[10:20] <cjwatson> evand: just saw the slideshow for the first time (it's been a while since I did a desktop installation) - very impressed
[10:20] <evand> hooray
[10:20] <evand> thanks
[10:20] <evand> though lots of credit goes to the community team that created the slides and the javascript transitions
[10:25] <cjwatson> pass it on :)
[10:26] <cjwatson> is it worth trying to space it out a little better? I noticed that the last slide is up from maybe 45% or so onwards
[10:27] <evand> will do :)
[10:27] <evand> I don't follow.  Can you explain that in another way?
[10:31] <cjwatson> well, there were a bunch of slide transitions up to about halfway through the installation progress bar
[10:31] <cjwatson> then the same slide was up for the entire second half of the progress bar
[10:31] <cjwatson> (the "Getting Help with Ubuntu" one, I assume the last slide)
[10:31] <evand> ahh, indeed
[10:32] <cjwatson> are they time-based transitions at the moment?
[10:32] <evand> indeed, 30 seconds
[10:32] <cjwatson> I wondered if maybe it should be max(30 seconds, 100% / number of slides)
[10:32] <cjwatson> or something like that
[10:33] <evand> yeah, a variation of that which ensures that slides aren't displayed for less than some minimum value (perhaps 10 or 15 seconds) would be fine with me
[10:33] <evand> I'll work on a fix
[10:38] <cjwatson> if it's max(30 seconds, 100% / number of slides), then 30 seconds is that minimum value ...
[10:44] <evand> right, I'd claim it was too early, but it's nearly 11
[10:49] <ogra> cjwatson, https://bugs.launchpad.net/bugs/423639 feel free to comment ;)
[10:57]  * cjwatson stares at icu a bit. Where on earth does it actually get the translations from?
[10:57] <cjwatson> ogra: hah, yeah
[10:57] <cjwatson> but Dylan raises an actual UI concern, damn him ;-)
[10:58] <ogra> heh
[10:58] <ogra> well, for you it is at 45% at least :)
[10:58] <ogra> for me its still partitioning if the last slide shows up
[10:59] <cjwatson> I commented
[11:02] <ogra> heh looping ... good idea
[11:02] <evand> hrm
[11:25] <cjwatson> dpm: you might be interested in the horror involved in the second paragraph of https://bugs.launchpad.net/ubuntu/+source/gfxboot-theme-ubuntu/+bug/408393/comments/15 :-)
[11:26] <cjwatson> some investigation was required to come up with all that ...
[11:26]  * dpm reads...
[11:26] <cjwatson> dpm: I suspect that perhaps translation teams ought to be encouraged to get the Unicode CLDR translated into their language quite early on, as it seems like quite a complex process
[11:26] <cjwatson> and there's a lead time
[11:31] <dpm> ok, yes. Thanks a lot for the pointer. I'll let translators know
[11:37] <evand> ick, the translations that come out of launchpad for ubiquity-slideshow-ubuntu are a bit of a mess.  Different directories for different translations of the same template ('.', 'accessibility', 'accessibility.html', where each directory has a few different po files)
[11:39] <dpm> evand: I've forwarded your question to #launchpad.
[11:50] <CIA-33> ubiquity-slideshow-ubuntu: evand * r138 html/ (170 files in 12 dirs): Updated translations from Launchpad.
[11:50] <CIA-33> console-setup: cjwatson * r113 ubuntu/debian/changelog: bug closure for previous commit
[12:31] <CIA-33> ubiquity-slideshow-ubuntu: evand * r139 html/debian/changelog: releasing version 5
[13:00] <CIA-33> ubiquity: cjwatson * r3439 ubiquity/gui/ (gtk/stepLocation.ui qt/stepLocation.ui): adjust timezone labels in .ui files to match translation templates
[13:20]  * evand considers automatic new translations importing into the usb-creator trunk branch.
[15:15] <cody-somerville> cjwatson, Do I have to do anything special to take advantaged of fuse-unionfs in karmic for live images or will casper take care of it?
[15:16] <ogra> advantage ?
[15:16] <ogra> heh
[15:16] <ogra> cody-somerville, there is no advantage in using unionfs-fuse, use aufs :)
[15:16] <cjwatson> what he said
[15:17] <cody-somerville> splendid
[15:17] <cody-somerville> So the only thing I need to do is get the new version of squashfs-tools and all should be well?
[15:17] <cjwatson> if you want to force unionfs-fuse, you can make sure unionfs-fuse is installed when building the initramfs and boot with union=unionfs-fuse; but it's not normally worth it
[15:17] <cjwatson> should be
[15:17] <ogra> cjwatson, seems you introduced some new bugs with the partman merge btw, i was just told manual partitioning is broken
[15:18] <ogra> (bugs on their way)
[15:18] <cjwatson> ogra: ok, I'll have a look when they arrive
[15:18] <cody-somerville> Do you know if the new squashfs module will work with the jaunty kernel?
[15:19] <cjwatson> you of course can't insmod the new module in a jaunty kernel; the ABI is different
[15:19] <cjwatson> you should ensure that squashfs-tools and the squashfs module in the image match, roughly
[15:20]  * cody-somerville nods.
[15:20] <ogra> cody-somerville, StevenK has some experience with out of sync karmic/jaunty squashfs
[15:20] <ogra> (its no fun)
[15:23] <cody-somerville> Is it possible to union a remote filesystem in d-i? like, for PXE net based installs
[15:23] <cjwatson> not out of the box
[15:24] <cjwatson> I think you would need to define the problem a bit more; we don't in general need union mounts to do network installations
[15:24]  * cjwatson -> out for a bit
[15:25] <evand> hrm, anyone know if there's a way of detecting what block device is used for the live media in the live CD environment?  I'd rather not just check for a /cdrom mount, as that strikes me as being too broad, but if there's no other way, so be it
[15:25] <cody-somerville> evand, cdrom-detect/cdrom_device
[15:26] <ogra> cody-somerville, you likely could easily create a bastard out of an ltsp client that fires up d-i
[15:26] <ogra> ltsp uses nbd exported squashfs images it mounts in a unionfs on boot
[15:28]  * cody-somerville nods.
[15:32] <evand> cody-somerville: I don't follow.  cdrom-detect/cdrom_device is not set in the booted live environment, nor do I see it referenced anywhere in casper.
[15:33] <cody-somerville> oh, ubiquity.
[15:34] <cody-somerville> I'm not sure about ubiquity bit in d-i cdrom-detect is responsible for finding and mounting the install media. Once it successfully does so, it sets cdrom-detect/cdrom_device and cdrom-detect/cdrom_fs
[15:34] <cody-somerville> *but
[15:34] <evand> indeed
[16:10] <cjwatson> evand: I think the mount point might be exported in casper.conf, so you could then use that
[16:11] <evand> it's not, but perhaps it should be
[16:12] <evand> I asked Scott if he knew of any way to follow the mounts back up to /cdrom, or if aufs provides any useful information here, but it apparently nukes anything remotely useful
[16:13] <evand> err rather, he suggested that we follow the mounts in /proc/self/mountinfo up to /cdrom, but aufs doesn't make that possible here
[16:32]  * cjwatson gets to the point where he can at least start thinking about debugging wubi/grub2/loopback into existence
[16:42] <cjwatson> xivulon: hmm, this is odd, do you have any idea why there are no .mod files in /ubuntu/disks/boot/grub/ ?
[16:48] <cjwatson> xivulon: I'm thoroughly confused why they aren't there, and I can't insmod ext2 without them of course ...
[16:53] <cjwatson> I guess I can just copy them in manually for now but it's a bit worrying
[16:56] <cjwatson> xivulon: ok, at least I can reproduce your problem now
[16:56] <cjwatson> now to see if I can do it with grub-emu or grub-fstest or something ...
[17:02] <cjwatson> ok, good, I can
[17:03] <cjwatson> it doesn't happen when I don't use ntfs+loopback
[17:05] <cjwatson> yay for grub-emu though, this would be totally impossible if I couldn't gdb-attach to the bootloader
[17:11] <cjwatson> xivulon: oh, I see the problem - something is rounding the disk size to 32 bits
[17:12] <cjwatson> or truncating rather
[17:16] <cjwatson> it's a grub ntfs implementation bug
[17:16] <cjwatson> it declares a file's size as grub_uint32_t
[17:29] <cjwatson> xivulon: OK, I think I've fixed it - I'll get the patch upstream
[17:40] <xivulon> xivulon you are a *
[17:40] <xivulon> ^ cjwatson
[17:40] <xivulon> haha
[17:41] <xivulon> cjwatson, re .mod files I think that grub-common is installed but not grub-pc
[17:42] <xivulon> which isn't needed since grub is already in the host partition (deployed by wubi)
[17:42] <xivulon> and the wubi version already has the required mods built-in
[17:43] <xivulon> cjwatson, thank you a bunch
[17:45] <xivulon> ps please have a quick look at the makefile if I need any extra mod
[17:48] <CIA-33> usb-creator: evand * r174 trunk/ (debian/changelog usbcreator/frontends/gtk/frontend.py): Ellipsize long text strings in the GTK+ treeviews (LP: #424883).
[18:25] <cjwatson> xivulon: hmm, you really need to have grub-pc installed I think - the configuration file format is not yet stable and it's important to keep the grub core image up to date
[18:25] <cjwatson> xivulon: is there any reason why installing grub-pc would break?
[18:26] <cjwatson> xivulon: in any case, though, grub-pc *is* installed here
[18:27] <xivulon> iirc (was long time ago') grub-install does not like having to deal with a disk image/loop device
[18:28] <xivulon> then I am wrong
[18:28] <cjwatson> mm. but there's no reason it shouldn't copy the .mod files. This is odd.
[18:28] <cjwatson> I think I might do a fresh install and arrange for grub-install to be set -x, so that I can see what's going on
[18:29] <cjwatson> so, I'm going to include this grub ntfs patch in the next upload. Is there anything else stopping us getting rid of the bind-mounted /boot?
[18:29] <cjwatson> beyond some configuration magic
[18:29] <xivulon> don't think so
[18:29] <cjwatson> rock on
[18:29] <xivulon> :-)
[18:32] <cjwatson> you'll want ext2 (or whatever) in that list in order to use loopback effectively
[18:33] <cjwatson> hmm, I think we need to arrange for that image to be upgraded occasionally, which is going to involve figuring out how to point it at wubildr.cfg on an ongoing basis
[18:33] <cjwatson> it really needs to be upgraded whenever grub-pc is upgraded in the loop image
[18:37] <xivulon> yes, that is a good point. The image will be in the /host root (wubildr), which is rw, so it's a matter of copying the winboot2 makefile into a lupin-support script (or grub-pc post-install)
[18:39] <cjwatson> well, ideally, we'd make grub-install figure it out for itself
[18:40] <xivulon> well I will play with the bindmount killing this w/e and see if I can also handle the wubildr upgrade
[18:41] <cjwatson> it really only needs to notice that / is loop-mounted
[18:48] <xivulon> hmm do we also need to upgrade wubildr.mbr?
[18:49] <xivulon> that requires grubutil/grubinst/grubinst to generate
[18:51] <rgreening> good-day evand
[18:58] <cjwatson> what's in wubildr.mbr?
[18:58] <cjwatson> or rather, what configuration file(s) does it read?
[18:58] <cjwatson> if any
[21:33] <cjwatson> xivulon: I think I have a working implementation of /etc/grub.d/10_lupin now (not tested in its entirety, though), but it relies on an extra grub2 patch that I just sent upstream
[21:33] <cjwatson> xivulon: so I'll keep plugging at it
[21:34] <cjwatson> I *really* want to get rid of that /boot bind-mount if we can
[21:35] <cjwatson> xivulon: http://paste.ubuntu.com/268771/ is the script implementation, and the subject line on grub-devel is "[PATCH] loopback vs. 'set root='"