Introduction

The central goal of this document is to provide the temporal database community with an extensive consensus test suite for temporal relational query languages that is independent of any existing language proposal. The test suite is not related to performance issues, but has a semantic focus and is intended to be an aid in evaluating the user-friendliness of temporal query languages. Thus, temporal query languages should ideally be able to express the included queries both conveniently and naturally. However, no definition of user-friendliness is included—this aspect is left to the individual users of the test suite.

The work that lead to this document was initiated in early 1993 when all researchers in temporal databases were invited to participate in creating an unbiased consensus test suite. An electronic mail distribution, tdbbenchmark@cs.arizona.edu, was used as the medium for the work on the test suite, and an initial working document (``The TSQL Benchmark'') was constructed by a total of 20 researchers. That document was presented at the ARPA/NSF International Workshop on an Infrastructure for Temporal Databases, held in Arlington, TX, June 1993, and was subsequently discussed among the 40 invited temporal-database researchers that attended the workshop. The present document is the result of the initial efforts and the efforts of the workshop participants, and as such it represents a consensus among a large fraction of the temporal database community.

The test suite consists of a database schema, an instance for the schema, and a set of queries on this database. The queries are classified according to a taxonomy, which is also part of the document. As a consequence of the central goal above, no existing temporal data models are used or mentioned. The database schema of the test suite is described using the ER model. The presented ER schema may be mapped to a set of relation schemas that fits a particular data model. With the exception of attributes illustrating user-defined time, the underlying temporal aspects are implicit in the description of the database schema. Of course, specific temporal data models might add explicit temporal attributes. The contents of the relations are described in natural language. The actual queries are also given only in natural language. The taxonomy is independent of any particular temporal query language.

The test suite is not intended as a substitute to other means of studying query languages, such as laboratory experiments with users or orthogonality studies. Rather, the test suite is intended as a complementary addition to the existing repertoire of query language evaluation techniques. It is emphasized that the test suite is not intended to constitute a metric for query language completeness, and as such it is not a substitute for a rigorous theoretical study of expressive powers of various temporal query languages. Such studies are still needed. While a sizable, or extensive, test suite was purposely constructed to ensure that a wide range of query language design aspects were covered, there is no formal basis for claiming that the list of queries is complete, or comprehensive. No such claim is made! It it emphasized that using the test suite as an advanced, quantitative scoring system for comparing languages makes little sense. Thus, one language is not necessarily superior to another just because one is capable of expressing more queries than the other.

In summary, the test suite may be understood as simply an unbiased list of queries. The queries are intended to aid in evaluating the user-friendliness of individual temporal relational query languages.

The presentation is structured as follows. Below, the intended scope of the test suite is defined. Sections , , and  are structured similarly. Each first presents design criteria, then presents a specific design. Section  concerns the database schema. The next section covers the database instance, and Section  concerns the classification scheme. The main body of the document is Section , which presents, using the classification scheme, approximately 150 queries.