Graph-Based Analysis of Highly-Configurable Systems

Todays's software systems are getting more complex every day and contain an increasing number of configuration options to customize their behavior. Developers of these highly-configurable systems face the challenge of finding faults within the variable source code and maintaining it without introducing new ones.

In order to understand variable source code of even medium-sized systems developers have to rely on multiple analysis techniques. However, current analysis techniques often do not scale well with the
number of configuration options or utilize heuristics which lead to results that are less reliable.

We propose an alternative approach for analyzing highly-configurable systems based on graph theory.

Both variability models, which describe a system's configuration options and their interdependencies, and variable source code can be represented by graph-like data structures.

Therefore, we want to introduce novel analysis techniques based on well-known graph algorithms and evaluate them regrading their result quality and performance during runtime.