**** BEGIN LOGGING AT Thu Dec 06 03:00:00 2007 Dec 06 13:02:01 Bartman007: ping Dec 06 13:25:17 Bartman007: when you're back, please test http://nbd.name/bb-osx.diff Dec 06 13:25:28 it should fix your libintl problems without breaking anything Dec 06 14:00:32 nbd: pong. it works. Dec 06 14:00:40 great Dec 06 14:00:47 experiencing 50%+ packet loss ATM, so I can't talk to well though Dec 06 14:01:36 ok Dec 06 14:01:38 fix is committed Dec 06 14:07:36 * loswillios kicks CIA-35 Dec 06 14:45:39 nbd, while your fixing bb you might as well fix it for everyone https://dev.openwrt.org/ticket/2811 Dec 06 14:46:27 unless your last commit does that, im building right now so i will find out soon Dec 06 14:50:42 Weedy: the broken commit was reverted Dec 06 14:50:45 Weedy: before i applied my fix Dec 06 14:50:55 ticket closed now Dec 06 14:53:59 \o/ Dec 07 00:03:08 nbd * r9664 /trunk/target/linux/ar7/files/drivers/net/cpmac.c: Dec 07 00:03:08 [PATCH 1/2] Fix race between device reset and start_xmit Dec 07 00:03:08 Signed-off-by: oliver@opencloud.com Dec 07 00:03:12 nbd * r9665 /trunk/target/linux/ar7/files/drivers/net/cpmac.c: Dec 07 00:03:12 [PATCH 2/2] Reimplement rx ring with EOQ markers to avoid reset storms Dec 07 00:03:12 Signed-off-by: oliver@opencloud.com Dec 07 00:03:16 nbd * r9666 /trunk/target/linux/ar7/base-files/etc/diag.sh: (log message trimmed) Dec 07 00:03:16 Minor patch to the AR7 diag.sh to make the status LED a bit more useful. From the patch: Dec 07 00:03:17 This setup gives us 3.5 distinguishable states: Dec 07 00:03:19 - Solid OFF: Bootloader running, or kernel hung (timer task stalled) Dec 07 00:03:21 - Solid ON: Kernel hung (timer task stalled) Dec 07 00:03:23 - 5Hz blink: preinit Dec 07 00:03:25 - Heartbeat: normal operation Dec 07 01:35:52 Weedy: what, you didn't like me OS X fix? who cares about linux? Dec 07 01:36:07 nah Dec 07 01:36:12 it fixed linux Dec 07 01:36:14 btw, did it really break for you, it still compiled fine on my linux boxes. Dec 07 01:36:51 (to be clear, busybox compiled fine w/ r9660 on linux) Dec 07 01:45:42 the problem was this: Dec 07 01:45:58 the busybox build system checked for libintl headers Dec 07 01:46:07 and enabled the use of libintl if those were present Dec 07 01:46:36 so for some reason on your system the test for libintl support succeeded, but it didn't link in the library Dec 07 01:46:58 while on linux systems it would detect properly when the library should be used and when not Dec 07 01:47:11 so your override forced the linking against libintl, even on systems where it isn't present Dec 07 01:47:19 which is what caused the build to fail on some systems Dec 07 01:48:04 so to fix this, i simply made sure it never attempts to use libintl, which works on all systems Dec 07 01:48:39 nbd: ah, so my alternative fix, ifeq ($(HOST_OS),Darwin) might have worked, but yours it much better. Dec 07 01:48:50 s/it/is/ Dec 07 01:49:20 it might have worked on some systems Dec 07 01:49:27 my osx compiled without all those fixes, btw. Dec 07 01:51:00 heh, interesting. I guess I will have to use some more 10.4 machines to isolate 10.5 specific issues. Dec 07 01:51:43 it may not even be osx version dependent Dec 07 01:52:04 depends on the amount of stuff that you have installed e.g. via fink or macports Dec 07 01:53:12 nbd: of course, I thinking of a more general sense. Dec 07 01:53:52 it would be nice to make sure I have a fast 10.4 box to test with, currently using a 550mhz G4 for that purpose. **** ENDING LOGGING AT Fri Dec 07 02:59:57 2007