The typical practice for building database applications is to use a direct representation, that is, directly map application concepts to the database structure. The direct approach is effective for applications with well-defined classes and attributes. However, it fails for applications with open-ended classes and attributes. Consider an application for managing equipment data. The pertinent data varies widely by equipment type. For example, storage tank data includes physical dimensions, materials of construction, and rated pressure. In contrast, pump data include flow rate, power consumption, pressure differential, and materials of construction.

Any attempt to hardcode equipment data is doomed to be fragile. New types of equipment and attributes would frequently arise, requiring model extensions, and disrupting the corresponding application.

With hardcoding, classes and attributes must be known at compile time as an application is being built. With softcoding, classes and attributes need not be known until runtime when the application executes. Softcoding lets both data and metadata (definitions of classes and attributes) be determined at runtime. Softcoding can be configurable, giving users the flexibility to define their own data structures.

• Jim X. Chen is an associate professor in the Department of Computer Science at George Mason University. He is currently the director of the Graphics Lab at GMU, visualization department editor of IEEE/AIP Computing in Science & Engineering, and general co-chair of IEEE VR2006. He served as program co-chair for IEEE VR (2002, 2003, and 2004) and guest editor for IEEE Computational Science & Engineering and PRESENCE. Chen’s research interests include computer graphics, virtual reality, visualization, networking, and simulation.
• Chen has taught about computer graphics and graphics tools for more than 10 years, including the use of JOGL (Java binding for OpenGL ) API. He is the author of Guide to Graphics Software Tools (Springer, 2002) and Computer Graphics and JOGL: OpenGL Programming in Java (in press) and several book chapters. He has also published over 60 research papers, holds two patents, and directed eight PhD dissertations; five of his students are currently professors in US universities.