• Summary
• Related Documents
  • Active Bibliography
  • Co-citation
• Version History

Abstract:

In general, the test data generation problem is equivalent to the Halting Problem; therefore, it is undecidable. This does not have to be the case for specific problem domains. We propose a solution to the test data generation problem for command-based systems, and we call our method Domain Based Testing (DBT). DBT uses Domain Analysis and a Domain Model to automate test generation. Domain Analysis was originally developed to support software reuse. It is one way to extract common information about a problem domain. The result of Domain Analysis is a Domain Model. Domain Models represent the reuse problem domain and they serve as a mechanism to create instances of the reusable components. Instead of using the Domain Model for reuse, we use it as a structure from which test cases can be generated. Part of the Domain Model for DBT includes the syntax and semantics of the command language. Historically, grammars have been successful at test generation for compilers. On the other hand, grammars have not been successful at general purpose test generation because of the combinatorial explosion of semantic rules that must be written and maintained. Domain Based Testing addresses these issues in two ways. First, command language syntax is separate from command language semantics. Second, the test generation process is divided into three phases, (1) Scripting, (2) Command Template Generation,

Citations