[PL-sem-jr] David van Horn: Confessions of a Computational Pervert (10/7)

[Aaron Turon]

David van Horn will be giving a PLJr talk on his work on control flow
analysis, and static analysis more broadly.  Due to schedule
conflicts, the talk will be held on *Tuesday*, 10/7, from 10:00 to
11:45, in WVH366.  Abstract follows.

--

Confessions of a Computational Pervert

Static analysis aims to predict properties of programs before they are
run, but how hard is this to do?  In this informal talk I want to
survey some results on lower bounds for program analysis of functional
languages.  I'll talk about type systems such as simple types, ML
types, and intersection types; and flow analyses such as simple
closure analysis, 0CFA, and kCFA.  In each case, I'll show how lower
bounds can be proved by perverting the analysis to do computational
work for you (like running your program).

The work on flow analysis constitutes my dissertation, so I want to
use this talk in part to share insights on completing a PhD.  There is
also, in my mind, plenty of work to be done in the area of flow
analysis for higher-order languages.  I'll sketch the current state of
the world and some ideas for future work.