Date of Award
Ph.D. in Engineering Science
H. Conrad Cunningham
J. Adam Jones
Transitional turbulence is a period of chaotic or unreliable variation in the state of a software system that results from changes in the system’s interconnected components. During these periods of instability, an external observer of the system’s state may “see” erroneous results. This is a problem that can affect visual user interfaces such as those in virtual and augmented reality applications and desktop or Web GUIs. In this research, we study two different reactive applications developed in C# on .NET. We reduce the transitional turbulence by augmenting the base applications with a dependency-graph-based event scheduling approach. The first study investigates desktop and Web GUIs. The second study investigates virtual and augmented reality applications built on the Unity3D game engine. The two studies use similar approaches, but both are somewhat embedded in the details of their applications and implementation platforms. In addition to presenting the two augmented applications, this dissertation characterizes the problem and its solution in a more general way. To do so, we use a design pattern to state the general problem-solution pair and enable it to be reused in similar contexts. We examine the two studies to identify their commonalities. We then unify the approaches by writing a new design pattern named Dynamically Coalescing Reactive Chains (DCRC). This dissertation both presents the new design pattern and records the systematic process we used to write it. To evaluate the design pattern and its usage, we apply it to the application in the first study as if we were approaching the application anew. The DCRC pattern facilitates the use of our approach for other applications and technologies and lays the foundation for further research on transitional turbulence and related software architecture issues.
Oliveira Marum, João Paulo, "DEPENDENCY-BASED REACTIVE CHANGE PROPAGATION DESIGN PATTERN APPLIED TO ENVIRONMENTS WITH HIGH UNPREDICTABILITY" (2021). Electronic Theses and Dissertations. 2122.