[02:26] <Robot101> I understood that my "task" job which was "start on runlevel [016]" would finish before other runlevel stuff would continue
[02:26] <Robot101> but it seems this isn't the case, so it gets terminated by the sysv task killer
[02:27] <Robot101> (I have a file sync from workstation back to server which would ideally be completed at system shutdown)
[02:29] <Robot101> (and also similarly, the reverse which needs to be done before gdm appears)
[03:48] <SpamapS> Robot101: runlevel is not waited on, so no, your task won't finish before other runlevel stuff will continue.
[03:49] <SpamapS> Robot101: what you probably want then is 'start on deconfiguring-networking'
[03:49] <SpamapS> Robot101: that is the latest possible moment before the network goes down
[03:55] <SpamapS> Robot101: IIRC sendsigs should actually wait because your task is in start/ goal state... it does not send kill to those processes.
[10:51] <Robot101> SpamapS: is runlevel special-cased by "task" somehow, and other events will be OK?
[11:12] <jodh> Robot101: not sure I understand your question. However, take a look at upstart-events(7) and you'll see why SpamapS suggestion will work (because deconfiguring-networking is a hook event that blocks).
[11:13] <jodh> Robot101: an alternative would be to have your job 'start on starting rc RUNLEVEL=[016]' which would block the starting of the rc job which causes the SysV shutdown sequence in this case.
[16:54] <toabctl> jodh, is there some spec about the dconf bridge you want to implement?
[16:55] <jodh> toabctl: it's partially done: https://code.launchpad.net/~stgraber/upstart/upstart-dconf-bridge
[16:55] <toabctl> jodh, yes. checked this already.
[16:55] <stgraber> still need to port it to gdbus, been busy with other things lately :)
[16:55] <toabctl> stgraber, maybe I'll help a bit
[16:56] <toabctl> stgraber, what about listening to the systembus and waiting for upstart to appear instead of an exit if upstart is not available on the bus?
[16:56] <toabctl> sessionbus, not systembus
[16:56] <jodh> toabctl: this bridge shouldn't be listening on the system bus - it's per user.
[16:57] <toabctl> jodh, yes. sorry
[16:58] <stgraber> toabctl: so, ultimately that bridge will be started by upstart itself once dbus is started, so there's no real need for the bridge to wait for the bus as upstart will guarantee it's spawned after dbus is ready (and respawn it if dbus dies for some reason)
[16:58] <toabctl> stgraber, ok
[16:58] <jodh> stgraber: could you add some details on this to the spec?
[17:44] <toabctl> jodh, stgraber how can I test the bridge? I guess I have to run upstart in my session..
[17:45] <stgraber> toabctl: right, do "mkdir ~/.init", then "init --session --confdir=~/.init/". That'll spawn an upstart instance on your session bus
[17:45] <toabctl> stgraber, does that work with upstart from raring? or should I install upstart from bzr?
[17:46] <stgraber> toabctl: works fine with upstart from raring
[17:46] <toabctl> great. thanks!
[17:48] <toabctl> funny. I get a lot of events when I resize a window because some programs (eg dconf-editor) store their window settings (height, width, maximized, ...) with gsettings.
[18:35] <toabctl> is there some html doc for libnih?
[18:59] <toabctl> stgraber, https://code.launchpad.net/~toabctl/upstart/dconf-bridge-gdbus
[19:00] <toabctl> stgraber, maybe a split of keyname and keypath would be good for the environment
[19:12] <toabctl> btw: where are the tests for the files in extra?