Effective Exploration and Visualization of Large Execution Traces

Abstract

Understanding the behaviour of a software system can be made easier if dynamic analysis techniques are used. Runtime information is typically represented in the form of execution traces. Raw traces, however, can be extremely large - often millions of lines long. In previous work, we presented a tool called SEAT (Software Exploration and Analysis Tool), which is a trace visualization tool that supports several features for rapid exploration of lengthy traces. In this paper, we present the new features supported by SEA T, namely, the ability to plug-in new trace filtering algorithms, a usable control widget called PictureTree for displaying traces, and several new views that display useful information about the trace under study.