High Level Efficiency in Database Languages

<p>The subject of this Ph.D. thesis is the design and implementation of database languages. The thesis consists of five articles:</p><p> </p><dl> <dt> [1] </dt><dd> Joan F. Boyar and Kim S. Larsen. Efficient Rebalancing of Chromatic Search Trees. In O. Nurmi and E. Ukkonen, eds., <em> LNCS 621: Algorithm Theory -- SWAT'92 </em>, pp. 151-164. Springer-Verlag, 1992. </dd><dt> [2] </dt><dd> Kim S. Larsen. On Aggregation and Computation on Domain Values. PB-414, Computer Science Department, Aarhus University, 1992. </dd><dt> [3] </dt><dd> Kim S. Larsen. Strategies for Expression Evaluation Using Sort-Merge Algorithms. PB-415, Computer Science Department, Aarhus University, 1992. </dd><dt> [4] </dt><dd> Kim S. Larsen and Michael I. Schwartzbach. Injectivity of Unary Queries With Computation on Domain Values. Computer Science Department, Aarhus University, 1992. Revised version of PB-311. </dd><dt> [5] </dt><dd> Kim S. Larsen, Michael I. Schwartzbach and Erik M. Schmidt. A New Formalism for Relational Algebra. <em> IPL </em>, 41(3):163-168, 1992. </dd></dl> <p>and this survey paper. In [5], a new query language design is proposed. The expressive power of the language is determined in [2] and all reasonable extensions are considered. In [3, 4], we focus on the optimization issue of avoiding unnecessary sorting of relations. The results in these papers are directly applicable to any algebra-based query language. In addition to the query language part, a database system also has to offer update facilities. The theory of standard tuple based updates is quite well developed in the sequential case. In [1], we discuss a new concurrent implementation of balanced search trees for that purpose.</p><p>This survey paper describes the results of the papers which form the thesis, and relates these results to each other and to the area in a broader sense than is customary in the introductions of individual papers. The paper is intended to be read in combination with the papers on which it is based.</p>