create in that form.
[09:09] <evand> They cannot however, change the details of reserved users, as it will not let them select one as a username.
[09:12] <evand> I'm leaning towards not letting them, as it might be more confusing than not letting them and they can always hit the back button and change it on the User Setup page.
[09:12] <evand> err than letting them
[09:12] <mpt> I don't see any mention of "reserved" on that page
[09:12] <mpt> Tell me what you mean by "reserved users"
[09:13] <evand> Yeah, I didn't know they existed until recently.  user-setup (the debconf module) has a list of reserved usernames such as root, nobody, cups, etc.
[09:13] <mpt> oh, ok
[09:13] <evand> Migration Assistant throws up an error if the user tries to select one of those as the user to import to.
[09:13] <mpt> ok
[09:13] <mpt> so
[09:14] <mpt> (a) if a Windows account just happens to have a login name the same as one of those reserved accounts, you autogenerate a different one
[09:14] <evand> not exactly...
[09:15] <mpt> (b) if someone tries to enter a reserved login name themselves, the error needs to be handled somehow
[09:15] <evand> If I set my default user as evan on the user setup page.  And then I go to the migration assistant page, select Evan from Windows, set the username to import to as evan, should I be able to change the fullname field and password field.
[09:15] <mpt> this is the old "display an error message" vs. "make continuing impossible" problem
[09:16] <mpt> ah
[09:16] <evand> (b) is handled already in the same manner as it is in User Setup (an error message alongside the field)
[09:16] <mpt> so maybe we have an ordering problem here
[09:17] <evand> Of the pages?
[09:17] <mpt> yes
[09:17] <mpt> Maybe the "Who are you?" page should be after the "Windows accounts" page
[09:18] <mpt> so the installer can ask you whether you in particular are represented by one of the migrated accounts, or whether you're someone new
[09:18] <mpt> Would that break anything?
[09:19] <mpt> Why did we go to the trouble of specifying "Migration-Assistant defaults seeding" in the first place, instead of moving migration assistance there?
[09:19] <evand> Not to my knowledge, but cjwatson would know for sure.  But we are going to be preseeding data for the user setup page via migration-assistant.
[09:19] <evand> hrm, I do not recall.
[09:19] <mpt> migration depends on partitioning
[09:20] <evand> right!
[09:20] <mpt> partitioning doesn't depend on who-are-you
[09:21] <mpt> so it could be Language, Location, Keyboard, Partitioning, Migration, Who are you, Summary
[09:23] <evand> It could.  I don't see that breaking anything, though cjwatson could disagree, I'm not sure.  How would we change Who are you?  A drop down box with the usernames created in migration-assistant?
[09:24] <mpt> yes
[09:26] <evand> imho, that would flow better, I'll run it by him or he'll probably see this.
[09:26] <evand> Thanks mpt
[09:27] <mpt> np
[11:33] <tepsipakki> cjwatson: the "udpkg --unpack" -spell works just fine if it is run from early_command :)
[11:34] <tepsipakki> meaning that postinst is handled when it is supposed
[11:37] <tepsipakki> and templates work as well
[11:37] <tepsipakki> i guess
[11:38] <tepsipakki> actually, I forgot to check that
[12:06] <tepsipakki> hmm, seems to work, yay
[01:19] <thom> cjwatson: is there anyway to get the configured swap partition from debconf? i can't see anything useful in debconf-get-selections --installer
[01:20] <thom> so i'm going to parse /proc/swaps in late_command, but i was hoping there was a less icky way
[03:42] <cjwatson> evand: I guess moving user-setup after the partitioner would be possible now; it wouldn't have been possible pre-edgy due to icky debconf interactions
[03:42] <cjwatson> (partman's debconf instance used to continue running during the summary page)
[03:42] <cjwatson> I think it'll surprise people, but ...
[03:42] <cjwatson> user-setup runs after partitioning in d-i, so shrug, I guess
[06:55] <evand> ok