Dynamic Test Composition In Hierarchical Software Testing

Abstract

Testing of large software systems is essential to assure the desired system reliability, but it involves large amounts of computer resources. It turns out that working in mid-range test granularity obtains very regular hierarchical test graphs, implying significant savings of testing-time, by dynamic generation of the strictly necessary tests. We formulate general analytic expressions for time savings and obtain the important result that savings increases with system size. The approach was demonstrated by actually implementing a testing system for large bundles of content files.