Common Threads in Incremental Data Flow Analysis: A Comprehensive Survey

Abstract

Incremental data flow analysis employs techniques that update the data flow information based only on the modified parts of the code, thus reusing a lot of previously computed information. Since most real-world software systems evolve with time, incremental analysis techniques provide an efficient, and often the only feasible alternative to a complete (re)analysis from scratch. We describe how the existing incremental analysis techniques fall under a common reset and recompute paradigm. This has two-fold benefits. First, it facilitates us to survey a wide range of incremental techniques based on how they adapt this paradigm. Secondly, it enables us to identify gaps and open challenges in the field of incremental data flow analysis, to guide future research in this area.