All hail Havoc... just kidding. I just want you to know I know you are a heck of a lot smarter and more versed in this than I am. On Wed, 2005-01-26 at 16:22 -0500, Havoc Pennington wrote: > On Wed, 2005-01-26 at 10:32 -0700, Trever L. Adams wrote: > > > > What do you all think? Is this possible (I am fairly sure it is)? Is it > > a good thing... I think so? > > I think it's possible and good yes, the key thing though is to get it > coded up. > > For the first you need either a system daemon that monitors the > utmp/wtmp/lastlog/whatever-it-is or you need each user session to > register somehow with a dbus-specific thing > This sounds good. I will look into doing this... maybe I can contribute. > For the second you probably want a generic "list jobs that are running", > "notify when jobs are complete" kind of interface and have cron, > up2date, etc. use it. > > Havoc Well, I would like this to be via dbus if possible, since everything seems to be using it. However, I don't want a generic list. I want programs to know they are critical and to have to acknowledge they are still alive (from time to time up2date locks up for example). Some sort of emit a message or respond to a check message. Maybe, if this is like multicast with dbus... shut down scripts ask "Anyone important alive" and the program respond "Sure, my name is up2date, pid blah blah blah, doing blah blah blah" When no one responds, shutdown continues. Sound good or am I foolish? Trever -- "What we Are is God's gift to us. What we Become is our gift to God." -- Unknown
