Cleanroom Software Engineering

1
Cleanroom Software Engineering

• Getting it right the first time

2
A Spectrum of Methods
More Formal
OCL and Z
Cleanroom
Traditional Models Waterfall, Spiral, Incremental
Agile Methods FDD and SCRUM
Less Formal
3
Characteristics of Formal Methods

• Well-defined specification language
• Typically based on set-theoretical concepts
• Emphasis on verification
• Of program correctness
• Of completeness of description
• Of refinements to different abstractions
• Testing/Debugging De-emphasized
• Process idea borrowed from manufacturingworth
  the cost to ensure process delivers acceptable
  products rather than remove defects to achieve
  quality

4
Why Use Cleanroom Process?

• Claims
• Verification and testing are synergistic
• Reasoning faults are easier to find than
  debugging faults
• Testing based on usage scenarios focuses on
  important errors
• Not all faults are of equal significance
• Bottom Line software developed under the
  cleanroom process has fewer errors

5
The Cleanroom Approach

• Based on the Incremental Process Model with the
  twist that formal verification is applied to
  engineering models and the code.
• Subprocesses
• System Engineering, Requirements Gathering, Test
  Planning, Box Structure Specification, Formal
  Design, Correctness Verification, Code
  Generation, Code Inspection and Verification,
  Statistical Usage Testing, Certification

6
Additional Points of Emphasis

• Why the cleanroom process is not widely used (pg
  797)
• Stereotyped perception, departure from standard
  practice, and process maturity
• How cleanroom differs from OO development (pg
  800)
• Statistical QC, mathematical verification,
  usage-driven testing.
• Functional Specification Types (pg 801)
• Black Box (concerned only with I/O)
• State Box (similar to a UML class diagram)
• Clear Box (procedural design, e.g., pseudocode)