On Sat, Jun 27, 2020 at 10:12 AM Roman Perepelitsa < roman.perepelitsa@gmail.com> wrote: > On Sat, Jun 27, 2020 at 7:01 PM Bart Schaefer > wrote: > > > > On Sat, Jun 27, 2020 at 12:27 AM Roman Perepelitsa < > roman.perepelitsa@gmail.com> wrote: > >> > >> Rounding up is the right thing to do for sleeps > >> and timeouts because it allows you to provide a guarantee. > > > > I would be mildly surprised if this doesn't happen as part of the OS > implementation of the operation. > > I've a tendency to use an overly direct language that's often > perceived by others (especially Americans) as aggressive. Perhaps I > was overcautious this time. No, that's not it. I just meant that the interval is ultimately implemented by calling nanosleep(), and I strongly suspect that nanosleep() rounds up.