On 25-8-2014 16:06, Ann Harrison aharrison@...
[Firebird-general] wrote:

>> On Aug 7, 2014, at 1:01 PM, "Lester Caine lester@... [Firebird-general]" <Firebird-general@yahoogroups.com> wrote:
>>
>> It does surprise me that the 100 microsecond resolution has been around
>> as I'm sure I've asked the question some time in the past!
>>
>
> Since 1984, approximately, in InterBase, and earlier in Rdb/ELN. For precisely the reason mentioned earlier: the 100 microsecond units (clunks) hold a day in 32 bits. Old-style dates, now Timestamps, are 64 bits. 32 time, 32 date.
>
> I believe that leap seconds disappear at the start of the next day. So if you're counting seconds and happen to bridge a leap second, you'll be wrong by a second. And probably timestamps at a leap second are messy.

Leap seconds 'disappear' in Unix time: It will count the second at the
end of the day, but then reset the count at the first second after the
leap second (or: count it again). So one way or another in unix time you
somehow lose the leap second. See also:
https://en.wikipedia.org/wiki/Unix_time#Leap_seconds

Mark
--
Mark Rotteveel