Generalized method for modeling complex ordered check constraints in a relational database system

Abstract

A system for enabling a user of a relational database to define and enforce ordered check constraints to insure that defined logical relationships are maintained between data values in pairs of adjacent rows when adjacency is defined by some ordering of the data. These inter-row conditions, here called 'ordered check constraints,' are expressed by a logical expressions which define a required relationship between the attributes of a given row and its adjacent row or rows. Arbitrarily complex expressions involving these sets of attributes can be formed to model the constraints of interest. These expressions can be created by the database in support of, for example, traditional primary key or uniqueness constraints, or they can be provided by the database user to model new more complex constraints such as a requirement that there be no gaps in a list of serial numbers. These constraints can then be efficiently supported in the presence of a b-tree or another ordered index structure by identifying any inserted, updated or deleted rows and evaluating the expression in the context of those rows and any associated adjacent rows. This approach is similar to that typically used in existing database systems to support primary key and uniqueness constraints, providing the user with the ability to model a much richer set of constraints using a generalized expression which extends and enhances the functionality of check constraints to permit relationships of data in different rows to be defined and enforced.