[12:13] <xivulo1> grub installation and mounting /host r/w where also on my list
[12:13] <cjwatson> mm, right
[12:13] <cjwatson> damnit :)
[12:13] <cjwatson> evand: looks safe enough, anyway
[12:14] <evand> ok
[12:16] <cjwatson> I'm going to look at ntfs-3g 1.826 tomorrow, since Debian has it
[12:19] <cjwatson> how were you planning to cope with the host filesystem potentially being dirty?
[12:19] <cjwatson> I suppose we just hope they shut it down cleanly
[12:20] <cjwatson> but ntfs-3g will refuse to mount "if the volume is scheduled for consistency check or the logfile is unclean" - I'm not comfortable with forcing that
[12:20] <superm1> is there no way to issue a chkdsk upon shutdown in windows?
[12:21] <cjwatson> Windows ought to generally leave the filesystem clean on clean shutdown
[12:21] <cjwatson> it's if you shut down in a more violent way (e.g. power failure, or just too impatient to wait for Windows to shut down all the way) that it's a problem
[12:21] <cjwatson> oh well, bed
[12:24] <evand> can't we pop up a message in the initramfs if we're unable to mount due to a dirty fs?  Something to the effect of, "run chkdsk /f in Windows then reboot"?
[12:24] <xivulo1> cjwatson, there is no chkdsk in ntfs-3g
[12:24] <xivulo1> it will refuse to mount
[12:25] <xivulo1> and I was stopping the init process with some message about ntfs being dirty and requiring chkdsk /r
[12:26] <xivulo1> in fact it will mount r/o (if memory does not fail me)
[12:29] <superm1> can wubi perhaps schedule a chkdsk for the next boot in windows then, and automatically reboot into windows to do that chkdsk /r?
[12:31] <xivulo1> hmm, that would require raising some sort of flag, which is difficult when the filesystem is r/o
[12:32] <superm1> ah right, and you can't even pass it to boot.ini with it being read only
[12:32] <xivulo1> exactly
[12:32] <evand> I believe it's in the registry, is it not?
[12:32] <cjwatson> ok, let's not worry about that
[12:32] <cjwatson> phase two at best
[12:34] <xivulo1> I guess that should be handled properly by ntfs-3g, I am sure that chkdsk is in their agenda anyway
[12:35] <xivulo1> cjwatson, is losetup command available during normal init?
[12:36] <cjwatson> yes
[12:36] <cjwatson> ok, I'm making initramfs-tools mount NTFS host filesystems read/write now
[12:37] <xivulo1> I was writing that myself
[12:38] <cjwatson> too late, sorry ;-)
[12:38] <cjwatson> -       if [ ${readonly} = y ] ; then
[12:38] <cjwatson> +       if [ ${readonly} = y ]  && \
[12:38] <cjwatson> +          ([ -z "$LOOP" ]  || [ "${FSTYPE#ntfs}" = "$FSTYPE" ] ); then
[12:38] <cjwatson>                 roflag=-r
[12:38] <cjwatson> and
[12:38] <xivulo1> how do I paste stuff here?
[12:38] <cjwatson> +               if [ ${readonly} = y ] ; then
[12:38] <cjwatson> +                       roflag=-r
[12:38] <cjwatson> +               else
[12:38] <cjwatson> +                       roflag=-w
[12:38] <cjwatson> +               fi
[12:38] <cjwatson> +
[12:38] <cjwatson> before the loopmount
[12:38] <cjwatson> if it's long, pick a random pastebin
[12:39] <cjwatson> anyway, really off to bed now, I've been up well before 7 for the last two days running, been down to London yesterday, and I'm exhausted ...
[12:40] <xivulo1> I'll send it to you by mail then
[03:32] <xivulon> cjwatson, evand, I was thinking about ways to detect language/keyboard from live cd and run into this http://people.warp.es/~xtor/blog/?p=25
[03:32] <xivulon> thought you might be interested
[03:32] <cjwatson> if it's implementable on the Windows side, I'm all for it
[03:33] <cjwatson> I don't want to mount the Windows filesystem that early from our installer though
[03:33] <xivulon> in the wubi side we already have that, I was thinking about the linux side
[03:33] <cjwatson> yeah, I'd rather not :)
[03:33] <evand> I believe trying to detect the keyboard and language and preseeding them in Ubiquity was something m-a was tasked with a long time ago
[03:34] <cjwatson> it's a neat idea but the ordering is really hard
[03:34] <cjwatson> and I think you have to ask anyway (consider a system you've got from the shop with Windows preinstalled, that you haven't configured at all), so it would only be useful for defalts
[03:34] <cjwatson> defaults
[03:35] <evand> indeed
[03:35] <xivulon> my line of thought is this: you now have cmdline arguments -- locale=XYZ and console-setup/layoutcode=XYZ
[03:36] <xivulon> you can have locale=detect console-setup/layoutcode=detect
[03:36] <xivulon> or detect_locale
[03:37] <xivulon> before the locale/layout are set if you have a detect flag, you run the heuristics.
[03:37] <cjwatson> I really don't think it's workable, sorry
[03:37] <xivulon> n.p. just an idea
[03:37] <cjwatson> I think it's great to make use of on the Windows side, but way too invasive from the Linux side
[03:38] <cjwatson> xivulon: btw I've fixed up the swap file thing, and am trying to work up the courage to tackle grub
[03:38] <xivulon> great, also please include the lupin-gutsy changes I submitted last night
[03:40] <cjwatson> +    disk=$(echo "${disk}"|sed 's/\//\\\//g')
[03:40] <cjwatson> ugh, use a different sed delimiter
[03:40] <xivulon> that's to replace /dev/sda1 with \/dev\/sda1 or sed gets confused
[03:40] <cjwatson> use a different delimiter
[03:41] <xivulon> didn't know you could do that with sed, yeah sounds neater
[03:41] <cjwatson> sed -i 's:LIDISK:$disk:g'
[03:42] <cjwatson> I uncommented modprobe fuse, the fuse module is there
[03:43] <xivulon> good
[03:43] <cjwatson> ${variablename} is only needed when you're following it immediately with something that might be confused for part of the variable name
[03:43] <cjwatson> in "$variablename" it's the quoting that's the important bit, generally
[03:43] <cjwatson> (random shell style point)
[03:45] <cjwatson> I'll fix the sed thing now
[03:45] <xivulon> lupin was my first "big" sh project, I normally try to "imitate" the style from other files I hack, but that is a less than perfect approach
[03:45] <cjwatson> most people do pick up sh organically
[03:47] <cjwatson> I've added a debian/changelog entry for your changes
[03:47] <xivulon> yesterday I also sent you some code re mounting /host rw
[03:48] <cjwatson> yes, I saw, I already did something else in initramfs-tools so it isn't urgent
[03:48] <cjwatson> your code prints a better message if it fails so I'll look at it at some point, but grub comes first :)
[03:49] <xivulon> there is also losetup use, as opposed to use mount -r -o loop, to avoid r/o loop device
[03:49] <cjwatson> is that important?
[03:50] <xivulon> I might be wrong, but if you create a loop device via the mount command with r/o option, the loop device is also locked ro and when you remount rw, the loop device does not follow
[03:51] <xivulon> Creating the loop device separately should address this (and it does not make any harm anyway)
[03:51] <xivulon> There was also the other command you mentioned last time, but I forgot which one it was
[03:51] <evand> blockdev?
[03:53] <xivulon> no it was something to set loop device settings
[03:54] <cjwatson> ok, I very much want hacks outside lupin to be minimal, so losetup in lupin is probably better
[03:54] <cjwatson> er, in initramfs-tools/scripts/local I mean
[03:54] <cjwatson> I've uploaded lupin 0.3 with your changes
[03:54] <xivulon> ahh
[03:55] <xivulon> thanks
[03:55] <cjwatson> I have such a headache :-/
[03:55] <cjwatson> how are we going to get grub4dos installed? I assume it isn't in a .deb
[03:57] <xivulon> It's installed by the windows frontend
[03:57] <xivulon> https://code.launchpad.net/~ubuntu-installer/wubi/gutsy
[03:57] <cjwatson> ah, so we just need an appropriate menu.lst and to suppress installing grub in the mbr
[03:57] <cjwatson> I have mail from you about the former, so that's fine
[03:57] <xivulon> yeah
[03:58] <cjwatson> might have to add an extra preseed parameter to grub-installer
[03:58] <xivulon> you can install grub in the loop file if you want
[03:59] <cjwatson> yeah, it's just pointless, right? :)
[03:59] <xivulon> yeah
[04:01] <xivulon> But menu.lst has to be generated
[04:01] <cjwatson> you know, groot_prefix is sort of an inappropriate name ...
[04:01] <cjwatson> groot is just for use by update-grub
[04:02] <cjwatson> let me have a look at exactly what they did in grub4dos
[04:02] <xivulon> I did not try I think they use groot
[04:02] <xivulon> It should be groot(hdX,Y)/path/to/windows/boot/folder
[04:02] <cjwatson> I hope not
[04:03] <cjwatson> well, let me check
[04:04] <xivulon> My understanding is that whatever is in groot is copied over as "root" when generating the menu sections
[04:05] <cjwatson> right, but my point is that's done by update-grub not by anything grub4dos should be doing
[04:05] <cjwatson> so hopefully they really support root_prefix
[04:05] <xivulon> In this case setting groot(hd3,2)/ubuntu/disks and running update-grub should set all menu entries to root (hd3,2)/ubuntu/disks
[04:05] <cjwatson> ah, what they did was allow 'root (hd0,0)/ubuntu/disks'
[04:06] <cjwatson> that's better
[04:06] <xivulon> Yeah that's what I meant
[04:06] <cjwatson> it was the groot bit that was confusing me
[04:07] <xivulon> Or at least that's what I asked for, not sure if they got it wrong, I did not try the new grub4dos yet (since there was no fuse).
[04:08] <cjwatson> where do you want the /boot directory to go?
[04:08] <cjwatson> /ubuntu/disks/boot/ ?
[04:08] <cjwatson> that would be convenient, I think
[04:08] <xivulon> That would be my guess
[04:10] <cjwatson> I do have to disable actually installing grub, because ordinary grub won't understand this root syntax and will fail
[04:11] <xivulon> The important thing is that update-grub works
[04:12] <xivulon> We might revert to use #root_prefix maybe?
[04:13] <cjwatson> no
[04:13] <cjwatson> not necessary
[04:13] <cjwatson> it's actually easier this way because no changes to update-grub are needed
[04:19] <cjwatson> in fact it's like 15 lines added to grub-installer, if this is right :)
[04:19] <cjwatson> still need to prod partman to bind-mount /boot, but I know where that wants to go
[04:30] <xivulon> it should also be in fstab
[04:31] <cjwatson> indeed so
[04:31] <cjwatson> hence, partman :)
[04:31] <xivulon> partman twice then
[04:33] <cjwatson> just the once
[04:33] <cjwatson> adding the appropriate thing to fstab also causes partman to mount it
[04:39] <xivulon> you're too smart for me :P
[04:42] <xivulon> By the way wubi can now be used to boot from live cd also when the bios boot order is adverse, and for people with no CD (hd-media style).
[04:43] <cjwatson> nice
[04:44] <xivulon> I plan to add an item in the size selector so that you have something like "No installation, just classic LiveCD, (read-only)"
[04:44] <xivulon> have to figure a concise msg to convey that
[04:50] <xivulon> I was thinking to in fact have 2 items: "Dedicated partition" will start boot the livecd in automatic mode but without preseeding the partitioning info
[04:51] <xivulon> and "Read only" which will simply boot the LiveCD for people with bios issues / no CD.
[04:56] <cjwatson> ok, grub-installer seems to be working now with the exception that entries for other operating systems are broken
[04:56] <cjwatson> not going to worry about that right now (those entries should just go away I think)
[04:57] <xivulon> it's not a big issue , since people will boot first from ntldr, and the grub menu will mostly be hidden (to avoid having 2 boot menus)
[04:57] <cjwatson> yeah
[04:58] <xivulon> I am really looking forward to next build
[04:59] <cjwatson> is there a wubi binary yet that I can slap onto the CD?
[05:00] <xivulon> The current nsis code "compiles" and produces a binary, but I haven't tested it at all, which means that it is probably broken
[05:01] <xivulon> If you have an ISO with the other changes in, I can do some tests and will create a working binary this w/e, so it can incorporated sunday/monday.
[05:02] <cjwatson> I'm not certain I can get lupin in by the end of today
[05:04] <xivulon> I'll give you a frontend anyway by sunday, that should at least set up grub4dos and the preseed file.
[05:04] <cjwatson> I tweaked the preseed file in that wubi branch a bit
[05:05] <xivulon> no problem
[05:06] <xivulon> In order to work with an ISO (as opposed to a CD), I need to be able to extract kernel/initrd from within the ISO (with a CD, they are copied over). That can be done with 7-zip, but I am not sure if I will have time to implement that.
[05:23] <cjwatson> ok, I've written a main inclusion report for lupin and (varying slightly from standard procedure) fast-tracked it into main
[05:26] <xivulon> * ago likes preferential treatment 8) *
[05:26] <evand> conflict of interest
[05:26] <evand> :)
[05:27] <cjwatson> a touch
[05:28] <cjwatson> it's what tight scheduling does to you :-/
[05:29] <xivulon> ps do you know +/- when installer matters are gonna be discusses in boston? I'd love to join but cannot take all week off
[05:29] <evand> speaking of tight schedule, can you NEW (if that's the correct terminology) gobuntu-artwork-usplash?  I made seb's requested changes and I think the licensing is correct.
[05:29] <evand> if you have time, of course
[05:31] <cjwatson> I have to take my family to the shops right now but will reply when I get back
[06:14] <cjwatson> xivulon: throughout, I should imagine. We don't tend to partition issues up very strictly by day
[06:15] <cjwatson> xivulon: the times when people are available are taken into consideration where possible
[06:15] <cjwatson> evand: I assume the older one should be rejected?
[06:15] <evand> I thought it already was yesterday?
[06:15] <cjwatson>   269061 | S- | gobuntu-artwork-uspl | 0.1                  | 20 hours
[06:15] <cjwatson>          | * gobuntu-artwork-usplash/0.1 Component: main Section: misc
[06:15] <cjwatson>   269453 | S- | gobuntu-artwork-uspl | 0.1                  | 45 minutes
[06:15] <cjwatson>          | * gobuntu-artwork-usplash/0.1 Component: main Section: misc
[06:15] <evand> ah yes
[06:16] <evand> reject the older one
[06:16] <cjwatson> right
[06:18] <cjwatson> evand: hmm, did seb ask you to split out the CC-BY-SA licence into COPYING-ART?
[06:19] <evand> cjwatson: he asked me to include the entire GPL, so I had to create COPYING.  To stay consistent I dumped the CC-by-SA out of copyright and into COPYING-ART.  Should I avoid doing that?
[06:19] <evand> so no, he did not ask me to do that
[06:24] <cjwatson> ok, this is sort of a weirdness in policy
[06:25] <cjwatson> the rule is that all licence text must be present in the copyright file (because there are tools to read it), but there's a special exception for anything that's in /usr/share/common-licenses/
[06:25] <evand> ohhh
[06:26] <cjwatson> I've accepted it, but could you revert the COPYING-ART bit for the next upload?
[06:26] <evand> will do
[06:26] <evand> I'm assuming I should do that before filing a MIR, correct?
[06:27] <cjwatson> nah
[06:27] <cjwatson> I punted it straight into main under the "obvious" category
[06:28] <evand> heh
[06:30] <evand> I don't suppose I could trouble you to update the desktop seed as well?
[06:34] <cjwatson> evand: done. binaries still need to be accepted mind you
[06:35] <evand> ah, indeed
[06:35] <cjwatson> ~1hr
[06:35] <evand> thanks!
[07:09] <CIA-18> ubiquity: cjwatson * r2213 ubiquity/ (d-i/manifest debian/changelog):
[07:09] <CIA-18> ubiquity: * Automatic update of included source packages: apt-setup 1:0.21ubuntu3,
[07:09] <CIA-18> ubiquity:  base-installer 1.81ubuntu2, grub-installer 1.24ubuntu3,
[07:09] <CIA-18> ubiquity:  partman-auto-loop 0ubuntu5, partman-efi 13ubuntu2, partman-target
[07:09] <CIA-18> ubiquity:  50ubuntu4.
[07:11] <evand> cjwatson: thoughts on having this instlux modification (see ubuntu-devel-discuss) on the CD for Hardy?  I'm leaning towards no as I think we have enough install options as it is, but I'm not strongly tied to that.
[07:12] <evand> I imagine they just want a button that says "reboot and install Ubuntu" on the open cd, or something similar
[07:13] <evand> I seem to recall discussing that option in addition to wubi, but it may be a false memory
[07:17] <cjwatson> didn't ago say earlier that he'd just added that option to wubi?
[07:18] <cjwatson> but yeah, I don't think we should have both this and wubi as distinct items, they need to be merged really
[07:19] <evand> oh, indeed he did
[07:19] <evand> sorry about that
[07:19] <evand> so we already have a solution
[07:19] <evand> great
[07:23] <CIA-18> ubiquity: cjwatson * r2214 ubiquity/debian/po/ (79 files): debconf-updatepo
[07:24] <evand> whoops, sorry about that
[07:30] <cjwatson> no worries, it's a task often best done just before release
[07:32] <evand> I'll add it to my notes on releasing a new version then
[07:37] <cjwatson> I only bother when lintian tells me newer-debconf-templates
[07:37] <CIA-18> ubiquity: cjwatson * r2215 ubiquity/debian/changelog: releasing version 1.5.13
[07:38] <cjwatson> otherwise I think it can end up bumping the date for the sake of it, though I'm not certain if that's checked
[07:39] <evand> ah, I'll keep an eye out for that
[07:48] <cjwatson> evand: gobuntu-artwork-usplash binaries accepted; will appear in the archive after the next publisher run
[07:48] <evand> nice, thanks
[08:14] <cjwatson> right, phew, I think that's everything on my installer-for-windows list for today done
[08:14] <evand> heh, enjoy your weekend
[08:15] <cjwatson> thanks, you too