**** BEGIN LOGGING AT Tue Apr 22 02:59:57 2008
Apr 22 06:29:21 <Bartman007> thepeople: pong
Apr 22 14:55:58 <stuffa_> I was wondering who was the developer who oversees all the network scripts. I particular I have a proposal that I would like to suggest we work on to improve them further. And I would like to work through my idea before I proceded any further - assuming you have a bit of time and the inclination - or I can just write them up in trac.
Apr 22 15:01:16 <stuffa_> if any one want to follwo this up you can email me chris at martin dot name
Apr 22 15:21:00 <loswillios> stuffa_: you should try the devel mailinglist
Apr 22 16:15:41 <CIA-7> juhosg * r10911 /trunk/target/linux/ixp4xx/ (3 files in 3 dirs): [ixp4xx] fix a compiler warning in the Sidewinder board specific code
Apr 22 19:22:09 <CIA-7> nbd * r10912 /trunk/package/base-files/files/etc/preinit:
Apr 22 19:22:09 <CIA-7> This patch re-adds the udev and hotplug2 initialization fixes originally
Apr 22 19:22:09 <CIA-7> present in /init but was removed when some code was moved from /init to
Apr 22 19:22:09 <CIA-7> /etc/preinit in rev 10870. This adds udev initialization and only handles
Apr 22 19:22:09 <CIA-7> hotplug2 initialization when hotplug2 is present.
Apr 22 19:22:12 <CIA-7> Signed-off-by: Yuji Mano <yuji.mano@am.sony.com>
Apr 22 19:33:01 <Bartman007> nbd: I don't understand the purpose of including multiple ntp servers in the ntpclient config with the way the current initscript is setup, it never progresses past the first server.
Apr 22 19:36:49 <nbd> Bartman007: ideally it should randomize them
Apr 22 19:37:30 <Bartman007> yeah, but with the way it is currently written it doesn't use anything but the first one
Apr 22 19:38:49 <nbd> yeah
Apr 22 19:38:49 <Bartman007> oh, I think I see why.
Apr 22 19:42:38 <matteo> nbd: who provided the openwrt.ntp.org alias?
Apr 22 19:42:50 <nbd> i asked for it
Apr 22 19:42:56 <Bartman007> matteo: it's part of the ntp pool project
Apr 22 19:43:25 <Bartman007> they request that vendors ask for their own aliases (so usage can be monitored)
Apr 22 19:43:36 <matteo> why the script pings the erver before using it?
Apr 22 19:43:46 <matteo> it could silently drop pings
Apr 22 19:43:58 <Bartman007> to see if it is up, otherwise it sleeps for 3 seconds and tries again...
Apr 22 19:44:11 <Bartman007> which is stupid.
Apr 22 19:44:26 <nbd> feel free to clean up that script
Apr 22 19:44:58 <Bartman007> I hoped to use some of it for ntpdate, but I'll have to cleanup ntpclient first.
Apr 22 20:00:12 <DDave> Does anyone know if I can use my nokia 3410 LCD screen to run on a openwrt?
Apr 22 20:00:29 <matteo> nbd: count=0	# unlimited
Apr 22 20:00:38 <matteo> this forses 0 as updates count
Apr 22 20:00:42 <matteo> and ignores teh config one
Apr 22 20:00:46 <matteo> i'll clean it as well
Apr 22 20:01:37 <Bartman007> matteo: yeah, alreaady fixed that, now I'm just trying to figure out how to fix the rest :-)
Apr 22 20:02:30 <matteo> ps | grep 'bin/[n]tpclient'
Apr 22 20:02:35 <matteo> i don't understand this
Apr 22 20:03:36 <matteo> to get if there is anoter running?
Apr 22 20:04:22 <Bartman007> yeah, though it looks like the [] is fluff.
Apr 22 20:05:56 <matteo> and what if we run it using path?
Apr 22 20:06:00 <matteo> without bin/?
Apr 22 20:06:47 <Bartman007> if bin/ is left out most of the time the ps entry from grep itself is found.
Apr 22 20:06:53 <Bartman007> s/from/for/
Apr 22 20:07:01 <matteo> hmm right
Apr 22 20:07:24 <matteo> what is the [n]?
Apr 22 20:07:32 <matteo> it matches all chars 'n'
Apr 22 20:07:36 <matteo> iirc
Apr 22 20:07:56 <Bartman007> yeah, so completely unneeded IMO.
Apr 22 20:08:39 <nbd> no
Apr 22 20:08:52 <matteo> no?
Apr 22 20:09:02 <nbd> because grep 'bin/ntpclient' also matches "grep 'bin/[n]tpclient'"
Apr 22 20:09:04 <nbd> erm
Apr 22 20:09:11 <nbd> it matches "grep 'bin/ntpclient'"
Apr 22 20:09:20 <matteo> ah
Apr 22 20:09:28 <nbd> while grep 'bin/[n]tpclient' does not match "grep 'bin/[n]tpclient'"
Apr 22 20:10:10 <Bartman007> oh... hahaha.
Apr 22 20:12:19 <matteo> then:
Apr 22 20:12:20 <matteo> ps |grep '[n]tpclient'
Apr 22 20:12:23 <matteo> is even better
Apr 22 20:12:28 <matteo> matches it ever
Apr 22 20:12:41 <Bartman007> won't that hit the initscript also?
Apr 22 20:12:47 <Bartman007> /etc/init.d/ntpclient
Apr 22 20:13:01 <matteo> damn, yes
Apr 22 20:13:27 <Bartman007> I'm not concerned with that line, it's while loop that is bugging me.
Apr 22 20:13:38 <matteo> why closing grep FD?
Apr 22 20:13:40 <matteo> there's q
Apr 22 20:13:44 <matteo> -ps | grep 'bin/[n]tpclient' >&-
Apr 22 20:13:50 <matteo> +ps | grep -q 'bin/[n]tpclient'
Apr 22 20:14:48 <Bartman007> the submitter probably did it out of habit, might not have known ab out -q
Apr 22 20:16:07 <nbd> we need to create some common code for dealing with start-stop-daemon in init scripts
Apr 22 20:16:11 <matteo> Bartman007: simply remove teh loop
Apr 22 20:16:17 <nbd> so we can get rid of all the ps | grep mess
Apr 22 20:17:30 <Bartman007> matteo: already done on my machine, I was just hoping I could get rid of the ping command also, but then I realized this script assumes continuous operation.
Apr 22 20:17:38 <matteo> nbd: a tool which writes pid in /var/run
Apr 22 20:17:38 <matteo> and waits for the child to die
Apr 22 20:17:52 <matteo> and when it dies, removes the pidfile
Apr 22 20:18:03 <nbd> yes, that's what start-stop-daemon does
Apr 22 20:18:21 <matteo> nbd: busybox has one?
Apr 22 20:18:26 <Bartman007> yes
Apr 22 20:18:29 <nbd> what i mean is writing some code so that you only have to set few variables and rc.common deals with the interaction with start-stop-daemon
Apr 22 20:18:32 <nbd> yes, busybox has it
Apr 22 20:18:37 <nbd> and iirc, i enabled it by default
Apr 22 20:18:46 <Bartman007> yes, enabled (in trunk at least)
Apr 22 20:18:56 <matteo> yes
Apr 22 20:19:02 <matteo> Usage: start-stop-daemon [OPTIONS] [--start|--stop] ... [-- arguments...]
Apr 22 20:19:27 <matteo> but then we'll have double the process number
Apr 22 20:19:36 <matteo> we need one, which runs all
Apr 22 20:19:44 <matteo> like apple's launchd
Apr 22 20:19:53 <Bartman007> matteo: start porting!
Apr 22 20:20:07 <matteo> there is launchd src?
Apr 22 20:20:12 <nbd> of course
Apr 22 20:20:19 <nbd> we shouldn't switch to launchd though
Apr 22 20:20:20 <Bartman007> darwin == open source.
Apr 22 20:20:25 <nbd> we should evaluate all the choices
Apr 22 20:20:32 <nbd> and then decide whether we use one of them or write our own
Apr 22 20:20:38 <Bartman007> solaris as SMF and ubuntu has upstart
Apr 22 20:20:43 <nbd> because there's more that we need in such a process than just handling services
Apr 22 20:20:45 <Bartman007> both xml based, woo!
Apr 22 20:20:53 <matteo> we'll write our own
Apr 22 20:20:58 <matteo> very simple and basic
Apr 22 20:20:59 <Bartman007> just what we need, giant bloated frameworks.
Apr 22 20:21:03 <matteo> it reads a conf file
Apr 22 20:21:11 <matteo> then starts all sequentially
Apr 22 20:21:19 <matteo> when one dies, restart it
Apr 22 20:21:21 <nbd> no
Apr 22 20:21:31 <matteo> then there will be a socket
Apr 22 20:21:39 <matteo> /var/run/init
Apr 22 20:21:53 <matteo> with which a tool can get proc infos
Apr 22 20:22:01 <nbd> we need something more advanced
Apr 22 20:22:13 <nbd> here's an example of what i mean
Apr 22 20:22:22 <nbd> first of all we need to differentiate between services and one time task
Apr 22 20:22:29 <nbd> we need dependencies between services
Apr 22 20:22:33 <matteo> sure
Apr 22 20:22:34 <nbd> and also virtual dependencies
Apr 22 20:22:38 <nbd> for instance firewall started
Apr 22 20:22:43 <matteo> dependencies? that's hard
Apr 22 20:22:46 <nbd> or link with default route up
Apr 22 20:23:03 <nbd> because you need dependencies not only when starting services initially
Apr 22 20:23:08 <nbd> but they are needed for restarting as well
Apr 22 20:23:14 <nbd> or sometimes for stopping
Apr 22 20:23:22 <nbd> so upstart might be somewhat close to what we need
Apr 22 20:23:29 <nbd> we could remove the xml though
Apr 22 20:23:31 <nbd> and replace it with uci
Apr 22 20:23:34 <nbd> that'll make it smaller
Apr 22 20:23:43 <nbd> since we have uci on the system already as a library
Apr 22 20:23:48 <nbd> now the tricky part
Apr 22 20:23:49 <matteo> nbd: that's an hard job
Apr 22 20:23:54 <matteo> ok, say
Apr 22 20:23:55 <nbd> i don't care
Apr 22 20:23:57 <nbd> it has to be done
Apr 22 20:24:00 <nbd> even if it's hard
Apr 22 20:24:12 <nbd> the tricky part is integrating it with the network setup system
Apr 22 20:24:29 <nbd> because services should be able to specify that they will only start if network connectivity is there
Apr 22 20:24:49 <matteo> flags *= NET
Apr 22 20:24:55 <matteo> flags += NET
Apr 22 20:24:55 <nbd> no
Apr 22 20:25:03 <Bartman007> nbd: if you are going to look at upstart you might was well look at gentoo's equivalent.
Apr 22 20:25:04 <matteo> a service has many flags
Apr 22 20:25:14 <matteo> Bartman007: init-ng?
Apr 22 20:25:28 <nbd> see what we need is not an init replacement
Apr 22 20:25:31 <nbd> we need a service manager
Apr 22 20:25:33 <Bartman007> matteo: I don't think that's it.
Apr 22 20:25:37 <nbd> something stateful
Apr 22 20:25:53 <nbd> and something that keeps track of the interfaces as well
Apr 22 20:25:57 <nbd> though only to a degree
Apr 22 20:30:23 <matteo> nbd: it will be started by preinit or by the kernel?
Apr 22 20:30:57 <CIA-7> nbd * r10913 /trunk/toolchain/gcc/Makefile: make sure the toolchain target dir really exists
Apr 22 20:31:00 <matteo> it will replace init, or will be a daemon too?
Apr 22 20:31:08 <nbd> it will replace init as well
Apr 22 20:31:42 <Bartman007> wait, I thought you said we don't need an init replacement?
Apr 22 20:32:54 <matteo> hehe
Apr 22 20:33:00 <matteo> that were the other nbd
Apr 22 20:33:10 <nbd> Bartman007: a service manager includes the functionality of init
Apr 22 20:33:14 <nbd> that's what i meant
Apr 22 20:33:20 <nbd> so it's more than just an init replacement
Apr 22 20:33:31 <matteo> so, we'll add inetd as well
Apr 22 20:33:35 <matteo> and here's launchd
Apr 22 20:33:57 <nbd> br
Apr 22 20:33:58 <nbd> brb
Apr 22 20:34:02 <Bartman007> matteo: 297KB on my box.
Apr 22 20:34:33 <Bartman007> granted that is a universal binary.
Apr 22 20:36:04 <matteo> wtf everyone here has a mac?
Apr 22 20:36:29 <Bartman007> yes, we are going to be dropping support for linux compilation soon ;-P
Apr 22 20:36:57 <matteo> i hate apple, they cheats on clients
Apr 22 20:37:05 <Bartman007> just kidding (for those reading channel logs)
Apr 22 20:38:06 <Bartman007> matteo: I jumped ship back during 2.6.20, some things just didn't work right, and they do on os x.
Apr 22 20:38:40 <Bartman007> that said, there are plenty of things that piss me off about os x, but it seems to be the os that hinders my productivity the least.
Apr 22 20:39:40 <matteo> have to reboot, my atheros card is starting to suffer
Apr 22 20:39:42 <matteo> bbl
Apr 22 20:44:20 <matteo> done
Apr 22 21:41:39 <CIA-7> olli * r10914 /packages/utils/sispmctl/ (. Makefile): [Packages]: Add utils/sispmctl (#3366)
Apr 22 21:49:33 <CIA-7> olli * r10915 /packages/utils/sispmctl/Makefile: [Packages] utils/sispmctl: Add description
Apr 22 23:32:34 <CIA-7> agb * r10916 /packages/net/rsync/Makefile:
Apr 22 23:32:34 <CIA-7> packages: Upgrade rsync to 3.0.2
Apr 22 23:32:34 <CIA-7>  -Disabled acl/xattr support
Apr 22 23:32:44 <CIA-7> agb * r10917 /packages/net/chrony/Makefile: [packages] fix chrony description formatting
Apr 22 23:32:54 <CIA-7> agb * r10918 /packages/net/ntpd/Makefile: [packages] split ntpd package into ntpd, ntpdate and ntp-utils
Apr 22 23:33:06 <CIA-7> agb * r10919 /packages/net/ntpd/Makefile: [packages] copy/modify openntpd's postinst code for ntpd, ntpdate and ntp-utils
Apr 22 23:33:21 <CIA-7> agb * r10920 /packages/net/ntpd/ (Makefile files/ntpd.init files/ntpdate.init):
Apr 22 23:33:21 <CIA-7> [packages] ntpd
Apr 22 23:33:21 <CIA-7>  -Split ntpdate off of ntpd's init-script, ntpd works properly without it
Apr 22 23:33:21 <CIA-7>  -Make ntpd start just after ntpdate (if both are installed)
Apr 22 23:33:22 <CIA-7>  -UCI config for ntpdate may be added, but ntpdate will be retired upstream in the future
Apr 22 23:35:29 <CIA-7> nbd * r10921 /trunk/package/madwifi/patches/345-minstrel_sampling.patch:
Apr 22 23:35:29 <CIA-7> madwifi: put an upper limit on the amount of sampling that minstrel should do when
Apr 22 23:35:29 <CIA-7> the link changes its status from very good to slightly bad. the previous version
Apr 22 23:35:29 <CIA-7> of the patch could have resulted in erratic rate control behavior when there
Apr 22 23:35:30 <CIA-7> was a lot of sampling backlog.
Apr 22 23:36:25 <CIA-7> agb * r10922 /packages/net/ntpd/Makefile: [packages] oops, forgot to package ntpdate.init
Apr 22 23:52:09 <CIA-7> matteo * r10923 /packages/net/ntpclient/files/ (ntpclient.config ntpclient.init):
Apr 22 23:52:09 <CIA-7> ntpclient: fix the script on avr32, remove useless loop
Apr 22 23:52:09 <CIA-7> and replace it with a '0' count value.
Apr 22 23:52:09 <CIA-7> don't ignore config file count setting
**** ENDING LOGGING AT Wed Apr 23 02:59:56 2008