No abstract

One of the reasons why large-scale software development is difficult is the number of dependencies that software engineers need to face: e.g., dependencies among the software components and among the development tasks. These dependencies create a need for communication and coordination that requires continuous effort by software developers. Empirical studies, including our own, suggest that technical dependencies among software components create social dependencies among the software developers implementing these components. Based on this observation, we developed Ariadne, a Java plug-in for Eclipse. Ariadne analyzes a Java project to identify program dependencies and collects authorship information about the project by connecting to a configuration management repository. Through this process, Ariadne can "translate" technical dependencies among software components into social dependencies among software developers. This paper describes the design of Ariadne, how it identifies technical dependencies among software components, how it extracts information from configuration management systems and, finally, how it translates this into social dependencies. Ariadne's purpose is to create a bridge between technical and social dependencies.

Ariadne is a novel visualization tool that allows end users to explore the socio-technical relationships in software development projects. Essentially the visualization is a variant of a social network graph. It is based on the observation that dependencies between software components create dependencies between the developers implementing those components. This relationship emerged in our own and other researchers' field studies of software projects. Large software development projects require management of dependencies by managers and developers to ensure the smooth coordination of work. We sought to evaluate our visualization to assess its utility. Although we had some informal trials with potential end users, we sought a deeper analysis before further refinement of the tool and evaluation on a larger scale. Usability inspection methods provided one potential avenue. Moreover, such inspection methods yield a kind of rationale not directly derived from human subjects evaluations. We report on the application of these inspection methods and discuss the implications of their results in the context of usability evaluations for visual interfaces.