Characterization of Software Design and Collaborative Modeling in Open Source Projects

Abstract

Software design is fundamental to developing high-quality, sustainable, maintainable software. Design languages, such as UML, have become the defacto standard in software design, but their infiltration in the mainstream practices remains vague. Recent studies suggest significant and increasing uptake in mainstream and open source spheres. Mining repositories and the software modeling artifacts often underpin the findings of these studies and focus on counting the instances of modeling artifacts as an indicator for adoption. This study aims to characterize this uptake in greater depth by focusing on analyzing the instances of models in open source projects. The goal is to uncover the profiles of developers who tend to create modeling artifacts, and those developers who maintain them throughout the project life cycle and to uncover the timelines of model creation and manipulation in reference to project evolution. This study sheds light on the nature of modelbased collaboration and interactions and characterizes the role of model-based artifacts well beyond mining their presence in open source repositories. The study finds that, despite the nominal increase in the presence of model-based artifacts, these artifacts are rarely maintained and are typically created by a small and unique set of practitioners. Models are often created early in the project life cycle and do not play any significant role in the collaborative development activities of the subject projects. Life span of these model files is relatively shorter than the code file life span. Unexpectedly, models tend to be more frequently updated and maintained when the project has a relatively fewer number of models.