Abstract
Almost all of the tools for software construction, optimization, parallelization, verification, security, and understanding, depend upon the foundation of flow analysis, a ubiquitous and much-studied component of programming language technology. But modern software consists of increasingly larger code-bases written in languages that are higher-order. If the programming languages community is going to meet the paramount challenge of enabling effective construction of reliable software, our tools must scale to large code-bases and accommodate the higher-order nature of languages. Therefore, the scalability and applicability of our tools depend upon the scalability and applicability of flow analysis.

In this talk, I will provide new insights into the limitations of higher-order program analysis and describe how to leverage these insights to design new approaches. I will sketch these new ideas and argue why they lower verification costs, accommodate sophisticated language features, and can reason about modular programs; all of which contribute to the wide applicability of flow analysis and the verification, optimization, and understanding tools that require flow analysis.