The objective of the Application Experiment in OASIS is to test the use of object-oriented system development methods for administrative information systems. We believe that the use of object-oriented technology in this application domain significantly will improve cost/benefit and quality assurance, both for the development and the maintenance phases. This assumption is based on the capability of object-oriented technology to support:
a) An evolutionary system development process,
b) Effective design and implementation of windows-oriented event-driven user interfaces,
c) The reuse of concepts and mechanisms established earlier,
d) Implementing systems according to a distributed architecture,
e) The adaptability of changes in the system due to future changes in the requirements.
The transition to object-oriented technology implies a major shift in thinking and can be expensive if not done in an optimal way. Thus it is important not only to analyse the effect of the technology transfer, but also to observe the different aspects of the development process itself. Based on these assumptions, we would like to focus the objective of the Application Experiment on the following:
Applying existing and commercially available methodologies for object-oriented analysis, design and implementation in baseline projects in the user organisations. This implies technological training and assistance by some technological supplier (SINTEF). The methodologies in question for analysis and design are the OMT (Object Modelling Technique by Rumbaugh et. al.), the OOA (Object Oriented Analysis by James Martin) and the Norwegian OORAM (Object Oriented Requirements & Analysis Method by Reenskaug et. al.). The implementation will mainly be done using established object-oriented languages (C++ and Smalltalk).
In administrative information systems the requirements connected with points b, c, d and e varies considerably from one company to another. Three companies with a relatively wide range of requirements are involved.
The reuse capabilities of Object orientation may be one of the mechanisms capable of mending the 'software crises': the widening gap between the possibilities offered by the rapidly increasing quality and efficiency of computer hardware production, and the rather slow increase in corresponding software production and software production capabilities.
Normally the effect of using 'software ICs' ( reuse of standardised software components / building blocks) is expected to be seen after 3 - 5 sequential projects. This experiment is cutting this to 2 non-parallel projects in each environment, hoping to be able to detect the effect of reuse by inspecting the software development process closely, instead of using statistical methods.
Applying combinations of these methodologies in baseline projects, we will show that object-oriented system development methodologies constitutes a better basis for development of administrative systems than do traditional ones.