Karl Reed

A preliminary survey on software testing practices in Australia

We present the findings of, to the best of our knowledge, the first survey on software testing practices carried out in Australian ICT industry. A total of 65 organizations from various major capital cities in Australia participated in the survey, which was conducted between 2002 and 2003. The survey focused on five major aspects of software testing, namely testing methodologies and techniques, automated testing tools, software testing metrics, testing standards, and software testing training and education. Based on the survey results, current practices in software testing are reported, as well as some observations and recommendations for the future of software testing in Australia for industry and academia.

A hypertext based software-engineering environment

HyperCASE, an architectural framework for integrating CASE tools under an extended hypertext system, is described. HyperCASEs objective is to provide a powerful, user-friendly, integrated development platform that can significantly raise productivity. Its specific goal is to support software developers in project management, system analysis, design, and coding. HyperCASE integrates tools by combining a hypertext-based user interface with a common knowledge-based document repository. It includes extensive natural-language capabilities tailored to the CASE domain. These are used in the interface to the software repository, providing an alternative to hypertext information management and interdocument navigation. English input can be analyzed during informal system-requirements specification, allowing a significant degree of automation for design and concept reuse at the earliest development stages. HyperCASEs three subsystems, HyperEdit, the graphical user interface, HyperBase, the knowledge base, and HyperDict, the data dictionary, are discussed.<<ETX>>

Requirements Classification and Reuse: Crossing Domain Boundaries

A serious problem in the classification of software project artefacts for reuse is the natural partitioning of classification terms into many separate domains of discourse. This problem is particularly pronounced when dealing with requirements artefacts that need to be matched with design components in the refinement process. In such a case, requirements can be described with terms drawn from a problem domain (e.g. games), whereas designs with the use of terms characteristic for the solution domain (e.g. implementation). The two domains have not only distinct terminology, but also different semantics and use of their artefacts. This paper describes a method of cross-domain classification of requirements texts with a view to facilitate their reuse and their refinement into reusable design components.

COCOMO-Based Effort Estimation for Iterative and Incremental Software Development

Incremental software development and delivery have been used in software projects in many ways for many years. Justifications for incremental approaches include risk amelioration, the management of evolving requirements, and end-user involvement. Incremental development, including iterative, incremental delivery, has become a norm in many sectors. However, there has been little work on modelling the effort in such development and hence a dearth of comparative analyses of cost models for incremental development/delivery. We attempt to rectify this by proposing a COCOMO-style effort model for incremental development/delivery and explore the relationship between effort and the number of increments, thereby providing new insights into the economic impact of incremental approaches to software projects.

The Use of a Meta-Model to Support Multi-Project Process Measurement

In todays environment, software companies are engaged in multiple projects delivered on heterogeneous platforms for a wide class of applications in disparate application domains. They are increasingly engaged in the co-development of software systems through joint software development projects including staff from partners and customers as well as their own. As a result, they must support multiple software development processes while trying to guarantee uniform levels of process enactment, and product quality across all projects. Our approach is capable of providing process measurement in a joint-project, multi-process model business environment. It is based on a simple meta-model for computing across-process, multiple-project metrics designed to permit monitoring of CMMI compliance. The open source tool Spago4Q has been developed to support our approach and is capable of producing the measurements needed for monitoring of a set of large-scale development projects using different process models, in a real industrial setting in Europe. The results support the view that that it will not always be possible to aggregate the same set of metrics across disparate process models.

Why we need a different view of software architecture

The definition and understanding of software architectures and architecture views still shows considerable disagreement in the software engineering community. The paper argues that the problems we face exist because our understanding is based on specious analogies with traditionally engineered artefacts. A review of the history of ideas shows the evolution of this understanding. A detailed examination is then presented of the differences that exist between the nature of the systems, the content of their large-scale representations, and how they are used in practice in the respective disciplines. These differences seriously undermine the analogies used to develop our understanding and this is discussed in terms of software engineering as a whole.

Computer-assisted analysis and refinement of informal software requirements documents

This paper describes RARE (Reuse-Assisted Requirements Elicitation), a method enabling software requirements engineers to process informal software requirements effectively. RAREs object is to assist analysts in transforming requirements expressed in natural language into a comprehensive collection of rigorous specifications that can be used as a starting point in software development. However, unlike other approaches to managing requirements documents, RARE focuses on the application of reuse-intensive methods of dealing with requirements documents, their contents and structure, and the processes involved in the analysis and refinement of requirements texts. The RARE method circumscribes an iterative process of planning, gathering and elaboration analysis, refinement, adaptation, integration and validation of requirements texts. The paper also describes the operation of IDIOM (Informal Document Interpreter, Organiser and Manager), a requirements management tool that supports the RARE method.

Towards a taxonomy for guiding multimedia application development

For research into multimedia development practices to progress, the actual processes and activities performed need to be recorded and compared. For accurate comparison, some agreed means of describing the material is needed. In other fields such as botany and zoology, taxonomies are constructed to facilitate such recording and comparison. We draw on the published literature on multimedia to obtain appropriate terms, and then constructs a taxonomy using traditional techniques, in particular, faceting. We compare our results with existing taxonomies and argue that the latters focus on aesthetics is not suitable for research into process and estimating issues. Finally, we discuss the use of the taxonomy as a basis for comparing multimedia projects, and as a basis for identifying the critical factors in multimedia project development. This will be used to support the development of improved process and estimating models.

EQUITY and the Problem of Return on IT Investment

We seem to need a better understanding of the process of exploring quantifying information technology yields--that is, the economic benefit flowing from an IT project proposal. As a result, the TCSE has decided to establish the IEEE International Conference on Exploring Quantifiable Information Technology Yields (IEEE EQUITY).

Automating Requirements Refinement with Cross-Domain Requirements Classification

Introduction of software reuse into the development process significantly alters the life-cycle model, and has a tremendous impact on all of its software development phases. This also includes the phase of requirements engineering, its activities and the tools used in the process. To deal with the consequences of incorporating reuse in the early phases of software development, we suggest the use of a special process model, capable of simultaneously meeting the goals of software reuse and the goals of requirements engineering. The RARE (Reuse Assisted Requirements Engineering) method proposes such a model. RARE places a particular emphasis on two aspects of requirements processing, i.e. requirements classification and domain mapping. Both of these aspects are especially helpful in the automation of requirements refinement into reusable designs.