Marc Roper - Overview of Current Projects

The evaluation and development of inspection-based defect detection technologies for OO software.
(In conjunction with Murray Wood and Alastair Dunsmore)

This work grew out of previous interests in the development and evaluation of testing techniques, and a critical examination of the impact of OO technologies on maintenance. Inspection techniques are widely used and effective approaches to defect detection but very few people have considered their application to OO systems. It appears that radically different approach to code inspection is required and a number of proposed techniques are being developed and evaluated.
Slides from our ICSE2002 presentation on this work.

The construction, verification and understanding of large OO architectures and frameworks.
(In conjunction with Murray Wood and Doug Kirk)

This work is currently concentrated on the area of frameworks - in particular the problem of how developers can quickly come to grips with such large pieces of software so as to be able to sensibly develop and deploy framework-based systems. This project is raising a number of related-issues, such as pattern-based documentation, reverse-engineering problems and a raft of testing and maintenance problems. This work is funded by the EPSRC - see Documenting Object-Oriented Frameworks: An Empirical Investigation for more details.

The application of metaheuristic techniques to software engineering problems.

This grew out of the challenge of developing test data generation tools. One of the approaches made use of genetic algorithms to automatically grow data to follow particular paths through a program. A number of other researchers in the UK were developing similar approaches and this grew into the EPSRC funded SEMINAL network which has widened the net to look at how a range of software engineering problems can be tackled by soft computing approaches.

Evaluation of testing technologies.

Despite the proliferation of testing techniques there exists a fundamental weakness - trying to establish how effective the application of a technique has been. For example, what percentage defects is the technique likely to find, what are the characteristics of these defects etc. etc. Some empirical work in this area is being carried out in conjunction with CERN .