1 Send mail on success and/or failure.
3 How to deal with stuck / long-running jobs?
5 - We should have instance/timeout settings that allows us to deal
6 with jobs that run too long:
8 job "foo" timeout 1 hour ... # timeout the job after an hour
9 job "bar" single instance ... # only one instance can run
10 job "bar" kill previous ... # kill previous if still running
12 Implement native dynlink.