State Machine Compiler

I am working in software testing, specifically automatic test cases generation. Among the existing forms of test cases, my focus is on the test cases that are composed of sequences of events such as _.event1.event2 eventx()

However, the events can be classified into: sensitive and insensitive. The latter does not affect the system's states, and hence, it can be ignored; while the former affects the states. Anyhow, the sensitive events in the test cases may lead to states explosion and there is a need to prevent that. Therefore, some techniques suggest using one variable to present states and group all similar states together such as using len variable in circular queue. Relatively, the states can be represented by using specific drawings such FSM.

For example, the test cases for circular queue may look like:

add(0).remove().add(1).Front()
add(0).add(1).remove().Front()

which produce the following states:

len=1, rear=0, front=0 and dataQ[0]=0
len=0, rear=0, front=1 and dataQ={0}
len=1, rear=1, front=1 and dataQ[1]=1

len=1, rear=0, front=0 and dataQ[0]=0
len=2, rear=1, front=0 and dataQ[1]=1
len=1, rear=1, front=1 and dataQ={1,0}

As can be seen, every addition/deletion produces a new state. A state is composed of 4 variables: len, rear, front and dataQ. The 1st three variables are integers while the dataQ is an integer array. Nonetheless, the states produced by different test cases can be identical which wastes effort and time. So, there is a need to optimize these states. The search techniques were suggested where the problem can be represented as a search problem and the technique is applied. If we consider Len as a state, then we will have: len=0; 0QSize. However, this does not represent the state but it suits for classifying the states into groups.

In terms of states representation, State Machine/Map Compiler (SMC) was suggested as a modeling mechanism that takes the state machines (i.e. FSM) drawing and generates the code in any preferred language. In SMC, the FSM is represented in a specific syntax (state---transition----next state) and saved in a file (.sm). This file will be compiled by SMC to generate a context class which includes definitions of states, transitions and actions in FSM but still need to be triggered by another class. This class has to call the transitions that modifies the state.

We had created that class and implemented all the methods with their transitions. However, the FSM used was based on 1 variable only (i.e. len). Besides, we are still looking for the SMC results as they will be the input for any search technique to be applied. Supposedly, the states generated by SMC can be used directly in the search technique but this is still questionable.