The Convergence of Query and Object-Oriented Languages

The area of custom database applications is a large and rapidly growing domain, as evidenced by the appearance of tools for developing web front-ends to databases (e.g. Microsoft Visual InterDev), the large number of commercial programming environments now providing \database wizards" (e.g. Symantec Visual Caf e, Microsoft Visual C++) and the emergence of database interconnectivity standards (e.g. JDBC, ODBC). Most sophisticated database applications are a combination of code in a full object-oriented host language such as C++ or Java and a series of queries written in a declarative language such as SQL or OQL. The combination of object-oriented and declarative languages, as well as of diierent underlying data representations, creates an impedance mismatch { the two languages have diierent data and execution models. This paper explores the compromises made by diierent methods of reconciling those models. We contrast embedded SQL languages (C++ with embedded SQL, SQLJ), which ooer static pre-compiled query capabilities, with dynamic implementations (ODBC, JDBC), which support ad-hoc queries. We also examine the eeects of diierent data models (relational, object-relational, and object-oriented) on the impedance mismatch problem, as well as their eeects on optimizing declarative queries. We conclude with an assessment of the strengths and weaknesses of the diierent approaches, and ooer suggestions about where the database eld may be moving.