Empirically Assessing the Impact of Dependency Injection on the Development of Web Service Applications

Abstract

Service-Oriented Computing (SOC) has been broadly conceived as the next big thing in distributed software development. The software industry has embraced SOC through Web Services -functionality that is accessible via ubiquitous protocols such as HTTP-. This technology provides the basis for reuse and interoperability of applications across the WWW. However, consuming Web Services is still an expensive task in terms of development costs, since developers still have to invest much effort not only into manually discovering services, but also on providing code to invoke them, which leads to software that is polluted with service-aware code and therefore is more difficult to modify and test. Recently, a technique that has become very popular for building software is Dependency Injection (DI), which allows applications to be far more testable and maintainable. In this paper, we quantitatively analyze some of the benefits and costs of DI for building Web Service applications. We base our experiments on a refined version of DI that combines text-mining, machine learning, and best practices from component-based software development to simplify the way Web Services are discovered and consumed. To our knowledge, this is the first study on the impacts of using DI in the context of SOC.