Alias analysis: Beyond the code

Though impressive advances have been made in the precision and scalabilty of alias analyses over the last 20 years, their applicability to real-world object-oriented programs has actually decreased. The growth in size of standard libraries and application frameworks has far exceeded scalability improvements in alias analysis - even analyzing a Hello world program in Java has become non-trivial due to enormous standard libraries. Precision gains from greater flow and context sensitivity have been countered by greater usage of reflective constructs in programs, leading analyses to either unsoundly ignore reflection or compute very coarse results. Similar complications are emerging for large JavaScript applications, and the lack of static types in such programs can make computing even a basic call graph difficult [7]. Given these trends, it seems unlikely that further incremental improvements to traditional alias analysis algorithms will be sufficient to address the challenges of analyzing current and future real-world codes. © Springer-Verlag Berlin Heidelberg 2013.