Professor Colin J Theaker, David J Leigh
Staffordshire University, Stafford, England

Neil Blackwood, Dr Robert Mason
Terrafix Limited, Stoke-on-Trent, England

This presentation considers the practical experiences of a commercial company when undertaking the move to an object oriented paradigm, and the impact that the paradigm shift has entailed, both in terms of the product quality, and the process for software development.

This work has been supported though a Teaching Company Scheme sponsored through the DTI and through the European Systems and Software Initiative (ESSI) as a Process Improvement Initiative (ESSI project: 27719 - PIOJAVA) undertaken collaboratively between the company, Terrafix Ltd and Staffordshire University.

The presentation will address the following issues:

• Background to the project.
  The company supplies many of the command and control systems for public services, such as the ambulance and police. Consequently their systems are highly interactive, real-time and safety-critical. This imposes particular constraints on the quality of the delivered product, and the needs of the company to provide auditable quality systems.
  These are particularly demanding system constraints which, it was felt, required a rigorous approach in the development of new systems architectures. The company objectives will therefore be presented, in terms of tangible measurable benefits.
• Impact on product and process
  A traditional approach using C/C++ had been used for software development prior to the project. It was envisaged that a move to an object oriented paradigm for both design capture and development would have an impact in terms of both the development process (particularly with respect to module reuse and distribution) and to the development product itself. The presentation will therefore introduce the anticipated positive impact of the paradigm shift, to identify areas in which an impact might be felt (and measured), and any initial negative aspects encountered.
• Experiment design
  As the company falls within the SME category (approx. 40 employees), it was feasible to consider a rigorous approach to the introduction of new paradigms, essentially treating the migration as a tightly controlled experiment for which the impact could be measured.
  The experimental design phase addressed two problems, namely the change of product quality, that could be assessed with reference to existing products of comparable functionality and complexity, and the impact of the paradigm on the development process, with particular regard to the need to retrain staff and introduce object oriented concepts.
  A metrics programme was introduced into the company to provide quantifiable information on both of these counts. This involved the analysis of a Reference Data Set, drawn from existing software modules, against which the object oriented software could be compared. Also the monitoring of the development of new object oriented modules was undertaken, over a time period and to include the induction of the staff and experiences of the development team in producing a new system. The experimental process therefore includes both a comparison with a 'control' and also monitoring of the development over time.
• Practical considerations for metrics collection
  As part of the experimental design, there were a number of significant practical considerations that had an immediate impact:
  A configuration management tool was introduced to enable the automatic monitoring of changes, complete with a recording of rationale and perceived impact.
  The issue of what metrics and how to collect them was a major issue.
  The paradigm shift could have had a significant effect on the personnel involved in development, but as the ethos within the company was that the programme was aimed at process and product improvement, rather than culpability, this was not an issue.
• Choice of metrics tools
  A number of tools were assessed and these will be introduced at the presentation. In particular, the correlation between measures designed for non-object oriented software and those specifically aimed at object oriented paradigms will be discussed. Also presented will be the choice of actual metrics, with their rationale.
• Reference data set.
  The reference data set includes measures from a set of real time programs developed within the company, including both application level software and deeply embedded software. Interesting characteristics have been identified within these, which provide a tangible research finding in their own right. These results will be presented at the Conference. The module characteristics will also be compared with the equivalent set of object oriented modules within the new implementation. These are being implemented in Java.
• Initial findings.
  As this is an on-going programme, results are emerging as the development proceeds. The current state of findings, particularly with respect to the object-oriented metrics, will be presented, in context.

Email: C.J.Theaker@soc.staffs.ac.uk

Email: D.J.Leigh@soc.staffs.ac.uk

Email: nb@terrafix.co.uk

Email: r.j.m@terrafix.co.uk