Study On Variation Point In Software Reuse Based On Functional Dependency

There are three fundamental problems that have not been solved in theory as well as in practice about Software Reuse: Whether do the masses of reusable components exist? How to identify the usability of the reused components? How to use the reused components? These three problems are relevant to the concept of Variation Point in Software Reuse. The Normalization Theory with Relational Data system, especially Functional Dependencies Theory is introduced to research the problems of Variation Point in Software Reuse. They are applied to Design Patterns and the Specification of the interfaces of Component Model in EntityBean on the next. While they are NOT Sufficient Condition and NOT Necessary Condition for the success of Software Reuse, they are meaningful to show and emphasis the problems of Variation Point in Software Reuse. The thought of Normalization in Relational Data system is to eliminate the improprieties in Data Dependencies with steps, that is, to some extents, to partition the Relation Schema, as well as the principle of design patterns in schema, One Thing One Place. That is, make a Relation with one conception, or one entity, or one relationship between the entities. So, if there is more than one conception in a Relation, they should be partitioned. In a word, Normalization is essentially the simplification of conceptions. The Reuse-Driven Software Engineering Business (RSEB)has been practiced in AXE,Ericsson,Sweden. Software Reuse based on RSEB traces the variation points and the corresponding varieties from UseCase components to the end-point where the components and the varieties are reused. A Method to Study on Variation Point in Software Reuse Based on Functional Dependency,called VRFD, is introduced. In theory, the problem of the first Normal Form is discussed firstly, and the following are the relationships between two respective Dependencies, Partial Dependency and Transfer Dependency, and Variation Point. On the bases of the theoretical study, the method is applied to Design Patterns and the Specification of the interfaces of Component Model in EntityBean. VRFD is applied to Design Patterns. First of all are the conceptions of Class Inheritance, Interface Inheritance, Abstraction and Encapsulation in the Object-Oriented Method. The following are 9 design patterns. With these design patterns, the discussion goes into further depth, talking about both the partial dependencies and the transfer dependencies; further more, the CONSISTENCY about the two Dependencies is explored. Then the transition to the conception of Responsibility-Partition is accomplished. At last is the opinion of studying Design Patterns. Applying VRFD to the Specification of the interfaces of Component Model in EntityBean is the end. While EJB is defined to be a distributed computing architecture based on Component, Java RMI, which is a fundamental part of EJB, is discussed. Then the conception of Responsibility-Partition is used to research the Specification of the interfaces of Component Model in EntityBean, which is the intuitive way to handle Variation Point in Software Reuse. Introducing Functional Dependency in Relational Data system to research the problem of Variation Point in Software Reuse, VRFD point out the equivalence of the two conceptions, handling the Variation Point with VRFD and Responsibility-Partition. Further more, with formalization language introduced, the problem of Variation Point in Software Reuse is explored in-depth.