• Summary
• Related Documents
  • Active Bibliography
  • Co-citation
• Version History

Abstract:

Each operation that involves code mobility can be divided into three steps: determining the code operation targets, transferring the code, and integrating the code into the target system. In a static system architecture, the target determination step can be done at compilation time. If the system architecture is dynamic, such that it is determined at run time, then the operation targets should be computed immediately prior to transferring the code. Following the target determination, the code is transferred by applying one of the design paradigms for code mobility, which extend the traditional client-server paradigm from data to code. Once transferred, the code can be integrated with the local target system by activating an instance of it, connecting it to existing data or code, or continuing its transfer over the network. Code mobility is supported by such environments as Java, Telescript, and D’Agents. However, current analysis and design techniques do not handle this concept satisfactorily. The standard object-oriented method, UML [8], enables modeling functionality through class services and message passing among objects. Concepts involving code mobility, such as Java applets, are modeled in several views using pre-declared extensions. Hence, modeling these concepts with UML is technology-dependent (e.g., specific to the Java language and its

Citations