[12:10] <jbailey> Hey - post Hoary, what compiler version should be used for the kernel?  Have folks moved to 3.4/4.0 yet?
[01:07] <dilinger> oh, you guys might be interested in this also
[01:07] <dilinger> http://wiki.debian.net/?KernelFirmwareLicensing
[01:07] <dilinger> i know you're also stripping out stuff like qla22xx
[01:08] <dilinger> looks like a few of them can be put back in, so long as you don't mind the firmware licenses being non-DSFG (of which numerous other firmware bits in the kernel aren't)
[01:10] <dilinger> jbailey: as far as how simple it would be to implement, i'd need to go through kernel-package a bit more closely
[01:10] <dilinger> and as far as the compiler, i recall people talking about using 4.0, but 4.0 is known to miscompile parts of the kernel
[01:10] <dilinger> (haven't actually tried 4.0 myself, yet)
[01:11] <jbailey> Ah, okay.  doko and I are writing the toolchain transition document, and need to figure out what compiler has to be around for the kernel.
[01:11] <jbailey> is 3.4 generally considered to work?
[01:30] <dilinger> you can probably assume whatever the latest released version of gcc will work ok
[01:31] <dilinger> if it doesn't, people will probably fix it quickly
[05:05] <zul> hey
[05:17] <fabbione> morning
[05:18] <fabbione> oh zul already left...
[05:18] <fabbione> ehhehe
[05:18] <fabbione> i know why his debian dir is blown away :)
[05:18] <fabbione> that's why added the point d) i leave this up to you :)
[05:18] <fabbione> there is the trick ;)
[05:19] <lamont> lol
[05:22] <fabbione> lamont: there is only file that is patched in the diff.gz
[05:22] <fabbione> exactly to avoid that hehhe
[05:24] <lamont> you evil man you
[05:25] <fabbione> well I got caught in the same problem when i did 2.6.9 :)
[05:35] <fabbione> lamont: so. do you know if he prepared the .orig.tar.gz?
[05:38] <lamont> dunnop
[05:39] <lamont> had just gotten home about the time he went to bed
[05:40] <fabbione> ehe ok
[05:40] <lamont> eth0: no IPv6 routers present
[05:40] <lamont> gonna have to do something about that one of these days
[05:40] <lamont> :)
[05:40] <fabbione> do you remember what is the dpkg-buildpackage option to override the distro?
[05:41] <lamont> does 2.6.12 have related in ip6tables, I wonder?
[05:41] <fabbione> so that dpkg-genchanges will report hoary
[05:41] <fabbione> lamont: not that i know off, but we will make it part of it
[05:41] <lamont> -D
[05:41] <fabbione> -Dhoary ?
[05:41] <lamont> yeah
[05:41] <lamont> see people.ubuntu.com/~lamont/uch
[05:42] <lamont> works hard to do the right thing for both native ubuntu, and debian packages
[05:42] <lamont> with uch -i support
[05:43] <fabbione> oh that wasn't it
[05:43] <fabbione> i lost libcairo because it was uploaded to ubuntu with the wrong distro
[05:44] <fabbione> i only need to rebuild without messing to much
[05:44] <fabbione> but i guess i can just edit the changes before signing them
[05:44] <lamont> how the hell did it get the wrong distro?  sbuild -dhoary would give you a hoary changes file...
[05:45] <fabbione> lamont: probably bvecause i builded it manually
[05:46] <fabbione> i try to fork buildd when it goes to gcc & co since they take really a long time
[05:46] <fabbione> and it must have been a Build-Dep i needed on the fly
[05:46] <fabbione> still gcc-4.0 has bad Build-Dep
[05:46] <fabbione> it should be versioned
[05:47] <fabbione> and i told doko
[09:59] <dilinger> hm
[11:15] <dholbach> heya
[11:15] <dilinger> fabbione: long live baz? :)
[11:15] <fabbione> ehhee
[11:24] <dholbach> i have to ask about the universe kernels again - don't want to do anything wrong
[11:25] <dholbach> i made up a list of packages to sync from debian for universe. would alsa-modules-i386, kernel-image-2.4.27-i386, kernel-patch-powerpc-2.4.27, kernel-source-2.4.27, pcmcia-modules-2.4.27-i386 -- kernel-image-2.6.10-amd64, kernel-image-2.6.10-i386, kernel-image-2.6.10-ia64, kernel-patch-powerpc-2.6.10, kernel-source-2.6.10 be ok? and what about kernel-image-2.4.27-ia64, kernel-source-2.6.11?
[11:25] <fabbione> dholbach: dude... SNMP :)
[11:26] <dholbach> *cry*
[11:26] <dholbach> i'm absolutely happy with the main kernels too
[11:26] <dholbach> i just want to have the universe kernel stuff SOMEHOW under control
[11:27] <dholbach> i'd be happy to chuck everything with *kernel* in the name completely out
[11:27] <dilinger> why include the universe kernels at all?
[11:27] <fabbione> dilinger: 2.4
[11:27] <dholbach> we'll need 2.4 - it was fabbione's wish! so fabbione: it IS your F.CKING problem
[11:27] <dilinger> ah
[11:27] <dholbach> so please advise me
[11:27] <dholbach> :-)
[11:27] <fabbione> dholbach: skip 2.6 please
[11:27] <fabbione> three is no need to
[11:28] <dholbach> fabbione: zul told me to include them
[11:28] <fabbione> and keep the minumum for 2.4
[11:28] <fabbione> aka no external modules like alsa
[11:28] <fabbione> dholbach: why?
[11:28] <dholbach> *shrug*
[11:28] <dholbach> what about pcmcia, alsa? ia64 as well?
[11:29] <fabbione> no pcmcia, no alsa.. ia64 ask T-Bone
[11:29] <fabbione> only a bare kernel
[11:29] <dholbach> ok, so kernel-image-2.4.27-i386, kernel-patch-powerpc-2.4.27, kernel-source-2.4.27
[11:30] <dholbach> nice, thanks
[11:30] <fabbione> we want to provide a minimum 2.4
[11:30] <fabbione> that's all the user need
[11:30] <dholbach> and i'll get back to T-None for 2.4-ia64
[11:30] <fabbione> otherwise it's their problem
[11:30] <dholbach> yeah
[11:30] <dholbach> i'll have a brief look over the kernel-patch-CRACK packages as well
[11:31] <dholbach> maybe we can get rid of them in one swish as well
[11:36] <dholbach> thanks fabbione 
[11:38] <fabbione> no problem
[11:38] <fabbione> or better. SNMP
[01:30] <jbailey> fabbione: Mmm.  Will you be really upset if the Breezy glibc doesn't run on 2.4?
[01:30] <fabbione> jbailey: not at all
[01:30] <jbailey> And any applications compiled with it...
[01:30] <fabbione> i am ok to kill 2.4 in breezy
[01:30] <jbailey> Cool. =)
[01:31] <fabbione> at that point they had 1 year to convert
[01:31] <fabbione> that is a reasonable amount of time
[01:31] <jbailey> doko and I were hashing out the breezy toolchain plan last night, and I've proposed making nptl the default set of threading headers.  If we're doing that, we may as well the set the minimum kernel version at 2.6.8 or so.
[01:32] <dholbach> jbailey: you know what will happen to gcc-2.95?
[01:32] <dholbach> jbailey: i'd love to chuck it out at some stage
[01:32] <dholbach> jbailey: but there are unfortunately packages that hard depend on it
[01:33] <jbailey> dholbach: Ask the kernel folks here.  I assume that gcc-2.5 is only kept around for them.
[01:33] <fabbione> dholbach: afaik i know the only thing that still requires (officially) gcc 2.95 is silo
[01:33] <fabbione> that i am anyway compiling with gcc-3.3 and seems to work
[01:33] <jbailey> silo does?  Ugh.
[01:33] <fabbione> jbailey: there is a note about compiling silo with more recent compilers...
[01:33] <jbailey> Maybe a nice summer project will be to port grub2 to sparc.
[01:33] <dholbach> i'm currently working on MorgueCandidates and checking all the reverse depends
[01:33] <fabbione> jbailey: if you want to blow away some dust from that code.. 
[01:33] <fabbione> jbailey: have fun.. really :)
[01:34] <fabbione> i am already scared enough to change bootloader on i386
[01:34] <jbailey> fabbione: 'kay.  I'd have to think about that from a copyright point of view.  If I dig into silo internals, I'm not clean for hacking grub2 anymore, although I could work with someone else to do it.
[01:35] <fabbione> jbailey: wtf????????
[01:35] <fabbione> is that somekind of retarded licence?
[01:35] <jbailey> No, it's the recommended steps for GNU maintainers on keeping copyrights clean.
[01:35] <jbailey> Not the license at all, but the code ownership.
[01:35] <dilinger> fabbione: i dunno about requiring gcc-2.95.  the theory is that recompiling it w/ gcc-2.95 reduces bugs
[01:36] <dilinger> fabbione: i'm not convinced it actually does, as i see the same bugs regardless of compiler
[01:36] <fabbione> dilinger: the problem with silo is slightly different
[01:36] <fabbione> jbailey: i will need to translate the Italian Bible of Code Obfuscation
[01:37] <fabbione> and hand a copy over to you
[01:37] <jbailey> *lol*
[01:38] <fabbione> well i think BenC wrote silo
[01:38] <fabbione> you could simply ask him
[01:38] <jbailey> fabbione: It's usually not that bad.  Mostly it's a matter of someone reading the code and then explaining in English how it works in a document.  Then someone else can take the very detailed english document and turn it into code.
[01:39] <jbailey> The final implementation will vary enough to be clean.
[01:39] <jbailey> And as a side effect, there's documentation. =)
[01:39] <dilinger> vil
[01:39] <dilinger> evil
[01:39] <fabbione> right.. but if BenC allows copy of the code...
[01:39] <fabbione> it will be faster
[01:40] <fabbione> also because BenC would have to write the documentation anyway
[01:40] <Mithrandir> jbailey: uhm, but silo is GPL?
[01:40] <Mithrandir> oh, code ownership, as you say.
[01:41] <jbailey> fabbione: True, but my luck getting anything useful out of benc.. (Like, say, even a reply by email) is not high.
[01:41] <jbailey> He'll even drop out of an email conversation..  And not reply for 6 months.
[01:41] <jbailey> Mithrandir: Yeah.  It's the part that sucks about GNU hacking, but has the nice result that when the lawyers come knocking, it's not at my door.
[01:42] <fabbione> jbailey: ok, i can try to contact him, since i know him for reasons outside Debian 
[01:42] <Mithrandir> jbailey: so you shouldn't ever look at non-FSF code if you ever want to write something similar?
[01:43] <dilinger> jbailey: yea, but how often do the lawyers come knocking?
[01:43] <jbailey> Mithrandir: Ideally, yes.  Or you need to radically alter the implementation.  The examples they give are things like if it's a two pass parser implementation, do it in one.  If they do all of the work in core, do it as a stream.
[01:43] <Mithrandir> jbailey: well, that can be just the shell, the algorithms may still be the same.
[01:44] <Mithrandir> though, I'm not sure you can actually have copyright on such small parts as "a linked list implementation", since it's not new in any way.
[01:44] <jbailey> fabbione: Cool.  If he's interested, I can ask for a disclaimer to be sent to him.  So just something saying that if we copy the stuff, it's not a big deal.  Weaker than a copyright assignment.
[01:44] <fabbione> jbailey: sure..
[01:45] <fabbione> i can try.. and let see..
[01:45] <jbailey> Mithrandir: Right.  In Canada and the US, you can't copyright the algorithm anyway.  It's just a way of making sure the code is sufficiently different.
[01:45] <jbailey> Mithrandir: I try not to think about European law too much, it's fundamentally different in enough ways that I don't pretend to know it.
[01:45] <Mithrandir> jbailey: heh, :)
[01:46] <fabbione> jbailey: do you remember his NON Debian email address?
[01:46] <fabbione> i think it was something like phunnyfarm.org?
[01:47] <jbailey> That sounds familiar.  /me tries to think if he's subscribed to a list where benc would've posted to.
[01:47] <jbailey> ISTR he lists his email addys in his .sig
[01:47] <fabbione> debian-sparc?
[01:48] <jbailey> I checked -private, and they're weblinks in his .sig.
[01:48] <fabbione> here is
[01:48] <fabbione> Sender: Ben Collins <bmc@phunnypharm.org>
[01:49] <fabbione> i was pretty close :)
[01:50] <fabbione> jbailey: ok mailed...
[01:50] <fabbione> let see if he answer back
[02:03] <zul> hey
[02:03] <fabbione> hey zul
[02:03] <zul> fabbione: :P
[02:03] <fabbione> 2.6.11.90 building here now...
[02:03] <zul> good good..
[02:03] <fabbione> zul: btw.. read the chan logs from this morning
[02:03] <fabbione> i arrived a few minutes after you left :)
[02:03] <zul> yeah i saw that im missing step 3 :)
[02:04] <fabbione> ehhee
[02:04] <zul> bastard
[02:04] <fabbione> btw.. bk is dead forever
[02:04] <zul> :)
[02:04] <zul> huh?
[02:04] <fabbione> no.. i told you: I will let you figure it out
[02:05] <zul> why is bk dead?
[02:05] <fabbione> http://kerneltrap.org/node/4966
[02:05] <dholbach> does our kernel come with kernel-patch-bluez applied?
[02:05] <fabbione> dholbach: 2.6 has bluez and we also added some more recent snapshot
[02:05] <dholbach> fabbione: thanks, will chuck out
[02:09] <zul> heh...i wake up and bitkeeper still sucks ;)
[02:10] <fabbione> eheh
[02:11] <zul> heh...all i get on my gentoo email is spam
[02:13] <Mithrandir> zul: same goes for Debian mails.  Spam and bug reports. :P
[02:14] <zul> i know..
[02:14] <zul> i just like to bitch about stuff :)
[02:40] <fabbione> amen.. building the last flavour...
[02:41] <zul> cool...then do the monkey dance :)
[03:07] <fabbione> lamont: ehhee
[03:07] <fabbione> later
[03:07] <lamont> about an hour or so
[03:07] <lamont> can't arrive on the neighbor's doorstep this early in the day...
[03:08] <fabbione> ehhehe
[03:08] <fabbione> i am going for a nap too
[03:10] <zul> lazy sods :)
[03:51] <fabbione> bah
[03:51] <fabbione> couldn't sleep
[03:52] <zul> you had dreams of pointers and arrays in your head right
[03:52] <fabbione> i just figured how to steal 3 IP addresses from my ISP instead :)
[03:52] <zul> im telling
[03:53] <Lathiat> fabbione: h aha
[03:53] <Lathiat> fabbione: the isps here notice when you do that
[03:53] <Lathiat> fabbione: riend of mine stoll half a class C once :)
[03:53] <fabbione> well basically they have an offer for a /30
[03:54] <fabbione> but they assume to configure that on the router eth
[03:54] <fabbione> that means that they leave to you only one ip
[03:54] <fabbione> wrong...
[03:54] <fabbione> i can reconfigure the router to use a pvt class on the eth
[03:54] <fabbione> route the /30 to the server
[03:54] <fabbione> and route each single ip to different machines in terms of /32
[03:54] <fabbione> bang
[03:55] <fabbione> 4 ips instead of 1
[03:55] <fabbione> traceroute would skip on hop...
[03:55] <fabbione> tough luck
[03:55] <Lathiat> just run a iptables queue script to fake the extra hop :)
[04:03] <fabbione> jbailey: i got an answer back from Ben
[04:04] <jbailey> fabbione: w00t!
[04:05] <jbailey> fabbione: and, and, and....? =)
[04:05] <fabbione> jbailey: and now i am explaining him the problem.. it was just a "hey dude.. are you still alive?"
[04:06] <jbailey> I'm guessing the answer was yes, rather than an automated "I'm spending a year dead for tax purposes". =)
[04:12] <fabbione> second mail with the info requests on the way
[04:14] <dilinger> alright, davem doesn't get his isos until he makes 280 support stop sucking
[04:14] <dilinger> bleh, ECHAN.  too many #*-kernel
[04:14] <zul> there is only 2 you have to worry about...well one really ;)
[04:17] <fabbione> dilinger: i am surprised nobody started a flame on LKML yet
[04:17] <fabbione> are they talking on irc?
[04:18] <zul> for bitkeeper?
[04:18] <fabbione> yeah
[04:18] <zul> heh i think everything has been said before ;)
[04:20] <dilinger> i think everyone's quietly celebrating ;)
[04:20] <dilinger> except linus & co
[04:21] <dilinger> hm, i should sleep, i need to be up in like 7 hours
[04:22] <fabbione> eheh good night
[04:26] <dilinger> night
[05:38] <zul> fabbione: at least the flames are going on slashdot
[05:38] <fabbione> zul: i don't really read /.
[05:38] <fabbione> it's too full of lusers
[05:39] <zul> hehe...well you wanted a flame war ;)
[08:15] (Mithrandir/#ubuntu-kernel) lamont_r: the local star cluster?
[08:15] (Mithrandir/#ubuntu-kernel) as in, galaxy cluster
[08:15] (lamont_r/#ubuntu-kernel) yeah... by itself, cluster tends to bring to my mind the military/fire term....
[08:16] (lamont_r/#ubuntu-kernel) but like dholbach  says, unlikely to happen anyway
[08:16] <lamont_r> that's the beauty of derivatives - they can have their own archive... :-)
[08:17] <Mithrandir> so what should ubuntu name the maintained parts of universe?  Mars?
[08:17] <dholbach> lamont_r: yeah i look forward to Ancient-buntu, they can have a lot of stuff out of universe
[08:19] <lamont_r> dholbach: yeah - we really need to get through a policy that packages that haven't been uploaded in 6 years really need to justify their continued existance, or they at least move into the 'blackhole' component
[08:20] <lamont_r> supernova?
[08:22] <zul> hey fabbione 
[08:24] <fabbione> yo
[08:39] <T-Bone> lamont_r: that could be an idea. Yet we may need an answer quickly for hoary :)
[08:41] <lamont_r> T-Bone: I believe 6 years would just about start to apply to 2.0 kernels now...
[08:41] <T-Bone> lamont_r: lol
[08:42] <T-Bone> maybe an absolute timeline isn't good. Some packages might be tagged "DEPRECATED" in other and more obvious ways
[08:43] <lamont_r> T-Bone: the question really is, what (if any) quality standards are the MOTU allowed to require for packages to remain in universe.  And that's an MOTU/sabdfl question, not a kernel-team question
[08:43] <T-Bone> lamont_r: true
[08:43] <lamont_r> although the answer does seem to be that dead/dying packages can be moved to the morgue
[08:43] <T-Bone> lamont_r: but then, kernel packages shouldn't be discriminated
[08:44] <T-Bone> depending on said QA, all or none should be removed
[08:44] <T-Bone> lamont_r: as long as you don't consider kernel as something special, which i do, btw
[08:45] <T-Bone> imo, kernel packages are very distribution specific, and since we have our own kernels, i don't think we should ship debian ones on the ground we're shipping all that debian has. This is confusing to users. Some might chose to use these packages which we don't support, end up with security issues and the like, and blame us for it
[08:47] <T-Bone> lamont_r: *that* question should be sorted out among us before poking sabdfl about it, imho
[08:49] <jbailey> lamont_r: The 2.0 kernel maintainer is a DD, I should suggest that he upload an update just to reset the counter ;)
[08:49] <T-Bone> jbailey: lol
[08:51] <zul> yay...i had to help up setting cvxs
[08:51] <zul> er...cvs
[08:52] <zul> the horrors i have seen.
[08:53] <lamont_r> jbailey: motu did an upload (I think), speciifcally to fix a katie reject because there was a file in the .deb with a 1980 timestamp or so
[08:53] <zul> heheh
[08:54] <jbailey> But..  But..
[08:54] <jbailey> I'm sure it was just some dark and dustry corner of emacs...
[09:43] <dholbach> fabbione: the powerpc-2.4-kernel failed to build - i'm not sure what to do about it...
[10:02] <dholbach> T-Bone: fire away - anything not listed on wiki.ubuntu.com/UniverseUnmetDeps?
[10:02] <T-Bone> lemme look
[10:03] <dholbach> althought the kernel-stuff should be out, after elmo purged
[10:03] <T-Bone> dholbach: hmm, i'm not talking about that
[10:03] <T-Bone> i'm talking about *build* dependencies actually
[10:03] <dholbach> ahhhhh
[10:03] <dholbach> yes
[10:03] <dholbach> sorry
[10:04] <T-Bone> eg, i have a xmms pluggin trying to build itself without debhelper
[10:04] <dholbach> oh nice
[10:04] <T-Bone> gives you the very funny failure "dpkg-buildpackage: command not found"
[10:04] <dholbach> erm... which one?
[10:04] <dholbach> i'd be delighted to fix it
[10:04] <T-Bone> lemme look
[10:04] <dholbach> we're just composing UniverseLastMinuteFixes
[10:05] <dholbach> so if you have some ultra-urgent stuff, tell us
[10:06] <dholbach> haha: dietlibc has libcruft/ - nice :-)
[10:07] <T-Bone> ------------------------------------------------------------------------------
[10:07] <T-Bone> dpkg-source: extracting xmms-infopipe in xmms-infopipe-1.3
[10:07] <T-Bone> sh: line 0: exec: dpkg-buildpackage: not found
[10:07] <T-Bone> dholbach: ^^^ that one :)
[10:07] <dholbach> nice
[10:07] <dholbach> will look into it
[10:07] <T-Bone> dholbach: http://buildd.slashdirt.org/logs/previous/k2000/xmms-infopipe_1.3-4_20050402-0734
[10:08] <T-Bone> dholbach: i have loads of issues that can probably be easily fixed, yet i don't know whether they affect only hppa or not. It seems that a good half of them should affect all archs, but i haven't found time to look on other builders' logs
[10:09] <dholbach> i'd like to get the ultra-urgent ones listed somewhere
[10:09] <T-Bone> dholbach: when breezy opens, you might want to educate me about the proper way to fix them, eg either simply uploading the fix or sending it to whoever care. I don't mind doing a bunch of simple uploads if that can help you guys
[10:09] <dholbach> so people know where to head first, when trying to help out
[10:10] <T-Bone> well i can't really tell you what's ultra urgent and what's not, since i'm running hppa autobuilders, and hppa isn't supported (yet) :)
[10:10] <dholbach> :-)
[10:10] <dholbach> package-wise
[10:10] <dholbach> stuff out of universe you need every day :-)
[10:11] <T-Bone> heh
[10:11] <T-Bone> well, unless lamont sends me his nice parsing scripts, i have limited pre-parsing done at http://buildd.slashdirt.org/*html
[10:12] <T-Bone> lemme clean these a bit so that they only show the last round of builds
[10:15] <dholbach> T-Bone: build-dep added
[10:16] <T-Bone> hehe
[10:16] <T-Bone> you're going to spend the night here if i start listing them all :)