Browsing by Subject "adaptivity"

Designing software for a variety of execution environments is a difficult task. This is due to a multitude of device-specific features that must be taken into account. Hence, it is often difficult to determine all the available features and produce a single piece of software covering the possible scenarios. Moreover, with varying resources available, monolithic applications are often hardly suitable and require to be modularized while still providing all the necessary features of the original application. By employing units of deployment, such as components, it is possible to retrieve required functionality on-demand, thus adapting to the environment. Adaptivity has been identified as one of the main enablers that allow leveraging offered capabilities while reducing the complexity related to software development. In this thesis, we produced a proof-of-concept (PoC) implementation leveraging WebAssembly modules to assemble applications and adapt to a particular execution environment. Adaptation is driven by the information contained in metadata files. Modules are retrieved on-demand from one or more repositories based on the characteristics of the environment and integrated during execution using dynamic linking capabilities. We evaluate the work by considering what is the impact of modular WebAssembly applications and compare them to standard monolithic WebAssembly applications. In particular, we investigate startup time, application execution time, and overhead introduced by the implementation. Finally, we examine the limitations of both, the used technology and the implementation, and provide ideas for future work.