Browse Prior Art Database

On Minimal Sets of Operations for Relational Data Sublanguages

IP.com Disclosure Number: IPCOM000128287D
Original Publication Date: 1978-Dec-31
Included in the Prior Art Database: 2005-Sep-15
Document File: 10 page(s) / 28K

Publishing Venue

Software Patent Institute

Related People

Leland L. Beck: AUTHOR [+3]

Abstract

A minimal set of relational algebra operators is constructed. A set of primitive operations on tuples is derived from this; it is shown that the primitive operations are necessary and sufficient for the implementation of a relationally complete data sublanguage. These operations should provide a common basis for implementing different data sublanguages with minimal dupli-cation of software function. In addition, they appear to form the core of an intuitively attractive user-oriented language.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 19% of the total text.

Page 1 of 10

THIS DOCUMENT IS AN APPROXIMATE REPRESENTATION OF THE ORIGINAL.

On Minimal Sets of Operations for Relational Data Sublanguages

Leland L. Beck

Department of Computer Science Southern Methodist University Dallas, Texas 75275

February, 1978

P~

Abstract

A minimal set of relational algebra operators is constructed. A set of primitive operations on tuples is derived from this; it is shown that the primitive operations are necessary and sufficient for the implementation of a relationally complete data sublanguage. These operations should provide a common basis for implementing different data sublanguages with minimal dupli-cation of software function. In addition, they appear to form the core of an intuitively attractive user- oriented language.

INTRODUCTION.

A great variety of data sublanguages incorporating the relational data model have been introduced in recent years [8]. These range from mathematics-like languages based on the relational algebra [6,10) and the

relational calculus [6,71 to those based on examples of the desired out-put [13). In general, no language or class of languages seems to be clearly superior to all the others. Instead, it seems probable that dif-ferent classes of languages are most suitable for different classes of users (111.

One means for comparing data sublanguages is the notion of relational completeness [7]. A language is relationally complete if it permits the expression of any query expressible in the relational algebra (or the re-lagional calculus). Proofs of relational completeness have been published for several data sublanguages [5,9,12].

There have been many experimental and prototype implementations of relational data sublanguages (see [8] for a comprehensive discussion and bibliography).- Because of the wide variety of language types, however, each of these implementations is directed toward a sptcific language (or class of languages). Furthermore, these implementations generally supply their own access methods and provisions for authorization, integrity con-trol, etc. This situation would require an installation that wished to switch from one,language to another to change a large amount of its soft-ware. If it wished to support two different languages (perhaps for two different classes of users), or to support a network or hierarchical DBMS in addition to the relational, there would be a great deal of duplication of software function. The resulting increases in cost, maintenance, etc., would obviously make this situation highly undesirable.

Southern Methodist University Page 1 Dec 31, 1978

Page 2 of 10

On Minimal Sets of Operations for Relational Data Sublanguages

In this paper, we shall consider minimal sets of operators for a rela-tionally complete data sublanguage. We shall then propose a set of primitive operations on tuples which is necessary and sufficient for the implementation of any relationally complete sublanguage. These operations should provide a common base upon which many differe...