Abstract
An exception-handling facility suitable for languages used to implement database-intensive information systems is presented. Such a mechanism facilitates the development and maintenance of more flexible software systems by supporting the abstraction of details concerning special or abnormal occurrences. The type constraints imposed by the schema as well as various semantic integrity assertions are considered to be normalcy conditions, and the key contribution of this work is to allow exceptions to these constraints to persist. To achieve this, solutions are proposed to a range of problems, including sharing and computing with exceptional information, exception handling by users, the logic of constraints with exceptions, and implementation issues. The use of exception handling in dealing with null values, estimates, and measurement is also illustrated.

This publication has 19 references indexed in Scilit: