Design and Implementation of a Virtual Machine for a Dynamic Object-Oriented Programming Language

Over the last years, dynamic, interpreted and JIT-compiled programming languages have become a primary choice for fields such as web development, data science, and artificial intelligence. The primary concept of these language interpreters is that they usually generate a set of machine code-like instructions (often called “bytecode”) instead of directly generating machine code for the targeted platform. This does not only allow programmers to follow program execution instantly but also to develop and deploy applications more efficiently, often in combination with agile practices. Their first simple functional prototypes are usually implemented in another high-level language such as Python, Java, or JavaScript due to larger flexibility for implementing algorithms and the language grammar and later converted to a low-level system language such as C or C++ with special emphasis on memory management and bytecode optimizations. This paper presents some of the common examples, approaches, and algorithms for implementing such systems with sufficient or acceptable portability and efficiency, and also describes the design and implementation of Stella, a simple, yet powerful and user-friendly dynamic programming language designed to provide an optimal infrastructure for scalable and safe object-oriented design.