> You can program a completely deterministic parallel system using a simple fork-join model, can't you? No physical or timing factors interfere with that, can t

> You can program a completely deterministic parallel system using a simple fork-join model, can't you? No physical or timing factors interfere with that, can they? How simple is "simple" What happens if process 1 crashes? What if it suffers a network disconnection and looks like it crashes, but then tries to reconnect later? What is the correct logic for completing the combined task?