Content

1. Overview
2. Test Case Design Strategies
   1. Black-box Testing
      1. Equivalence Class Partitioning
         1. Guideline for Selecting Equivalence Classes
         2. Develop Test Cases
      2. Boundary Value Analysis

Overview

• Common levels of testing

        Unit/Component test: defects in a single component
    ->  Integration test: defects in their interactions
    ->  System test: evaluating usability, performance, ...
    ->  Acceptance test: validation - did we build the right system?
  

• Goal
  • Effective: reveal as many defects as possible
  • Efficient: reveal defects with minimum costs

Test Case Design Strategies

Black-box Testing

Functional or specification-based testing; no knowledge of inner structure of code.

Equivalence Class Partitioning

Partition the input domain into equivalence classes where all members of a class will be processed the same way.

• Select 1 test case per equivalence class
  • Assume a defect revealed by a member of a class will be revealed by all others
  • If a member doesn't reveal a defect, no other members will, unless they are some subset of other equivalence classes

Guideline for Selecting Equivalence Classes

1. Range values
   • 1 valid covering the range
   • 2 invalid outside the range
2. Set of values
   • 1 valid covering all set of values
   • 1 invalid outside the set

Develop Test Cases

• Develop test cases for all valid ECs; a case may cover >= 1 EC (single fault assumption)
• Develop test cases for all invalid ECs; they are covered individually (the correct EC can mask the defect in another)

Single Fault Assumption

Failures are NOT the result of >= 2 faults occuring together.

Sufficient to partition into the number of ECs. Otherwise (multi-mode), partition into the product of all ECs.

Boundary Value Analysis

Probe on boundaries, where mistakes happen often, rather than elsewhere -> efficiency.

• Boundary value
• Value just below boundary
• Value just above boundary