That would only add one extra process instance with each call. The pipe makes it add 2 extra processes with each call, making the number of processes grow exponentially instead of only linearly.
Edit: Also, Im not at a computer to test this, but since the child is forked in the background (due to &), the parent is free to exit at that point, so your version would probably just effectively have 1-2 processes at a time, although the last one would have a new pid each time, so it would be impossible to get the pid and then kill it before it has already replaced itself. The original has the same “feature”, but with exponentially more to catch on each recursion. Each child would be reparented by pid 1, so you could kill them by killing pid 1 i guess (although you dont want to do that… and there would be a few you wouldn’t catch because they weren’t reparented yet)
That would only add one extra process instance with each call. The pipe makes it add 2 extra processes with each call, making the number of processes grow exponentially instead of only linearly.
Edit: Also, Im not at a computer to test this, but since the child is forked in the background (due to &), the parent is free to exit at that point, so your version would probably just effectively have 1-2 processes at a time, although the last one would have a new pid each time, so it would be impossible to get the pid and then kill it before it has already replaced itself. The original has the same “feature”, but with exponentially more to catch on each recursion. Each child would be reparented by pid 1, so you could kill them by killing pid 1 i guess (although you dont want to do that… and there would be a few you wouldn’t catch because they weren’t reparented yet)