Common optimization for timers with event loops

I discuss a common optimization with timer functionality in event loops implementations.

For a concrete example to reference in the above discussion have a look at kqueue which is part of FreeBSD (the core of Mac OS X):

  int kevent(int kq, const struct kevent *changelist, int nchanges,
	 struct	kevent *eventlist, int nevents,
	 const struct timespec *timeout);

Notice how it takes a timeout argument? The same is true for the Berkley sockets select call:

int select(int nfds, fd_set *restrict readfds,
           fd_set *restrict writefds, fd_set *restrict exceptfds,
           struct timeval *restrict timeout);

Different implementation with the same core underlying need. Make sense?