[00:12] cody-somerville: same bootstrap? [00:12] base system created by debootstrap [00:13] right, but same as what? [00:13] like, instead of rebuilding it each time [00:13] oh, you mean continuous upgrades? [00:14] it's probably mostly ok, but the reasons we don't do that in Ubuntu proper include: [00:15] * it's only about 45 minutes to build the whole thing from scratch again, which isn't a big deal when run daily [00:15] * upgrades often require a bit of babysitting, whereas installations tend to work unattended provided that the packages aren't broken [00:16] * fresh installs produce cleaner images - there's often a bit of cruft left over from upgrades [00:16] * cody-somerville nods. [00:16] * it's easier to debug things when a fresh install goes wrong, because there are fewer variables: all that matters is the current state, not some hard-to-reproduce older state [00:17] so we just focused on automating the rebuild-from-scratch and then tend not to worry about the fact that it takes a bit longer [00:17] Okay. [00:20] I read somewhere that if a package isn't listed in /casper/filesystem.manifest-desktop than casper will uninstall it [00:20] With the last few installs I've done, that file hasn't exist at all [00:21] err... not casper [00:21] ubiquity [00:21] if that file doesn't exist, does ubiquity skip that? [00:23] yes, it does, but are these stock Ubuntu images? it's a serious problem if that file doesn't exist [00:24] No, these are images I've generated [00:24] ok, I recommend fixing that :) [00:24] one consequence of not fixing this is that the installer will still be on the installed system [00:24] which is likely to cause confusion [00:24] livecd-rootfs spits out a manifest-desktop at the right time [00:25] lh has a hook system so I imagine it'll be very easy to do the same [00:25] you're going to keep on running into this stuff [00:25] evaluate whether you gain more from live-helper than you would from not having to run into problems in the field and reimplement the same fixes we applied years ago [00:26] Thats indeed my goal [00:26] there's almost certainly stuff in debian-cd that avoids subtle installer bugs too [00:28] since live-helper is an official project of Debian and was used to produce some of the images for the last release of Debian I'm guessing those sort of things would be addressed by lh as well [00:29] assuming identical installers ... [00:29] It uses d-i [00:29] it uses Debian's d-i [00:29] How custom is Ubuntu's di? [00:30] not trivially [00:30] One option lh supports is have two partitions on the install medium - the livecd and one with d-i [00:30] we have some different components and a reasonably extensive collection of other changes [00:30] When it comes to install time, it'll copy over the livecd fs instead of the normal install [00:30] the thing I don't want is to end up de facto supporting live-helper because you need it and ask lots of questions about it :-) [00:31] if you're going to use live-helper, I think its problems need to be yours [00:31] copy over the livecd fs> which relies on a Debian d-i component, live-installer, that is not supported in Ubuntu because we use ubiquity for that purpose [00:32] it would not surprise me in the least if live-installer didn't work in Ubuntu without significant customisation; it will *certainly* not produce the same result as ubiquity out of the box [00:33] it sounds like you're assuming that live-helper's advertised features are offered by it standalone and that you can plug Ubuntu into it and get the same features, which I do not believe to be the case [00:59] cjwatson, Well, so far so good. I created a UNR jaunty cd last night and used ubiquity to install it [00:59] cjwatson, All it required was for me to install lh and and type two commands after reading a man page. [01:04] well, except for things like the installer still being installed on the target system ;-) [01:04] I do realise that live-helper is a lot easier to set up [01:05] It also scales [01:05] I'm just concerned that I don't even know the rough list of things that are likely to break in subtle ways; all I can do is think back through the things we've fixed in the image building tools in the past [01:06] pkgsel: cjwatson * r134 ubuntu/debian/ (changelog postinst): [01:06] pkgsel: If installing basic language packs fails the first time round, probably [01:06] pkgsel: because they aren't on the CD, try again once we have the final [01:06] pkgsel: sources.list in place (LP: #348393). [01:06] you may have the same d-bus machine id on all booted images, for example [01:06] that kind of glitchy sort of thing that you wouldn't immediately notice [01:06] long-lived systems accumulate bug fixes [01:08] * cody-somerville nods. [01:09] if I were you I would compare the image file by file with that produced by our software, both inside and outside the squashfs [01:09] any difference is a potential bug [01:09] aside from trivial rubbish like gzip timestamps [01:10] thats a good idea === kralph is now known as kenyon [01:27] libdebian-installer: cjwatson * r187 ubuntu/debian/control: set Vcs-Bzr for Ubuntu [01:32] libdebian-installer: cjwatson * r188 ubuntu/ (debian/changelog src/system/subarch-arm-linux.c): Added subarchitecture imx51 for Freescale iMX51 SoCs (LP: #345534). [01:48] libdebian-installer: cjwatson * r189 ubuntu/debian/changelog: merge from lp:~mcasadevall/libdebian-installer/imx51-support (mostly applied already from a patch, but taking the merge to clean up metadata) [01:51] base-installer: cjwatson * r356 ubuntu/ (8 files in 3 dirs): merge from lp:~mcasadevall/base-installer/imx51-support [02:01] debian-installer: cjwatson * r1073 ubuntu/ (6 files in 4 dirs): merge from lp:~mcasadevall/debian-installer/imx51-support === ogra_ is now known as ogra === davmor2 is now known as davmor3 === davmor3 is now known as davmor2 [10:32] oem-config: evand * r639 oem-config/ (6 files in 4 dirs): [10:32] oem-config: Merge in the new KDE time zone map from ubiquity. Thanks Roman [10:32] oem-config: Shtylman. === davmor2 is now known as bolt_ === bolt_ is now known as davmor2 [11:56] ubiquity: cjwatson * r3129 ubiquity/ (debian/changelog ubiquity/components/install.py): [11:56] ubiquity: Fix ubiquity/install_bootloader handling to check the seen flag rather [11:56] ubiquity: its value when deciding whether to override it with the UI's value (see [11:56] ubiquity: LP #348660). [11:56] Launchpad bug 348660 in ubiquity "ubiquity unsets ubiquity/install_bootloader=false at some point during installation" [High,Confirmed] https://launchpad.net/bugs/348660 [12:15] ubiquity: cjwatson * r3130 ubiquity/ (5 files in 3 dirs): [12:15] ubiquity: Outside automatic mode, default the UI's "Install boot loader" question [12:15] ubiquity: to the value of ubiquity/install_bootloader (see LP #348660). [12:15] Launchpad bug 348660 in ubiquity "ubiquity unsets ubiquity/install_bootloader=false at some point during installation" [High,Confirmed] https://launchpad.net/bugs/348660 [12:16] ubiquity: cjwatson * r3131 ubiquity/debian/changelog: rearrange changelog, since the last two commits fix LP: #348660 [13:05] cjwatson, https://code.launchpad.net/~persia/choose-mirror/armel+lpia/+merge/4774 [13:06] ok [13:13] persia: your branch is based on intrepid; I fixed that in jaunty already a while back ... [13:14] ubiquity: evand * r3132 ubiquity/ (debian/changelog ubiquity/frontend/gtk_ui.py): [13:14] ubiquity: Make sure that the before and after partitioning bars use the same [13:14] ubiquity: colors for the same partitions (LP: #289324). [13:17] ubiquity: evand * r3133 ubiquity/ubiquity/segmented_bar.py: Initially set the partition being resized to the maximum size it can be. [13:18] cjwatson, Hrm. Sorry. I'll refresh. Dunno how I got the wrong *release* in my branch. [13:19] persia: odd that you said you reproduced it on jaunty though. Is it possible that something else was wrong? [13:19] Very likely so. [13:19] just trying to think of what else might break in a similar way [13:20] nothing is coming to mind ... [13:22] I didn't personally run the alternate install. I'll track it down: perhaps it was just something odd about the way that the custom d-i being built was built. === persia_ is now known as persia [16:02] evand: should wubi work on 64bit vista? [16:06] evand: So do you know if wubi will work on vista 64 bit at all? === ogra_ is now known as ogra [17:16] ubiquity: evand * r3134 ubiquity/ (3 files in 3 dirs): [17:16] ubiquity: Fix the placement of the partition bars in scrolled windows when [17:16] ubiquity: necessary. === nxvl_ is now known as nxvl === tjaalton_ is now known as tjaalton [21:00] user-setup: cjwatson * r166 ubuntu/ (debian/changelog user-setup-apply user-setup-ask): [21:00] user-setup: Don't offer encrypting the home directory if the selected user's home [21:00] user-setup: directory already exists (LP: #321345). [23:38] ubiquity: evand * r3135 ubiquity/ (debian/changelog gui/glade/stepPartAuto.glade): Don't let the partition bars eat up any extra space. === evand_ is now known as evand [23:40] lool,persia: you can ignore my request for logs on bug 288320 [23:40] Launchpad bug 288320 in ubuntu-cdimage "Networkless Ubuntu MID install hangs at "Scanning archive..."" [Medium,Fix released] https://launchpad.net/bugs/288320 [23:53] casper: TheMuso * r600 trunk/ (3 files in 3 dirs): [23:53] casper: * scripts/casper-bottom/30accessibility && ubiquity-hooks/30accessibility: [23:53] casper: - Make the special case disabling of pulseaudio per user, and not system [23:53] casper: wide. [23:53] casper: - Also disable PulseAudio for the Braille profile. [23:59] cjwatson, whats the magic keyword to make d-i include hdd-like block devices in its search for the cd?