Rosetta: Compiling SysML v2 Behavior into Lingua Franca Modal Reactors

Systems Modeling Language (SysML) v2 standardizes the modeling of complex, reactive, time-dependent systems, but its Kernel Modeling Language (KerML) foundation gives behavior a declarative semantics: it characterizes which executions are valid without prescribing how a model is run. There is consequently no standard path from a SysML v2 behavioral model to deterministic, deployable reactive software, and today that gap is closed by hand. We present Rosetta, a compiler that translates the state-based and structural fragments of SysML v2 into Lingua Franca (LF), a deterministic coordination language whose programs are networks of reactors that communicate over ports in logical time. Unlike simulation-oriented targets, where determinism rests on coding discipline, LF fixes a single deterministic execution, and its modal reactors give the mapping a native state-machine abstraction. The translation maps state machines and part hierarchies construct-by-construct onto LF modal reactors. The constructs without a direct counterpart—state hierarchy, parallel regions with completion joins, timed triggers, cross-level priority, and deep exits—drive the encoding. We evaluate Rosetta on ten models, each equipped with in-model checks to ensure the generated program compiles and runs end-to-end. The programs reproduce their timed traces deterministically and are 1.4–3.3× the size of their SysML sources, an expansion Rosetta synthesizes rather than the engineer. Rosetta is available at https://github.com/glacier-project/sysmlc-rosetta.