The development of software for high integrity systems requires methods to provide assurance of the quality of the systems, relative to performance, conformance to standards or reference specifications, and dependability. One aspect of the many methods for measuring assurance relies on examining and testing the software product against reference data for a class of products and for a specific architecture within that class. The purpose is to prevent or detect software faults during development and potential system faults and failures before they result in operational failure. The Information Technology Laboratory's (ITL) Reference Data Repository of Software Fault and Failure Data currently contains a prototype query capability on project and fault data, and a draft handbook about fault types and recommendations for preventing and detecting faults. Graphics and text summarzing current data are available in a presentation.

Non-disclosure agreements with ITL protect industry, Federal agencies, and universities who provide the data for the repository. ITL is continuously seeking additional contributors to this effort. Interested parties should contact Michael Koo.

• a Fault Handbook providing information about fault classes for specific languages, or attributes and includes prevention and detection methods for the fault classes.
• the EFFTool available to industry to assist in their collection and analysis of fault and failure data for individual projects. The EFFTool, for either unix or NT systems, has been in the public domain since early 1998.
• a public data collection of project data (especially fault data) from private industry and Federal agencies.
• Related Information