[06:50] <fabbione> morning
[06:51] <fabbione> lamont: you around by any chance?
[07:17] <fabbione> jbailey: ping?
[07:34] <fabbione> Inconsistency detected by ld.so: rtld.c: 1075: dl_main: Assertion `_rtld_local._dl_rtld_map.l_libname' failed!
[07:34] <fabbione> AMEN
[07:50] <fabbione> JaneW morning 
[08:00] <JaneW> morning fabbione - good week-end?
[08:00] <fabbione> JaneW: it was ok... nothing spectacular.. and you?
[08:08] <JaneW> fabbione: ok, mine was great thanks - didn;t touch a computer once ;)
[08:09] <fabbione> ehhe i almost didn't :)
[12:20] <TheMuso> 
[12:24] <chmj> ?
[12:29] <TheMuso> apologies. Flicking through Irssi windows and bumped some other keys. :(
[03:46] <lamont> fabbione: ack
[05:08] <fabbione> lamont: i need kernel-wedge 2 installed in the hppa chroot
[05:08] <fabbione> i am going to do some heavy build orgies tomorrow
[05:23] <mx|gone> who takes care of the linux-wlan-ng patch?
[05:26] <lamont> fabbione: will a simple dist-upgrade do the trick?
[05:27] <lamont> 119 upgraded, 11 newly installed, 0 to remove and 6 not upgraded.
[05:27] <fabbione> lamont: if you have enough breezy pkgs yes.. but you can just install that specifically.. it's a _all.deb
[05:27] <fabbione> mx|gone: it depends...
[05:27] <lamont> The following packages have been kept back:
[05:27] <lamont>   apt dselect html2text libosp4 sp tetex-bin
[05:28] <mx|gone> fabbione: who would I ask to update the patch to a recent svn snapshot?
[05:28] <fabbione> mx|gone: no svn snapshots.. only releases..
[05:28] <lamont> hrm... ew.
[05:28] <mx|gone> fabbione: damn... the last release has some problems that are fixed in the snapshot...
[05:28] <lamont> what gcc are we using to build the kernel?
[05:29] <fabbione> mx|gone: but usually you file an "enanchement" bug
[05:29] <fabbione> lamont: gcc-3.4
[05:29] <lamont> dpkg -l gcc-3.4
[05:29] <lamont> No packages found matching gcc-3.4.
[05:29] <lamont> hrm... gonna fix that
[05:29] <lamont> the only thing bothering me is that the gcc-3.4 we have generally available for hppa is from hoary (3.4.3-9ubuntu3)
[05:29] <fabbione> well for hppa there is gcc-3.4-hppa or something..
[05:30] <lamont> so don't do any user-space builds. :-)
[05:30] <fabbione> i am not going to :)
[05:30] <fabbione> i need to build the kernel
[05:31] <lamont> devel/linux-source-2.6.12_2.6.12-3.3: Uploaded by buildd+bld-18 [optional:out-of-date] 
[05:31] <lamont> but seems to need NEW love
[05:31] <mx|gone> fabbione: ok, later today I'm going to submit a patch to the linux-wlan-ng guys that will make it work with iwconfig completely... if that gets accepted, would you guys still need a release?
[05:31] <fabbione> mx|gone: yes
[05:31] <mx|gone> suck, ok
[05:33] <lamont> fabbione: dist-upgrading now
[05:34] <fabbione> lamont: no rush.. i am done for the day
[05:34] <fabbione> and i will crash it tomorrow :)
[05:35] <lamont> hehe
[05:35] <lamont> dist-upgrade included kernel-wedge_2.05ubuntu1, iirc
[05:35] <fabbione> you probably want to warn your friends that i am going to push the cluster a bit :P
[05:35] <fabbione> just in case nagios will make the machines look redish ;)
[05:38] <fabbione> jbailey: i have been doing more thinking about initramfs modules shipped directly by our kernel
[05:38] <fabbione> jbailey: i don't think it's a good idea to do it
[05:38] <fabbione> or better.. there are some cons i didn't think about
[05:39] <fabbione> like enlarging the package of N MB
[05:39] <fabbione> that's seems to be quite a lot
[05:40] <fabbione> in order to make these initramfs files we will need to change make-kpkg...
[05:40] <fabbione> becuase at least custom compiled kernels should have the same treatment
[05:40] <fabbione> but than you hit a bigger issue
[05:41] <fabbione> what if driver foo in our kernel is either compiled in or not available in custom kernel?
[05:41] <fabbione> what the initramfs creation tool is going to do in that case?
[05:41] <fabbione> (since we feed it with some lists)
[05:46] <fabbione> jbailey: i agree with you, but there are some cases i am thinking of custom kernels where the user actually might want to be able to add drivers to the initramfs
[05:46] <fabbione> we need to try to make these operations simpler
[05:46] <fabbione> instead of more complicated
[05:47] <jbailey> Right.
[05:47] <jbailey> Adding drivers to the initramfs is as easy as adding modules.
[05:47] <jbailey> err
[05:47] <jbailey> names to /etc/mkinitramfs/modules
[05:47] <jbailey> But it doesn't solve the annoyance of having to add to make-kpkg
[05:48] <fabbione> exactly
[05:48] <fabbione> i am really not 100% sure what's the best in the long run, but i think for breezy it's better we keep initramfs as close as possible to initrd
[05:49] <fabbione> it will give people time to look at it and see 1) what problems it has 2) how it can evolve
[05:49] <fabbione> 3) and clearly how much better it is compared to initrd ;)
[05:54] <lamont> jbailey: in other news. ../net.c:948: error: field `nl' has incomplete type
[05:54] <lamont>  on at least 2 architectures
[05:54] <fabbione> same here :)
[05:54] <fabbione> lamont: what package?
[05:54] <fabbione> i remember i have seen it... i don't remember where
[05:54] <lamont> strace
[05:54] <lamont> and others
[05:54] <fabbione> yeah
[05:55] <fabbione> ../net.c: In function 'printsock':
[05:55] <fabbione> ../net.c:948: error: field 'nl' has incomplete type
[05:55] <fabbione> ../net.c:950: confused by earlier errors, bailing out
[05:55] <fabbione> same on sparc
[05:55] <fabbione> lamont: do you know if elmo did setup logs@ ???
[05:55] <lamont> and i386, ppc, and ia64.  not sure what the story is with amd64...  guess I could go look
[05:55] <jbailey> lamont: Sorry, which package?
[05:55] <fabbione> strace
[05:56] <lamont> ah, building on crested, which is down...
[05:56] <lamont> that could explain it
[05:56] <fabbione> lamont: amd64 is fucked for other reasons too
[05:56] <lamont>  /usr/include/resolv.h:110: error: field `nsaddr_list' has incomplete type
[05:56] <fabbione> (see ocfs2-tools FTBFS)
[05:56] <jbailey> lamont: I fixed hfsplus, haven't done the other two yet.
[05:56] <lamont> that's iputils, fwiw
[05:56] <fabbione> it's also missing iptables
[05:56] <fabbione> and iproute
[05:56] <lamont> iptables has need of some love, iirc
[05:57] <lamont>  /usr/include/linux/config.h:1:2: #error "Compilation aborted. Please read the FAQ for linux-libc-headers package."
[05:57] <lamont>  /usr/include/linux/config.h:2:2: #error "(can be found at http://ep09.pld-linux.org/~mmazur/linux-libc-headers/doc/)"
[05:57] <fabbione> i have seen a bunch of those in universe too
[05:57] <jbailey> Right, those are all broken applications.
[05:58] <jbailey> I'll fix iptables, and do up a FAQ sheet for the universe bits.
[05:58] <lamont> woot
[05:58] <jbailey> Including config.h is bad.
[05:59] <fabbione> including kernel headers in apps is BAD
[05:59] <fabbione> full.
[05:59] <jbailey> Mmm..  I only half agree.
[05:59] <jbailey> If you're using kernel facilities directly, using kernel headers makes sense.
[05:59] <jbailey> If only the kernel would provide a decent set.
[05:59] <fabbione> jbailey: linus doesn't agree with you :)
[05:59] <fabbione> even for that half
[06:00] <fabbione> and i tend to agree with him :)
[06:00] <jbailey> Right.  Linus has said that there's no promises in the userspace to kernel space ABI, which is a bit frightening.
[06:00] <fabbione> exactly.. kernel headers changes a lot
[06:00] <fabbione> so it might build today
[06:00] <fabbione> tomorrow is broken again
[06:01] <jbailey> Should - who should provide the headers then?
[06:01] <jbailey> Certainly not glibc, it's not providing the facility.
[06:01] <fabbione> glibc ?
[06:01] <jbailey> Also, that doesn't allow for systems using other C libraries.
[06:02] <jbailey> The only thing might be if glibc were to provide a wrapper for every syscall that iptables, iproute and friends want to call.
[06:02] <fabbione> anyway.. it's time to go and start preparing dinner
[06:02] <jbailey> But that's a bit excessive.
[06:02] <jbailey> Enjoy your dinner. =)
[06:02] <fabbione> jbailey: tbh this is one of the few things i don't really care...
[06:03] <fabbione> given that different userland upstreams can keep up with the kernel..
[06:03] <fabbione> when they break i blame upstream for userlend
[06:03] <fabbione> land even
[06:03] <jbailey> fabbione: Right.  But this is where the conflict is.  We can't say that including kernel headers is bad until someone is stepping forward to say "Link against library FOO with headers FOO instead" and noone has done that.
[06:04] <fabbione> jbailey: i am not sure.. i think linus and other did that.. but after the usual 2/3 messages out of the endless thread i get bored and delete the thread
[06:04] <fabbione> it's a topic that shows up at least once a month on lkml
[06:05] <jbailey> I haven't seen anyone doing that.  I wish they would.  All I see is a hot potato being tossed around.
[06:07] <fabbione> :)
[06:07] <jbailey> Mmm potatoes.
[06:07] <fabbione> yesterday i did grill peppers, obergine and squash...
[06:08] <fabbione> with a bit of garlic and oil of olive...
[06:08] <fabbione> (on top)
[06:08] <jbailey> We finally got a stove on friday.
[06:08] <jbailey> It's hard to cook in 30 weather.
[06:08] <jbailey> Hopefully we'll get a bbq this fall for grilling.
[06:08] <fabbione> today i am up for tomatoes with rice and meat with cheese topping...
[06:09] <fabbione> but i think you can make easily a vegan variant out of it
[06:09] <jbailey> I'll probably do a fruit salad and a tofu scramble for lunch.
[06:09] <jbailey> Sure. =)  I'm good at those.
[06:09] <jbailey> We'll have to have you over for dinner next time you're in Montral. =)
[06:09] <fabbione> jbailey: can you actually eat rice, can't you?
[06:09] <jbailey> Yup
[06:09] <jbailey> I usually have rice at least once a day.
[06:09] <fabbione> than do this:
[06:09] <fabbione> take the tomatoes, open only the top
[06:09] <fabbione> empty it and keep the inside
[06:10] <fabbione> boil some rice
[06:10] <fabbione> fry some mushrooms...
[06:10] <fabbione> mix the rice, the mushrooms and the inside of the tomatoes
[06:10] <Lathiat> take some frogs legs
[06:10] <fabbione> mix them very well
[06:11] <fabbione> and fill the tomatoes again with that
[06:11] <fabbione> put in the oven....
[06:11] <fabbione> enjoy
[06:11] <jbailey> Any spices in with the mixture?
[06:11] <fabbione> the best would be with mozarella on top, but well...
[06:11] <fabbione> jbailey: you can use rosmarine and basilicum
[06:11] <jbailey> Cool.
[06:12] <fabbione> perhaps i would put rosmarine together with the rice mix
[06:12] <fabbione> and fresh basicilum on top (instead of the cheese)
[06:12] <jbailey> Angie dislikes mushrooms, but I suspect that diced aubergine might be nice in there.
[06:12] <fabbione> jbailey: i am pretty sure you can use whatever veggy you like best in the mixture...
[06:12] <jbailey> That sounds terribly good. =)
[06:13] <fabbione> try and let me know :)
[06:14] <fabbione> i eat them with meat, this is a fast omnivor2vegan regexp ;)
[06:14] <fabbione> but it might actually work :)
[06:14] <fabbione> and with this i am going to cook...
[06:14] <fabbione> later fellas
[06:14] <jbailey> 'bye Fabio. =)
[06:16] <lamont> fabbione: gsyprf11's chroot updated
[06:28] <lamont> fabbione: btw, redhat-cluster-suite links shared libs with ld instead of gcc.  bad R-C-S.  Either that needs to be fixed, or you need to buy jbailey a pallet of his favorite, which might _still_ not be enough to fix the issue.
[07:36] <jbailey> fabbione: 
[07:36] <jbailey> missing module eata_pio
[07:36] <jbailey> command exited with status 1
[07:36] <jbailey> make: *** [binary-udebs]  Error 2
[07:36] <jbailey> This doesn't sound like the bug you're reporting.
[07:36] <jbailey> Any chance of a more reduced testcase?
[07:36] <jbailey> Or do you have something poisoned in your ccache somehow?
[08:51] <fabbione> lamont: yes i know it does, but linking with gcc seems to create non-PIC code
[08:51] <fabbione> jbailey: i have no idea why it works for you
[08:51] <fabbione> but definetely it doesn't for me
[08:51] <fabbione> i can try to flush the ccache..
[08:55] <lamont> fabbione: it shouldn't....
[08:55] <lamont> linking with ld means that it needs to be aware of whatever additional libraries gcc sticks on the ld command line, if it wants to work.
[08:55] <fabbione> lamont: apparently it does :)
[08:55] <lamont> specifically, hppa needs an extra lib passed to ld for shared libs to work
[08:56] <fabbione> lamont: we did try to use gcc
[08:56] <fabbione> and it ended up with lintian/linda complaning about non-PIC libs
[10:26] <mxpxpod> fabbione: it looks like linux-wlan-ng is going to make a release this week :)
[10:28] <fabbione> mxpxpod: ok
[10:28] <mxpxpod> fabbione: do you want me to let you know when it happens since I keep up on it?
[10:29] <fabbione> up to you. we have an automatic tool that goes around and check for us
[10:29] <mxpxpod> ah, ok