Date of Award

May 2023

Degree Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Engineering

First Advisor

Tian Zhao

Committee Members

John Boyland, Christine Cheng, Jun Zhang, Dexuan Xie

Keywords

Asynchronous Programming, JavaScript, Reactive Programming, RxJS

Abstract

Functional reactive programming (FRP) is a programming paradigm that utilizes the concepts of functional programming and time-varying data types to create event-driven applications. In this paradigm, data types in which values can change over time are primitives and can be applied to functions. These values are composable and can be combined with functions to create values that react to changes in values from multiple sources. Events can be modeled as values that change in discrete time steps. Computation can be encoded as values that produce events, with combination operators, it enables us to write concurrent event-driven programs by combining the concurrent computation as events. Combined with the denotational approach of functional programming, we can write programs in a concise manner.

The style of event-driven programming has been widely adopted for developing graphical user interface applications, since they need to process events concurrently to stay responsive. This makes FRP a fitting approach for managing complex state and handling of events concurrently.

In recent years, real-time systems such as IoT (internet of things) applications have become an important field of computation. Applying FRP to real-time systems is still an active area of research.For IoT applications, they are commonly tasked to perform data capturing in real time and transmit them to other devices. They need to exchange data with other applications over the internet and respond in a timely manner. The data needs to be processed, for simple analysis or more computation intensive work such as machine learning. Designing applications that perform these tasks and remain efficient and responsive can be challenging.

In this thesis, we demonstrate that FRP is a suitable approach for real-time applications. These applications require soft real-time requirements, where systems can tolerate tasks that fail to meet the deadline and the results of these tasks might still be useful.First, we design the concurrency abstractions needed for supporting asynchronous computation and use it as the basis for building the FRP abstraction. Our implementation is in Haskell, a functional programming language with a rich type system that allows us to model abstractions with ease. The concurrency abstraction is based on some of the ideas from the Haskell solution for asynchronous computation, which elegantly supports cancelation in a composable way. Based on the Haskell implementation, we extend our design with operators that are more suitable for building web applications. We translate our implementation to JavaScript as it is more commonly used for web application development, and implementing the RxJS interface. RxJS is a popular JavaScript library for reactive programming in web applications. By implementing the RxJS interface, we argue that our programming model implemented in Haskell is also applicable in mainstream languages such as JavaScript.

Download

Share

COinS