> There is not much use for this "feature" to begin with, since
> 'sleep' by its very definition cannot be used for high-precision
> sleeps.

On an unloaded system of any speed, there are at least two orders of
magnitude between the practical inaccuracy of the sleep command's
implementation and "one second".  Sleeping for half a second or a
tenth of a second can be quite useful.

> and because it encourages people to write non-portable scripts

Any neat feature we add to a command has this problem, e.g. the find
-0 and xargs -0 change.  It still has value for scripts which only run
on NetBSD, and it has value if other systems follow our lead.  (Which
they won't do for market reasons, but which they may do for technical
reasons.)

But I do think that perry shouldn't have made the change
unilaterally.  (Maybe he didn't; his log messages seems to indicate
that he did.)