Version 1.0 is released. It is working reasonably well.
Things left to be done:
1. Time alarms should persist across crashes. The way to do this is to write 
   out the alarm when it is being set to /usr/spool/alert/<username>??
   and read it in upon startup. Maybe even 1 file will be enough?

2. Host watch. The function that exists in host.c has to be incorporated in.  
   A host has 2 states, UP or DOWN. Once the state changes from UP to DOWN,
   we check it only once every 15 or 30 invocations of the alertd main alarm.
   Since this happens once every minute a granularity of 15-30 minute
   before you know that a particular host is up results.
   Once the state changes from DOWN to UP we switch to checking every
   five invocations of the alertd main alarm. This means that ~5
   minutes may pass before you know that a host is down. This is
   necessary because one clearly doesnt want to do this every time.(too
   much overhead and I cant figure out a simple way to do this
   asynchronously with low overhead - the crufty CHAOS times out only
   after 15 seconds..)
   
3. The gdate routine should take defaults intelligently. If we say at
   8pm to set an alarm for 6 am it should default to the next day for example.

Sundar (Jan 20,1986)
