Graduate Research Talk: Why Objects Die: Improving Program Tracing Tools

April 14, 2009
12:00-1:00 pm
Halligan 127

Abstract

Garbage Collection has considerable software engineering benefits, but not with out significant impact on performance. Improving the performance situation requires more research and, in order to accelerate this research, program traces are commonly used. However, existing tracing tools do not provide enough information to answer the question “What caused this object to die?” in a way that relates to the programmer's view of the program. In this talk, we discuss modern garbage collection algorithms, existing tracing tools, and our implementation of the Mearth Tracing Tool.