Large software evolution, migration, or reengineering projects usually require a combination of different techniques to analyze, reverse engineer, transform, and visualize (legacy) software systems under evolution. As each project has different goals, toolchains supporting their processes need to be tailored individually to their specific requirements. Many tools exist, yet mostly only implement a single technique, and are usually not designed for interoperability. Therefore, for each project, a toolchain has to be built by selecting the techniques required, finding appropriate tools implementing them, and then integrating these tools. With little to no means of interoperability, this involves creating a lot of glue code and data transformations to “wire up” all tools in the desired ways, a tedious and error-prone task. It yields brittle and inflexible toolchains, as extending or changing the toolchain, or swapping one tool for an alternative implementation, will require to also write new glue code. Consequently, this code is also non-reusable, as it is usually hard-wired to specific interfaces of the tools glued together.

The lack of tool interoperability is recognized by the software evolution research community as a general challenge of the field. So far, existing approaches were aimed at integrating on a low-level via exchange file formats, building a closed platform without the ability to easily incorporate existing tools, or leveraging similarities of certain sets of tools, limiting application range.

The proposed PhD thesis puts forward the Sensei-approach (Software EvolutioN SErvice Integration), aimed at improving software evolution tool interoperability, and largely automate toolchain integration. Based on the fact that a large body of software evolution tools exist, yet they lack sufficient interoperability means to be easily integrated into the required, tailor-made toolchain, the following two main objectives are derived:

• Enabling software evolution practitioners to easily build toolchains tailored to their project-specific needs, focusing on the techniques to be employed, and the processes to be supported, while being as implementation-agnostic as possible, and abstract from interoperability issues.

• Enabling tool developers to easily build tools with standardized, interoperable interfaces, or extend existing tools, with as little limitations to implementation technology choices as possible.

The approach taken towards these objectives is based on viewing software evolution techniques as services, to abstract from interoperability issues. It entails

1. surveying tools and techniques, and compiling them into a catalog of standardized software evolution services,
2. utilizing existing, component-based technology to provide an integration framework using the catalog as a basis, and
3. providing a means to describe software evolution processes in terms of coordinated services, and to automatically generate toolchains based on the integration framework.

For validation, Sensei will be applied to real-world projects, e.g. for Q-MIG, a joint venture of University of Oldenburg’s software engineering group and pro et con GmbH, aimed at investigating the quality of software under migration, and building an integrated toolchain to support it. Furthermore, Sensei is thought to be applicable beyond the field of software evolution. To put this claim to the test, it is planned to apply the approach to at least one different field, e.g. toolchains supporting the development of energy-efficient applications, as well as a self-application.