[LEAPSECS] Focus in the debate, alternative proposal

Warner Losh imp at bsdimp.com
Fri Jan 7 11:20:18 EST 2011


On 01/07/2011 05:54, Ian Batten wrote:

> When the leap second is applied is irrelevant for 24x7 operations. So

> the issue you raise only affects daytime only shops. What stops leap

> seconds being applied at local 2am? Are there really non 24x7

> processes where a 1s offset from UTC for less than one day, with the

> clock remaining monotonic, are an issue? People who get their leap

> seconds by slewing the clock to match NTP or a local reference don't

> care, only people who step their clock in response to a lead second

> announcement, so who are these people?


The ITU standard says it is applied at the end of the day UTC time, not
local time. Everybody has to agree when to apply it. This whole notion
of "what would it hurt to be 1s off" is the wrong way to think about
things and leads to crappy implementations of leap seconds in the first
place. What makes it wrong is that it is *WRONG*: UTC is supposed to be
the same everywhere.

Put another way, if I'm a trader in China trading in New York, I can't
be a second ahead or behind of them. Great effort is made to being
within microseconds of the time. The converse is also true: if I'm a
trader in New York executing a trade in China, I need up to the second
information, and if I'm a second off, bad things happen. If you look at
the news, you'll find that microseconds do matter. See
http://www.infoworld.com/d/the-industry-standard/hackers-find-new-way-cheat-wall-street-everyones-peril-699
if you think that it doesn't matter.

Warner



More information about the LEAPSECS mailing list