Concept Location and Program Comprehension in Service-Oriented Software

The maintenance process of service-oriented software might be different from the maintenance of software built using established technologies. This research aims at an investigation of existing maintenance methods for concept location and program comprehension and their applicability to new service-oriented environments as well as development of new appropriate methods. This paper outlines the goals and shortly discusses how these can be achieved. 1. Research Objectives To discover a demanded PhD research topic, twelve maintainers with various experiences and from various domain areas were interviewed and observed in their daily work. Usually, maintainers obtain their tasks as customer messages in the textual form. They need to find a corresponding piece of code where problem domain concepts are implemented in order to locate problem code or to find a starting point for program comprehension. Maintainers look for both real world concepts (like invoice or customer) and technical concepts (like handling of a variable or an implementation of certain functionality). This process is an essential part of maintenance and is called concept location [10]. Inflexible software is a bottleneck when changing business processes within an enterprise what results in the loss of competitive advantage. One of the approaches dealing with repeatedly changing customer requirements is the iterative program development, enhancement and maintenance. However, for the iterative approach the rendering of a smooth maintenance process is highly important [10]. Many researchers and industry specialists aim at improvement of software flexibility and scalability and at acceleration software development. One of the proposed solutions is the service-oriented architecture (SOA). However, in many cases the implementation of SOA concepts results in high code complexity, hardly comprehensible architecture and tangled code organization. Consequently, this can complicate the maintenance of such systems. Even experienced interviewees reported that they have difficulties with the maintenance of new complex solutions. To support program comprehension and concept location activities this research project analyzes either established maintenance methods can be reused under new conditions and how these should be accommodated if necessary. This research will help maintainers by providing methods and tools for dealing with code complexity of SOA-based systems and will significantly improve the overall maintenance process. Maintenance costs can be estimated by up to 90% of the total cost of software [2]. Hence, a research accelerating maintenance of new architectures is very important. The first hypothesis is that the process of concept location and program comprehension depends on the selected architectural approach. The second hypothesis is that existing methods for concept location are limited applicable in maintenance of new systems. The final hypothesis is that a use of appropriate tools for SOA-based software will increase maintenance performance. Each maintenance method [1, 7, 9, 12] has its area of applicability. However, discussions of these methods’ advantages and disadvantages cover mostly only well established architectural paradigms like object-oriented design or legacy software [7, 12]. Several researchers have investigated how software design impacts the maintainability and have compared specialties of procedural, object-oriented and aspectoriented design (encapsulation, inheritance, etc.) [2, 5, 6]. Although a number of methods have been proposed, the maintainability of the SOA remains unattended. The difficulty of concept location in new architectures is often underestimated. Neither deep research nor industry experiences have confirmed or denied a better maintainability of new architectures yet. Many challenges are still open.