Search Based Testing

This is creating test cases with a goal. For example, to increase coverage of a test suite, we need to generate test cases that exercise certain statements or follow a path.

Components

1. Search algorithm
2. Representation: Encoding of the problem solution
3. Search operators: Modifications of encoded solutions
4. Fitness function: Measure how good a candidate solution is
5. Test Execution: Execute the tests
6. Instrumentation: Collect traces for fitness calculations during execution

We often use Hillclimbing for our search algorithm which we cover in Fuzzing.

A fitness function can be:

• Code coverage from JaCoCo
• Mutation score from Pit
• Genetic Algorithm
  • Here:
    • Generate the initial population
    • Compute fitness
    • REPEAT
      • Selection
      • Crossover
      • Mutation
      • Compute fitness
    • UNTIL population has converged