[01:12] <mjg59> fabbione: I just got an oops in inotify
[01:12] <mjg59> I tried unmounting an ntfs partition
[01:13] <mjg59> (this is 2.6.12 as of Wednesday)
[01:14] <jbailey> There's no __NR_sigaction on x86_64.  Does the 32-bit emulation on amd64 promise to do all to 32bit syscalls?
[01:15] <Mithrandir> jbailey: I would think so, yes.
[01:16] <Mithrandir> it's supposed to be a 100% emulation
[01:16] <jbailey> Oh good. =)
[01:16] <jbailey> So now all we need is glibc headers that DTRT.
[01:16] <jbailey> Mithrandir: Is there any hope of us doing multiarch at least a little bit for Breezy? =)
[01:17] <Mithrandir> jbailey: you mean "a little bit more"? :)  I'm hoping to get some time allocated for actually coding up a bit once I'm done with my thesis
[01:17] <Mithrandir> doing binutils + gcc support should either already be there or easily fixed.
[01:18] <jbailey> Mithrandir: I'd love to see at a base just have gcc look in the multiarch directories for -m64 and -m32 as well as the standard directories.
[01:18] <jbailey> We can then take the current biarch packages and have it put things in the right places.
[01:18] <jbailey> Including all the header files and such.
[01:18] <Mithrandir> jbailey: get doko to apply a set of fixes from http://arch.err.no/index.cgi/tfheen@idi.ntnu.no--2005/gcc-debian--multiarch--3.4 and you're set.
[01:18] <jbailey> Nothing from dpkg would need to be touched, basically just gcc, glibc, lkh, ncurses and zlib.
[01:18] <jbailey> And it's all stuff we could do in our sleep today.
[01:19] <Mithrandir> that's easy enough to do, sure.
[01:19] <jbailey> I suspect that doko's current libjava on 32 bit arch problem will give him incentive to say yes. =)
[01:19] <Mithrandir> heh :)
[01:20] <Mithrandir> this won't give you coinstallability until dpkg is fixed, though
[01:20] <jbailey> He's running into the fact that i386 and amd64 aren't like ppc, sparc and s390 in glibc.
[01:20] <Mithrandir> oh, fun
[01:20] <jbailey> Noone in glibc land ever intended it to be a biarch arch AFAICT.
[01:20] <jbailey> So the headers don't seem to easily cope with each system.
[01:21] <Mithrandir> which is the reason for the crack in amd64-libs-dev to divert headers and such?
[01:21] <jbailey> Partly, but mostly amd64-libs crack is due to the fact that Sarge wasn't shipping with amd64.
[01:21] <Mithrandir> that too
[01:21] <jbailey> So no compiler support, so no way to make proper biarch packages.
[01:22] <Mithrandir> true
[01:22] <Mithrandir> it's also aiming for a bit more than ia32-libs-dev which is "this is enough to give you a crosscompiler"
[01:33] <jbailey> Right. =)
[01:34] <jbailey> So assuming doko agrees, that patch is all that needs to be applied to 3.4 to make it look in /usr/include/${GNU_ARCH} and /usr/lib/${GNU_ARCH} before the regular search paths?
[01:36] <Mithrandir> you need to fix up binutils too, but yes.
[01:37] <jbailey> Right - for ls/.
[01:37] <jbailey> ld.
[01:37] <Mithrandir> in the cases of people calling ld directly, for instance.
[01:37] <Mithrandir> gcc DTRT and passes the right parameters, but it's not always enough
[05:55] <fabbione> morning
[06:14] <fabbione> mjg59: i have an inotify update pending for the next upload.
[06:14] <fabbione> mjg59: but i can't upload yet.
[06:24] <zul> hey
[09:40] <zul> blah
[11:40] <jbailey> Can the ia64 kernels run i386 binaries if the appropriate libraries are hanging around?
[11:41] <doko> jbailey, they should
[11:42] <jbailey> doko: I'm just wondering how many multiarch setups we'll need.  I imagine the trouble you're running over with OOo is going to have to be repeated for ia64 at some point.
[11:46] <doko> hmm, who does run OOo on ia64? ;-)
[11:46] <Mithrandir> doko: insane people.  That means, everybody who uses ia64.
[11:47] <jbailey> Mithrandir: That's not fair.  I don't use OOo anywhere, much less on my ia64. =)
[11:48] <Mithrandir> jbailey: well, you're a glibc hacker so you're insane, hands down. :-)
[11:48] <jbailey> ;P
[11:49] <Mithrandir> actually, I didn't say that everybody with an ia64 used ooo, just that they were insane.
[11:49] <doko> Mithrandir: amu reports 100% CPU usage when installing the new bash package on amd64, can you confirm that?
[11:49] <Mithrandir> doko: just installing it?
[11:49] <doko> no, using it
[11:49] <Mithrandir> give me a few secs
[11:51] <Mithrandir> just by launching it and running some commands?
[11:51] <Mithrandir> if so, can't duplicate.
[11:52] <doko> hmm, strange, maybe you could ask amu tomorrow
[11:52] <Mithrandir> I'll try to remember.
[11:54] <doko> thanks