Charla: Towards a Development Process for Maturing Delta-Oriented Software Product Lines

Sven Schuster is a PhD student at University of Technology Braunschweig in Germany.
16 Agosto, 2017 - 12:00
Auditorio "Philippe Flajolet" Sala 303, Edificio Poniente, B
Prof. Éric Tanter



A Software Product Line (SPL) exploits reuse-in-the-large to enable customization by explicitly modeling commonalities and variabilities of closely related software systems. Delta-Oriented Programming (DOP) a flexible implementation approach to SPL engineering, which transforms an existing core product to another desired product by applying transformation operations. By capturing product alterations related to configurable functionality within delta modules, DOP closely resembles a natural process of software development, which proves beneficial in early stages of development. However, increasing complexity for a growing SPL in later development stages caused by the invasiveness and implicit variation points of DOP drastically impairs maintenance and extensibility. Hence, a process utilizing the invasiveness of DOP in early development stages and restricting it in later stages would allow to properly mature growing delta-oriented SPLs. Moreover, ever-increasing complexity asks for a means to migrate into less invasive development approaches that are more suited for large-scale configurable applications. In this talk, I propose a development process including explicit variability points, metrics and refactorings, as well as a semi-automatic reengineering process of a delta-oriented SPL into a framework-based SPL. I sketch this development process with its constituents and



Sven Schuster is a PhD student at University of Technology Braunschweig in Germany. His research interests revolve around software product lines and variant-rich software systems. He is especially concerned with support for the development of large-scale delta-oriented software product lines. Other topics of interest include model-driven software and language engineering as well as variability-aware design patterns.