Performance Measurement of C Software Product Lines

Software Product Line (SPL) are the answer to the rising demand for configurable and cross-platform systems. For such a system with just 33 configurable features, there are already more possible derivable variants than humans on our planet. Maintaining and analyzing these highly configurable software systems can be a difficult task. It is not uncommon that performance-related issues, especially in the maintenance phase, are a major risk to the longevity of a project. Variability encoding is the transformation of compile-time variability into load-time variability and is a technique that can be applied to SPL for further analysis. This work introduces the combination of performance measuring functions and variability encoding in order to gain performance related information for individual features and compositions of features. We also propose a method to make performance predictions program configurations by utilizing the previously mentioned feature data that was obtained through analyzing other configurations.