Towards assessing the complexity of object migration in dynamic, feature-oriented software product lines

  • Authors:
  • Stephan Adelsberger;Stefan Sobernig;Gustaf Neumann

  • Affiliations:
  • Institute for Information Systems and New Media, WU Vienna, Austria;Institute for Information Systems and New Media, WU Vienna, Austria;Institute for Information Systems and New Media, WU Vienna, Austria

  • Venue:
  • Proceedings of the Eighth International Workshop on Variability Modelling of Software-Intensive Systems
  • Year:
  • 2014

Quantified Score

Hi-index 0.00

Visualization

Abstract

Dynamic Software Product Lines (DSPLs) implement features of a product family, from which products can be derived and reconfigured at runtime. This way, systems can alternate their configurations without service interruption. The activation and deactivation of features at runtime pose challenges for the implementation of a DSPL, in particular for handling object states such as runtime changes to object-scoped variables, their value assignments, and the variable properties. To quantify the complexity of this object migration, we propose a systematic code-level measurement approach which harvests feature implementations and the corresponding variability models for code introductions responsible for critical changes to object states. We have applied our measurement process tentatively to data sets representing 9 SPLs implemented using Fuji. This way, we arrived at first insights on object-migration complexity in SPLs. For example, we observed that the number of feature-specific object states is distributed very unequally in Fuji SPLs, with a few objects having an overly complex map of potential object states and the majority of objects potentially seeing transitions between 1 and 5 object states. We also evaluated different tactics of applying SAT solvers to analyze variability models in this context.