While the SQL/ODMG standardization signals the eminent emergence of a SQL-like query language for object databases, an adequate object algebra for query processing at the lower level is still missing. Hence, the quest for a decent object algebra is very much current and justified. The contemporary proposals that address this issue fall short of capturing the two basic concepts of object-orientation - inheritance and encapsulation. It is our contention that these two concepts need to be modeled at the algebraic operator level. Models that consider these two concepts as orthogonal to the query processing issue fail to provide the much anticipated object level abstraction of data and consequently defeat the essence of object-orientation at the outset. In this talk we propose an object algebra based on nested relational model that supports set based processing of data. The underlying model captures the notion of complex objects - sets, tuples and aggregation via object reference, object ids, inheritance with overriding and methods as in reflective relational programming. As an orthogonal extension, the proposed algebra can be non-intrusively tailored to capture the notion of encapsulation. Our approach decouples the conceptual view of the object model from its implementation concerns. Finally, we show that the well known relational model is a special case of our model and that most of the extended operators of our algebra can be simulated in its relational counterpart.

WHERE: TEC 340