Walker, Robert (University of Calgary)

Purpose

Many industrial software systems are written in multiple programming languages and leverage varioustechnologies for intercommunicating between the disparate parts; we call these heterogeneous systems. Unfortunately, software development tools and their underlying analysis techniques are created to operate in homogeneous environments; while versions could be built for heterogeneous systems in principle, practical limitations arise: (a) poor portability to novel combinations of languages/technologies, (b) poor scalability to ever larger heterogeneous systems, and (c) poor evolvability when an expensive analysis has to be repeated in its entirety for even small changes to the software system. All these issues lead to prohibitively high costs when moving to enterprise-scale heterogeneous situations, causing industrial development to fall back on error-prone and inefficient manual approaches. To address these issues, we propose to meet four research objectives: (1) incrementally maintainable change impact analysis; (2) modular dependency analyzis in heterogeneous systems; (3) scalable encoding of modularized dependency graphs for fast analysis; and (4) integration between the three results and with an industrial toolset. Our industrial partner, Find it EZ Software, will obtain a novel solution for supporting analysis of enterprise-scale heterogeneous systems, not addressed by other solutions in this space. As a result, their customers (in the software development industrial) will be better able to modify their systems, leading to fewer_x000D_