Abstract

This talk summarises the ideas of points-to and alias analysis which are increasingly used in modern compilers. It then introduces the ad-hoc approaches and approximations used in practice (for example the degree to which approaches are interprocedural) and notes these are often justified as being ‘for scalability’ but this phrase is often used without scientific evaluation.

We then discuss various long-term research efforts into program analysis at IIT Bombay. Although these started as distinct research efforts, they have converged towards a single agenda of precise and efficient pointer analysis (note that both precision and efficiency are necessary for practical scalability). In hindsight, a common theme that has emerged is that the quest for precision need not conflict with a quest for efficiency. With more careful modelling, it may well be possible to achieve them together.

This talk is relevant for an audience at multiple levels: at a practical level, it describes some interesting research investigations in program analysis. At a conceptual level, it contradicts the common wisdom that compromising on precision is necessary for efficiency. At a philosophical level, it highlights serendipity at work in seemingly distinct strands pursued over a prolonged duration weaving themselves into a unified whole.