Introduction

The key features of the EYEDB OODBMS are:

• standard OODBMS features: persistent typed data management; client/server model; transactional services; recovery system; expressive object model; inheritance; integrity constraints; methods; triggers; query language; application programming interfaces ...
• language orientation: a definition language based on the ODMG Object Definition Language (ODL); a query language based on the ODMG Object Query Language (OQL); several manipulation language bindings (at least C++ and Java),
• genericity and orthogonality of the object model: inspired by the SmallTalk, LOOPS, Java and ObjVlisp object models (i.e. every class derives from the class object and can be manipulated as an object); type polymorphism; binary relationships; literal and object types; transient and persistent objects; method and trigger overloading; template-based collections such as set, bag and array; multi-dimensional and variable size dimensional arrays,
• support for large databases: databases up to several Tb (tera-bytes),
• efficiency: database objects must be directly mapped within the virtual memory space; object memory copy must be reduced to the minimum; clever caching policies must be implemented,
• scalability: programs must be able to deal with hundred of millions of objects without loss of performance.

We describe below how EYEDB meets these requirements. Section 2 introduces the storage manager subsystem. In section 3, the object model is exposed. Sections 4 and 5 describe the EYEDB object definition language and query language. Sections 6 and 7 deal with the C++ and Java bindings.