Data flow analysis algorithms gather facts about the use and definition of data in programs. Incremental data flow analysis only calculate information affected by the program changes and avoid total recalculation. This research investigates the application of incremental techniques to the analysis of large software systems. The research plan is: o to complete performance studies of an elimination-based incremental update algorithm using ISMM, a prototype implementation for interprocedural analysis of C programs, o to empirically study the behavior of a new hybrid iteration-elimination incremental data flow analysis algorithm, o to develop more precise analysis techniques to deal with aliasing through pointers in C-like languages, o to study issues involved in scaling up incremental data flow analysis techniques for programming-in-the-large, o to extend previous empirical studies of the calling structure of C programs, their typical parameter and pointer usages.
