Task-specific source code dependency investigation

Abstract

We present a simple, visual approach to help developers view and navigate structural dependency information, aimed specifically at pragmatic reuse tasks. Our visual approach, implemented as the Gilligan tool, uses standard GUI widgets (such as lists and editors) that developers are familiar with. Gilligan represents complex dependency data in a simplified format, appropriate for investigating reuse tasks. We present a small-scale, semi-controlled experiment that indicates that the approach permits more accurate identification of relevant structural dependencies with a lower time investment, as compared to traditional manual approaches. Last, we discuss the potential for the approach to aid in other specific software understanding tasks.