Skip to main content Link Menu Expand (external link) Document Search Copy Copied


Simple systems can be expressed accurately as linear relationships–straight forward cause and effect. More complex systems, however, transform those linear relationships into dynamic feedback loops where change to one variable’s value impacts several different parts of the system.

Our state chart is that simple system, states are either on or off. But when specifying what triggers them to go on or off we end up with a dynamic system instead.

Stocks are the first building block of a dynamic system. They are resevoirs of resources. In our REPL model we care about two scenarios:

  1. That the orchestrator goes from pulling the container to standing it up correctly
  2. That the cache always has enough memory to store references to active servers.

So to do that we create the following stocks

For the cache

def resources = stock{
    blocks: 0, // Used blocks of memory 
    table: 0, // Number of items in the table

For the orchestrator

def pool = stock{
    instances: 0,
    loading: 0,

Possible Values

In Fault, stocks can be numeric values (floats or integers), booleans, or unknown/uncertain types.

The Fault compiler treats all numeric values as doubles (Real numbers in SMT, essentially floats) in order to be friendlier to the solver.

For more information about different types of values see the section on data types