We propose a new semantics for shared-memory parallel programs that gives strong guarantees even in the presence of data races. Our local data race freedom property guarantees that all data-race-free portions of programs exhibit sequential semantics. We provide a straightforward operational semantics and an equivalent axiomatic model, and evaluate an implementation for the OCaml programming language. Our evaluation demonstrates that it is possible to balance a comprehensible memory model with a reasonable (no overhead on x86,~0.6% on ARM) sequential performance trade-off in a mainstream programming language.

We present a case-study of using OCaml within a large product development project, focussing on both the technical and nontechnical issues that arose as a result. We draw comparisons between the OCaml team and the other teams that worked on the project, providing comparative data on hiring patterns and crossteam code contribution.