Fujitsu's Automatic Test-Generation to Cap Excessive Testing in Agile Software Development

SUNNYVALE, CA: Fujitsu Laboratories announces the development of automatic test-generation technologies to limit excessive testing work and enhance efficiencies for agile development. Fujitsu’s new technology is a transformative approach to developing software as it incrementally adds small-scale functions and then one function after another, finally making them available as a complete solution.

In conventional automatic test-generation technology, when source code enhancements are made– sometimes previously available tests cannot be used to test the enhancement. Implying that it is necessary to generate and add new tests. In agile development, generally there are frequent revisions and when the number of added tests becomes excessive it becomes a problem to handle test cases, check test results and maintain test cases.

Fujitsu Laboratories have efficiently analyzed the relationships between input variables and the revised sections of the source code, so that test code can be reused at the input variable level. Using this technique, Fujitsu Laboratories have developed automatic test-generation technologiesthat limit the increase in the number of test cases due to source-code revisions.

In a trial that applied Fujitsu’s technologies to a revision of open source software with approximately five thousand lines of source code, Fujitsu Laboratories confirmed that it was possible to dramatically reduce the increase in test code to 1/24th of previous levels. This newly developed technology will enable efficiencies in development through test automation and also shorten overall development times.

Key features of the technology include:

Technology that reuses test code at the input variable level

With this newly developed technology, as the test code is rewritten at the individual input variable level, it is possible to regenerate test cases efficiently.  Fujitsu Laboratories developed test-code regeneration technology that can limit the increase in test cases by revising the test code only for input variables that are related to the changed sections of source code.

 With previous technology, in response to source-code revisions, each test case had to be reused as is, a newly regenerated test case had to be added, or the original had to be replaced, and all such choices had to be made at the test case level.

Technology to efficiently find the input variables related to a line of source code

Identifying the input variables related to revised sections of source code requires a large amount of memory. Fujitsu Laboratories developed a technology that can find the input variables related to a source code line while greatly reducing the amount of memory used, by representing the input variables related to a line of code as paths in a graph and partially sharing these paths with other lines of source code.