Simulation using Processes

HERE IS WHAT PROCESSES CAN DO:

HOW TO DEFINE A PROCESS:

1. Create a process class as follows:

   TYPE
       GIANTptr = ^ GIANT;
       GIANT    = OBJECT (procdescr)
          field1:  )
          .....    )  Globally accessible variables of this process type
          fieldN:  )
         constructor init (name:pname, index:integer, ..[specific inits] ) ;
           {--- OPTIONAL may beef up standard init --
            IF SO, MUST CALL ProcDescr.Init (name, index) !!}
         procedure lifecycle; VIRTUAL;   {MUST define a life cycle}
             [ other special methods ]
       end; 
   

   The minimum requirement is:

           Giant = object (procdescr)
               procedure lifecycle; VIRTUAL;
           end;
   

2. Define the process' life cycle as a procedure:

           procedure GIANT.LifeCycle;
           var
              [any private local variables]
           begin
              [do this ...]
              ....
              [... do that]
           end; 
   

   {processes may exhaust life cycle, in which case they terminate themselves, and are removed by the monitor}
3. Create instances, and initialize, using NEW and Init:

       VAR
            bigboy : GiantPtr;
   
       NEW (bigboy, init('Rumble-Buffin', 42, ... ));        
       RESUME (bigboy);        { to put on ready queue }
   

HOW the "MONITOR" WORKS:

       Monitor.RunSimulation (time_to_simulate, number_of_samples)

1. the main program, If the simulation time or number of samples has expired, or
2. the first process on the ready queue, or if none,
3. the first process on the sleep queue, and
   • advances the clock to its time, and
   • puts the deferred processes on the ready queue.

Is this DISCRETE OR CONTINUOUS simulation?

• a discrete-event process would advance the clock by arbitrary, perhaps large, time intervals.
• A (quasi-)continuous process would advance the clock by small "time-slices," in fact
• the same process could do both, at different stages of its life cycle.