Hi Mateusz,<br><br>For question 2, appearently if any worker process dies, it will also cause the supervisor to be killed. That's why the unlink is there. For example, if the supervisor process is started, and you call a worker process with incorrect number of arguments it will cause the supervisor to be killed.
<br><br>You are right about question 3, it makes a lot of sense now.<br><br>For question 1, I tried out the code from the book, and I still don't understand why the spawn is there fully. You could be right about the quicker initialization, but I think there's another reason for it. If I call the start() from the shell, I can't access the worker process that is managed by the supervisor. In fact, I have no idea how I can access the worker process at all. However if I call
<span class="q">start_in_shell_for_testing() in the shell, I can access the worker process like usual.<br><br>Aaron<br></span><span class="q"></span>