Cacti: a front end for program visualization

We describe a system that allows the user to rapidly construct program visualizations over a variety of data sources. Such a system is a necessary foundation for using visualization as an aid to software understanding. The system supports an arbitrary set of data sources so that information from both static and dynamic analysis can be combined to offer meaningful software visualizations. It provides the user with a visual universal-relation front end that supports the definition of queries over multiple data sources without knowledge of the structure or contents of the sources. It uses a flexible back end with a range of different visualizations, most geared to the efficient display of large amounts of data. The result is a high-quality, easy-to-define program visualization that can address specific problems and hence is useful for software understanding. The overall system is flexible and extensible in that both the underlying data model and the set of visualizations are defined in resource files.