Incremental Points-to Analysis for Java via Edit Propagation.

Points-to analysis is a static analysis technique which computes the relationships between the program variables and the heap references. It has been widely used in program optimization, program understanding, and error detection. Inclusion-based points-to analysis computes the points-to sets in a program by translating the program into a set of inclusion constraints on the points-to sets and then solving them to yield the desired results. Yet the analysis faces a difficulty in that a program can be frequently changed in its development, and great efforts may be exhausted to re-generate the inclusion constraints and re-solve them. In this paper, we extend the inclusion-based points-to analysis to an incremental one called Inc-PTA. The essential idea of Inc-PTA is to sum up the program changes into an editscript of a sequence of successive edits, and then to propagate the edits to the constraints followed by taking a demand-driven points-to analysis of the program. We also discuss about the correctness of Inc-PTA, and believe that Inc-PTA can provide with a cost-effective solution to incremental points-to analysis.