WebHosting Paid by #1Payday.Loans


   The ROCK Linux project has been discontinued in 2010. Here are the old data for the historical record!

[00:13] _BoS_ (n=BoS@dslb-084-059-198-193.pools.arcor-ip.net) left irc: Remote closed the connection
[00:13] _BoS_ (n=BoS@dslb-084-059-200-245.pools.arcor-ip.net) joined #rocklinux.
[06:48] stf^rocklinux (n=user@heim-033-137.raab-heim.uni-linz.ac.at) left irc: Remote closed the connection
[07:56] kasc_ (n=kasc@dslb-084-060-105-122.pools.arcor-ip.net) joined #rocklinux.
[08:06] kasc (n=kasc@dslb-084-060-096-026.pools.arcor-ip.net) left irc: Read error: 110 (Connection timed out)
[08:06] Nick change: kasc_ -> kasc
[08:32] <blindcoder> moin
[08:43] <netrunner> moin, 2
[08:44] <netrunner> stf: stage 1 iirc
[08:45] <netrunner> yep, 1-glibc23
[08:57] <blindcoder> bash-3.1 released
[09:29] jamatik (n=jamatik@dslb-084-059-018-213.pools.arcor-ip.net) joined #rocklinux.
[10:32] jamatik (n=jamatik@dslb-084-059-018-213.pools.arcor-ip.net) left irc: "Leaving"
[12:02] blindcod1r (n=blindcod@dslb-084-059-061-226.pools.arcor-ip.net) joined #rocklinux.
[12:02] blindcoder (n=blindcod@dslb-084-059-041-189.pools.arcor-ip.net) left irc: Nick collision from services.
[12:02] Nick change: blindcod1r -> blindcoder
[12:22] <netrunner> why are the installed initfiles not in the flist? (hal, dbus, ivman)
[13:58] <netrunner> aha. needs to be done in postmake
[14:24] <mnemoc> postflist?
[14:26] <netrunner> I just executed install_init in the .conf. but it needs to be in one of the hooks.
[14:26] <mnemoc> weird
[14:26] <mnemoc> ah... you did it before build_this_package :D
[14:27] <netrunner> udev has an init script but it is not installed :)
[14:27] <mnemoc> on t2 that was automatized.... you could do the same
[14:27] <mnemoc> .init, .desktop, .cron..... are always treated in the same way
[14:28] <netrunner> hm. should be. I think .desktop are automatically treated. I wonder why .init not ...
[15:10] <netrunner> who here is running svnserve/dav_svn combo?
[15:26] <mnemoc> only dav_svn here
[15:27] <mnemoc> i can't live without authz :p
[15:45] stf^rocklinux (n=user@heim-033-137.raab-heim.uni-linz.ac.at) joined #rocklinux.
[16:43] <daja77> where is madtux when you want to tease him
[16:44] <mnemoc> :)
[17:12] <daja77> ah on msn ^^
[17:14] <daja77> and it works ;)
[18:15] Action: netrunner is happy about kde 3.5 ... finally automatic media recognition
[18:20] <netrunner> and the new kopete now has almost all annoyances removed
[18:21] <netrunner> what's his msn id?
[18:22] Action: netrunner enriching his contact list
[18:29] <stf^rocklinux> netrunner: did you find a solution for building glibc 2.3.6 in stage 1?
[18:30] <netrunner> stf^rocklinux: have not looked for one. but I think clifford's patch should be modified to apply only in stages >4
[18:30] <netrunner> hm. another thing is that this unwind thing needs to be added to gcc, but I don't know how.
[18:30] <stf^rocklinux> ok, I'll take a look
[18:31] <stf^rocklinux> afaics it is already in gcc, but in stage 1 glibc can't detect it
[18:32] <netrunner> stf^rocklinux: I'll then try to start a build with my suggested modification.
[19:04] markuman (n=supermar@p50927A89.dip.t-dialin.net) joined #rocklinux.
[19:17] markuman (n=supermar@p50927A89.dip.t-dialin.net) left irc: "https://www.linwiki.org - wiki around os linux!! #linwiki"
[19:21] <netrunner> stf: that unwind thing seems not to be dependent on anything, as I see reading the configure file
[19:25] <SMP> netrunner: this is caused by a crt1.o symlink that you will find somewhere under build/<...>/ROCK/
[19:25] <SMP> netrunner: I mean, the symlink point to a nonexistent file
[19:26] <netrunner> SMP: cool. suggested fix?
[19:27] <netrunner> I mean, it happens in a new build...is this a new problem?
[19:27] <SMP> I don't even know what causes this, let alone how to fix it
[19:27] <SMP> that's what I was going to ask. since when does this happen?
[19:28] <netrunner> SMP: well, I did sm sync and applied cliffords nptl patch, as well as fakes patches for package selection.
[19:28] <netrunner> I further selected gcc 4.0.1
[19:28] <SMP> I have the NPTL patch but not fakes pkgsel stuff
[19:29] <netrunner> I'll try with gcc 3.4
[19:29] <SMP> using gcc 4.0.2 here as well
[19:30] <netrunner> hm. my running system has /usr/lib/crt1.o (which the bad link points to) from glibc23:dev
[19:31] <netrunner> as well as /usr/i586-unknown-linux-gnu/lib/crt1.o from gcc34:dev
[19:32] <netrunner> started build with gcc34 only
[19:32] <stf^rocklinux> netrunner: which configure file do you mean?
[19:33] <netrunner> stf^rocklinux: glibc23. but SMP already suggested the problem is caused by a dead symlink.
[19:34] <stf^rocklinux> the forced unwind detection is done in the glibc nptl/sysdeps/pthread/configure file
[19:35] <stf^rocklinux> it compiles a small test program that includes unwind.h and calls _Unwind_GetCFA()
[19:39] <stf^rocklinux> unwind.h is part of gcc (34 and 40) and _Unwind_GetCFA() is in libgcc_s.so and libgcc_eh.a in stage 2 gccs (or in libgcc.a in stage 0 gcc)
[19:41] <stf^rocklinux> hm, I reverted Cliffords patch to glibc.conf in stage 1 and glibc 2.3.6 compiled fine
[19:43] <SMP> this has nothing specific to do with that unwind stuff
[19:43] <SMP> stf^rocklinux: find build/$target -name crt1.o -ls
[19:46] <stf^rocklinux> hm, there's no broken crt1.o symlink in my build 
[19:47] <SMP> in the one that just built the glibc?
[19:47] <stf^rocklinux> maybe because 1-glibc23 already built...
[19:47] <stf^rocklinux> yes
[19:50] <stf^rocklinux> I'll start a new build to check this...
[19:58] <SMP> I think I know this ...
[20:02] <SMP> when I started looking at NPTL, I had to bootstrap a glibc without it in stage 1, then build with NPTL in a later stage
[20:03] <SMP> so, to build a glibc with NPTL you need to have ... a glibc installed
[20:04] <stf^rocklinux> isn't it a problem if some programs are compiled against a glibc with NPTL and others are not?
[20:04] <SMP> well first of all it only affect the threading part of it
[20:06] <SMP> then I think this isn't so weird as it looks on first sight, because I think we should ship a plain libc in /lib and put the TLS version into /lib/i686
[20:06] <SMP> which then isn't a problem (even in stage 1), because you alread yhave the plain libc
[20:10] <stf^rocklinux> but then all programs would use the plain libc by default, which would leave the TLS version unused?
[20:11] <SMP> no, ld-linux.so knows when to use /lib/{tls,i686}/* instead of the ones in /lib
[20:11] <netrunner> same problem with gcc34
[20:12] <stf^rocklinux> ah, ok
[20:12] <netrunner> no crt1.o
[20:13] <stf^rocklinux> what's so fancy about NPTL and TLS anyway? ^^
[20:13] <netrunner> nptl should make threading much faster.
[20:14] <netrunner> tls ... top level security? 
[20:14] <SMP> well, more scalable, more standards-compliant
[20:14] <stf^rocklinux> nice
[20:14] <SMP> netrunner: thread-local storage
[20:14] <netrunner> aha.
[20:32] Action: netrunner writing 3 weeks of blog :)
[21:15] <netrunner> damn. I forgot all the names of the people I met there :(
[23:04] <blindcoder> moin
[23:05] <netrunner> moin blindy
[00:00] --- Sun Dec 11 2005