Guaranteeing Configuration Validity in Evolving Software Product Lines

Abstract

Software Product Lines (SPLs) are an approach to capture families of closely related software systems in terms of commonalities and variabilities where individual variants are defined by configurations of selected features. Specific (partial) configurations may be of particular importance to SPL manufacturers, e.g., if they are very popular or used by major customers. SPLs are subject to evolution, which may inadvertently break existing configurations, e.g., if a previously selected feature does no longer exist. This is problematic as it may delay or completely prevent creation of previously existing important variants causing monetary loss and customer dissatisfaction. In this paper, we present a method to lock specific configurations to ensure their validity during evolution of the SPL. For this, we present Temporal Feature Models (TFMs) and dedicated evolution operations as a semantic-enriched first-class notion for evolution of feature models, which we use to assess the impact on existing configurations. Using the presented method, it is possible to guarantee that locked configurations remain valid during SPL evolution and make statements on which part of the evolution would break the configurations. CCS Concepts •Software and its engineering → Software notations and tools; Software evolution;

175% Modeling for Product-Line Evolution of Domain Artifacts

“…e )} returns the pairs of versions θ i and feature conditions φ θ i e for which the element is contained in the version-specific 150% model with annotation α θ i 150 (e) = φ θ i e . As an element e is contained in different SPL versions θ i , the associated annotations φ θ i e potentially differ as well, e.g., due to restrictions in which variants the element should be contained or due to feature model evolution [31,32,37]. To allow for valid feature annotations in a 175% model, for each θ i , we require a respective feature model FM θ i SPL to ensure a valid mapping in α 175 .…”

“…In literature, SPL evolution is mainly tackled with feature models [7,14,31,32,37]. Gamez et al [14] use cardinality-based feature models to capture and reason about evolution, whereas Bürdek et al [7] apply model differencing.…”

“…Gamez et al [14] use cardinality-based feature models to capture and reason about evolution, whereas Bürdek et al [7] apply model differencing. In contrast, Pleuss et al [32] with EvoFM, Seidl et al [37] with hyper feature models, and Nieke et al [31] with temporal feature models propose extended feature models to capture the variability and the evolution by the same formalism. Compared to 175% modeling, those approaches only focus on feature models and are not adaptable for SPL domain artifacts.…”

“…Comprehensive tool support for SPL evolution coping with feature model as well as artifact evolution is provided by Dar-winSPL [30], DeltaEcore [38], and SPLEmma [34]. We are currently working on the tool support for 175% modeling and plan to also incorporate temporal feature models [31] for handling the feature model evolution similar to DarwinSPL.…”

“…Existing approaches consider SPL evolution on the feature model level [31,32,37], by adopting delta modeling [15,24,38], by using aspect-oriented programming [2,3,13], or specify SPL evolution templates [29]. However, to the best of our knowledge no approach exists in the literature for the annotative category of variability modeling techniques such that version as well as variant information is captured together in a single model, i.e., elements are annotated with both version and variability conditions.…”