**** BEGIN LOGGING AT Wed Apr 15 02:59:57 2009
Apr 15 12:57:58 <phaidros> thanks :)
Apr 15 14:11:38 <iSteve> geoff-: ping
Apr 15 16:25:48 <geoff-> iSteve: hi
Apr 15 16:26:23 <iSteve> geoff-: two things
Apr 15 16:26:29 <iSteve> geoff-: a) in backlog, and via email, I've sent you migration guide
Apr 15 16:26:40 <geoff-> yes
Apr 15 16:27:12 <iSteve> geoff-: b) I have explored udevtrigger -> udevadm trigger in detail, and it appears that it should not be complicated writing a standalone app using the trigger-related part of their sourcecode... which is, in my opinion, absolutely the best solution for coldplugging software
Apr 15 16:27:55 <iSteve> geoff-: I should find time to write proof of concept either by friday or, more likely, early next week
Apr 15 16:29:10 <geoff-> iSteve: I've been too busy with othere things to work on the udev stuff
Apr 15 16:29:26 <iSteve> okay
Apr 15 16:29:51 <geoff-> I planned to look at your migration guide, then try to fixup my hotplug2 patch
Apr 15 16:30:17 <geoff-> but my main interest is petitboot
Apr 15 16:31:13 <geoff-> it uses udev to get hotplug events
Apr 15 16:31:27 <iSteve> hmm... and why not hotplug2? :)
Apr 15 16:32:46 <geoff-> does hotplug2 support writing to sockets in the rules?
Apr 15 16:33:25 <iSteve> sadly, no -- it's more designed to distribute to other applications... however, it's a trivial addition
Apr 15 16:33:27 <geoff-> SUBSYSTEM=="block", RUN+="socket:/tmp/petitboot.udev"
Apr 15 16:33:39 <geoff-> that is what we use
Apr 15 16:34:13 <iSteve> and what does it do, exactly? sends the serialized event to the unix socket?
Apr 15 16:35:10 <geoff-> yes
Apr 15 16:36:04 <iSteve> that'd be a trivial addition
Apr 15 16:36:14 <iSteve> may I inquire what do you guys use this for?
Apr 15 16:36:27 <geoff-> it is a bootloader
Apr 15 16:36:41 <geoff-> we try to mount the device
Apr 15 16:36:48 <geoff-> then look for a .conf file
Apr 15 16:37:40 <iSteve> so you have a daemon that reads from the unix socket and performs this logic, instead of running an application every time you hit a block device?
Apr 15 16:37:51 <geoff-> yes
Apr 15 16:37:55 <iSteve> cunning
Apr 15 16:38:10 <iSteve> I'll add serializing events into unix socket
Apr 15 16:38:21 <geoff-> it also gets events from udhcpc and polls for removable media
Apr 15 16:38:24 <iSteve> (and into a file)
Apr 15 16:38:43 <iSteve> any other feature like this you guys use?
Apr 15 16:39:29 <geoff-> not really, just need the block events to start the processing
Apr 15 16:39:40 <iSteve> actually...
Apr 15 16:40:10 <geoff-> and we do a udev trigger at atartup to get the cuurent devices
Apr 15 16:40:41 * iSteve ponders he would allow simple 3rd party commandset extension for hotplug2
Apr 15 16:40:47 <iSteve> so you wouldn't actually need two daemons
Apr 15 16:40:56 <iSteve> but I guess that'd be against unix principles
Apr 15 16:41:05 <iSteve> anyway... I'll add the event serialization
Apr 15 16:41:14 <iSteve> and I will write PoC of the udevtrigger
Apr 15 16:48:35 <geoff-> iSteve: here is the udev code: http://pastebin.com/mf166d64
Apr 15 16:48:54 <geoff-> iSteve: start at udev_init()
Apr 15 16:50:00 <geoff-> an event goes to udev_process()
Apr 15 16:53:13 <iSteve> mhm, so udev relays the event verbatim
Apr 15 16:53:29 <iSteve> writing into unix socket should be 10 lines of code addition into hotplug2 :)
Apr 15 16:55:08 <geoff-> yes, I belive it is essentially the same info as in the notification mechanisms
Apr 15 16:56:03 <geoff-> as in the other mechanisms
Apr 15 18:13:59 <freezer> hi
Apr 15 18:14:21 <freezer> anyone knows about this https://dev.openwrt.org/changeset/14556
Apr 15 18:14:34 <freezer> looks like it doesnt work
Apr 15 18:14:50 <freezer> RAM performance on routerstation is still poor on current trunk
Apr 15 19:22:24 <xxiao> freezer: what is the benchmark that you're using?
Apr 15 19:24:26 <xxiao> two months ago i was doing the ar71xx, the performance is bad, juhosg added that fix and cpu_bench had a decent output http://oldwiki.openwrt.org/HardwarePerformance.html
**** ENDING LOGGING AT Thu Apr 16 02:59:57 2009