Analyzing the impact on identifiable defects in a code independent of typescript

Abstract

Javascript has become the most widespread programming language used to build software for many platforms, including web, mobile, backends, hardware, and desktop applications. It is a dynamically typed programming language that gives freedom to ignore types and build applications quickly. Still, Javascript is a poor language for identifying bugs at the compile time and maintaining large applications because of the dynamic behavior. The developers can use Typescript to add type syntaxes on top of JavaScript as a solution. However, not enough empirical studies exhibit how Typescript impacts detecting detects in applications. We decided to follow an empirically quantified "what-if" style of experimentation. We collected merged javascript bug-fix pull requests from selected open-source projects. Then we selected candidate bugs by applying our predefined criteria such as discarded pull requests with more than five files, code-refactoring, and configuration changes. We manually added Typescript annotation to the buggy code base and checked whether Typescript could detect the defects at the compile type. We assessed 500 bug-fix pull requests over five projects and identified that using Typescript over Javascript could have prevented 22.7% of bugs. According to our literature review, this is one of the few studies related to Typescript identifying bug impact. We believe this study will be significant evidence to consider using Typescript over Javascript in the future to reduce the significant number of bugs. This result will influence developers to adapt to the Typescript from Javascript. However, Typescript is not a silver build for Javascript because developers have to add extra code and complexity to the codebase. So, More research on Typescript is needed. In the future, we plan to explore the cognitive complexity of applying Typescript and the number of required lines to annotate. Furthermore, we plan to use a Typescript converter for annotating to increase the number of sample bugs to analyze.