**** BEGIN LOGGING AT Fri Oct 01 02:59:58 2010 Oct 01 04:48:07 build #2 of brcm63xx is complete: Failure [failed compile_4] Build details are at http://tksite.gotdns.org:8010/builders/brcm63xx/builds/2 Oct 01 06:39:14 build #2 of octeon is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/octeon/builds/2 Oct 01 07:06:09 hello Oct 01 07:10:27 juhosg * r23165 /trunk/target/linux/ar71xx/ (13 files in 6 dirs): Oct 01 07:10:27 ar71xx: initial support for jjPlus JWAP003 Oct 01 07:10:27 Patch-by: Cezary Jackiewicz Oct 01 07:10:27 (the GPIODEV code has been removed, the sysfs interface Oct 01 07:10:27 can be used to control the GPIO lines instead - juhosg) Oct 01 07:10:28 juhosg * r23166 /trunk/include/package-ipkg.mk: package-ipkg.mk: use the BOARD variable in opkg cmdline if ARCH_PACKAGES is not defined Oct 01 07:25:01 nbd: Do you know about a patch for hotplug2? I have a problem that it seems solved by a patch found on the net but I see nothing about in Backfire. To try I should apply the patch to the file package/base-file/files/etc/init.d/boot/patch-file.patch It is sufficient I create a directory with the patch and put the file inside? Oct 01 07:27:59 nbd: I see that there are three patche in ... /hotplug2/patches in kamikaze omitted in backfire. This patch don't refers to any of these three cases. Oct 01 07:58:51 nbd: pong Oct 01 09:52:57 by the time im done making 'debwrt' work correctly ill have it patched in openwrt lol. Oct 01 10:06:55 philipp64|laptop: Did you find a solution for the customizer scripts? I'm also interested in that. I currently do it by unpacking the rootfs and repacking it again. Oct 01 10:26:33 hmpf. uclibc doesn't have _Exit()? Oct 01 10:26:50 That's a standard-C library function Oct 01 10:41:18 mb__: _exit() exists Oct 01 10:41:32 we should ressurrcet libnotimpl Oct 01 10:45:36 blogic: Yeah, however _Exit() is the official standard C function Oct 01 10:46:22 i know Oct 01 10:52:11 mb * r23167 /packages/Xorg/app/pwrtray/Makefile: pwrtray: Fix CPU hog bug Oct 01 10:59:23 wouldn't that be a one line fix? something like #define _Exit(x) _exit(x) ? Oct 01 11:00:32 ah I see, they are in different headers defined Oct 01 11:02:26 mb__: _Exit() is defined in uclibc's stdlib.h Oct 01 11:08:54 KanjiMonster: My app fails to compile, though. It includes stdlib. Oct 01 11:27:22 mb__: the definition is the same as the one in eglibc's stdlib.h (same #idef guards, too) Oct 01 11:32:54 so it must be something else in uclibc Oct 01 12:19:38 build #2 of atheros is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/atheros/builds/2 Oct 01 12:46:25 hotplug2 poblem... Someone can help me ? Oct 01 12:54:05 yes Oct 01 12:54:09 explain it please Oct 01 12:56:53 Oh, thank john... It's a strange trick. I already investigated alot and found nothing more thant that everything seems right. Ok: kernel in the target starts correctly and load from sdcard the uImage. After the kernel had been loaded starts init script that call preinit. Oct 01 12:57:45 preinit work while hotplug2 is called then the console stop to communicate, don't enter in console and init stops whithout executing any other command line. Oct 01 12:58:25 in the meanwhile, I see that the system is running because if I insert some usb peripheral (and remove repetitely) the console signals correctly the detection. Oct 01 12:59:17 ok Oct 01 12:59:33 are you sure that console=ttySX is set correctly ? Oct 01 12:59:44 yes, Oct 01 12:59:51 and is /etc/inittab set properly ? Oct 01 12:59:58 what ttySX are you using ? Oct 01 13:00:06 cshore_: ping Oct 01 13:00:20 it is set because I see the messages also those echoes I put to debug preinit bash script. Oct 01 13:00:29 ttyS0 Oct 01 13:00:52 hmmm Oct 01 13:01:05 ::sysinit:/etc/init.d/rcS S boot Oct 01 13:01:05 ::shutdown:/etc/init.d/rcS K stop Oct 01 13:01:05 tts/0::askfirst:/bin/ash --login Oct 01 13:01:05 ttyS0::askfirst:/bin/ash --login Oct 01 13:01:05 tty1::askfirst:/bin/ash --login Oct 01 13:01:06 These are the only needed lines in inittab Oct 01 13:01:07 ;( Oct 01 13:01:34 and hotplug is enabled in the Kconfig ? Oct 01 13:02:08 now I see, but there is no way It's not. Oct 01 13:02:26 CONFIG_HOTPLUG=y Oct 01 13:02:38 so it is set ? Oct 01 13:02:57 can you show me kernel command line it uses ? Oct 01 13:03:05 ah no Oct 01 13:03:14 it finds the rootfs otherwise preinit would not work Oct 01 13:03:18 phew Oct 01 13:03:23 cshore_: rewrote the init code Oct 01 13:03:28 i never looked at it yet Oct 01 13:03:33 best ask him how to debug it Oct 01 13:03:38 what kconfig I must see ? Oct 01 13:03:55 CONFIG_HOTPLUG Oct 01 13:04:10 but there are more then one kconfig Oct 01 13:04:16 vi build_dir/linux-lpcxxx/linux-2.6.xyz/.config Oct 01 13:04:21 in that file Oct 01 13:04:26 is CONFIG_HOTPLUG=y ? Oct 01 13:06:17 Joh: build_dir//linux-2.6.32/arch/Kconfig ? Oct 01 13:06:34 no Oct 01 13:06:42 build_dir//linux-2.6.32/.config Oct 01 13:09:33 sure, it's set. In effect during make I see "building hotplug2" in packages list. and the package is set with "*" in make menuconfig Oct 01 13:11:53 strange Oct 01 13:13:06 yes very strange. And I can't find any previous notices. Oct 01 13:13:42 All references to hotplug2 are based on the previous bugs and problems, already fixed in kamikaze, not backfire. Oct 01 13:13:56 probabily is a very silly question, but very good hidden. Oct 01 13:14:44 /sbin/hotplug2 --coldplug --set-rules-file /etc/hotplug2-init.rules Oct 01 13:14:54 This is the line where preinit stops. Oct 01 13:15:09 without errors, it seem a forever wait loop. Oct 01 13:15:27 but the preinit file is the standard that is in the git from owrt 10 Oct 01 13:15:57 ok Oct 01 13:16:21 is udevtrigegr in the image ? Oct 01 13:16:25 udevtrigger Oct 01 13:16:43 wait I check config, but I think none. Oct 01 13:16:48 add it Oct 01 13:16:58 using make menuconfig ? Oct 01 13:16:59 i think it is needed for coldplug if i am not mistaken Oct 01 13:17:01 yes Oct 01 13:17:12 ok just a moment. Oct 01 13:18:15 sorry udevtrigger there is, as a matter of fact is udev that is missing, because it's substituted by hotplug2 specifically for embedded systems. Oct 01 13:18:47 I read already somewhere that starting both udev and hotplug2 may result in a very slow system for timeout conflicts... Oct 01 13:19:27 but in effect, missing udevtrigger, needed by coldplug may result in an error "not found something" during execution. Oct 01 13:19:43 here seems that hotplug starts then wait for something that never happens. Oct 01 13:20:42 hmm Oct 01 13:22:12 Hotplug in preinit is called twice: the first is the line before, the second is no-coldplug in --persistent mode. Oct 01 13:22:38 Now I tried to set --no-persistent the first time, else this is the default condition, but to see what happens. Oct 01 13:27:42 is there a way I can bypass hotplug2 to see more ? Oct 01 13:28:24 without hotplug2 you wont see much Oct 01 13:28:42 but the system starts the console ? Oct 01 13:29:06 Can I recompile hotplug2 with some print message to see what can be ? Oct 01 13:36:09 Joh: The following line Oct 01 13:36:10 size=$(awk '/MemTotal:/ {l=5242880;mt=($2*1024);print((s=mt/2)l)?mt-l:s}' /proc/meminfo) Oct 01 13:36:19 is before the hotplug2 call. Oct 01 13:36:39 I find strange that MemTotal 5.242.880 Oct 01 13:36:53 can you explain to me just to be sure ? Oct 01 13:37:06 (this is the standard file from OpenWRT Oct 01 13:40:48 Alicemirror: 5.242.880 is supposed to be the upper limit Oct 01 13:41:04 for tmpfs reserved space Oct 01 13:42:02 ok the specific case uImage is 7.3 Mb and the system has 32Mb. Can be it incremented? Becaue the only difference in previous kamikaze version is that uImage there was 5 Mib. Oct 01 13:46:13 jow_laptop: more like 5MB is the lower limit of how much *not* to allow to be used for the ramfs, right? If I understand it right, then size is (total - ~5MB) if total is between ~5MB and ~10MB, else size is total/2 Oct 01 13:49:01 so I must put a smaller number that 5Mib to increase reserved memory. Oct 01 13:49:14 Can it have sominfluence on the hotplug2 problem? Oct 01 13:50:17 Alicemirror: how much mem does your device have? Oct 01 13:50:30 32 Mb Oct 01 13:51:10 then it should default to ~16 Mb as the upper limit Oct 01 13:51:37 Ok, now I try to see what happens. Oct 01 13:52:28 whats does the MemTotal line in /proc/meminfo say? Oct 01 13:53:59 28804 Oct 01 13:54:22 shown in the old kamikaze build because this new version stops etc... Oct 01 13:55:48 KanjiMonster: oops, right Oct 01 13:56:56 ok I rebuilt backfire with 16Mb Oct 01 13:57:30 and (because the system stops) put echo $size immediately after that line. preinit set size to ~14 Mb Oct 01 13:57:39 but it stop again... Oct 01 14:01:01 14 Mb looks right Oct 01 14:01:09 I'm sure. Oct 01 14:08:22 KanjiMonster: I tried to temporary comment the two hotplug2 calls in preinit to see whats happen beyond. Oct 01 14:08:51 The script continue (then there is a recursive error "Error reading fixed information") Oct 01 14:09:03 If it can be helpful... Oct 01 14:15:13 I tried another short roundtrip... Moved the two hotplug2 lines down in the script. Oct 01 14:15:34 Ok, preinit exectues all then stops calling hotplug2. Oct 01 14:17:55 stintel: can you try a patch for me to see if it helps with the multiple cards regdomain issue? Oct 01 14:18:30 stintel: http://nbd.name/530-ath9k_multiple_cards_fix.patch Oct 01 14:19:12 nbd: certainly Oct 01 14:21:23 nbd: ath5k needs something similar probably too; until I reload the module the user regdomain isn't effective (but my wifi is more for the "I'm too lazy to plug the cable in" case, so it doesn't have high prio ;) Oct 01 14:22:05 KanjiMonster: i'm not sure it's necessary yet Oct 01 14:22:36 could be module load order problem, too; haven't really looked into it yet Oct 01 14:23:49 compiling Oct 01 14:24:16 something else I noticed recently is that even after rmmod ath9k then insmod ath9k, my 5GHz ap wasn't visible on my laptop Oct 01 14:24:26 but I did not have that before Oct 01 14:25:07 someone told here or in #openwrt that he had issues after restarting hostapd that there was only 1 beacon every 30s (iirc), maybe I hit that issue too Oct 01 14:25:33 still have to check in monitor mode what happens Oct 01 14:25:48 or I just messed up my pigtails last time I opened my rspro Oct 01 14:26:40 stintel: on my rspro, reloading ath9k messes up the card Oct 01 14:26:45 it needs a reboot to function again Oct 01 14:26:55 didn't notice this particular issue on any other platform Oct 01 14:27:12 ic Oct 01 14:27:40 btw, my x86-kvm image compiled with 4.4.1+cs+graphite is still running, 9d+ uptime now Oct 01 14:27:56 and my rspro is also still running it Oct 01 14:28:15 so stability looks ok Oct 01 14:28:28 as far you can decide that after 9 days ofcourse :) Oct 01 14:45:50 nbd: very nice, after flash with image with your patch and reboot, I have 2 master mode interfaces, and correct output in iw list Oct 01 14:46:23 interesting Oct 01 14:46:53 need any proof ? :) Oct 01 14:47:15 nope, it's fine Oct 01 14:47:27 ok Oct 01 14:47:34 thanks a lot, btw :) Oct 01 14:47:37 no problem Oct 01 14:47:45 unfortunately i still don't know why the patch helps Oct 01 14:48:08 heh Oct 01 14:48:08 xMff: could you tell me what is the current approach for getting multicast routing between subnets? Oct 01 14:48:08 the regd code is supposed to be able to handle shared channel lists just fine Oct 01 14:56:34 Memphiz: I don't know, would need to google that myself Oct 01 14:56:50 Memphiz: if anything is missing in openwrt to support that, tell me Oct 01 14:57:07 KanjiMonster: In the bin target directory after the compilation, I see that hotplug2 is "hotplug2_1.0-beta-1_.ipk Oct 01 14:57:28 well i'll give smcroute a try (building a package for openwrt) ... it seems very outdated ... but i only want to route one multicast group for my dlna server ... Oct 01 14:57:31 It's normal this version? Seems strange, I even read about as a stable release, not beta. Oct 01 14:58:01 Memphiz: there is a package called igmp-proxy, this sounds to me like that should do that Oct 01 14:58:13 KanjiMonster: mhh ahh ... thx Oct 01 14:58:32 Memphiz: often used with iptv and vdsl iirc Oct 01 14:58:45 Alicemirror: at least that's the version openwrt uses Oct 01 14:59:07 Ok, thank you. I'm already investigating the problem. Oct 01 15:00:07 Kanji: where I find locally the souces of the files I build for my target? I see that in dl directory are all compressed files for sources. Isn't there a build directory with sources ? Oct 01 15:02:28 KanjiMonster: do you know how i can figure out if a running kernel has mroute (multicast routing) support? think this is needed for igmpproxy Oct 01 15:02:41 Alicemirror: build_dir Oct 01 15:02:53 Memphiz: its enabled since ages for openwrt Oct 01 15:02:58 cool Oct 01 15:03:10 Yes I supposed, but there aren't other than compiled and config, patches and more. Oct 01 15:03:15 then i will try it know (configuration seemed a bit to easy ... lets see if it works) Oct 01 15:03:19 .c and .h there aren't Oct 01 15:04:12 Alicemirror: they should be, e.g. for normal packages in build_dir/target-/-/ Oct 01 15:04:30 thanks. Oct 01 15:15:11 nbd * r23168 /trunk/package/mac80211/patches/520-ath9k_multiple_cards_fix.patch: ath9k: fix regulatory issues with multiple cards Oct 01 15:16:54 KanjiMonster: thanx ... it works ... had to add some iptables rules on the router and now my tv can see my dlna stuff *yeah* Oct 01 15:17:10 great :) Oct 01 15:18:42 ping xMff Oct 01 16:40:05 acoul * r23169 /trunk/ (11 files in 3 dirs): tools/mtd-utils: update to mtd-utils-20101001, fix FreeBSD (and hopefully Darwin) build issues. Oct 01 16:44:43 acoul * r23170 /trunk/package/grub/Makefile: package/grub: revert accidental/unessesery grub commit Oct 01 17:13:06 KanjiMonster: I have news on the problem hotplug2. I debugged and inserted in the main function a set of printout to know where we are and what does happens. I see the following: Oct 01 17:13:06 1) main() do a set of initialization calls then enter in a while loop to explore all current processes (hotplug2 is calleg by init.d) Oct 01 17:13:06 2) The loop is at the end of hotplug.c >> while (process) { ... } Oct 01 17:13:06 3) The first thing that do is to get size calling recv function as in the following line: Oct 01 17:13:06 size = recv(settings->netlink_socket, &buffer, sizeof(buffer), 0); Oct 01 17:13:06 4) The parameters before the call to recv are the following: Oct 01 17:13:07 socket 3 Oct 01 17:13:07 buffer is "/lib/libc.so.o" Oct 01 17:13:08 sizeof buffer is 2560 Oct 01 17:13:08 5) Everything die. Oct 01 17:16:57 Alicemirror: I have to admit, I don't even know what problem you are debugging (symptoms, what platform, etc) Oct 01 17:18:16 d/win 37 Oct 01 17:19:02 blogic: no such file or directory. Oct 01 17:19:13 ;) Oct 01 17:21:24 :) Oct 01 17:22:00 morning Oct 01 17:22:15 wow Oct 01 17:22:28 lo groz Oct 01 17:22:40 been a while since i've poked in, but, i've got some patches that i'm just cleaning up Oct 01 17:22:48 oh boy Oct 01 17:22:50 and a bit out of touch with the process these days Oct 01 17:22:57 whatz u need to know Oct 01 17:22:59 well, here's the issue i had to deal with Oct 01 17:23:07 already got a ssh key for svn ? Oct 01 17:23:16 or been offline that long :) Oct 01 17:23:23 i have a net5501, all is good, but, i want to run the exact same binaries on another box, which uses a usb jumpstick to boot Oct 01 17:23:29 problem is, with the jumpstick, and rootwait Oct 01 17:23:48 block2mtd doesn't see the jumpstick, it initializes AFTER block2mtd Oct 01 17:24:03 ok Oct 01 17:24:06 I added some stuff into the block2mtd driver, to retry finding it's device Oct 01 17:24:11 if it failed initially Oct 01 17:24:14 did you set rootdely ? Oct 01 17:24:18 so, it finds it after the usb loads Oct 01 17:24:22 rootdelay even Oct 01 17:24:32 yes i did, problem is, tha tkicks in AFTER the block2mtd driver gets it's parameters Oct 01 17:24:37 and decides 'the disk doesn't exist' Oct 01 17:24:41 ah ok Oct 01 17:24:49 so i added a timer hook, so it'll retry for 30 seconds if it fails Oct 01 17:25:00 now, the exact same binary works, on a usb jumpstick in my via board Oct 01 17:25:05 as it does on my net5501 Oct 01 17:25:20 i've also got another set of patches Oct 01 17:25:38 which update gdb to recent 7.2, and add in the 'multi threaded debugging capability' to uclibc Oct 01 17:25:52 so i can remotely debug with gdbserver, on target, for multi threaded programs Oct 01 17:25:59 ok Oct 01 17:26:03 and it's tested on x86, brcm47xx, wrt160nl Oct 01 17:26:19 send them to blogic@openwrt.org so i can commit them if you dont have svn access Oct 01 17:26:34 haha, i've been offline for so long, no keys these days Oct 01 17:26:48 but, i've started a new 'play' project, that's gonna use this stuff heavily for a while Oct 01 17:26:49 i thought so :D Oct 01 17:27:02 so, i'm ramping up on the state of the world again Oct 01 17:27:21 building up a full remotely operated setup, all done over network, for an observatory Oct 01 17:27:43 gonna host it on a via board with 4 usb, 4 serial, and 2 ethernet connections Oct 01 17:27:51 which boots from a usb jumpstick Oct 01 17:28:10 what is a jumpstick ? Oct 01 17:28:22 umm, little usb flash thingy you plug into a usb port Oct 01 17:28:28 and it pretends to be a disk drive Oct 01 17:30:00 also known as 'flash drive' or 'pen drive', or who knows what all else Oct 01 17:30:09 ah ok Oct 01 17:30:22 But the problem it presents is simple Oct 01 17:30:38 block2mtd inits before the usb system detects the flash drive in the usb port Oct 01 17:30:51 and returns an error saying '/dev/sda2 not present' Oct 01 17:31:04 and then rootwait sits forever waiting for /dev/mtdblock0 Oct 01 17:31:39 but with this timer hook added, block2mtd tries again,and finds it after usb stuff initiliazes, and the flashdrive is recognized Oct 01 17:32:37 i couldn't make it wait with rootwait or whatever, cuz, block2mtd (compiled into kernel, along with usb storage) initializes before usb storage Oct 01 17:32:42 and i couldn't find a way to change that Oct 01 17:34:18 i see Oct 01 17:34:55 my fully remote deployment will likely be on the net5501 soekris board Oct 01 17:35:04 but my 'develop and test it all' will live on the via board here at home Oct 01 17:36:00 and it does make it easy to experiment with squashfs + jffs2 images on the x86, just dd into a flashdrive, and boot the flashdrive Oct 01 17:36:07 and it all 'just works' Oct 01 17:37:21 I'll email that one later today, after I clean out a few debug prints Oct 01 17:37:59 the other one I gotta do a little more fussing with, for now for my own use, I have it adding the thread_db library into the pthread ipk Oct 01 17:38:13 need to fix that up so it's a separate package to be selected Oct 01 17:38:26 is only needed if you are going to remotely debug multi-threaded programs Oct 01 17:38:32 with gdbserver Oct 01 17:51:40 ping glp Oct 01 17:51:54 glp: both routers work....now to get OpenWRT on them Oct 01 17:52:15 is it worth adding a new leds-XXX.c file for a new platform, or is there a generic way to do that now? Oct 01 18:00:13 philipp64: you can use gpio-leds Oct 01 18:28:20 nbd: ping Oct 01 18:30:09 philipp64|laptop: 19:59 < blogic> philipp64: you can use gpio-leds Oct 01 18:30:24 philipp64|laptop: what arch is it ? are the leds attached to a led IP or gpios ? Oct 01 18:56:16 blogic: it's an x86, with a cs55353, and they are GPIO-based LED's. Oct 01 18:58:35 KanjiMonster: pong Oct 01 19:09:58 philipp64|laptop: in that case gpio-leds Oct 01 19:37:07 blogic: can you give me more details? like another machine to use as an example? Oct 01 19:47:57 philipp64|laptop: yes, 1 sec Oct 01 19:53:40 build #2 of at91 is complete: Failure [failed compile_2] Build details are at http://tksite.gotdns.org:8010/builders/at91/builds/2 Oct 01 19:54:41 philipp64|laptop: https://dev.openwrt.org/browser/trunk/target/linux/ar71xx/files/arch/mips/ar71xx/mach-tl-wr841n.c#L68 Oct 01 20:16:18 blogic: that helps. I can just put the button and LED sections, right? Oct 01 20:19:10 yes Oct 01 20:19:21 and then you need to register it in the setup code Oct 01 20:19:54 1 sec let me find an example Oct 01 20:21:02 https://dev.openwrt.org/browser/trunk/target/linux/ifxmips/files-2.6.33/arch/mips/ifxmips/common/devices.c#L24 Oct 01 20:24:57 I wouldn't need both, right? just one or the other? Oct 01 20:30:23 no Oct 01 20:30:31 you need the struct with the led defines Oct 01 20:30:41 and then the platrom call to register it Oct 01 20:47:27 lines 1-61 of devices.c then? Oct 01 21:37:40 build #2 of ep93xx is complete: Success [build successful] Build details are at http://tksite.gotdns.org:8010/builders/ep93xx/builds/2 Oct 01 22:00:26 build #2 of iop32x is complete: Failure [failed compile_3] Build details are at http://tksite.gotdns.org:8010/builders/iop32x/builds/2 Oct 01 22:15:19 <|KanjiMonster|> mb__: I don't know much about uclibc, but it looks like it never was "enabled": http://git.uclibc.org/uClibc/tree/libc/stdlib/stdlib.c?h=0_9_30#n769 (it was directly committed that way) Oct 01 22:16:13 KanjiMonster: hehe. Yeah Oct 01 22:16:43 you could try uncommenting that it Oct 01 22:16:52 -it Oct 01 22:17:18 Well, I already changed my software. I was just surprised that it doesn't support the standard function. Oct 01 22:35:03 nbd: thanks again, now that I don't have to rmmod and insmod ath9k, I can see my AP at 5GHz again Oct 01 22:37:13 might set my 2.4GHz AP to HT20 rather than HT40 now so I'm not polluting half of the 2.4GHz band Oct 01 22:37:23 seems that most windows drivers wont use HT40 on 2.4GHz anyway Oct 01 23:22:05 is it possible to quickly select a package without entering menuconfig, like "make packages=+python" ? Oct 01 23:22:57 echo CONFIG_PACKAGE_foo=m >> .config && make package/foo/compile V=99 Oct 01 23:24:04 xMff: thanks Oct 01 23:54:06 nunojpg: that doesn't do dependencies though Oct 01 23:54:17 IIRC Oct 02 00:25:51 cshore: would be nice to have a make target that rebuilt the dependencies non-interactively... i.e. pulled in everything that was needed that might have been left out. Oct 02 00:26:28 philipp64|laptop: yep but I have no idea how to implement it Oct 02 00:27:32 I mostly work on the systems stuff (not kernel and not build system)....things like preinit and packages like freeswitch Oct 02 00:35:17 philipp64|laptop: make oldconfig Oct 02 00:35:42 http://www.explosm.net/comics/2189/ :D Oct 02 02:38:31 jow * r23171 /trunk/scripts/feeds: [scripts] feeds: restore previously active feed after install_package(), fixes install all case for packages with inter-feed dependencies **** ENDING LOGGING AT Sat Oct 02 02:59:57 2010