In a previous study six software verification tools have been applied to a representative space software package. The findings reported by each tool have been compared in order to derive footprints regarding fault identification. In a continuation three more tools were applied to the previously selected application software and to another application together with two tools previously used in order to broaden the base of evaluation. More aspects were considered regarding the evaluation of results: an additional evaluation criterion was added and a comparison of reported defects with the outcome of unit tests was performed. Due to a higher degree of formalization and automation the manual evaluation effort could be decreased while extending the number of considered reports and the number of tools. The encountered evaluation and verification issues are discussed in detail. All results together shall provide a detailed view on the defect identification capabilities of the considered tools w.r.t. current software base. Altogether, the high quality of reports as obtained in the previous study was not obtained again: in context of a different set of tools and another (object-oriented) language a lot of trivial reports were observed.

Testing as a method of software verification is limited in that it can only prove the presence of defects, not their absence. To be useful, a large number of test cases may be needed, a strategy that is often in conflict with project constraints such as available time and funds. Test automation may be considered as an interesting approach to alleviating this conflict. However, test automation requires accurate and computer-accessible information about the system to be tested, both in terms of the interfaces by which the system is to be stimulated as well as the desired properties of these interfaces. Within the FASTII activity (FAST=Flow-optimised Automated Source-code based Testing) the possibility of deriving this information from available requirements and design documents is being investigated. Preliminary results of this investigation as well as suggestions for future changes in the process are presented in this paper.