[time-nuts] Loran-C & French Clocks
magnus at rubidium.dyndns.org
Thu Mar 19 00:33:21 UTC 2009
M. Warner Losh skrev:
> In message: <49C17D52.6050703 at rubidium.dyndns.org>
> Magnus Danielson <magnus at rubidium.dyndns.org> writes:
> : I am still wondering who would set up NTP servers that provides UT1 time
> : in order to realize GMT over NTP. It would not be all that difficult as
> : UTC-UT1R is published regularly with advance estimates which could be
> : smoothed out and interpolated properly.
> I've never heard of such a beast... At least there wouldn't be any
> leap seconds in that kind of setup :)
Indeed. It could also be the only way to use NTP for GMT when needed for
legal activities without hacking away on the full protocol, only the
statum 1 server(s) would need to be hacked.
The side-effect is that it would be POSIX compliant if UT1 was accepted
instead of UTC in the time-mapping and a generous interpretation would
allow for it.
Poof and leap seconds would cease to be a problem and it would follow
the law of some countries. UTC does not need to be refined. :)
It is however evil to the bone as UTC and UT1 servers can't be mixed
(and should not be that) in the same set of sources because they would
be false-tickers to each other.
Also, it would break the NTP definition, but not necessarilly the
protocol and its implementations, except for details in the servers.
It would not take too much of hacking to achieve such a beast. There's
even a hacker-value in it I suspect.
More information about the time-nuts