@presidentsdaughter said:A value returned by POSIX's time(), which is no. of seconds since epoch, includes UTC leap seconds?
No. UTC is civil time, which assumes that leap seconds don't exist. When a leap second happens, it's usual to either just correct the clock immediately (not normally a big problem) or push the adjustment through over a period of time (e.g., adjust by one millisecond per second over the following 1000 seconds). Either of these works well for most applications. The others typically need GPS time anyway, or are NTP implementations that are written by people who ought to know WTF they are doing…