[01:36] <zul> merde
[01:39] <zul> the new system of a down is gooooood..
[02:22] <zul> hmm...
[02:22] <zul> Linux bart 2.6.15-1-686 #1 SMP Sun Nov 13 11:44:52 EST 2005 i686 GNU/Linux
[02:23] <zul> couple of hardware regressions on my laptop though
[02:23] <zul> ipw2200 couldnt load the firmware and usplash didnt work
[02:28] <BenC> usplash worked for me on my laptop and G4
[02:28] <BenC> is that the buildd build?
[02:29] <zul> my own build
[02:29] <BenC> from my git?
[02:29] <zul> my git
[02:29] <BenC> but I mean does it have all the stuff from my git tree?
[02:29] <zul> no wait...hmm...just a sec..
[02:29] <zul> yeah
[02:30] <zul> yours and mine
[02:35] <BenC> any messages in kern.log about unresolved symbols in any modules?
[07:40] <fabbione> hmm how do you cherry pick with git?
[08:51] <fabbione> mjg59: ping?
[12:23] <mjg59> fabbione: Hi?
[12:24] <fabbione> hey
[12:24] <fabbione> mjg59: i was looking at one commit Ben did on .12
[12:24] <fabbione> after breezy release
[12:24] <fabbione>   * external-arch-i386-kernel-reboot_reboot-thru-bios: Remove
[12:24] <fabbione> i assume this was coming from you
[12:24] <mjg59> Yeah
[12:24] <mjg59> We added a DMI patch to only do it on the machines that needed it
[12:25] <fabbione> is it a whishlist for dapper or something absolutely important for breezy?
[12:25] <mjg59> Some machines in Breezy don't reboot
[12:25] <mjg59> Dropping that patch would probably fix them
[12:25] <fabbione> ok so it is not even sure it will fix the problem, right?
[12:26] <mjg59> I don't have any of the hardware concerned
[12:26] <fabbione> ok
[12:26] <fabbione> than i will kill the change
[12:27] <fabbione> if it was sure and important i would have snicked in for -security
[12:27] <fabbione> but given the status i will keep it as it is from breezy
[12:27] <fabbione> there might be a -update for breezy anyway
[12:28] <Mithrandir> mjg59: iirc, Karianne's laptop (HP something) doesn't reboot cleanly, so I guess I can have her test it.
[12:29] <mjg59> Mithrandir: No, that one works now
[12:30] <Mithrandir> mjg59: oh, ok
[03:33] <zul> BenC: sorry my wife was in a pissy mood last night and decided to take it out on me
[03:33] <BenC> that's what wives are for
[03:34] <zul> yeah so uspalsh doesnt work for me on my laptop
[03:51] <BenC> any messages about vga16fb loading?
[03:55] <zul> didnt see..
[03:56] <zul> ill have to check tonight when my laptop is powered back up
[03:58] <BenC> ok
[04:17] <jbailey> Is the git in dapper useful for upstream kernel trees now?
[04:20] <zul> well it was announced on lkml
[04:22] <jbailey> Which it, sorry?
[04:23] <jbailey> I don't follow lkml all the time.
[04:25] <BenC> jbailey: 0.99.x is suggested
[04:25] <BenC> 0.99.8 is what I have
[04:25] <BenC> 0.99.9 something or other is latest
[04:25] <BenC> I don't use cogito though, so I can't answer any questions about that :)
[04:26] <jbailey> Oh?  What do you use/
[04:30] <BenC> just the git-* commands
[04:30] <jbailey> But not packaged?
[04:32] <BenC> fabbione: ping
[04:33] <BenC> jbailey: not sure, but I don't think so
[04:33] <jbailey> 'k
[04:34] <BenC> fabbione said he was going to do it
[04:34] <BenC> cogito contains the git commands, so maybe he was just going to update that
[04:35] <jbailey> fabbione: Didja, didja, didja?  huh huh huh?
[04:35] <jbailey> =)
[04:36] <jbailey> BenC: The linux-libc-header guys have been using a separate svn for their 'user abi' headers.  I'm going to try and push them into using git and also keep that tree available.
[04:36] <jbailey> Easier to merge the pieces that probably actually ought to be, and easier to track new changes.
[04:37] <BenC> yeah, would make things a lot easier
[04:58] <zul> jbailey: im trying to package it
[04:58] <jbailey> zul: Cool.  Do you need help?
[05:06] <jbailey> ROAR
[05:06] <jbailey> "GIT 0.99.9i aka 1.0rc2 is found at usual places."
[05:06] <jbailey> Is a stupid line to have at the top of an announcement mail.
[05:06] <fabbione> BenC: pong
[05:06] <BenC> fabbione: re the reboot patch, it was tested by users
[05:07] <fabbione> BenC: ok.. it will wait -updates if we will ever do one
[05:09] <zul> jbailey: not really its my first real debian package so its best that i learn how to do it and get you to look over it 
[05:09] <BenC> zul: did you use debmake? :)
[05:09] <zul> nope
[05:10] <fabbione> BenC: did you actually uploaded k-p too?
[05:10] <BenC> fabbione: yeah, and build-dep's on it
[05:10] <fabbione> coooool
[05:10] <fabbione> if i can get somebody to build a breezy ppc d-i for me
[05:10] <fabbione> i might even get my powerbook to install
[05:11] <BenC> I'm trying to convince my wife to let me dual-boot ubuntu on her g5 so I can some extra testing there
[05:11] <zul> BenC: i bought a new hard drive for my wife on condition that she dual boots
[05:11] <fabbione> BenC: you have a ppc or two, rigth?
[05:11] <BenC> I need to get an amd64 machine so I can test boot that...I hate just saying "it built"
[05:11] <jbailey> zul: Fair 'nuff.
[05:11] <jbailey> zul: Are you updating cogito or just doing git?
[05:11] <BenC> fabbione: two G4's (one my desktop) and a G5
[05:12] <fabbione> ah
[05:12] <fabbione> BenC: you just offered volunteer to do a d-i build for me
[05:13] <fabbione> because i can't believe in 36 hours into my ppc experience and i still can't install linux :)
[05:13] <BenC> you can't install breezy?
[05:13] <fabbione> BenC: are you running breezy?
[05:13] <fabbione> BenC: no
[05:13] <fabbione> apple did change PCI IDs on the latest powerbooks
[05:13] <BenC> yeah, breezy with dapper kernel (oooh, aaah, bad)
[05:13] <fabbione> so now on davis
[05:13] <fabbione> i have a .12 with the patch 
[05:14] <doko> BenC: can silo be built with something newer than gcc-2.95?
[05:14] <fabbione> the kernel and udebs are already built
[05:14] <fabbione> doko: silo is already building with gcc-4.0
[05:14] <fabbione> BenC: i need to build d-oi
[05:14] <BenC> doko: not last I checked
[05:14] <fabbione> d-i
[05:14] <fabbione> but davis has no B-D
[05:14] <fabbione> and nobody around that can install them
[05:15] <BenC> not sure I can do that in a reasonable time frame
[05:15] <fabbione> BenC: so would it be possible for you to build it for me?
[05:15] <doko> fabbione, BenC: *lol*
[05:15] <BenC> oh, it only breaks on sparc32 with newer gcc, so ubuntu can do that :)
[05:15] <fabbione> re: silo has been fixed to build with gcc-4.0 by David
[05:16] <BenC> fabbione: oh, right, I need to merge that patch into silo
[05:16] <fabbione> BenC: the d-i build takes like 15 minutes.. please
[05:16] <fabbione> i really really want to try to install this machine
[05:16] <doko> ok, thanks, one compiler less ...
[05:16] <BenC> fabbionne: it's not the build, it's all the stuff I have to download :)
[05:16] <fabbione> BenC: it's 3 udebs
[05:16] <BenC> really?
[05:17] <fabbione> yes
[05:17] <BenC> email me the specifics of what I need
[05:17] <fabbione> ok
[05:18] <BenC> I'm used to the old installer build I guess, where it needed a mostly complete base mirror
[05:20] <fabbione> nah
[05:20] <fabbione> it downloads some stuff from the net
[05:20] <fabbione> but not much
[05:22] <fabbione> BenC: you got mail
[05:23] <fabbione> dude if you can get that done asap, i am going to put one portrait of you in my personal "Heros of my life" section in the temple
[05:31] <zul> jbailey: just git
[05:35] <jbailey> zul: git itself appears to have debian packaging already in it.
[05:36] <BenC> fabbione: ok, as long as it's atleast an 8x10 portrait
[05:39] <fabbione> BenC: ehhehe
[05:43] <BenC> sweet, kernel-package is built, so 2.6.15-2.2 should follow soon
[05:44] <BenC> fabbione: at the very least, if I can get my e3k up this weekend, I can atleast do some builds for 2.6.15 kernels
[05:45] <BenC> the builds go pretty fast when you can through 6 cpu's and 6gigs of ram at it
[05:45] <BenC> s/through/throw/
[05:48] <fabbione> eheh i know
[05:49] <fabbione> i have been building security kernels on a 2 x dual core amd64 with 8GB of ram today
[05:49] <fabbione> pretty fast
[05:51] <zul> jbailey: yeah i know fabbione said it sucked
[05:52] <zul> fabbione: pretty fast?!?!
[05:52] <zul> thats a bit of an understatement
[05:52] <fabbione> as everything else that has a debian/ dir upstream
[05:52] <fabbione> zul: it did start swapping approx 512MB with make -j750
[05:52] <fabbione> = the sucks
[05:52] <fabbione> with -j500 was okish
[05:53] <zul> heh...can i give you my ssh key ;)
[05:53] <fabbione> give it to Mithrandir 
[05:54] <fabbione> it's his playtoy
[05:54] <zul> hmmm..
[05:55] <zul> im off to lunch
[05:55] <jbailey> Lunch sounds like a lovely idea.
[06:05] <zul> with meat ;)
[06:06] <fabbione> thai vegetarian ++
[06:06] <fabbione> jbailey: that resturant was really nice..
[06:07] <fabbione> i wouldn't mind to go there again
[06:43] <fabbione> BenC: you can stop build d-i (if you started)
[06:53] <BenC> you have it?
[06:53] <fabbione> yes
[06:53] <BenC> ok
[06:53] <fabbione> i got Znarl to install b-d on davis
[06:55] <BenC> ok, cool
[07:01] <jbailey> BenC: Got a moment for a stream of git questions?
[07:01] <BenC> sure
[07:02] <jbailey> I'm looking at putting lkh into git.  What I've noticed others have done is have a working branch and then a branch marked "for-linus" or some such for merging back in.
[07:02] <jbailey> Are those truly just branches in the same .git file?
[07:02] <jbailey> Or do they turn into multiple repositories that get merged back and forth?
[07:05] <BenC> same .git/
[07:06] <BenC> you can merge from head to a branch, and branch to head
[07:06] <BenC> all in the same repo
[07:06] <jbailey> Okay, cool.
[07:06] <BenC> so you can cherry-pick commits from head, into the branch, and do a pull request to linus, or whatever
[07:07] <jbailey> Okay.
[07:07] <jbailey> When doing commits, is it hard to go back afterwards and break things into better changesets?
[07:08] <jbailey> At this point I need to just get it done, but I'd like to go back afterwards and start to split things up into changes that I think would be accepts back upstream, and changes that haven't a hope of going in.
[07:08] <BenC> well, let's say you do a large commit, then later you want break it down into smaller ones...
[07:08] <jbailey> Right.
[07:08] <BenC> first you'd rebase the commit to the head (make the chageset against the latest code, instead of your older branch point)
[07:09] <BenC> then you could pull the diff, and revert the changeset
[07:09] <BenC> and re-apply into changesets as needed
[07:09] <BenC> it's not very straight forward, but it is possible
[07:09] <jbailey> Okay, so it does understand the concept of "take this changeset out".  Cool.
[07:09] <BenC> but if you have commits that depend on the large one, it gets ugly fast
[07:09] <BenC> yeah
[07:10] <jbailey> Since it's just futzing with include files, it shouldn't be too bad for now.
[07:10] <BenC> the commit still shows in the log, but the revert does too
[07:10] <jbailey> I'm sure they're probably used to that from folks new with the tool. =)
[07:11] <BenC> yeah, I have a few reverted commits in the ubuntu tree :)
[07:16] <jbailey> Argh, I think I'm still confused.
[07:16] <jbailey> So for my code that is never intended for Linux, I can just commit this to my checkout, yes?
[07:16] <jbailey> s/Linux/Linus
[07:17] <jbailey> Then when it comes time to hand changesets back, do I make a new branch then?
[07:19] <BenC> yes
[07:19] <BenC> you can make a branch at any time
[07:19] <BenC> best to pull things into the branch just after you check them into the head
[07:23] <zul> dont do what i did
[07:39] <jbailey> Actually.
[07:39] <jbailey> Does that include getting up this morning?
[07:39] <jbailey> I could've avoided that easily enough.
[07:54] <BenC> I tried to avoid it, but the rest of the house wasn't having it
[07:59] <dilinger> i managed to put it off for a few hours
[08:07] <dilinger> hm
[08:08] <dilinger> looks like we're skipping breezy on our desktop systems at work :/
[08:11] <BenC> going straight to dapper? :)
[08:11] <dilinger> probably
[08:12] <dilinger> the problem is, we're running sarge on our servers, and dealing w/ the differences between gcc-3.3 and gcc-4.0 is a pain
[08:12] <dilinger> w/ our in house software
[08:12] <dilinger> oh well
[08:19] <zul> jbailey: i didnt branch and that probably fucked things up
[08:20] <zul> actually i dont remember what i did
[08:21] <jbailey> dilinger: Is that going straight to dapper in 6 months, or is that, "Whee, let's ride the rollercoaster" on the way there?
[08:24] <dilinger> jbailey: ooh, rollercoaster?
[08:29] <jbailey> dilinger: Minor C++ abi transition, kernel update.  glibc update, you know.  The little things.
[08:29] <dilinger> oh
[08:29] <dilinger> going straight to dapper in 6mo
[08:29] <dilinger> sticking w/ the same gcc/libc6 for now
[08:30] <dilinger> my boss was actually talking about "breezy for sarge"
[08:30] <dilinger> recompiling most of breezy against the older gcc/libc6
[08:32] <jbailey> Might not work in the case of the C++ bits.
[08:32] <jbailey> Sounds like what you might want is a breezy server with a sarge chroot for now.
[08:32] <dilinger> yea
[08:32] <dilinger> i recommended the chroot
[08:32] <dilinger> but it's just easier to stick w/ hoary for now
[08:33] <dilinger> than to change how people work
[08:40] <jbailey> It's all good.
[08:41] <jbailey> The whole point of the long support cycle is to let you choose when to upgrade.
[08:41] <jbailey> You're only in trouble if you're still running warty and can't upgrade soonish.
[08:42] <jbailey> Do oldworld macs use PCI?  I thoguht they did.
[08:42] <jbailey> The only thing I think that the "SCSI drivers built into kernel" guy on u-d should need is to set his /etc/mkinitramfs/initramfs.conf to MODULES=dep
[08:42] <jbailey> infinity: ^^
[08:42] <jbailey> infinity: I'll let you take that so that I actually can start pretending like initramfs-tools isn't mine anymore. =)
[08:45] <dilinger> jbailey: the main problem is all the backporting i'm going to have to keep doing
[08:47] <jbailey> dilinger: Hmm.  The automatic backports and stuff aren't enabled for Hoary?
[08:47] <dilinger> there's automatic backports?
[08:48] <dilinger> i remember mako talking about the idea at UDU, but..
[08:49] <jbailey> I thought that it got done.  I might be mistaken, though.
[08:49] <jbailey> I know that there's supposed to be something crazy with backports going on/.
[09:03] <fabbione> dilinger: remembet also that we don't support hoary -> dapper upgrades. You might have to jump via breezy anyway
[09:42] <zul> later
[09:44] <dilinger> fabbione: given how badly the 2 hoary->breezy upgrades went for me, i'm not that concerned :)
[09:45] <fabbione> badly? ;)
[09:45] <dilinger> yea
[09:47] <fabbione> badly how?
[09:47] <dilinger> #19477, #19478
[09:47] <dilinger> that was for my upgrade
[09:47] <dilinger> a coworker did an upgrade that failed in various other places, but i couldn't file bugs because he needed his machine
[09:48] <fabbione> -ENOBUGZILLA atm
[09:49] <dilinger> misc packages that were missing conflicts/replaces
[09:50] <fabbione> from within main?
[09:50] <fabbione> or from universe?
[09:52] <dilinger> main, i believe
[09:53] <dilinger> i'd have fixed them myself if i had some sort of upload privileges; they seem like easy fixes
[09:53] <dilinger> instead i just fixed them in a local repository
[11:08] <BenC> mjg59: ping
[11:09] <BenC> mjg59: just filed a bug report on usplash, it needs to load softcursor.ko first to work with 2.6.15, and the module path changed
[11:13] <dilinger> ok
[11:30] <infinity> BenC : It does?... I just upgraded to 2.6.15 here, and usplash worked.
[11:31] <infinity> BenC : Or is this PPC specific, perchance?
[11:31] <BenC> no, this was i386
[11:31] <BenC> it may "work", but it's broken early on
[11:31] <infinity> BenC : Also, if you're not yet aweare, the ipw2200 driver is looking for firmware named 'ipw-2.4-*', but we're shipping 'ipw-2.3-*'... A quick rename and my wireless worked again.
[11:32] <BenC> the first load attempt in the usplash initramfs script doesn't work because fbcon.ko needs softcursor.ko
[11:32] <infinity> \o/
[11:32] <BenC> infinity: yeah, that's next on my list
[11:32] <infinity> I'll look into the usplash issue.
[11:32] <infinity> Was this with vga16fb?
[11:32] <infinity> Maybe I dodged the bullet with vesafb.
[11:33] <BenC> doesn't matter if it's vga16fb or vesa
[11:33] <infinity> Oh, also, damn you.  You didn't switch to the New World Order of update-initramfs in your postinst.
[11:33] <BenC> I tried both
[11:33] <infinity> But we can fix that on your next ABI bump.
[11:33] <BenC> heh, yeah, I forgot :)
[11:33] <BenC> almost every upload between now and near 2.6.15 release will be an ABI bump
[11:34] <infinity> I assumed as much.
[11:34] <BenC> but other than ipw2200 it worked ok for you?
[11:34] <infinity> We'll have a ridiculously high ABI magic number in the package names.
[11:34] <infinity> Oh well.
[11:34] <infinity> Yeah, it's working okay right now.
[11:34] <infinity> Too early to say it's working "well", but it hasn't crashed yet.
[11:34] <BenC> it atleast booted, so that's promising

[11:35] <infinity> I guess today will have to be LRM day for me.
[11:35] <BenC> inifinty: oh, and damn you, I've been watching family guy every night since I got back, trying ot watch them all :)
[11:35] <infinity> Especially if you want to update the linux-meta package.
[11:35] <BenC> productivity killer
[11:35] <infinity> Family Guy is good for the soul.
[11:35] <BenC> yeah, that would be nice
[11:35] <BenC> I have a machine that needs nvidia
[11:36] <infinity> Legacy or newskool?
[11:36] <infinity> I can always go for some testing.
[11:36] <BenC> fairly new
[11:36] <infinity> Zofia's machine is a 6800GT, so it can use both drivers.  But I'd love a "legacy-only" machine to get some testing.
[11:36] <BenC> 0000:02:04.0 VGA compatible controller: nVidia Corporation NV18 [GeForce4 MX 4000 AGP 8x]  (rev c1)
[11:37] <BenC> not sure what's considered legacy, but that may be it
[11:37] <BenC> it's a mame machine, so I can give it some good testing :)
[11:37] <infinity> Not quite legacy yet, no.
[11:37] <Mithrandir> nope, I think you need to have GF2MX or older.
[11:37] <infinity> Might be in another year.
[11:38] <infinity> But your machine can run both drivers, so you can do some testing of that setup for me.
[11:38] <infinity> I wonder if I'm going to have to wait for jbailey's glibc/lkh upload, followed by the new X building correctly, before I can update the ATI shit to something shiny and new.
[11:39] <infinity> Oh well, I can do nvidia at any rate, and build for the new kernel.
[11:41] <BenC> definitely, I'll test whatever I can
[11:45] <infinity> Hrm, core hours start in 15 minutes.  Guess I should find some breakfats before I start working for real.
[11:45] <infinity> Or, some breakfast.
[11:46] <dilinger> nice fatty bacon? :P
[11:48] <infinity> Or some fat breaks.
[11:50] <dilinger> are those like beats?
[11:52] <infinity> Well, like breakbeats, sure.