Software development environments should include query handlers. Query handlers based on the relational database model are attractive because the model provides a uniform, non-procedural approach to query writing. There are two drawbacks to using the relational model to support query handlers in software development systems: (1) Standard relational database systems require that all information be stored in relations; however, the data structures used by existing software development

more »

... are generally non-relational, and it is impractical to replace them with relations. (2) The standard relational operators are not powerful enough to express certain important classes of queries. In [10] we proposed a model of editing environments, based on the use of relationally-attributed grammars, that supports a relational query facility. We introduced a new kind of relations, implicit relations, and a new approach to query evaluation to handle queries that use implicit relations. In this paper we illustrate the utility of implicit relations in contexts other than relationally-attributed grammars. We extend the definition of implicit relations and show how they can be used to support relational query facilities in software development environments without giving up the use of nonrelational data structures. Implicit relations can also be used to provide non-standard relational operations such as transitive closure. I N T R O D U C T I O N It is easy to see the benefits of including query facilities in components of software development environments, such as language-based editors, debuggers, and version-control managers. However, existing systems that provide query facilities generally do so in a limited way; the user is restricted to a pre-defined set of queries, and query answers cannot be used as inputs to further queries. This is because the query facilities are implemented in an ad hoc manner. A better approach would be to take advantage of current database technology, basing the query facility on the relational database model [4]. Under this model, arbitrm-y queries can be written by