Kurzfassung

Software engineering processes require facilities to handle both, large amounts of simply structured data, e. g. measurement results, as well as complex structured design objects, e. g. software modules or specifications at different levels of abstraction. Furthermore, large objects, e. g. texts or images, must be efficiently managed. As diverse as the information/data structures are, as diverse are the processing characteristics. While measurement data is evaluated at the database server,...Software engineering processes require facilities to handle both, large amounts of simply structured data, e. g. measurement results, as well as complex structured design objects, e. g. software modules or specifications at different levels of abstraction. Furthermore, large objects, e. g. texts or images, must be efficiently managed. As diverse as the information/data structures are, as diverse are the processing characteristics. While measurement data is evaluated at the database server, the actual, usually versioned, objects under design are reasonably processed nearby the application, i. e. within an object buffer located at a client (the software engineer´s workstation). The latter can best be supported by checkout/checkin primitives which, in turn, embody descriptive querying facilities for the provision/propagation of processing contexts. These wide spectrums of modeling/representation as well as processing requirements is not sufficiently met by current database technology - neither by relational nor by object-oriented systems. Only the combination of relational and object-oriented features which is currently a major focus of database research (buzzword "object-relational") promises sufficient flexibility to support the complete ranges. Especially the extensibility property associated with object-relational database systems is extremely promising, because it goes far beyond what is known from object-oriented systems. The extension interface of an object-relational system does not only allow the user to specify new data types w.r.t structure and behavior (methods), but also allows to perform a deeper integration by exploiting additional database services as, for example, tailored access paths and adapted query optimization. Although there are already commercial products available calling themselves "object-relational", it is by no means clear on how to optimally synthesize the relational and object-oriented worlds in order to achieve adequate modeling power and acceptable performance. Thus, this project aims at: Providing a deeper understanding on what integration of the relational and the object-oriented paradigms means, and delivering a set of basic object-relational modeling concepts which, in turn, allow to create database schemata and to perform schema evolution. Here, specially the upcoming SQL3 standard is taken into account. Proposing a client/server architecture for object-relational systems, giving the application the flexibility to choose the location (client or server) for given processing steps, and offering corresponding facilities for application programming. Developing data extenders tailored to support software engineering processes, e.g. versioning/configuring extenders or extenders for the evaluation of measurement data, and providing (data-control) primitives supporting the (higher level) software-process control, i. e. designflow and cooperation management. » weiterlesen» einklappen