[09:07] <makx> Keybuk: have you split out udev into hooks?
[09:07] <Keybuk> yes, our udev package will ship its own udev hook script and normal scripts
[09:08] <infinity> (But he hasnt made those uploads yet, so looking for patches is useless)
[09:08] <Keybuk> I've also taken the hook out of initramfs and the code itself out of the source
[09:08] <makx> nice
[09:08] <Keybuk> infinity: I'm gonna do them today
[09:08] <infinity> \o/
[09:08] <Keybuk> is lrm ready?
[09:08] <infinity> I'm doing LRM today.
[09:08] <makx> #341014 new udev related rc bug.
[09:08] <infinity> BenC is doing one last kernel upload tomorrow before we do -meta.
[09:09] <Keybuk> cool
[09:09] <Keybuk> makx: yeah, there's a zillion copies of that one -- doesn't affect us :)
[09:09] <infinity> (Which will be an ABI bump and another LRM, but that's no big deal)
[09:09] <makx> Keybuk: haven't looked at it yet.
[09:09] <makx> didn't know there are more than this one.
[09:09] <infinity> Keybuk : Is there any chance of us converging with Debian's udev again, or are we forking too far to come back at this point?
[09:09] <Keybuk> makx: afaict it's a user using a udev that requires 2.6.15 on 2.6.14
[09:10] <Keybuk> infinity: forking it and never looking back
[09:10] <Keybuk> upstream is our rock for udev now
[09:16] <makx> Keybuk: any recommendation how i should handle that bug?
[09:16] <makx> but udev keeps it away atm.
[09:16] <makx> d-i needs initramfs-tools in testing.
[09:16] <Keybuk> Depends: linux-image-2.6.15-<abi> would work
[09:27] <Kamion> kernel tomorrow? OK, Flight 2 is officially next week, not this week
[09:27] <Kamion> I'm not going to try to get it out two days after a major kernel upgrade
[09:31] <makx> hmm newer udev works on my box
[09:31] <makx> even with 2.6.14 *strange*
[09:33] <infinity> Kamion : Wimp. :)
[09:33] <infinity> Kamion : After some of my other tasks are settled, CD buildability/installability is moving up my personal priority list.
[09:45] <infinity> Kamion : Have you decided what to do about cdebconf and libgtk+2.0-directfb-dev?
[10:04] <makx> bahh could reproduce while recreating initramfs
[10:11] <Kamion> infinity: not yet
[10:11] <Kamion> inclined to rip it out of cdebconf for the moment until it stabilises upstream, though
[10:52] <fabbione> Kamion: can i safely assume that architectures that have special /boot requirements are properly coded in partman-auto recipes?
[10:53] <fabbione> it looks to me that it is a "true" statement and that the recipes are done in such a way that / is always on an ext3 fs
[10:57] <fabbione> if so, that will simplify my life of a magnitude
[11:11] <makx> Keybuk: i've fixes for udev 0.76, will integrate your separation after.
[11:15] <Kamion> fabbione: the partman-auto recipes shouldn't create anything incorrect
[11:15] <Kamion> by themselves
[11:16] <fabbione> Kamion: ok.. i did recheck and it looks like i am right.. in that case i can save a few tons of code :)
[11:58] <Mithrandir> Kamion lshttp://people.ubuntu.com/~tfheen/bzr/cdebconf-keystep/
[11:58] <Mithrandir> grr
[11:58] <Mithrandir> Kamion: http://people.ubuntu.com/~tfheen/bzr/cdebconf-keystep/ and http://people.ubuntu.com/~tfheen/bzr/cdebconf/ might be interesting.
[11:58] <Mithrandir> they should be pullable now
[11:59] <Kamion> I'll have a look later; buried in syslinux right now
[11:59] <Mithrandir> sure, have fun. :-)
[12:02] <Kamion> unlikely :-/
[12:02] <Kamion> "please apply 3000 lines of assembly patch, kthxbye"
[12:03] <Kamion> +.nosusedata:
[12:03] <Kamion> NO, THIS ISN'T DISTRO-SPECIFIC AT ALL, WHY DO YOU ASK?
[12:03] <Mithrandir> it's early enough that you can't use C sensibly?
[12:03] <Kamion> syslinux is mostly assembly, yes
[12:03] <Kamion> there's some C in the higher-level areas, but this is all very low-level
[12:04] <Kamion> +lxrc_data      db ' SuSE='
[12:04] <Mithrandir> heh
[12:18] <Mithrandir> Kamion: it's debian-cd which generates the syslinux configuration for the install CDs, right?
[12:22] <Kamion> Mithrandir: currently, yes
[12:22] <Kamion> colin.watson@canonical.com--2005/debian-cd--ubuntu--0
[12:22] <Mithrandir> cheers
[12:26] <Mithrandir> I think I have a somewhat ugly hack for it to reboot after checking.  Set MENU to /bin/cdrom-checker;/sbin/reboot
[12:34] <fabbione> i think i got partman-auto-lvm to do exactly what we want
[12:34] <fabbione> Kamion: is the latest parted (with lvm support on ppc) in dapper?
[12:36] <Kamion> fabbione: yes
[12:37] <Kamion> <cjwatson@cairhien ~>$ bash
[12:37] <Kamion> <cjwatson@cairhien ~>$ MENU='foo;id'
[12:37] <Kamion> <cjwatson@cairhien ~>$ exec echo $MENU
[12:37] <Kamion> foo;id
[12:37] <Kamion> Mithrandir: I don't think that will work, no
[12:37] <fabbione> Kamion: thanks
[12:38] <Kamion> Mithrandir: shouldn't we just arrange for cdrom-checker to exit 12 in some mode? that will cause rootskel to reboot
[12:38] <Mithrandir> Kamion: oh, it does?  If so, that's a trivial change.
[12:39] <Mithrandir> isn't export MENU=${1:-/usr/bin/main-menu}
[12:40] <Mithrandir> bloody unuseful, since it's always called from inittab?
[12:44] <Kamion> you can run debian-installer by hand when debugging stuff
[12:44] <Kamion> in fact I often do :)
[12:45] <Kamion> yes, it's useless in the default path
[12:45] <Mithrandir> yes, but that piece of code means you can't do my hack which was just to set MENU on the syslinux command line
[12:45] <Mithrandir> so I'm wondering if it should leave $MENU alone if it's set.
[12:46] <Mithrandir> any thoughts on that?
[12:46] <Kamion> but your hack wouldn't work anyway because variable expansion doesn't work the way you need
[12:47] <Kamion> oh, without ; you mean?
[12:47] <Kamion> sure, it probably should
[12:47] <Mithrandir> true, so cdrom-checker would need a small adjustment as well
[12:47] <Kamion> anyone here know what the SuSE= kernel argument does on SuSE kernels?
[12:48] <Mithrandir> to exit 12 instead of EXIT_SUCCESS if strcmp(argv[0] ,getenv("MENU")) == 0 or something like that.
[12:48] <Kamion> that would work ...
[12:49] <Mithrandir> it's a hack, but I don't think it's undesirable.
[12:50] <Mithrandir> I'll just commit it upstream, I can't see any reason not to take that.
[12:50] <Keybuk> Kamion: does /lib/modules/.../modules.ofmap exist and contain data for you?
[12:51] <Kamion> Keybuk: not on Debian; should I reboot into dapper?
[12:52] <Keybuk> it may need new module-init-tools
[12:53] <Keybuk> ok, I won't worry for now
[12:54] <Kamion> oh, use-case for having proc in fstab; upgrading chroots and being too lazy to type 'mount -t proc proc /proc' (some packages require mounted /proc to upgrade cleanly)
[12:56] <Keybuk> true, but then we'd also need /sys in there
[12:56] <Keybuk> not to mention /dev
[01:02] <Kamion> 'mount /sys' seems to work without it being in fstab
[01:03] <Kamion> (hmm, maybe /proc would too, haven't tried)
[01:03] <Kamion> the minimal set of device nodes debootstrap sticks in /dev is generally enough to get by for just upgrading stuff
[01:03] <Kamion> you don't really need hardware devices there, just the virtual ones
[01:04] <Keybuk> I guess
[01:04] <Keybuk> I'm not too concerned which way we go
[01:04] <Keybuk> if we have /proc in fstab, we should have /sys too, both with a note that everything is ignored and it's just there so you can "mount /proc"
[01:26] <infinity> Kamion : People mount proc from inside their chroots?... I mount proc from the base system into all the chroots.
[01:34] <fabbione> GO LVM
[01:34] <fabbione> DIE DIE DIE
[01:58] <Mithrandir> Kamion: would you rather that I backport my changes (so we have to merge them later), or should I just upload the packages in Debian and wait for them to trickle through to us?  It's rootskel and cdrom-checker.
[02:02] <Kamion> Mithrandir: just do it in Debian, I think
[02:02] <Kamion> particularly since we're in sync with rootskel right now
[02:02] <Mithrandir> ok, sure
[02:03] <Mithrandir> do we sync installer components automatically?
[02:04] <Kamion> yes
[02:27] <Kamion> coo, syslinux supports assembling multiple initramfs pieces now
[02:33] <fabbione> we need to switch to .15
[02:33] <fabbione> or i will be forever doomed to build custom d-i
[02:33] <jbailey> fabbione: What's blocking it?
[02:33] <fabbione> jbailey: that my ppc is not supported in .12?
[02:34] <fabbione> so i had to patch .12 to get a kernel that can actually install on my ppc
[02:34] <jbailey> Kamion: I think it has since at least OLS.
[02:34] <jbailey> Kamion: HPA and I talked about it there.
[02:50] <fabbione> bah
[02:50] <fabbione> pal still hates me on ppc
[02:51] <fabbione> ok.. it's tomorrow's work to get it fixed
[02:56] <Mithrandir> Kamion: some people might think you're slightly obsessive in your branding when you are changing Debian to Ubuntu inside comment strings in the source. ;-)
[02:57] <Kamion> :-)
[02:57] <fabbione> hmmm
[02:57] <Kamion> yeah, should probably drop those patches
[02:57] <fabbione> Kamion: do you think you can help me a second?
[02:58] <Kamion> sure
[02:58] <fabbione> for some reasons a d-i build from today on ppc
[02:58] <fabbione> doesn't have libselinux
[02:58] <fabbione> and lvm2 basically doesn't work
[02:58] <fabbione> using the last daily install on i386 works.. and i am assuming it has libselinux
[02:58] <Kamion> it needs libselinux in the udeb build? that's fucked
[02:58] <Kamion> there's no point having libselinux in d-i and I'd rather it weren't there
[02:58] <fabbione> pvs: error while loading shared ..
[02:58] <fabbione> ok
[02:59] <Kamion> there's no selinux udeb
[02:59] <fabbione> so i guess i need to play with lvm2 to make 2 different builds
[02:59] <Kamion> yes
[02:59] <fabbione> that's ok
[02:59] <fabbione> it's only linked
[02:59] <fabbione> "only"
[02:59] <Kamion> if it's linked as a shared object (and if not, why not?) then there needs to be a udeb
[03:00] <fabbione> either i got the merge wrong
[03:00] <fabbione> (lvm2)
[03:00] <fabbione> or we need that
[03:00] <fabbione> but right now i am too tired to figure it out
[03:01] <fabbione> the new pal should work just fine once this is fixed
[03:46] <Keybuk> Kamion: so, what do you reckon?  should I upload new udev and friends now, or wait until wednesday morning
[03:46] <Keybuk> (on the basis that I'm not contactable at all tomorrow)
[03:47] <infinity> Will it break the current default kernel?
[03:48] <infinity> If 2.6.12-9 continues to work right until I update linux-meta, upload it now.
[03:48] <infinity> Otherwise, drop the upload somewhere where I can get at it, and I'll do it all together.
[03:50] <Keybuk> it'll braek
[03:50] <infinity> Now I tihnk I'm going to dirnk myself into a stupor and attempt to get some stress-relieved sleep.
[03:50] <infinity> Keybuk : Then sign the uplooad, put it somewhere, and mail me the location.
[03:50] <infinity> Keybuk : I'll make the upload together with linux-meta.
[03:53] <Keybuk> http://people.ubuntu.com/~scott/packages/new-udev
[03:55] <jbailey> Keybuk: How brave do we have to be? =)
[03:55] <Keybuk> works for me :)
[03:56] <jbailey> Do I need to be on 2.6.15?
[03:57] <jbailey> Last I checked that was still b0rked on ppc64
[03:57] <jbailey> dpkg-source: error: file udev_076.orig.tar.gz has size 4903 instead of expected 281735
[03:57] <Keybuk> yes, you do
[03:57] <jbailey> Keybuk: How are you planning to handle the migration from Breezy to Dapper?
[03:57] <Keybuk> uh, try now
[03:57] <Keybuk> "migration" ?
[03:58] <jbailey> Making sure that a person has the new kernel loaded.
[03:58] <jbailey> Works this time, thanks.
[03:59] <Keybuk> it checks whether /sys/class/mem/null/uevent exists, if not unmounts /dev and puts the old static one back
[03:59] <jbailey> Ah, cool.
[03:59] <Keybuk> it seemed a sane one to look for <g>
[04:00] <jbailey> Ooo, cuddly.  It provides the initramfs-tools hooks itself.
[04:00] <Keybuk> yes
[04:01] <jbailey> Nice.  The hooks are quite clean.
[04:02] <jbailey> I'll ask Ben about 2.6.15 when he's on later, and then test it on ppc64 and sparc for you.
[04:04] <Keybuk> right, gonna have lunch, then go down south -- back wednesday
[04:43] <makx> hello jbailey
[04:43] <makx> i've got an upldate for the debian version.
[04:43] <makx> new udev upstream..
[04:44] <makx> closes 1 rc bug, haven't packaged yet.
[04:44] <jbailey> makx: Cool.
[04:44] <makx> will ping you in one hour when it's on mentors.
[04:44] <jbailey> makx: Note that Ubuntu and Debian udev packages are about to diverge heavily.
[04:44] <makx> yes please tell me
[04:45] <jbailey> makx: It might be worth taking a look at some of the hacks that Scott's done in his repository.  It splits all of the udev stuff into the udev package.
[04:45] <makx> you are stabilizing?
[04:45] <makx> i know
[04:45] <jbailey> For Debian, you might want to take those hooks and package them into the initramfs-tools package directly.
[04:45] <makx> is it uploaded already to ubuntu?
[04:45] <makx> spoke with Md he will take those hooks in middle term
[04:46] <makx> enough running after udev.. ;)
[04:47] <makx> uugh missing /dev/input/mouse
[04:47] <makx> has your kernel also modular mousedev
[04:49] <jbailey> Yes.
[04:51] <makx> :(
[05:23] <makx> Keybuck: where can i find latest udev?
[07:04] <makx> jbailey uploaded 1 min ago initramfs-tools 0.41 -> http://mentors.debian.net/debian/pool/main/i/initramfs-tools/
[07:04] <makx> bug reporter did test that version with udev from incoming
[07:05] <makx> solved all issues for him+me.
[07:05] <makx> again an high urgency upload.
[07:07] <makx> also vorlon is letting initramfs-tools 0.40 into testing
[07:07] <makx> as it works against the udev which is in etch.
[07:07] <makx> :)
[11:35] <makx> Kamion: initramfs-tools migrated to testing :)
[11:36] <Kamion> hooray, we can revert that hack after the mirror push then
[11:42] <fabbione> Kamion: still around?
[11:44] <Kamion> fabbione: yes
[11:44] <fabbione> Kamion: do you think we can schedule a 20/30 minutes work together tomorrow?
[11:45] <Kamion> yeah, probably. I have to go into town at some point though
[11:45] <fabbione> other than between 10:30 UTC and 12:30 UTC ( i have to go to the dentist)
[11:45] <fabbione> yeah you tell me when it's best for you outside that slot and i am fine
[11:45] <fabbione> there are some issues with pal i need to look at together with you
[11:46] <fabbione> because i am not sure 100% how to address them
[11:57] <Kamion> 15:00 UTC?