**** BEGIN LOGGING AT Fri Jan 12 02:59:57 2007 Jan 12 09:28:19 [florian]: yup Jan 12 10:37:59 nbd * r6084 /trunk/ (3 files in 3 dirs): Jan 12 10:37:59 Fix the annoying base-files autorebuild bug. Jan 12 10:37:59 When checking file mtimes in directories, timestamp.pl was following symlinks. Jan 12 10:37:59 This broke on base-files, because there are symlinks to /tmp and to /proc/mounts in the build directory. Jan 12 10:38:39 this was really bugging me :) Jan 12 11:07:53 <[florian]> malbon: yo Jan 12 11:07:54 <[florian]> hi nbd Jan 12 11:08:06 [florian]? Jan 12 12:29:18 nico * r6085 /packages/net/tcp_wrappers/patches/ (4 files): reorder patches to fix tcp_wrappers build Jan 12 15:37:22 <[florian]> {Nico}: thanks for the ordering stuff Jan 12 15:42:48 <{Nico}> [florian]: np Jan 12 15:52:13 <{Nico}> [florian]: http://nthill.free.fr/images/florian/ Jan 12 15:52:39 <{Nico}> some images of the racks (waiting for you) Jan 12 16:25:52 <[florian]> {Nico}: yeah ! Jan 12 16:32:22 <[florian]> http://www.f-x.fr/wikini/wakka.php?wiki=Bcm6348PinOut Jan 12 17:10:05 florian * r6087 /packages/net/igmpproxy/Makefile: Add a default configuration file Jan 12 17:28:11 in my /etc/resolv.conf is a "search lan" although in /etc/dnsmasq.conf is "local=/bude.lan/" and "domain=bude.lan". I guess that's a bug Jan 12 17:28:49 should be "search bude.lan" I guess Jan 12 17:45:04 * crazy_imp still hopes for a rt2x00 package :) Jan 12 17:46:16 we all do :) Jan 12 17:46:57 I think I could do it as a normal package but the kernel package template makes it more difficult Jan 12 17:51:39 * crazy_imp should setup a small distcc farm Jan 12 18:00:27 Unless you have a bunch of hw already I suspect it's faster to set up one super-fast box Jan 12 18:00:34 and probably cheaper Jan 12 18:00:51 unless you modify the high-level build to drop the distcc granularity to a package per node Jan 12 18:00:54 vs. a file per node Jan 12 18:05:41 i have enought networkbootable hw :) Jan 12 18:09:01 i'd split up the granularity tho... I have a hard time keeping 4 high-power CPUs loaded on small files, I think you'll see a boost distributing package-level granularity vs. file level. Do some timing, if you do, I'm curious. Jan 12 18:10:21 I can do a full kamikaze build, not counting host toolchains, in 6m on a single box, is my argument :) (just timed it) Jan 12 18:10:53 is that with make -j highnumber and ccache Jan 12 18:10:58 no ccache Jan 12 18:11:08 dual proc dual core 2ghz athlon X2 Jan 12 18:11:14 with -j6 Jan 12 18:11:27 four cores in total ? Jan 12 18:11:31 yeah Jan 12 18:11:59 what board allows 2 x athlon X2 ? (I thought you should need opteron for more than one) Jan 12 18:11:59 just a generic Big x86 box. Just providing my counter-argument against a compile farm if you don't already have the hardware sitting around unused Jan 12 18:12:26 though power usage still enters into it, depending on your situation Jan 12 18:12:54 h3sp4wn: opteron x2, sorry. 270 cores. Jan 12 18:13:14 h3sp4wn: don't actually know the board, one of the few boxes i didn't build. let me see if dmi has it. Jan 12 18:13:25 opteron cpus are feck expensive though. Jan 12 18:13:44 50% more (twice the cache though) Jan 12 18:14:09 murb: I think this box was around $2.5k. Competitive numbers for the speed. Jan 12 18:14:30 h3sp4wn: Looks to be a Supermicro H8DA8/H8DAR, according to the DMI data Jan 12 19:55:47 {Nico}: i reopened https://dev.openwrt.org/ticket/1180 could you please update linux-igd with the patch & upnpd.conf **** ENDING LOGGING AT Sat Jan 13 02:59:57 2007