Extending the Object Query Language for Transparent Metadata Access

Abstract

Many applications in Object Databases (ODB), for example, schema management tools, CASE tools, database development tools and integration wrappers, need extensive queries over both application data as well as metadata. However queries over metadata via OQL, a de-facto standard for object query languages defined for the ODMG 2.0 Object Model, are tied to low-level implementation details of the underlying schema repository of the database system. Hence, they are neither portable nor easily usable, requiring the application developer to have detailed knowledge of the proprietary structure of the schema repository. In this paper, we propose an extension of OQL, called MetaOQL, to address this limitation. Our proposition of MetaOQL offers several benefits: (1) it is a natural extension of OQL in terms of both its syntax and semantics; (2) it removes the dependency of metadata queries on the particular schema repository, hence providing uniformity and portability of metadata queries across different ODBs; (3) it supports transparent navigation over the metadata thus offering ease of use; (4) unlike OQL, it hides metadata querying details from the users hence the queries can be simplified and more easy to read and understand. We have also investigated implementation strategies for MetaOQL. In particular, we propose a translation strategy from MetaOQL to OQL as a preferable solution compared to development of a special-purpose MetaOQL processor. The translation strategy offers the advantage that the MetaOQL queries can be retargeted to work on top of any existing ODB engine equipped with OQL with minimal effort. Furthermore, all OQL query optimization strategies can thus still be brought to bear in our extended system.