It’s efficient because it beats kqueue while reporting events accurately.

A proper benchmarking program is in the works, however manual testing does show only minimal resource usage.

For more, see this issue: https://github.com/e-dant/watcher/issues/10