bocpy brings behavior-oriented concurrency to Python via cowns and @when behaviors, eliminating explicit locks and condition variables.
Key Takeaways
Cowns (concurrent-owned variables) enforce single-interpreter access using Python’s cross-interpreter data API, enabling safe multi-interpreter shared state.
@when decorator declares resource dependencies per behavior; the runtime schedules execution automatically once all required cowns are free.
Declaration order determines scheduling priority, so cook_omelette runs only after all ingredient prep behaviors complete, without manual synchronization.
Built-in Matrix class stores dense float64 data in C memory outside the Python heap, enabling zero-copy cown sharing across subinterpreters.
Erlang-style send/receive with selective matching and timeout callbacks provides lock-free cross-interpreter messaging backed by a C concurrent queue.
