Inferring the Principal Type and the Schema Requirements of an OQL Query

Abstract

In this paper, we present an inference algorithm for OQL which both identifies the most general type of a query in the absence of schema type information, and derives the minimum type requirements a schema should satisfy to be compatible with this query. Our algorithm is useful in any database application where heterogeneity is encountered, for example, schema evolution, queries addressed against multiple schemata, inter-operation or reconciliation of heterogeneous schemata. Our inference algorithm is technically interesting as it concerns an object functional language with a rich semantics and complex type system. More precisely, we have devised a set of constraints and an algorithm to resolve them. Our resulting type inference system for OQL should be useful in any open distributed, or even semi-structured, database environment.