PP-P: Extraction of Mini Apps

Code complexity is a major challenge when developing and using tools. Developing tools that analyse or transform real world code becomes overwhelming from the technical side alone if they would have applied on full production software from the very beginning, the same holds for teaching how to use the tools.

Instead small excerpts of real software, so called Mini-Apps, that have certain, well-defined characteristics that are of interest for the tools to be developed, are used. Their extraction requires at the moment expert knowledge and manual work.

The goal of this part project is to develop a method and tool for the semi-automatic Mini-App extraction. The main challenges are to manage the overhead of measurement-based kernel identification techniques as well as the adaption of existing modeling techniques for kernel identification. Further, existing slicing techniques need to be extended to be suitable for extracting useful Mini-Apps. The workflow of the approach to be developed in this part project is shown below: