[13:38] <Keybuk> sadmac2: yeah, I'd thought of having a big table for lookups of nih_alloc contexts
[13:38] <Keybuk> one advantage to that would be the big table could be "pointer" referenced
[13:38] <Keybuk> so nih_alloc could adopt pointers allocated without it
[13:39] <Keybuk> jMCg: in fact, you hit the nail right on the head
[13:39] <Keybuk> the reason that Upstart doesn't (right now) have "run as user" support is because I want to do something a little more different and clever using capabilities and cgroups
[13:39] <Keybuk> the "run as user" feature will actually run things in a full PAM session for that user (as if they logged in and ran it)
[13:39] <Keybuk> whereas the "non-root service" feature will be much more useful
[13:44] <jMCg> Keybuk: that's very cool -- is there an ETA for that? Or anything I can help with (or code I can steal from you, to use for my long due port of mod_privileges to Linux)?
[13:57] <Keybuk> jMCg: sometime in the next 18 months
[15:58] <sadmac2> Keybuk: are you getting to reading the triggers thing I posted on upstart-devel?
[15:59] <sadmac2> Keybuk: I'm sure you're still swamped :) its just taking a lot of code to implement, and I'd hate to do all of it just to have you slap it down :)
[15:59] <sadmac2> With luck it will at least give us something to talk about in Portland
[16:00] <Keybuk> sadmac2: I haven't read it yet
[16:00] <Keybuk> I'll try and read through it later today :)
[16:00] <sadmac2> cool :)
[21:53] <pquerna> hi, i have a question about upstart... does anyone use it more like.. DJB's Daemontools for managing a small set of services? (not the whole os...)
[21:54] <sadmac2> pquerna: upstart doesn't like not being pid 1 unless you compile it specially
[21:54] <pquerna> yeah, I'm just looking for something somewhat more.. modern to replace daemontools. upstart seems like the coolest kid on the block though.
[21:56] <pquerna> $new_job uses a series of shell scripts spawning things into screen instances.... not exactly what i'd like to use forever :)