WG211/M10Kaestner

A Variability-Aware Module System by Christian Kaestner

Variability in software systems makes it challenging to develop and
analyze these systems. Especially in software product lines, in which we
can generate up to 2^n products by selecting from n features, it is no
longer feasible to check the generated products. Instead researchers
have investigated variability-aware analysis that checks the whole
product line. They found many approaches to analyze a whole product
line, but most of them require closed-world reasoning and are not
modular. In this context, we discuss a design of a variability-aware
module system. Each module can contain variability and can be (type-)
checked in isolation. We outline how open-world and closed-world
analysis play together, how to integrate annotative and compositional
approaches, and how a (usually global) feature model fits into this all.
We currently explore an implementation for C scaling for the Linux kernel.