Hyper-linked Software Architectures for Concurrent Engineering

In this paper, the idea of considering software architecture (SA) as a tool for the gradual introduction of Concurrent Engineering (CE) issues in the industrial area is explored. SAs are placed among the first steps in the development of a large software system. Using them following the principles of CE implies some requirements on the supporting infrastructure, that can be met by hyper-linked information systems. The quickly growing World Wide Web (WWW) system fits all the requirements expressed, and makes a realistic choice for a gradual introduction of SA in production environments. It also serves as an integration framework for tools dealing with SA reasoning. For these reasons, it can be used to provide services oriented towards helping designers in the discovery, communication, and evaluation of software designs in an incremental way. SOFTWARE ARCHITECTURES Software architecture (SA) has been recently described as "a communication technology for the development of software systems, instead of a software engineering technology" (Brown 1996). We can find almost as many SA definitions as researchers are, but almost all of them coincide to include in the definition both models of the system being built, as well as constraints or guidelines for design (Shaw and Garlan 1996). This coincidence reflects a common understanding about the final users of SAs: the people developing the system. To be more specific, groups of people working on the developing system. The main uses of the SA can further clarify its actual objective as a communication technology. SAs are used for teaching newcomers the basics of systems being built. In this case, the main advantage of using a SA is that it gives an abstract overview of the system, helping developers to find their working context (this is especially needed when “programming in the large”). It is also worth noting that SA has been proposed by some authors as the foundation of Software Engineering teaching, in the form of paradigmatic solutions for well-known problems, using architectural styles (Buschmann et al. 1996). In the initial phases of development, SA helps to understand the system requirements and guide the discussion on features and resources required. In fact, the initial design decisions are proven to be crucial for the future system quality, and they must be taken with the agreement of different stakeholders (perhaps with different backgrounds). Then, SA acts as a "common ground" to all of them, facilitating the discussion in terms easily understood by all of the group. The assessment (on objective or subjective basis) of the SA with respect to quality requirements (mainly concentrated on non-functional ones) is an important use, often done by a group of experts working on the same model. For this matter, we propose to link all the information regarding the objective or subjective decision making process to the SA models. SA can be used as a reengineering vehicle for the maintenance and improvement of the current systems. It is a recognized fact that some of the main efforts towards SA research are driven by large companies with long life span systems that would need to be improved, and the very first step towards that is obtaining the "conceptual integrity" (Brooks 1996) of these very large systems. It would also be of primary importance to obtain the design trajectory for the decisions taken on the system during its life cycle. In this aspect, the usage of reengineered models to discover structural or behavioral patterns (Mendoza and Kramer 1997) could lead to obtain the “hidden” architecture. Unfortunately, there is no evidence towards a possibility of solving this automatically at this moment, so the only choice is human inspection (Eixelsberger et al. 1997). Again, the documentation nature of SA appears. The model portion of the SA is valid in the development process to assign work loads or to define subsystems to be developed by different work groups. But the guidelines included in the SA are even more important for the controllable and repeatable design and implementation of the system. Initially, these guidelines can be handled as a "Frequently Asked Questions" list, or as a company’s style book. Our proposal is to give access to this information in a context-sensitive way, and offering several views depending on its intended use. The SA evolves in time, acting as a design trace on the development of systems of related domains. Initially, the SA composed by the Architecture Description Language (ADL) models (Carnegie Mellon University 1997b) and the guidelines should be enriched adding information about previous systems developed. Then, the SA should be