|
The recommended methodology for use with Model-driven Testing tools involves the GOTCHA test generator and Spider test execution engine, as follows:
- Write and debug a behavioral model of the software with testing directives. The model is coded in the GOTCHA Definition Language (GDL). GDL is a procedural language similar to Pascal. You can use any text editor to write GDL. Because the model developer relies on the software specification to develop the behavioral model, defects related to inconsistency, incompleteness, ambiguity, and lack of clarity in the specification are discovered before the code is implemented.
- Write a testing interface between the model and the application under test. You can write the testing interface in a language understood by an existing test execution engine or by the Spider test driver. The testing interface bridges the gap between the abstract behavior model and the actual implementation in order to run the tests. For instance, you can write the test interface in Perl, in which case, the toolset provides tools to assist in translating the test to Perl scripts. However, MDT supports its own test execution engine called Spider. Spider is tailor made to run tests generated from the model. In this case, the test interface is composed of XML test execution directives and optionally proxy classes coded in Java or C++. Since the test interface developer relies on the documented or the implied interface to the implementation, defects in the interface are discovered at this stage.
- Generate one or more abstract test suites (ATS) derived from the behavioral model, using one or more of the GOTCHA test generation algorithms and/or different test generation directives.
- Review the abstract test suite using the Test Suite Browser tool. If necessary, modify the model, test constraints, or other test generation directives.
Review the model and testing interface with the development team and test team. Further specification defects are discovered at this review.
- Execute the test suites against the software unit under test, using an existing test execution engine such as Mercury's WinRunner, or using the Spider test driver. Spider executes the abstract test suite and writes the results to a standardized Suite Execution Trace (SET).
- Review the Suite Execution Trace with the Test Suite Browser. Coding defects are discovered at this phase. Observe the test results and, if necessary, augment or restrict the model or interface and repeat steps 4-6.
| |
|
