Activity: Participating in or organising an event types › Participation in workshop, seminar, course
Description
Adopting a middle-ground view between Michael Jackson's ("features are
behaviours”) and Christian Kaestner's ("features are configuration options")
definitions, we consider features as units of variability specified within a feature
model. Feature models can be used to document valid choices (called
configurations) formed by combinations of features. Such configurations can
either relate to desired elevator behaviours’ interactions (normal use, emergency
use, etc.) or to viable Linux kernels. The number of configurations derivable from
a given feature model grows exponentially with the number of features, making
the testing process inherently difficult. To harness combinatorial explosion of the
number of configurations to be considered, we propose to sample them by
computing t-way interactions from the feature model. We present initial
experiments and a search-based approach maximising dissimilarity between
configurations. This approach mimics combinatorial interaction testing techniques
in a flexible and scalable manner.