Testing research directions: an industrial perspective

1
Testing research directions an industrial
perspective

• Ashok Sreenivas
• TRDDC, Pune

2
Primary problems

• Still very expensive
• 40-50 of time/effort
• Guarantees on quality very poor
• What kind of defects cannot occur?
• What have I tested?

3
Test evolution

• All software has very long lives
• Maintenance / enhancement cycles
• Developing regression test-suites
• Test selection
• Some work done on test prioritisation
• How to evolve test-suites for such systems?
• Testing often inadequate and ad-hoc
• Completeness, coverage

4
Use of formal methods

• Precise descriptions
• Can verify correctness at each step
• Possibly other aspects than correctness too
• Guarantees on quality of testing
• Quality improvement
• Tools are a possibility
• Cost reduction
• The test-oracle problem?

5
Domain specific formalisms

• Formal spec idea not new!
• Unfortunately, has not worked
• General purpose formalisms may be too general!
• Domain specific formalisms may be useful
• Telecommunications SDL, MSCs
• Help in productivity and quality improvements
• Systems with a persistent component?