On Thu, May 22, 2008 at 05:08:44PM -0700, Stephen Hahn wrote:
>   I would do it this latter way as well.  Two configuration sources, one
>   execution path.  You'll also need to spin up a thread or other means
>   of listening for restarter events.  I believe we have a trivial
>   restarter skeleton somewhere...

Today cron puts jobs in separate process contracts and abandons them (it
still, of course, waits for the process it kicks off).

I don't see any reason to change that, save, perhaps, that watching for
events like core dumps, SIGKILL, ... and being able to report them to
the user would be a nice addition.

Nico
-- 

Reply via email to