PhD Oral Proposal: Chang Hwan Peter Kim

Date: December 3, 2012
Time: 11:00 am
Place:  ACES 6.442
Research Supervisors: Don Batory and Sarfraz Khurshid

Title of Dissertation: Systematic Analyses for Testing
Software Product Lines

Abstract:
A Software Product Line (SPL) is a family of programs in which each program consists of code fragments defined by a unique combination of features. An SPL allows structured development of common and different code fragments between the programs, which can significantly reduce both the time and cost of software development.  Checking an SPL against a test requires examining each program's execution against the test, which conventionally requires running each program from start to finish. This is expensive because an SPL represents a number of programs that is potentially exponential in the number of features. Our thesis is that an SPL's feature-oriented structure can be exploited by automated techniques based on static and dynamic analyses to reduce the execution space without introducing a prohibitive overhead. We present a suite of four complementary techniques that embodies this idea. Two, namely focused test execution and focused property monitoring, employ static analysis to prune feature combinations that are irrelevant for testing and runtime monitoring. The other two, namely shared execution and deferred execution, employ dynamic analysis to interleave multiple executions as a single execution, executing bytecode instructions common to the different executions just once. Evaluation of these four techniques demonstrates that they can be effective and advances the idea that despite the combinatorics of an SPL, its structure can be exploited by automated analyses to make testing more efficient.