[ntpwg] Further to the timestamping issue

[M. Warner Losh]

In message: <FB53E558-8FA8-4A51-9C96-493F171DAB22 at noao.edu>
            Rob Seaman <seaman at noao.edu> writes:
: M. Warner Losh wrote:
: 
: > That only matters if you think that having civil time that's closely  
: > tied to the mean solar time is important.
: 
: I'll keep this short since LEAPSECS provides the proper place for such  
: discussions.  There are any number of related threads over the years:
: 
: 	http://six.pairlist.net/pipermail/leapsecs
: 
: > After all, there's very few people that live at the meridian where  
: > the localtime is defined.
: > Today, my clock is just about an hour off from mean solar time due  
: > to daylight savings time.
: 
: This confuses periodic and secular issues.  A static or seasonal  
: offset is one thing.  A quadratically accelerating drift is another.   
: The world is riddled with dependencies on the length of the solar day.

A quadratically accelerating drift doesn't fit leap seconds well
either...  A few hundred years and we'll need leap seconds way too
often to be practicable without major changes...  Also, I'm not
confusing periodic and secular issues.  To keep noon constant, you
just move the local offset by an hour when it has time has drifted
enough...  It just expands the tolerance...

: > I know Rod loves leap seconds.
: 
: Rather I care about mean solar time.  Leap seconds are just a means to  
: an end.
: 
: > UTC can do leap seconds more often than twice a year.
: 
: NTP already handles this case, but it is a different issue than I was  
: discussing.

The ntp protocol does.  The last ntpd I looked at specifically
filtered out all leap second indicators that weren't done on the last
day of June or December.

Warner