Lecture: Seven Principles of Testing

Introduction

• Seven guiding principles for software testing
• Important to understand these principles for effective testing

Principle 1: Testing Shows the Presence of Defects

• Testing demonstrates defects but cannot prove absence of defects
• Cannot claim a software is defect-free
• Reduces risk of failure but cannot eliminate it
• Example: "White swans" analogy

Principle 2: Exhaustive Testing is Impossible

• Testing all combinations of inputs/preconditions is unfeasible
• Instead, testing efforts focus on risks and priorities
• Use test design techniques like boundary value analysis, equivalence partitioning
• Examples:
  • One-digit numeric field with 68-70 tests
  • 15 input fields with 5 possible values each requiring an impractical number of tests

Principle 3: Early Testing

• Start testing activities as early as possible in SDLC
• Early defect detection is cheaper and more efficient
• Helps prevent costly fixes later
• Defined objectives for different phases:
  • Integration/system defects during development
  • User acceptance testing (UAT) to meet end-user requirements
  • Maintenance testing to ensure no new defects are introduced (regression testing)
  • Production testing to assure availability and reliability

Principle 4: Defect Clustering

• A small number of modules contain most defects
• Common in complex/tricky code areas
• Use historical defect data for risk assessment and planning

Principle 5: Pesticide Paradox

• Repeated use of the same tests will find fewer new bugs over time
• Regularly review, revise, and add new test cases
• Shift focus once 'hotspots' are cleaned up

Principle 6: Testing is Context Dependent

• Different testing approaches for different types of software
• Examples:
  • Family website: ad hoc testing
  • E-commerce site: more thorough testing, use of tools, and stringent guidelines
  • Safety-critical applications: strict guidelines and comprehensive testing

Principle 7: Absence of Error Fallacy

• Defect-free software is not useful if it doesn't meet user needs
• User satisfaction should be the goal
• Software should be usable and fulfill user expectations

Conclusion

• Recap of seven principles:
  1. Presence of defects
  2. Exhaustive testing impossibility
  3. Early testing
  4. Defect clustering
  5. Pesticide paradox
  6. Context dependency
  7. Absence of error fallacy