=== superm1 is now known as superm1|away === superm1|away is now known as superm1 [05:37] ubiquity: evand * r2857 ubiquity/ (d-i/manifest debian/changelog): [05:37] ubiquity: Automatic update of included source packages: debian-installer-utils [05:37] ubiquity: 1.59ubuntu3, grub-installer 1.32ubuntu6, hw-detect 1.63ubuntu3, [05:37] ubiquity: user-setup 1.20ubuntu7. [05:53] ubiquity: evand * r2858 ubiquity/debian/changelog: releasing version 1.10.0 [06:38] cjwatson: Is there a historical reason why grub-installer and lilo-installer use sfdisk rather than parted? [06:52] cjwatson: also, if you have a free moment, would you give the latest usb-creator a go to see if bug 269052 is still present for you [06:53] (I cannot reproduce it) [07:48] Is passwd/auto_login as a boolean preseed option truly ubiquity-specific, or does it inherit from d-i ? I'm not seeing it the relevant part of the installation-guide, but wanted to double-check. [07:53] persia: it's not ubiquity specific, it's in user-setup. See user-setup-apply in the user-setup source package. [07:54] evand: Thank you. So "user-setup passwd/auto_login boolean true" would be the correct string? [07:55] whatever passwd/auto-login boolean true [07:55] dash, not underscore [07:55] and whatever because aiui, the ownership field isn't really important when preseeding [07:56] Ah. Right. Is the value of "whatever" not important? [07:56] that's my understanding, but cjwatson might prove me quite wrong. [07:57] Well, I was advised to use d-i for whatever for everything in my last preseed file. I suspect I'll want it here as well, as user-setup seems a fairly standard d-i component. [07:59] ubiquity definitely doesn't care as casper just uses debconf-set-selections on the preseed file [07:59] er, it's late. That's irrelevant. Ignore my previous statement. [08:00] and ubiquity doesn't seem to even check the value of whatever when trying to comprehend the preseeding. [08:01] Right. Found it. Thanks. [08:02] anytime [08:02] Oh, and just in case you haven't been told enough lately: ubiquity is *wonderfully* flexible and themeable, and I've been pleasantly surprised by how well it has worked in each of the environments I've tried it. [08:03] Thank Colin, he laid the foundation, I'm just here to build flimsy walls ;) [08:04] Thanks though [08:13] hrm, figuring out an EBUSY in mount in the initramfs where lsof isn't available. [08:13] This should be fun. [08:16] evand: sfdisk> historical state rather than historical reason, I think. They ought to use parted or a libparted-based program. [08:17] ok [08:17] thanks [08:17] evand: ok, I'll have a look at 269052 when I can [08:17] persia: "d-i" as the owner would be right [08:17] So there are no relevant processes running, there's nothing mounted below /cdrom, and yet I cannot remount it ro (when persistent is set in the kernel command line) [08:18] the owner field isn't directly important when preseeding, but it does matter for proper garbage-collection of the debconf database further down the line [08:18] cjwatson: Excellent. Thank you. [08:18] noted [08:19] evand: bind-mounts maybe? [08:19] So by using "d-i" as owner, one can identify the installation-only stuff, or by using "tasksel" one can identify the task-selection stuff? [08:19] And then (non-udeb) packages register their own directly? [08:21] hrm, perhaps this?: [08:21] /dev/sda1 on /cdrom [08:21] /dev/sda1 on /casper-rw-backing [08:22] everything else is unmounted (save /sys, /proc, /dev and /) [08:23] debian-installer: cjwatson * r966 ubuntu/debian/changelog: [08:23] debian-installer: No-change rebuild to pick up new components, including disabling the [08:23] debian-installer: broken e1000e driver. [08:24] persia: d-i is magic in that questions with that owner don't get copied to the target system. (Disclaimer: I haven't checked recently that ubiquity honours this ...) [08:25] persia: for everything else, setting an owner means that those questions will get unregistered when the owning package is purged [08:25] * persia checks on a very recently ubiquity-installed system [08:26] debian-installer: cjwatson * r967 ubuntu/debian/changelog: releasing version 20080522ubuntu16 [08:29] Indeed. None of the d-i entries are shown by debconf-get-selections on a ubiquity-installed system. [11:44] superm1, lupin-casper is missing from mythbuntu, that is required to use it with wubi [15:13] superm1, after installing I am left with a ubuntu user [15:59] xivulon, do you need a disk mastered with lupin-casper then, or can you test around it? [16:00] xivulon, yeah i'm aware of that happening re: the ubuntu user, but not sure how to take care of it atm [16:04] superm1: What's happening with the ubuntu user? [16:04] persia, on mythbuntu disks the entire home directory for the "ubuntu" live session user isn't getting wiped [16:05] it comes from a side effect of the way that we build the disks i believe right now [16:05] superm1, would be better to have lupin-casper in the iso although I can work around [16:05] superm1: Are you copying /rofs ? [16:06] * persia thought /home/ubuntu came from casper at boot time [16:06] persia, just like regular ubuntu operates in copying files [16:06] strange. [16:06] * persia grabs a mythbuntu image. [16:06] its because i believe during our ISO generation something gets added - a background i believe [16:07] it showed up a while ago, and its been off my radar for a bit - but one of the deterants that has kept from switching to the livecd.sh that is used for making canonical disks [16:08] xivulon, is lupin-casper automatically removed when ubiquity is done, or would this be something that had to be added to code to do so? [16:09] superm1: You don't use livecd-rootfs livecd.sh ? [16:09] superm1, lupin-casper only affects the casper initrd adding extra boot options, so after installation it is not used since a regular initrd gets installed [16:10] persia, not currently - that was one of the things that i was going to convert over to this cycle, but with virtualbox being broke for such a long time, i didn't have a method to test CDs [16:10] other than burn and bring to real hardware [16:10] ah xivulon [16:11] that said we do install lupin-support at the end of the installation... [16:12] persia, we do a decent job at emulating most of it's functionality though atm [16:12] so that packages has to be on the iso [16:12] superm1: Hrm. Well, I'll look at the structure, and contents of the squashfs, but I suspect the migration might fix the issue. [16:13] persia, yeah i'm pretty sure it would. so hopefully for jaunty can take another stab at trying to migrate [16:13] I know that changing from Moblin Image Creator to livecd-rootfs fixed *lots* of issues with Ubuntu MID. [16:14] superm1: I only find the alternate CD. From where should I pull a daily? [16:14] persia, the daily mirror isn't public right now (we're migrating servers). grab alpha 6 [16:14] from mythbuntu.org [16:14] it will try to offer you a mirror local to you [16:16] OK. I'm not having luck with navigation. [16:16] Ah. Found it. [16:16] http://mythbuntu.org/download/?file=mythbuntu-8.10-alpha6-desktop-i386.iso [16:25] superm1: Yep. It's in your squashfs, which is why it gets there. [16:25] persia, well would you perhaps have a good recommendation on how to have a different background just during install then without making modifications to casper? [16:26] Nope. That would be a casper thing. [16:26] then most definitely will need to be deferred to jaunty [16:27] Well, it doesn't necessarily need modifications to core casper: you could just drop the additional casper scripts on the ISO. [17:42] nm resets static ip info and goes back to dhcp [17:44] the above is in mythbuntu not sure yet if the same holds on standard intrepid (settings get reset at reboot) [18:12] get the same behaviour on vm [18:13] xivulon, well we dont do anything special in NM [18:14] so i'd imagine same bugs holds true to both [18:20] added a comment to #5364