Composable, Demand-Driven Incremental Computation

Many researchers have proposed programming languages that support incremental computation (IC), which allows programs to be efficiently re-executed after a small change to the input. However, existing implementations of such languages have two important drawbacks. First, recomputation is oblivious to specific demands on the program output; that is, if a program input changes, all dependencies will be recomputed, even if an observer no longer requires certain outputs. Second, programs are made incremental as a unit, with little or no support for reusing results outside of their original context, e.g., when reordered. To address these problems, we present λ ic , a core calculus that applies a demand-driven semantics to incremental computation, tracking changes in a hierarchical fashion in a novel demanded computation graph. λ ic also formalizes an explicit separation between inner, incremental computations and outer observers. This combination ensures λ ic programs only recompute computations as demanded by observers, and allows inner computations to be composed more freely. We describe an algorithm for implementing λ ic efficiently, and we present ADAPTON, a library for writing λ ic -style programs in OCaml. We evaluated ADAPTON on a range of benchmarks, and found that it provides reliable speedups, and in many cases dramatically outperforms prior state-of-the-art IC approaches.