Update Languages vs. Query Languages

Abstract

Two kinds of basic operations can be performed on a database: update it, i.e. modify the data values stored in it, and query it, i.e. ask questions about its data value. In this work we deal with update and query languages for the relational database model, introduced by Codd in [1], Some query languages were developed since the origin of this model, e.g. the first-order relational calculus and the relational algebra. A precise concept of completeness, based on computability, for relational query languages was defined by Chandra and Harel in [2]. Updates in the relational database model were considered systematically as a research topic in [3]. Restructuring of relational databases was considered in [4, 5]. In [6, 7] a family of relational update languages was defined. The languages introduced in [6, 7] are complete update languages (deterministic and non-deterministic). By completeness it follows that these languages are also capable of expressing all computable queries, as defined in [2]. However, the constructs chosen in [7] do not necessarily reflect this, but capture the essence of update operations. The completeness proof in [7] is very detailed and built from scratch.