[15:39] <lcapriotti> hi, what would be the best option to start an interactive script at boot with upstart?
[15:40] <lcapriotti> ie something similar to 'su -c "/usr/bin/ascript" -l' ?
[15:41] <sadmac2> lcapriotti: you would need to run it with the console output stanza
[15:42] <lcapriotti> ... any online examples ?
[15:42] <lcapriotti> tks btw
[15:43] <sadmac2> start on startup
[15:43] <sadmac2> console output
[15:43] <sadmac2> exec /usr/bin/ascript
[15:43] <sadmac2> ^^there's an example :)
[15:59] <lcapriotti> cool, it works :) tks a lot
[16:02] <lcapriotti> now for a more advanced one: what if I need to stop usplash before the script?
[16:03] <sadmac2> lcapriotti: you would add:
[16:03] <sadmac2> pre-start exec /whatever/thing/controls/usplash --stop
[16:03] <sadmac2> or something to that effect
[16:03] <sadmac2> not an Ubuntu guy ;)
[16:04] <lcapriotti> nice, seems easy once you know what you are doing ;)
[16:06] <Keybuk> or add "on starting your-script" to /etc/init/usplash.conf ;)
[16:06] <Keybuk> pre-start exec stop usplash
[16:06] <Keybuk> would work too
[16:08] <lcapriotti> I prefer not to modify a system script, I'll go for "pre-start exec stop usplash"
[16:09] <lcapriotti> tks a million to both of you!
[16:09] <sadmac2> np
[20:29] <milanbv> Keybuk: Hi again!
[20:30] <milanbv> I've finally come to working on upstart support in the stb without having had the occasion to talk with you about it :-)
[20:32] <milanbv> ...and sadly it doesn't seem we share the same availability hours
[20:37] <Keybuk> stb?
[20:39] <milanbv> system-tools-backends -;)
[20:39] <Keybuk> ah right
[20:39] <milanbv> have you seen bug 435935?
[20:40] <Keybuk> no, I haven't seen that bug
[20:41] <milanbv> that's basically about implementing a rough upstart support, so that services-admin can be shipped in Karmic without breaking things
[20:42] <milanbv> I have a patch that works with init.d scripts in Karmic (not upstart jobs yet)
[20:42] <milanbv> I'd need some feedback to know if that's good enough that I can release the stb 2.8.2, so that it can go into Karmic
[20:47] <Keybuk> Karmic is almost past Beta
[20:49] <milanbv> yeah, but that one kind of problematic
[20:49] <milanbv> we can't break services-admin more than it is now :-)
[20:51] <milanbv> sorry, my network connection is really bad
[20:51] <milanbv> (public WiFi, not even allowed to use IRC directly)
[20:51] <milanbv> did you say something since 21:48?
[21:34] <ion> Hi Keybuk :-)
[21:40] <Keybuk> hi
[22:18] <ion> Is there a way to subscribe to all current and future Upstart blueprints in Launchpad?
[22:21] <Keybuk> probably not
[22:21] <Keybuk> blueprint is very underdeveloped
[22:21] <Keybuk> I don't even use it now
[23:42] <ion> Any interesting decisions/news/something from LPC? :-)
[23:43] <ion> The most recent news at the linuxplumbersconf.org: LPC Registration opens 7:30AM PT Wednesday (come early!)
[23:44] <Keybuk> haha
[23:44] <Keybuk> no, it wasn't a great conf this time
[23:44] <Keybuk> casey and I hung out a bit
[23:44] <Keybuk> bounced a few ideas
[23:45] <ion> Has anyone came up with a good name for the ‘like while, but start on starting and stop on stopped’ stanza for upstart+1 yet? :-)
[23:45] <Keybuk> nope
[23:45] <Keybuk> but I am going to get rid of pre-start, post-stop, etc. from the state machine :p
[23:46] <ion> Why?
[23:46] <Keybuk> because we don't need them anymore
[23:46] <Keybuk> the blocking stuff added to events really gets rid of the need
[23:47] <Keybuk> so, let's say we simplify a job down to just a process, and events
[23:47] <Keybuk> we have
[23:47] <Keybuk>  [ pre-start event ] [ post-start event ] [ pre-stop event ] [ post-stop event ]
[23:47] <Keybuk>                      [ process being run here              ]
[23:47] <Keybuk> how would we do a pre-start script?
[23:47] <Keybuk> easy!
[23:47] <Keybuk>   on job pre-start
[23:48] <Keybuk>   task
[23:48] <Keybuk>   exec ...
[23:48] <Keybuk> that would block the pre-start event, so block the process being run ;)
[23:48] <Keybuk> which is exactly what we want
[23:49] <ion> In that case, it would be nice to be able to put multiple job definitions to a single file, so you’d have the stuff that e.g. does mkdir -p /var/run/dbus in the same file as the stuff that runs dbus.
[23:50] <ion> Ah, i might have misunderstood. It would still be possible to have pre-start, post-stop stuff in the job definition, it would be just an abstraction that works based on those events?
[23:50] <Keybuk> exactly!
[23:51] <Keybuk> even better, it'd be possible to put *any* custom sub-job in the definition
[23:51] <Keybuk> if we make that really easy, and work well
[23:51] <sadmac> Keybuk: the only downside is the hideous things I know people will do with this :)
[23:51] <Keybuk> sadmac: they can do them already
[23:51] <sadmac> Keybuk: on started pre-start job
[23:51] <Keybuk> just in separate job files ;)
[23:51] <Keybuk> sadmac: I kinda like that <g>
[23:51] <Keybuk>   while apache pre-start
[23:52] <Keybuk> (while the apache pre-start script is running)
[23:52] <ion> :-)
[23:52] <sadmac> that one's ok
[23:52] <sadmac> actually there'd be no started pre-start, since that would imply a started started, which would imply a started started started...
[23:53] <Keybuk> or pre-start implies pre-start pre-start implies pre-start pre-start pre-start
[23:53] <Keybuk>   on portmap pre-start pre-start pre-start pre-start
[23:53] <Keybuk> we should rename that to "turtle"
[23:54] <ion> ‘while’ for ‘start on started, stop on stopping’; the Old English word ‘hwile’ for ‘start on starting, stop on stopped’
[23:55] <sadmac> Keybuk: rename which?
[23:55] <Keybuk>   on portmap turtle turtle turtle turtle turtle turtle turtle turtle
[23:55] <Keybuk> turtles all the way!
[23:55] <sadmac> gdb.fail++
[23:56] <sadmac> Keybuk: ah
[23:56] <ion> ‘surrounding’?
[23:56] <sadmac> ion: you mean as a name for "before"?
[23:57] <ion> I’m not sure i like ‘before’. It only means, well, before, which is only the starting point.
[23:57] <sadmac> ion: me neither, we've been trying to rename it for months :)
[23:58] <sadmac> Keybuk: actually, we may not need before anymore with the new model
[23:58] <sadmac> Keybuk: while dbus / while dbus running
[23:59] <Keybuk> something like that yeah