[12:28] <ion_> Sounds good.
[10:35] <madduck> Keybuk: is it just me or why did the installation of upstart just pull in a finger client?
[10:35] <madduck> are you using finger?
[10:36] <madduck> i am currently challenged by the task of authenticating users via unix sockets
[10:36] <madduck> http://lists.alioth.debian.org/pipermail/netconf-devel/2007-June/000128.html
[10:36] <madduck> and since upstart also uses unix sockets (i think), this attracted my attention.
[10:36] <madduck> anyway, if you or anyone has some input to the issue in the mailing list post, i would love to hear it.
[10:37] <madduck> but for now: sorry for spamming the channel
[10:38] <shawarma> madduck: I was recently made aware of SO_PEERCRED.
[10:38] <shawarma> madduck: Doesn't that solve your problem?
[10:38] <shawarma> -> /msg
[10:39] <madduck> someone else just said this
[10:39] <madduck> i am reading up on this now. -> /msg
[10:39] <Keybuk> madduck: I've no idea why it would pull in a finger client
[10:39] <madduck> shawarma: or #netconf/irc.debian.org
[10:39] <Keybuk> it shouldn't do, there's not one in the dependencies
[10:39] <madduck> Keybuk: i can't figure it out either. maybe it's the scottish beer in my head?
[10:39] <Keybuk> maybe
[10:40] <Keybuk> upstart uses SO_PEERCRED for authentication of commands
[10:40] <Keybuk> (one of the reasons it uses unix domain sockets and not a fifo
[10:40] <Keybuk>  the other being the ability to pass file descriptors from initctl)
[10:40] <madduck> i am prototyping in python, so i hope it can do it.
[10:40] <madduck> i'll read up now.
[11:24] <Keybuk> wasabi: in particular "Eat My Data: How everybody gets file I/O wrong" :p
[02:04] <Keybuk> freudian typo, that
[04:04] <Keybuk> http://codebrowse.launchpad.net/~keybuk/upstart/main/revision/scott%40netsplit.com-20070612135558-5cp54lvyxc5pwax5?start_revid=scott%40netsplit.com-20070612135714-innzwjnyivysns66
[04:28] <ion_> Whee
[05:01] <AlexExtreme> \o/
[05:02] <AlexExtreme> "Retain loading from /etc/event.d" << does that mean that it'll use jobs in both event.d and the new jobs.d?
[05:02] <Keybuk> yes
[05:03] <AlexExtreme> cool, that'll make upgrading easier ;)
[05:09] <Keybuk> they'll fight over job names right now
[05:11] <AlexExtreme> that's no problem for me, really
[05:12] <AlexExtreme> anyway, /me gets on with porting profiles to the new code
hmm, i like your new blog theme</ot>
[05:28] <Keybuk> the port should be quite easy, since now most of the hard work is taken care of for you by conf.c
[05:31] <AlexExtreme> yep
[05:33] <AlexExtreme> brb, testing latest bzr code
[05:34] <Keybuk> 732 is current
[05:36] <AlexExtreme> yep, got that
[05:36] <AlexExtreme> seems to work. although, moving all the jobs to jobs.d from event.d before booting the new /sbin/init wasn't one of my greatest ideas :p
[05:37] <AlexExtreme> had to sync and reboot -f
[05:37] <Keybuk> lol
[06:15] <AlexExtreme> Keybuk: should CONF_PROFILE also go into the ConfItemType enum in conf.h?
[06:17] <Keybuk> yup
[06:17] <Keybuk> actually, you'll want that to be CONF_PROFILE_DIR, no?
[06:17] <Keybuk> and then add a case for it and parse directories of profiles?
[06:17] <Keybuk> oh
[06:17] <Keybuk> right, CONF_PROFILE_DIR in ConfSourceType
[06:17] <Keybuk> and CONF_PROFILE in ConfItemTypoe
[06:17] <Keybuk> :p
[06:20] <AlexExtreme> ok
[06:20] <AlexExtreme> :)
[06:52] <AlexExtreme> hmm
[06:52] <AlexExtreme> Keybuk: is it right that there's a Job entry specifically defined in the ConfItem struct?
[06:52] <Keybuk> it's in a union
[06:53] <Keybuk> so there should be a Profile *profile there as well
[06:53] <AlexExtreme> ahh
[06:53] <Keybuk> so in the appropriate case bits you refer item->profile
[06:53] <AlexExtreme> right, forget i said anything :p