[01:25] <xivulon> cjwatson partman autoloop fails to mount /host when an iso is used
[10:03] <cjwatson> sigh, xivulon, that was what I already told you
[01:09] <tepsipakki> does the "pending Cell support" allow native installation on PS3?
[01:15] <cjwatson> we had that in feisty ...
[01:15] <cjwatson> it expands it to some more machines
[01:17] <cjwatson> well, when I say feisty I mean with a few extra bug-fixes; see http://cdimage.ubuntu.com/custom/20070608-feisty-ps3/ and http://cdimage.ubuntu.com/custom/20070614-xubuntu-feisty-ps3/
[01:17] <tepsipakki> oh, ok.. I don't own one yet, but every installation doc I've encountered mentions that fedora needs to be installed first
[01:17] <cjwatson> it doesn't
[01:17] <tepsipakki> but maybe they were outdated
[01:17] <tepsipakki> right
[01:17] <cjwatson> my PS3 has never had Fedora on it
[01:18] <tepsipakki> heh, I take your word for it :)
[01:19] <xivulon> cjwatson could you pls have a quick look at the patches I sent this morning and in case merge them?
[01:19] <cjwatson> xivulon: I will do, just doing other things first
[01:19] <xivulon> sure
[02:35] <cjwatson> xivulon: if the find_iso mount needs to be visible to partman-auto-loop, I'm leaning towards reinstating the change to just call it /host
[02:35] <cjwatson> I like keeping the number of top-level names as small as I can
[02:42] <xivulon> cjwatson the issue is that if you have the ISO on /host and target the same device (which is by far the most common case), that is fine
[02:44] <xivulon> but... if you have the ISO on /host and target another device, autopartition loop will try to mount that other device as /host (if memory does not fail me)
[02:46] <xivulon> I would suggest to rename /isodevice -> host and IIF autopartion loop likes to use /host change that instead (since that top level folder is only used during installation)
[02:47] <xivulon> yep
[02:47] <cjwatson> oh, hmm, I see
[02:48] <xivulon> #189 mount -o move /target /host
[02:48] <cjwatson> no in that case I think I prefer /iso or /isodevice for the ISO
[02:48] <xivulon> ok
[02:48] <cjwatson> because /host is what we end up using on the installed system for the same purpose as /host during installation
[02:48] <cjwatson> so it should have the same name
[02:49] <xivulon> makes sense
[02:49] <cjwatson> I'm not sure about pushing this in for beta mind you
[02:50] <xivulon> I was thinking the same thing
[02:50] <cjwatson> don't get me wrong, it seems correct, but I'm trying to use my release team points sparingly for things that are critical :)
[02:50] <xivulon> But on the other side we had very little testing on wubi/lupin...
[02:51] <cjwatson> I know, and I'm desperate to get it tested, but at this point I think we only have time to test the CD case
[02:52] <xivulon> I'd still put the partman auto loop stuff in, since it will have no effect on a normal installation
[02:53] <xivulon> And can have people test using the ISO + the downloaded wubi.exe
[02:54] <xivulon> Same goes for lupin of course
[02:55] <xivulon> Lupin patches should be included, since they should have no impact on a normal installation (and if they do, you do want to know now)
[02:55] <cjwatson> "should have no impact" is the sort of phrase that automatically strikes terror into release managers
[02:56] <xivulon> What I mean is that if they do have an impact, that is a bug, and it should be discovered now
[02:57] <cjwatson> erm
[02:57] <cjwatson> I see where you're coming from, but wubi needs to work in the default case by beta; that is the highest priority
[02:57] <cjwatson> the possibility of introducing bugs that might need us to respin is very very bad
[02:58] <cjwatson> because it either means delaying the Ubuntu beta or removing wubi from it
[02:58] <cjwatson> we will have time after beta to fix this sort of thing
[02:58] <cjwatson> but if wubi isn't in place and working by beta, it isn't going to be in the final release
[02:58] <xivulon> Let me rephrase, partman auto-loop and lupin do work as far as I can tell in the default case (when there is no wubi.exe)
[02:59] <cjwatson> right, that's great and that's enough for beta
[02:59] <cjwatson> we can land these fixes immediately afterwards
[03:01] <cjwatson> I'm already getting away with a lot by getting it into place this late, and I can't stretch much further
[03:01] <cjwatson> (and I realise that was my fault at least as much as anyone else's)
[03:01] <xivulon> cjwatson, I understand, don't worry
[03:02] <cjwatson> in particular I'm not confident of the partman-auto-loop changes
[03:02] <cjwatson> what you sent is definitely wrong because it will break any preseeding that involves more than one partition
[03:02] <xivulon> They have not been tested to date
[03:02] <cjwatson> so I think it deserves some care
[03:04] <cjwatson> plus, it's not really the same device
[03:04] <cjwatson> actually, no, ignore that bit
[03:04] <xivulon> I do not think that multiple partitions with loopinstallations is much of an issue
[03:04] <cjwatson> sorry, I'm confused, your change should be fine
[03:05] <cjwatson> I meant like multiple .disk files
[03:05] <cjwatson> but it's not an issue here
[03:05] <cjwatson> still, anything that makes my head hurt this much I don't want to ram through for beta :)
[03:06] <xivulon> what is a realistic plan?
[03:07] <cjwatson> xivulon: we go with CD-only for beta and land the changes immediately after beta
[03:07] <cjwatson> there is a period between beta and release when we go back to "merely" feature freeze
[03:07] <cjwatson> and that's a reasonable time to land this sort of thing
[03:07] <xivulon> doesn't that mean that wubi should not be in the final?
[03:08] <cjwatson> but right now, we're in beta freeze, and the only fixes we can make are those which are critical to the beta
[03:08] <cjwatson> xivulon: I don't see why, if it works in the must-have-a-CD-in-the-drive mode for beta
[03:09] <xivulon> You mean we only enable the CD-only mode
[03:09] <cjwatson> right
[03:09] <xivulon> So we only have wubi
[03:09] <cjwatson> blink?
[03:09] <cjwatson> oh, I'm not religious about ripping it out, we can release-note that something doesn't work
[03:09] <cjwatson> that's ok
[03:10] <cjwatson> let me reiterate
[03:10] <cjwatson> we have six days
[03:10] <cjwatson> there will be bugs left at the end
[03:10] <xivulon> cjwatson I do not mind whatever you decide.
[03:10] <cjwatson> the key is to get the most important paths tested and working
[03:11] <xivulon> I am only need a bit of planning, for instance if there is no loopinstallation support, I'll have to resume the dynamic patching mechanism in the standalone version
[03:11] <cjwatson> so with my archive and release hat on I'm working to get the unionfs fixes through
[03:11] <xivulon> which means I have to ship my own initrd and so on...
[03:11] <cjwatson> with my cdimage hat on I'll be running round rebuilding images this afternoon
[03:11] <cjwatson> can you not live with needing a daily build slightly after beta?
[03:12] <cjwatson> I mean it's not like the images have worked for a while ...
[03:13] <xivulon> I think more in terms of the final, beta or not is not a big issue at all.
[03:14] <cjwatson> or are you going to have users beating down your door asking why wubi-installer.org doesn't support 7.10 beta?
[03:14] <xivulon> No that's not an issue
[03:14] <xivulon> I can have the wubi.exe point at a daily build instead of the beta
[03:14] <xivulon> That is what it does now
[03:15] <xivulon> The issue is that if there is no loppinstallation support in the final ISO, I will not be able to point them anywhere else
[03:16] <xivulon> And I will have to maintain my own initrd again wich patches the ISO on the fly.
[03:18] <cjwatson> ok, I'm keen to avoid you doing that
[03:19] <cjwatson> and I don't think it will be necessary
[03:19] <cjwatson> the lupin patch is in bzr, and the partman-auto-loop patch is in place in my local tree now pending me thinking about it a bit more
[03:19] <cjwatson> it should all be ready for right after beta
[03:20] <xivulon> another minor issue was that ubiquity seemed to ignore atomatic-ubiquity flag and I just landed on the desktop (did not check whether xinit-ubiquity was in place)
[03:22] <xivulon> cjwatson, it would be good if the initrd could support some external hooks "pre_mount=/path/to/premount.sh" and "post_mount=/path/to/postmount.sh"
[03:24] <xivulon> But I guess it's late for that too
[03:41] <cjwatson> automatic-ubiquity should be fixed finally in 1.5.16
[03:41] <cjwatson> taken a few goes, that
[03:41] <cjwatson> talk to me about hooks after beta :)
[03:42] <xivulon> thanks
[03:43] <xivulon> the hooks are an emergency thing so that if I can override a few initrd/iso files without having to ship a whoule new thing
[03:43] <cjwatson> nod
[03:43] <cjwatson> somebody else asked for that too I think - will try to come up with something generic
[03:45] <xivulon> cjwatson, do you think you can push out an ISO build with all the patches?
[03:45] <xivulon> I will be travelling this w/e and will not have much time (and probably zero connectivity), having an ISO to play with might help
[03:47] <xivulon> By the way, lupin-casper already provides external hooks... ;P
[03:52] <cjwatson> you mean all the other patches up to now, or the post-beta ones?
[03:53] <cjwatson> I can't do post-beta stuff yet, but we'll be building new CDs today as soon as all the necessary stuff is in the archive
[03:53] <xivulon> cjwatson, I mean that current lupin version supports importing early_command.sh and late_command.sh
[03:53] <cjwatson> 14:45 <xivulon> cjwatson, do you think you can push out an ISO build with all the patches?
[03:53] <cjwatson> I was replying to that
[03:54] <xivulon> re ISO build, yes I mean post beta stuff, with any patch that might be relevant to lupin/wubi (including automatic-ubiquity)
[03:55] <xivulon> re Lupin, now that I think of it, Lupin can support early_command.sh. That is enough to override the ISO.
[03:55] <xivulon> So even if you decide not to push in partman-autoloop, I can add it later on, provided lupin is available in the ISO
[03:56] <cjwatson> lupin will be there
[03:57] <cjwatson> automatic-ubiquity should work, but still needs testing
[03:57] <xivulon> Even in the beta?
[03:57] <cjwatson> same way it is in current daily builds ...
[03:57] <cjwatson> the lupin patches we've been talking about today, unlikely
[03:57] <cjwatson> but lupin in general, sure
[04:24] <xivulon> cjwatson, to simplify testing in the coming days, I'd like you to reconsider having the latest lupin patches (with that I can patch partman on the fly, but changing the initrd is not as convenient), also it would be good to have all other "post-beta" patches in one place
[04:26] <xivulon> Ideally the patches should be as an override folder that can simply be extracted on top of the mounted ISO by early_command
[04:26] <xivulon> Would something like that be feasible?
[04:28] <cjwatson> xivulon: you already have early_command so there is nothing to do there, as far as I know
[04:29] <cjwatson> ok, *sigh*, I'll see if I can push through the /isodevice move-mounting stuff
[04:29] <xivulon> Thanks, without that, I'd need to push out a separate initrd
[04:32] <xivulon> As for the post-beta folder, that has not to be anything formal, just a folder where you dump files and packages that did not make the ISO, which probably means partman-auto-loop and automatic-ubiquity.
[04:34] <xivulon> That ultimately might reduce your workload since you'd have to worry about a single ISO, and it would speed and simplify lupin/wubi testing, since we could easily apply changes on the fly and do not have to wait for special ISOs
[04:36] <cjwatson> no, I don't want to do that sorry
[04:36] <cjwatson> we'll just do updated builds
[04:36] <cjwatson> early_command hooks can of course create whatever they want themselves
[04:37] <xivulon> I think we are saying the same thing.
[04:38] <xivulon> You just do official builds, simply let me know what relevant lupin packages you left out
[04:38] <cjwatson> I can certainly mention fixes
[04:38] <xivulon> I'll take care of the override folder
[04:43] <cjwatson> lupin uploading with the /isodevice changes
[05:16] <xivulon> nice
[05:28] <xivulon> When will the ISO with that be ready?
[05:29] <cjwatson> publisher + publisher + CD build + random slack for shit that goes wrong
[05:29] <cjwatson> oh and include live filesystem build
[05:30] <cjwatson> + time for Soyuz to wake up and schedule builds + binary build time
[05:30] <cjwatson> no less than 2.5 hours, probably a bit more
[05:30] <xivulon> That's more than good enough, I am leaving tomorrow morning so I can download tonight
[05:32] <xivulon> Can you also put somewhere your local patched version of partman autoloop and auto ubiquity?
[05:32] <cjwatson> like I said, automatic-ubiquity should work in the next build, as far as I know
[05:33] <xivulon> missed that bit, great -1
[05:33] <cjwatson> I'll give you the codebrowse link for partman-auto-loop once I've committed it and you can use that
[05:34] <xivulon> Anything else I'd need you can think of? lupin-support is on the ISO?
[05:34] <cjwatson> yeah
[05:35] <cjwatson> I'm off for a while, since the publisher is churning on manual and CDs will build after that; evand has my phone number if the universe is collapsing
[05:35] <xivulon> Thanks a lot for everything, and sorry for the trouble
[05:35] <cjwatson> it's ok
[05:35] <cjwatson> I hope this build will work :)
[05:35] <xivulon> me too
[05:43] <xivulon> By the way, in anyone wants to try wubi/lupin, to go around current issues, burn a CD, start wubi normally and select 4GB installation, before rebooting edit /ubuntu/install/grub/menu.lst and remove find_iso=XYZ
[05:45] <xivulon> Do not boot with the CD inserted, pop it in when requested
[05:46] <xivulon> This will force use a physical CD instead of an ISO on HD which at the moments creates problems. Haven't tried that personally but in theory you should be able to go through the loopinstallation.
[08:52] <evand> bug #1: no mouse when starting straight into ubiquity
[09:17] <xivulon> cjwatson, do you have updated partman-auto-loop sources?
[09:20] <cjwatson> no
[09:20] <cjwatson> evand: blink
[09:21] <xivulon> you mentioned you patched your local branch could you pls email me autopartition-loop?
[09:22] <evand> cjwatson: yeah, not sure what's causing it yet.  Trying to get it running in automatic mode first.
[09:22] <cjwatson> xivulon: please, when I am confident in it I'll check it in and e-mail it to you
[09:23] <cjwatson> I don't want to mail about half-broken copies and waste your time
[09:23] <xivulon> cjwatson, it's only because I'll be away this w/e and I am not sure about network connectivity, so I could at least work on something
[09:23] <xivulon> does not matter if it's half done, better that today that something on monday
[09:24] <cjwatson> http://people.ubuntu.com/~cjwatson/tmp/autopartition-loop but if it's broken I warned you :P
[09:24] <cjwatson> er, in a moment, scp hasn't finished
[09:24] <xivulon> great
[09:25] <cjwatson> it's there now
[09:25] <evand> speaking of broken autopartition-loop ;)
[09:26] <cjwatson> evand: what broke?
[09:26] <evand> it's sitting there.  It doesn't seem to like me stracing it either.
[09:26] <evand> but that's probably an error on my part
[09:26] <xivulon> are you testing with loopinstallation?
[09:27] <xivulon> check if /host is mounted
[09:27] <evand> xivulon: it is mounted
[09:27] <evand>  /host/ubuntu/disks/root.disk
[09:27] <xivulon> what I mean, is /host on a device or is sitting inside of the ramdisk?
[09:28] <xivulon> do you see it in /proc/mounts
[09:28] <evand> ah, it's on /
[09:28] <xivulon> then it's blocking because you are out of ram probably
[09:28] <cjwatson> right, question is why it didn't get mounted
[09:28] <evand> you're good :)
[09:28] <cjwatson> did wubi sit for ages downloading an iso?
[09:28] <xivulon> That is the patch I submitted
[09:29] <xivulon> it's not mounted because the target partition is the one hosting the ISO
[09:29] <xivulon> which means that the target partition is already mounted
[09:29] <cjwatson> xivulon: not so much, it's a bug if it gets that far
[09:29] <xivulon> which means that mount.d/70basic fails
[09:29] <evand> yeah, I'm using the old wubi.exe.  I'll update.
[09:29] <cjwatson> it shouldn't be using an ISO, it should be using the CD
[09:29] <evand> sorry for the false alarm
[09:29] <cjwatson> ah, that would be it
[09:30] <cjwatson> xivulon: no, mount.d/70basic failed for an entirely different reason which I fixed
[09:30] <cjwatson> nothing to do with the ISO being on the target partition
[09:30] <evand> though, I did find one more bug
[09:30] <evand> we need to set UBIQUITY_AUTOMATIC in environ
[09:30] <cjwatson> I thought ubiquity did that itself
[09:30] <cjwatson> if you pass --automatic
[09:31] <evand> it does
[09:31] <evand> but we're not passing --automatic in ubiquity-dm, are we?
[09:31] <cjwatson> oh, but we don't call it with --automatic
[09:31] <cjwatson> I wonder why not
[09:31] <cjwatson> I think that's a bug
[09:32] <xivulon> evand, I did not understand if you are using an ISO or the CD
[09:32] <cjwatson> evand: what does it break?
[09:32] <cjwatson> because I saw the "checking installation" progress bar when I tried it
[09:32] <cjwatson> which made me assume that it was in automatic mode
[09:32] <evand> oh, hrm
[09:32] <evand> that does mean it's in automatic mode
[09:32] <cjwatson> oh, except maybe I was running ubiquity by hand
[09:32] <cjwatson> you might want to ignore me
[09:33] <evand> xivulon: I was using an iso, I'm going to grab the latest wubi now to fix that.
[09:33] <cjwatson> I'll upload a fix tonight
[09:33] <CIA-18> ubiquity: cjwatson * r2250 ubiquity/debian/ (changelog init):
[09:33] <CIA-18> ubiquity: * Use the --automatic option when running ubiquity from the init script
[09:33] <CIA-18> ubiquity:  (thanks, Evan Dandrea).
[09:33] <xivulon> cjwatson, is it now possible to target the partition hosting the ISOs now?
[09:34] <cjwatson> xivulon: if booting from the CD
[09:34] <xivulon> which means no
[09:34] <cjwatson> correct
[09:34] <evand> thanks cjwatson
[09:34] <xivulon> So evand you can run Wubi and pick 4GB
[09:34] <xivulon> Then before rebooting edit install/boot/grub/menu.lst
[09:34] <xivulon> and remove find_iso=*
[09:35] <xivulon> That will be identical to a real wubi loopinstallation except that the CD will be used
[09:35] <xivulon> Reboot without the CD and only insert that when requested to do so
[09:36] <evand> xivulon: ok, thanks
[09:36] <xivulon> cjwatson, out of curiosity what was the mount.d bug you referred to?
[09:36] <cjwatson> xivulon: partman-auto-loop wasn't reading all the output from fstab.d scripts
[09:37] <cjwatson> so it started a second one before the first one had finished, and they fought
[09:37] <xivulon> ah
[09:37] <cjwatson> oh, hmm, sorry, I was confusing fstab.d and mount.d; you were right about mount.d/70basic
[09:37] <cjwatson> sorry, I have a lot of things going on at once and this is only one of them :-/
[09:38] <xivulon> I am always right 8)
[09:38] <xivulon> brb
[09:38] <evand> cjwatson: not to add something else to the pile, but did you notice that jdong is still experiencing unionfs issues?
[09:38] <CIA-18> ubiquity: cjwatson * r2251 ubiquity/ (configure configure.ac): bump to 1.5.17
[09:38] <evand> he's updating the original bug report with his logs, so I imagine pkl will see that.
[09:40] <cjwatson> evand: yeah, I didn't notice them in my own tests so it isn't universal, but do explicitly mention it to pkl in case he can make something of it
[09:40] <evand> will do
[09:40] <evand> I didn't notice it either, fwiw
[09:41] <cjwatson> don't rely on bug mail, he's on the kernel team so probably gets stacks of it
[09:41] <evand> well, his name was also mentioned with it a few times in #ubuntu-devel, but I'll msg him directly
[09:43] <cjwatson> I'd mention it in the private #kernel
[09:43] <cjwatson> less noise
[09:43] <evand> ...I swear, windows updates occur more often than gutsy updates
[09:45] <xivulon> back
[09:48] <xivulon> what are the default boot parameters of "safe graphic mode"?
[09:54] <cjwatson> xivulon: this thing is still trying to download an ISO
[09:54] <cjwatson> today's build
[09:55] <evand> odd, not here unless it did it quite quickly
[09:55] <cjwatson> I can't afford to download the ISO every time, the last test took hours
[09:56] <cjwatson> it says it's rev 291
[09:56] <cjwatson> xivulon: safe graphics mode adds the 'xforcevesa' parameter
[09:57] <cjwatson> xivulon: is there anything I can do to help you debug why this can't seem to detect the CD?
[09:57] <evand> cjwatson: yeah, same here.  Odd.
[09:57] <cjwatson> I was booting without network connectivity (more or less by chance, I'd carried the laptop downstairs so I could make coffee and watch it) and of course it got stuck
[09:58] <evand> hrm, I'll give it another try with networking disabled as soon as I see if it gets any further on the ubuntu end this time.
[09:58] <xivulon> I am having a look
[09:59] <xivulon> if you want to debug launch wubi --debug
[09:59] <cjwatson> yep, if it's connected to a network it starts downloading
[09:59] <cjwatson> ok
[09:59] <xivulon> you should see lots of message bogs
[09:59] <xivulon> it will look for the CD twice
[09:59] <xivulon> at the beginning
[09:59] <xivulon> and after you click install
[09:59] <xivulon> if it fails to detect it it will give you an explanation
[10:01] <cjwatson> says 'isvalid Ubuntu 7.10 "Gutsy Gibbon" - Alpha i386 (20070921.1)' 'cdinfo=Ubuntu 7.10 "Gutsy Gibbon" - Alpha i386 (20070921.1) cdversion=7.10 cddistro=Ubuntu cdarch=i386 cdcodename=Gutsy Gibbon cdsubversion=Alpha'
[10:01] <cjwatson> (continuing)
[10:01] <cjwatson> 'success Ubuntu 7.10 <blah>'
[10:01] <cjwatson> 'cddrive=E:\'
[10:01] <cjwatson> and then pops up setup
[10:02] <cjwatson> wish I could c&p this
[10:02] <cjwatson> ah, makeimg fails
[10:02] <cjwatson> Using device \\.\E:
[10:02] <cjwatson> Error: The process cannot access the file because it is being used by another process.
[10:03] <cjwatson> the thing it's trying to run is ...\makeimg.exe -o C:\ubuntu\install\gutsy-desktop-i386.iso -i E:
[10:03] <xivulon> is the sleep 3000 in there?
[10:03] <cjwatson> didn't see any sleep 3000
[10:03] <cjwatson> but I didn't look at the code
[10:03] <cjwatson> it's rev 291
[10:03] <xivulon> checking
[10:03] <cjwatson> why does it need to copy the .iso to the hard disk?
[10:04] <xivulon> If you use read-only mode or loopinstallation, so that the CD can be ejected and we can avoid booting from CD
[10:04] <cjwatson> I wasn't though, I'd selected 7 GB
[10:04] <cjwatson> (the default here)
[10:05] <xivulon> yeah that is loopinstallation
[10:05] <cjwatson> oh
[10:05] <xivulon> the size is calculated based on free disk space
[10:05] <cjwatson> perhaps we could just skip that for now and document the awkwardness, then?
[10:05] <cjwatson> unless you have a good idea on how to fix this
[10:06] <xivulon> sleep 3000 is there it was sufficient in my case, but I was using CD emulator
[10:06] <cjwatson> what's the other process using the CD, do you think?
[10:07] <xivulon> The only time we use the CD is to read .disk/info
[10:07] <cjwatson> I have Explorer open ...
[10:07] <cjwatson> could it be that?
[10:07] <xivulon> possibly
[10:08] <evand> http://www.microsoft.com/technet/sysinternals/FileAndDisk/Filemon.mspx ?
[10:08] <cjwatson> nope, closing Explorer didn't help
[10:09] <xivulon> Let me try to do a buidl with sleep 10000
[10:09] <xivulon> make it 30000
[10:09] <cjwatson> well, what is it that's accessing the disk?
[10:09] <cjwatson> if it's wubi itself, then I don't see how sleep would help
[10:09] <cjwatson> that should all be synchronous, surely
[10:09] <xivulon> hmm it did help in my case
[10:10] <cjwatson> are you certain that you're properly closing the CD device?
[10:10] <xivulon> I am only using the default readfile command
[10:10] <xivulon> will check if there is a second command to use close cleanly
[10:10] <evand> mouse works again, false alarm there
[10:11] <xivulon> cjwatson, instead of me increasing the sleep time, simply use a --debug, and leave the last message open for 30 secs
[10:11] <xivulon> alternative workaround is to download the ISO and place it in the same folder where wubi.exe is
[10:12] <xivulon> The ISO has to be called gutsy-desktop-i386.iso
[10:12] <cjwatson> which last message should I leave open?
[10:12] <cjwatson> I'm running out of time, my wife wants me downstairs for DVD watching
[10:14] <xivulon> The one before "makeimg" do not remember which one it is
[10:14] <xivulon> Hmm I do call FileClose (wubi/inspector/detect_iso.nsh #113) and I cannot think of any other time when I use the CD
[10:15] <cjwatson> wubi itself is running from the CD
[10:15] <cjwatson> wouldn't that be enough to lock it?
[10:15] <xivulon> ah that's more likely
[10:16] <xivulon> can you try running off hd?
[10:16] <cjwatson> yeah
[10:17] <cjwatson> it's sitting at "Retrieving installation files" ...
[10:18] <xivulon> it takes a few secs
[10:18] <xivulon> it has to copy the full CD
[10:18] <cjwatson> oh, of course
[10:18] <xivulon> we will have a proper progress bar
[10:18] <cjwatson> ok
[10:18] <cjwatson> this doesn't sound workable if wubi can't copy the CD because it's locked though
[10:19] <cjwatson> the standard mode of operation will be to run wubi from the CD
[10:19] <xivulon> I was thinking of a workaround
[10:19] <cjwatson> we may have to come up with a different option
[10:19] <evand> copy the executable to the drive then run it from there, in code?
[10:19] <cjwatson> anyway, I really have to run for a bit, back later
[10:19] <xivulon> for instance wrapping wubi in a 7z selfextracting file
[10:19] <xivulon> so when you run wubi it first extracts to some tmp folder and then runs it
[10:19] <cjwatson> evand: I'd rather have wubi get the system to boot from the hard disk next time, bypassing the CD
[10:20] <cjwatson> same way instlux can get the system to boot from the CD next time
[10:20] <cjwatson> or perhaps leave a note somewhere on the disk that causes gfxboot to go straight to booting from the hard disk
[10:20] <cjwatson> that sounds doable
[10:20] <cjwatson> ok, wubi completed, I'll try the rest later :)
[10:20] <cjwatson> thanks
[10:20] <xivulon> good
[10:20] <xivulon> next time it will be quicker since it will reuse the ISO
[10:21] <evand> ok
[10:22] <xivulon> cjwatson, just try to set the bios to start from HD, then select Use CD in wubi and see if it works well for booting
[10:27] <xivulon> We will have a look to see if the requirement of the CD not being in use can be loosened up
[10:28] <xivulon> instlux and debian installer by the way work the same way as wubi when it comes to booting: grub4dos + local kernel/initrd
[10:56] <superm1> cjwatson, could you point me at what's involved with making a mythbuntu task?
[10:56] <superm1> i saw how ubuntu studio's show up in synaptic now
[10:59] <superm1> is it just a matter of adding the task to the top of the seed file in bzr and poking someone?
[11:57] <cjwatson> nope, partman-auto-loop still hosed
[11:57] <cjwatson> I know why though, it's using move-mount syntax that busybox likes but util-linux doesn't
[11:57] <cjwatson> I'll hack around it
[12:01] <evand> I can't even get that far.  No matter what I do, I can't seem to get the host system mounted at startup.
[12:03] <cjwatson> it's not supposed to be mounted when you boot the live filesystem
[12:03] <cjwatson> partman-auto-loop deals with mounting it, at least when booting from CD
[12:04] <evand> at least it's broken anyway :)