In the termination detection barrier, the state is set to active be-fore stealing the task; otherwise the stealing thread could be declared inactive; then it would steal a task, and before setting its state back to active, the thread it stole from could become inactive. This would lead to an undesirable situation in which all threads are declared inactive yet the computation continues. Can you devise a terminating executor pool in which the state is set to active only after successfully stealing a task?
Already registered? Login
Not Account? Sign up
Enter your email address to reset your password
Back to Login? Click here