Efficient Instrumentation Strategies for Software Feature Location

Description

Software maintenance continues to be the most expensive part of the software life cycle. The cost of software change is high costs because of the need to locate and understand the existing code to make changes safely and cleanly. The Software Reconnaissance technique has proved effective in earlier projects but its application at some SERC affiliates has been hampered by the difficulties in tracing execution of large, long-running, multi-process, time-sensitive, systems. This project is developing and evaluating instrumentation methods to make Reconnaissance feasible for such systems.