[03:25] <cjwatson> sigh, I broke ntfs in d-i
[03:47] <xivulon> cjwatson, I sent you a couple of emails, keyboard detection should be better now
[03:47] <cjwatson> ok, your mail to me is pretty reliable, so you don't need to tell me on IRC when you've sent me mail, thanks :)
[03:48] <xivulon> something not on the email, if you want to build wubi yourself, you have to edit scripts/build.sh and add prerequisit to the case statement at the bottom
[03:49] <xivulon> then run "make prerequisites" and "make"
[03:49] <xivulon> first command will setup wine + nsis
[03:50] <xivulon> "make test" to run the dialog in debug mode
[03:54] <xivulon> The reason I split keyboard versions in a separate ini file is that windows uses an 8-digit hex, and did not want to remap the entries in keymaps.ini
[03:56] <xivulon> I mean keyboard variants
[03:57] <cjwatson> I wonder if wubi could be built on our buildds with that
[03:57] <cjwatson> I sort of assumed it needed a Windows installation
[03:57] <xivulon> wine + nsis is fine
[03:57] <cjwatson> hmm, I don't think our buildds are allowed to wget random stuff from the web though
[03:57] <cjwatson> so that wouldn't work
[03:57] <xivulon> we have one plugin in c# at the moment though
[03:58] <xivulon> cjwatson, wget is only a convenience script, you can install nsis from within wine manually
[03:58] <xivulon> provided the building server has wine with nsis installed (and mono if you want to recompile wubibcd) it should be fine
[04:00] <xivulon> The makefile is sub-par at the moment, the only thing in common with a proper makefile is the name
[04:01] <xivulon> The nsis plugins we use are not recompiled at the moment
[04:08] <cjwatson> yeah, that's what I mean, I don't see our buildd administrators preinstalling wine and nsis in the lifetime of this universe :)
[04:35] <xivulon> I don't really blame them, anyway now at least you can rebuild the frontend if you need to
[04:38] <cjwatson> good stuff
[04:55] <xivulon> cjwatson, the nsis version is a bit dated, I will update it later on (there are some strange warnings when using the new version I have to sort out) and upload a new nsis tar
[05:01] <evand> yikes, unionfs or python seems to have major breakage in the most recent daily live cds.
[05:06] <evand> http://evalicious.com/syslog
[05:07] <superm1> okay so that wasn't just me
[05:08] <superm1> we changed a lot in our builds a day or two ago, and then i booted one and it was going all to hell
[05:08] <evand> heh
[05:08] <superm1> i'd point likely towards unionfs
[05:08] <superm1> because there a re unionfs errors all over dmesg
[05:18] <superm1> well the only unionfs related bugs i'm finding that were filed thus far: https://bugs.edge.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/138866
[05:24] <evand> it only seems to happen when calling python, but I imagine that's just luck of the draw as python wouldn't be able to dereference a NULL pointer in the kernel.
[05:24] <superm1> well it was happening to me actually with apt too
[05:24] <evand> ah
[05:24] <superm1> with some other weird things like mmap unable to msync
[05:25] <superm1> and then not being able to find files that were on the filesystem
[05:39] <cjwatson> evand,superm1: yeah, several of us are seeing that - pkl's looking at it
[05:39] <evand> ah, thanks
[05:39] <superm1> as evand just mentioned in #ubuntu-kernel https://bugs.launchpad.net/ubuntu/+source/linux-source-2.6.22/+bug/138915
[05:39] <superm1> has both of our dmesg/syslog attached
[05:52] <cjwatson> I got my new laptop today, with a Windows installation on it which I plan to use to test wubi/lupin
[05:53] <cjwatson> (though I've already installed Ubuntu separately of course ;-))
[05:56] <evand> http://www.vmware.com/products/converter/ ?
[05:57] <cjwatson> just running natively
[05:57] <cjwatson> the Vista Home Basic licence forbids running it in a VM
[05:58] <cjwatson> which is a bit control-freaky but there you go :-/
[06:00] <evand> ah, I forgot about that
[06:00] <xivulon> cjwatson... ...on vista? :P
[06:01] <cjwatson> will wubi not work on vista?
[06:01] <xivulon> Well I should shoutup since my laptop came with premium, I contributed to ms coffin more than colin
[06:01] <xivulon> cjwatson it should
[06:01] <cjwatson> xivulon: could we have a stable link for the latest version of wubi? like Wubi-7.10-minefield.exe -> Wubi-7.10-minefield-0.4.exe
[06:02] <cjwatson> coffer, not coffin :-)
[06:02] <cjwatson> (contributing to MS' coffin is a *good* thing ...)
[06:02] <xivulon> well talkin of coffer wubi helped took away 200K users and counting... So I guess it's a good investment...
[06:02] <cjwatson> yeah, I would have got one of Dell's Ubuntu laptops but they weren't shipping Core 2 Duos with Ubuntu in the UK
[06:03] <thom> what did you end up with? HP?
[06:03] <xivulon> same here, I ended up with a samsung, (also because on last 2 dell laptops power jacks desoldered)
[06:04] <cjwatson> thom: dell latitude d830
[06:04] <cjwatson> we'll see how it goes
[06:04] <thom> oh right; is that the T60 sized one, or smaller?
[06:04] <cjwatson> gar, what is wrong with cdimage? it's supposed to be downloading wubi but isn't
[06:05] <xivulon> if you do not have kids stumbling on the power cable they are fine machines
[06:05] <evand> laptop refreshment plan, only 2 years and 9 months away :)
[06:05] <cjwatson> thom: dunno how big a t60 is. it's 15", fairly chunky but I like big laptops
[06:06] <thom> yeah, similar to powerbook sized
[06:06] <xivulon> cjwatson, I cannot do the link from the office (no ssh), will do tonight
[06:07] <cjwatson> ah, the download breakage is my fault
[06:07] <cjwatson> xivulon: thanks, I'll just bump it to 0.4 for now
[06:14] <cjwatson> thom: yeah, about two inches higher than the 15" PowerBook and maybe half an inch deeper
[06:15] <cjwatson> (but the height is mostly because the monitor is hinged differently)
[06:15] <thom> nods
[06:33] <CIA-18> ubiquity: evand * r2220 ubiquity/bin/ubiquity: Fix test.
[06:45] <cjwatson> xivulon: I wonder if it's worth having the "-tribe5" subversion thing, which we'll just have to update all the time? We don't do that for our native installers
[06:47] <cjwatson> xivulon: grub4dos installation doesn't seem to have worked here. Anything I can do to investigate?
[06:47] <cjwatson> I just get thrown into my normal grub menu
[06:48] <cjwatson> ah, I need to boot into Windows
[06:48] <cjwatson> ok, that's convoluted
[06:48] <cjwatson> bugger, fell over in initramfs
[06:51] <xivulon> cjwatson, if you have grub, you have to select Windows
[06:51] <xivulon> wubi adds an option to the windows bootloader, not to grub
[06:53] <cjwatson> yeah, I worked it out while you were pinged out
[06:53] <cjwatson> it's fallen over in the initramfs now and I'm debugging
[06:54] <cjwatson> whoa, we're using klibc /bin/mount? when did that happen?
[06:55] <xivulon> if you can boot, and preseed and boot options are correct, the frontend job is done.
[07:02] <xivulon> re version name, what do you suggest?
[07:04] <cjwatson> perhaps just "Ubuntu 7.10 alpha"
[07:05] <cjwatson> ok, lupin is failing because fuse uses the -i option to mount which klibc mount doesn't support
[07:05] <cjwatson> easy fix
[07:08] <xivulon> Sounds good, do you want to have something like the launchpad revision in there?
[07:18] <cjwatson> nope
[07:19] <cjwatson> well I mean I suppose you could but it would just have to be removed for release :)
[07:19] <cjwatson> I dunno, up to you, I just feel it's better without too much information that has to be updated
[07:30] <xivulon> np
[07:31] <xivulon> How is the installation going? In the last few days I only looked at the frontend.
[07:35] <cjwatson> 17:53 <cjwatson> it's fallen over in the initramfs now and I'm debugging
[07:35] <cjwatson> about that well
[07:35] <cjwatson> I think I've beaten on klibc sufficiently to fix that
[07:42] <cjwatson> ok, at least it's booting now
[07:42] <cjwatson> unionfs is known to be a bit screwed at the moment though so I don't know how far we'll get
[07:46] <cjwatson> up to X
[07:47] <cjwatson> hmm, suspend and hibernate options still present
[07:50] <cjwatson> automatic-ubiquity didn't work, init script isn't symlinked properly
[07:51] <evand> hrm, using fuse, ntfs directories show up as DT_UNKNOWN.
[07:53] <cjwatson> xivulon: the name /tmpmountpoint isn't important, is it? if I rename that to /host then the suspend/hibernate inhibition should work right
[07:53] <CIA-18> ubiquity: cjwatson * r2221 ubiquity/debian/ (changelog rules): * Call update-rc.d so that automatic-ubiquity stands a chance of working.
[07:54] <xivulon> it's just a throwaway folder, make sure it's something not colliding with other scripts
[07:54] <cjwatson> OK, /host should be fine then
[07:55] <xivulon> not that it is called more than once, and for each call and for each partition /tmpmount is mounted/unmounted
[07:55] <xivulon> note
[07:56] <cjwatson> that's ok
[07:56] <cjwatson> as long as it's left mounted at the end
[07:57] <xivulon> It's left mounted unless the last call does not find what it was looking for
[07:57] <cjwatson> right
[07:58] <cjwatson> oh, hang on, am I crazy?
[07:59] <cjwatson> there's no reason why that fuse mount in particular should inhibit suspend/hibernate, since it isn't helping to implement the root filesystem
[07:59] <cjwatson> I'll revert that, sorry
[08:02] <xivulon> that is only used to find files, and the mount is always r/o
[08:02] <cjwatson> user_folder=#UserFolder=/Users/cjwatson^M
[08:02] <cjwatson> hmm
[08:03] <cjwatson> so preseeding itself is fine with CR/LF files but lupin's direct fiddling with the preseed file isn't ... looks like I spoke slightly too soon :(
[08:04] <xivulon> ah yes, I forgot about that
[08:04] <cjwatson> xivulon: doesn't /tmpmountpoint need to be unmounted after we're finished with it, regardless?
[08:04] <cjwatson> because otherwise it's not possible to mount the same filesystem from the live session
[08:06] <xivulon> hmm, often you want to do something with the file, so either you umount within find_path() and then remount in the code calling find_path or you have to remember to umount after calling findpath
[08:06] <cjwatson> yeah, it's clearly a little fiddly
[08:06] <cjwatson> I'll sort something out
[08:08] <xivulon> for instance, when you use it within find_iso, then the filesystem hosting the ISO must not be unmounted
[08:08] <cjwatson> right
[08:09] <xivulon> At the moment when find_path scans a folder which is already mounted it does not try to remount it, that is to make it work with find_iso.
[08:09] <cjwatson> I think a find_path_cleanup function called at the end of each client script would be sufficient
[08:12] <xivulon> yes, just make sure that find_iso is not affected, that is mounted on top of /isodevice (which overrdies the default /tmpmount folder)
[08:14] <cjwatson> yep
[08:18] <xivulon> one more thing, wait_for_devs, should be called the first time find_path is used, at the moment wait_for_devs is within find_iso, but that is not appropriate since you might use find_preseed without find_iso
[08:22] <xivulon> something like [ -e /var/run/devs_ready ]  || wait_for_devs (within find_path) and then "touch /var/run/devs_ready" (within wait_for_devs)
[08:23] <cjwatson> better, just wait_for_devs in both places and do the test inside wait_for_devfs
[08:23] <cjwatson> devs
[08:23] <cjwatson> ok, I'll do that once bzr deigns to respond
[08:23] <cjwatson> s/bzr/my local wireless network/ I think actually
[08:25] <xivulon> then call wait_for_devs from find_path and add the test within wait_for_devs
[08:25] <cjwatson> indeed, done :)
[08:25] <cjwatson> parallel development ;)
[08:27] <cjwatson> ?
[08:31] <xivulon> have a sun-fan friend who often says "we start to think alike" and I usually reply "not sure I like that"... nevermind
[08:32] <xivulon> like a pavlov response
[08:51] <cjwatson> OK, lupin 0.5 fixes all the problems I can find here
[08:51] <xivulon> cl
[08:52] <xivulon> cjwatson, not sure find_path_cleanup is ok in find_iso
[08:52] <cjwatson> why not?
[08:53] <xivulon> livemedia expects a filesystem path, and that path will be something like /isodevice/ubuntu/install/xyz.iso
[08:53] <xivulon> but if /isodevice is unmounted livemedia will not find any iso
[08:53] <cjwatson> oh
[08:53] <cjwatson> ok, good point
[08:56] <cjwatson> fixed
[08:57] <cjwatson> (lupin 0.6)
[08:58] <CIA-18> ubiquity: cjwatson * r2222 ubiquity/ (configure configure.ac): bump to 1.5.14
[09:08] <CIA-18> ubiquity: cjwatson * r2223 ubiquity/ (d-i/manifest debian/changelog):
[09:08] <CIA-18> ubiquity: * Automatic update of included source packages: grub-installer
[09:08] <CIA-18> ubiquity:  1.24ubuntu4.
[09:11] <xivulon> for the thinkpad fans out there: http://lenovoblogs.com/insidethebox/?p=98
[09:13] <CIA-18> ubiquity: cjwatson * r2224 ubiquity/debian/po/ (79 files): debconf-updatepo
[09:16] <cjwatson> somebody please remind me to do a full ubiquity translation update from Rosetta before beta
[09:20] <evand> will do
[09:21] <CIA-18> ubiquity: cjwatson * r2225 ubiquity/debian/changelog: releasing version 1.5.14
[09:21] <cjwatson> evand: not to mention to show you how to do it, but that can probably wait :)
[09:22] <cjwatson> (it's a nightmare of twisty little scripts in my ~/bin/)
[09:22] <evand> haha, I'm beginning to see a pattern here
[09:22] <evand> twisty little scripts everywhere
[09:24] <cjwatson> I *try* not to ...
[09:26] <cjwatson> yay, new laptop unpacks packages like greased lightning
[09:27] <cjwatson> I like having I/O worth a damn
[09:28] <evand> haha
[09:29] <evand> I wish I could just fold my desktop.  I'm tempted to just get one of those iMac carrying cases for the next sprint.
[12:05] <xivulon> cjwatson, did you manage to install at the end?
[12:06] <cjwatson> xivulon: no, I fixed everything I could but lupin was a bit too buggered really. I plan to give it another go with tomorrow's image
[12:06] <cjwatson> going to bed now
[12:08] <xivulon> I am updating some wubi frontend code, should be there by tomorrow
[12:08] <xivulon> late on tonight
[12:08] <cjwatson> ok, drop me a line if you have a new wubi url for me
[12:11] <xivulon> sure