[02:13] <joejaxx> superm1: yes i am
[02:14] <superm1> hey joejaxx .  I wanted to ask if you could look over the current packaging i have for our metas.  i can't figure out what is happening
[02:14] <superm1> why ./update won't work
[02:15] <superm1> the LP page for the meta is here: https://code.launchpad.net/~mythbuntu/mythbuntu/mythbuntu-meta and the seed here https://code.launchpad.net/~mythbuntu/ubuntu-seeds/mythbuntu.gutsy
[02:43] <superm1> joejaxx, this is the output that I am getting when trying to build: http://paste.ubuntu-nl.org/31854/
[02:44] <superm1> I had thought it was originally the version of germinate i was using (since i'm on feisty), so I backported the gutsy version of germinate and debootstrap to feisty and installed both of those
[10:18] <cjwatson> superm1: odd, looks like it got a STRUCTURE file with a blank line in it or something
[10:18] <superm1> cjwatson, yes that's what it ended up being
[10:19] <superm1> i was banging my head on the desk for a long time today
[10:19] <superm1> and caught that
[10:19] <cjwatson> ah yes, I see that
[10:19] <cjwatson> I'll fix germinate to ignore that
[10:19] <superm1> i'm trying to determine now how come the contents of the standard seed aren't being included
[10:20] <superm1> even though they are listed in the STRUCTURE file under the seed i wanted them to be dependent on
[10:20] <cjwatson> (these aren't relevant to that question, but) I'd suggest switching the order of live and standalone in STRUCTURE, and including a supported seed at the end ... supported is the only seed that's kinda hardcoded
[10:21] <superm1> oh the order actually matters then?
[10:22] <cjwatson> yes, it can make a slight difference - always list seeds from inner to outer
[10:22] <cjwatson> your seed_map/minimal is broken, because you don't have a minimal seed
[10:23] <cjwatson> I would suggest changing your seeds rather than changing update.cfg - it's better to be branched off the most recent possible Ubuntu seeds
[10:23] <cjwatson> err, "because you don't have a required seed" I mean
[10:23] <superm1> i was modeling it off of i think xubuntu or another that didn't have a required either
[10:23] <cjwatson> your seed_base is also outdated - should be /~ubuntu-archive/seeds/ now
[10:23] <superm1> and then was getting confused as to which way to follow
[10:23] <cjwatson> and you need to update archive_base/blah
[10:24] <cjwatson> xubuntu has required
[10:24] <cjwatson> it looks like you modelled it off feisty
[10:24] <superm1> possibly
[10:24] <cjwatson> DDTT :)
[10:25] <superm1> you were meaning in my update.cfg, that it was outdated?  I have seed_base: http://people.ubuntu.com/~ubuntu-archive/seeds/
[10:25] <superm1> listed
[10:25] <cjwatson> that's not committed yet then?
[10:25] <superm1> it should be
[10:26] <cjwatson> I was looking on codebrowse.launchpad.net
[10:26] <superm1> whoops wasn't pushed
[10:26] <superm1> that would make a bit of a difference wouldn't it :)
[10:28] <superm1> nonetheless though i still don't have a required seed listed, so i'll have to add that in
[11:03] <superm1> cjwatson, in switching things around to more closely model the ubuntu.gutsy branch and ubuntu-meta package, somehow this came along: http://paste.ubuntu-nl.org/31882/.  I'm a bit confused by it
[11:05] <cjwatson> superm1: take boot out of the live seed's inheritance list (boot is WEIRD)
[11:06] <cjwatson> superm1: and take it out of seeds: in update.cfg
[11:06] <cjwatson> superm1: change 'seed_map/minimal: minimal' to 'seed_map/minimal: required minimal' now that you have a required seed
[11:06] <superm1> well its out of the update.cfg
[11:06] <superm1> and i changed the seed_map to required minimal
[11:06] <cjwatson> this is part of the reason there's a separate ship-live in the Ubuntu seeds which inherits from boot
[11:06] <superm1> (locally)
[11:07] <cjwatson> you can't use boot in anything that generates a metapackage
[11:07] <superm1> oh
[11:07] <cjwatson> germinate gets too confused by having to follow the kernel dependency chain
[11:07] <cjwatson> and livecd.sh installs a kernel anyway, so it's not necessary
[11:08] <superm1> well the way that we bootstrap the mythbuntu disks is a bit different, so ideally a kernel should be listed in our metas
[11:08] <superm1> if possible
[11:08] <cjwatson> you can't
[11:08] <cjwatson> not without substantial code changes. I went down this road a year or two back
[11:09] <superm1> is there something wrong with listing something like linux-image-generic [i386]  [amd64] 
[11:09] <superm1> and then another line for linx-image-powerpc [powerpc]  ?
[11:09] <cjwatson> your call, don't come to me if it breaks
[11:09] <superm1> haha
[11:09] <cjwatson> I recommend against
[11:10] <superm1> perhaps just listing the kernel in our build script will suffice then
[11:10] <cjwatson> like I say, it tends to produce wrong results in some places
[11:10] <superm1> i see
[11:11] <cjwatson> oh and including them in the metapackage makes ubiquity do the wrong thing too
[11:11] <cjwatson> well, including them in live might not have that effect
[11:11] <cjwatson> including them in minimal or required or whatever would definitely break
[11:11] <superm1> well i was going to be including them in standalone
[11:11] <superm1> and then ubiquity was going to pick off live
[11:11] <superm1> so i assumed that standalone could keep them
[11:12] <cjwatson> that breaks because there isn't necessarily just one kernel to be installed everywhere
[11:12] <superm1> right
[11:12] <cjwatson> take the case of powerpc, where you need to install -powerpc or -powerpc64 depending on the system
[11:12] <cjwatson> kernels have to not be in metapackages for that reason
[11:12] <superm1> and i386 you can get -generic or -i386
[11:12] <superm1> okay this makes more sense then
[11:13] <superm1> unfortunate though :(
[11:26] <superm1> okay i got it working again now that i added the ship-live.  Thanks cjwatson .  I"m gonna get to bed now
[11:31] <cjwatson> sleep well
[01:13] <cjwatson> so close to having partman-auto-loop working!
[02:26] <evand> awesome!
[02:27] <cjwatson> uploaded, too
[02:51] <cjwatson> evand: I just disabled multiverse for Gobuntu - noticed somebody had commented on that on the mailing list
[02:52] <evand> cjwatson: ah, thanks
[02:53] <cjwatson> evand: how goes the automation work? we should be able to try that out in conjunction with partman-auto-loop soonish
[02:53] <cr3> cjwatson: dude, many thanks for all the help you provided to Yongkang
[02:53] <cjwatson> evand: grub and initramfs probably won't work right yet though ...
[02:53] <cjwatson> cr3: no trouble, I'll reply to your mail about general policy in a bit
[02:53] <cjwatson> cr3: (basically I only took it because there was contention that it was a bug)
[02:54] <evand> cjwatson: good, just need to figure out why one of the partman questions isn't getting asked when I have the graceful page skipping on, possibly more issues related to partman.
[02:54] <cr3> cjwatson: I had a couple questions while reading the thread: 1. is it possible to presseed so that the installation is done on an existing partition? 2. is it possible to preseed so that an existing partition is resized?
[02:55] <cjwatson> cr3: 1) no, longish-standing wishlist for partman-auto; 2) I don't think so at the moment - the resize UI isn't really flexible enough
[02:56] <cjwatson> partman preseeding is mostly designed for deploying lots of Linux-only systems, and it does show rather
[02:57] <CIA-19> ubiquity: cjwatson * r2167 ubiquity/ (d-i/lists/any debian/changelog): * Add partman-auto-loop.
[05:43] <cjwatson> argh, debugging initramfs changes on a loop-mounted root filesystem is tedious
[05:44] <evand> heh
[05:45] <evand> partman issues in graceful page skipping are resolved, now I'm just cleaning things up a bit, looking for outstanding issues, and beating migration-assistant mercilessly
[05:45] <cjwatson> rock
[05:48] <cjwatson> ok, that was doing quite well until fsck failed
[05:48] <cjwatson> ugh!
[05:48] <cjwatson> "Error writing block 810 (attempt to write block from filesystem resulted in short write)."
[05:49] <cjwatson> in an infinite loop
[05:53] <glledo> Hi, in the advanced partitioner I'm using an existing partition for the / that is already formatted. If I just change its mountpoint it complains that "The file system on /dev/bla has not been marked for formatting"
[05:54] <glledo> I guess that if I change the mountpoint to something like / it should mark it to format, right? : P
[05:56] <evand> wishlist bug? or is that a design decision, cjwatson?
[05:57] <cjwatson> hang on, just looking for the bit of UI you need to twiddle
[05:58] <cjwatson> glledo: which release of Ubuntu?
[05:58] <glledo> is a gutsy daily live cd
[05:58] <glledo> let me see
[05:59] <glledo> I downloaded 19th of this month
[05:59] <cjwatson> oh, yeah, right, that warning
[05:59] <cjwatson> you really do have to format /
[06:00] <cjwatson> back up anything you need from there first
[06:00] <glledo> ok, so is a design decision
[06:00] <cjwatson> it's probably a valid wishlist bug that it should check to see if it's empty first before emitting that warning
[06:01] <cjwatson> (it doesn't have the filesystem mounted when it emits that warning though, so it's not trivial to do that check)
[06:01] <glledo> or probably just show a warning and then proceed to mark it as formattable
[06:01] <cjwatson> same problem
[06:01] <cjwatson> evand: partman/check.d/12system_partitions_formatted, BTW
[06:02] <glledo> ok, I have to leave, I'll check what did you decide tomorrow ; )
[06:02] <evand> ah, thanks cjwatson
[06:03] <evand> glledo: can you file a bug on this?
[06:19] <superm1> evand, could you merge me once more.  i've got another set of lirc changes in ubiquity
[06:19] <superm1> i renamed the branch so it shows up on code.lp.net/ubiquity now too
[06:19] <evand> superm1: will do
[06:19] <superm1> here is the link to the lp page: https://code.launchpad.net/~mythbuntu/ubiquity/mythbuntu-ubiquity  thx :)
[06:23] <cjwatson> damnit, why is write() failing with -EPERM at exactly the same spot (not the start) of this loop-mounted image every time?
[07:06] <glledo> evand: https://bugs.launchpad.net/ubuntu/+source/ubiquity/+bug/129268
[07:13] <evand> glledo: thanks
[07:13] <glledo> no prob
[07:13] <CIA-19> ubiquity: evand * r2168 ubiquity/ (8 files in 5 dirs): Merged with ubiqutiy.mythbuntu
[07:18] <superm1> evand, out of curiosity, how do you and cjwatson handle your commits to the ~ubuntu-installer/trunk branch?  Are you both bound to it to make sure you don't step on each other's toes?  Or do you guys always bzr pull before committing something?
[07:19] <cjwatson> I bind to it
[07:19] <evand> I am bound as well
[07:20] <superm1> so when multiple people work on the exact same launchpad branch its better to operate that way, whereas if you merge in multiple different branches from other locations, then it's better to be unbound and just do merges
[07:21] <evand> there will always be merges.  I just use bind to avoid forgetting to push.
[07:21] <superm1> ah
[07:22] <superm1> yea i kept doing that last night too
[07:22] <superm1> so in your guys' case then, you always bzr pull or bzr merge each other's changes?
[07:24] <cjwatson> neither, if you're bound you use bzr update
[07:24] <cjwatson> it stops you even committing if you're out of date
[07:24] <superm1> ah that's how to do it then
[07:24] <superm1> ok
[07:24] <cjwatson> it's also fine to be bound if you merge branches from other locations
[07:25] <cjwatson> as evand says, it's convenient for avoiding forgetting to push
[08:04] <xivulon> hi cjwatson, saw your email, was just reading your code
[08:07] <cjwatson> xivulon: ok
[08:08] <cjwatson> it's certainly not all the way there yet, but I think it's worth an intermediate look
[08:08] <xivulon> just came back from holidays (the boss vetoed any cpu-based equipment)
[08:09] <xivulon> so sorry if I left you all the burden
[08:09] <cjwatson> not a problem, I expected to be writing partman-auto-loop myself anyway as it was not easy
[08:11] <xivulon> was just reading that
[08:12] <cjwatson> preseeding and having that in the regular images should be a win
[08:12] <xivulon> q
[08:12] <xivulon> do you have to preseed the partition number?
[08:12] <cjwatson> yes
[08:12] <cjwatson> gotta know where to put the images :)
[08:13] <xivulon> that might be an issue to find that out from within windows
[08:13] <xivulon> I mean mapping windows drive letters to linux partitions
[08:13] <cjwatson> mm, we might be able to come up with some other way to specify that
[08:13] <cjwatson> anyway, off for the evening, please follow up by e-mail
[08:13] <superm1> is that where this automatic preseeding will be used?  setup.exe?
[08:13] <cjwatson> that's one use for it
[08:14] <cjwatson> and the most immediately urgent, but not the only one
[08:14] <superm1> probably also for OEMs
[08:14] <superm1> have a good evening cjwatson
[08:14] <cjwatson> OEMs> maybe, depends on their setup
[08:14] <xivulon> quick q
[08:14] <xivulon> is the host folder normally mounted r/o correct?
[08:15] <cjwatson> xivulon: I'd expect so on normal boot
[08:15] <cjwatson> but e-mail, please :)
[08:15] <xivulon> sure
[08:16] <cjwatson> the initramfs-tools default is to mount the root filesystem read-only; the way I've done it the host filesystem just follows that
[08:16] <xivulon> I agree
[08:16] <cjwatson> but it could probably be made to be otherwise if need be
[08:16] <cjwatson> e.g. if it's a problem for making other use of the Windows filesystem from Linux
[08:16] <xivulon> just wanted to note that in that case you might have to remount
[08:16] <xivulon> r/w
[08:16] <cjwatson> that's true
[08:17] <xivulon> which means adding an /host entry to fstab
[08:17] <xivulon> and changing checkroot
[08:17] <cjwatson> fstab's very much not sorted out yet
[08:18] <xivulon> is the build going to be ready tomorrow correct?
[09:29] <cjwatson> xivulon: should be
[09:30] <superm1> build of ubiquity you mean, releasing 1.5.7?
[09:32] <cjwatson> no
[09:32] <cjwatson> well, that's not what I was talking about. I assumed xivulon was asking about the alternate CD build
[09:35] <superm1> ah
[10:08] <xivulon> what is an efficient way to test? do I need to create an alternate iso every time?
[10:09] <superm1> xivulon, i think cjwatson is afk, he announced away in -devel, so you might want to mail
[10:18] <xivulon> oj
[10:18] <xivulon> ok