On 3/28/2017 1:24 PM, Tony Finch wrote: > Joe Touch <touch@xxxxxxx> wrote: >> When UTC adds a leap second, nothing different happens to POSIX time. > Clearly not. Either it jumps backwards at the start of the leap second, or > at the end of the leap second, or it stops for a second, or it runs slow > for some period of time in a controlled manner (e.g. leap smear) or an > uncontrolled manner (NTP swinging around in a wild effort to resync). POSIX time just keeps ticking. It is only in relation to UTC that you would notice anything. And it isn't POSIX that is jumping - it is UTC. > The POSIX formula that specifies the translation from UTC to time_t > implies that it jumps back at the end of the leap second. > http://pubs.opengroup.org/onlinepubs/9699919799.2016edition/basedefs/V1_chap04.html#tag_04_16 That equation is clearly introduced as "A value that approximates..." Leap seconds and the difference between 86400 seconds/day and SI definitions are where the approximation errs. Joe