[LEAPSECS] What is GMT?

[Michael Sokolov]

Warner Losh <imp at bsdimp.com> wrote:


> (1) NTP is defined by the RFC to be UTC today,


Yes, but which version of UTC?  Some NTP server operators may very
well choose to continue serving a leaping version of UTC (what I call
UTC1972) well past 2017, for a couple of reasons:

a) because this "redefined UTC" is no longer a valid form of UT, some
may very well regard the redefinition as illegitimate and continue
using the old definition;

b) the existing RFC said "UTC" when this term unambiguously meant a
time scale with leap seconds, hence some may very well choose to
interpret the RFC as specified a leaping timescale regardless of what
someone else later does to the UTC acronym.  And even if someone later
writes a new RFC that explicitly specifies a leapless ITU2012
timescale, operators of existing NTP servers built to the old RFC
could use the age of their existing equipment as casuistical
justification for continuing to serve a leaping continuation of
UTC1972 instead of ITU2012.

Folks, the leaping version of UTC will NOT stop in 2017 no matter what
the ITU may wish.  If Monsier Gambis is unwilling to continue sending
leap second bulletins from his home PC during non-work hours, I'm sure
someone else will, and anyone who is enfused with a passionate love
for leap seconds and a burning desire to continue using them will
surely continue acting upon LS bulletins (and configuring his/her NTP
servers to do so) whether they come from IERS or from somewhere else.


> Why should this time be any different than the last time?  Leap seconds

> were rushed in, at the last minute with little coordination or cooperation

> in a manner that left hard feelings for years.


The difference is that with the vastly enlarged number of stakeholders,
the protests can be a lot louder.  The best protests are those done
with actions rather than just words, and that means continuing to
operate real-life systems according to the old definition,
intentionally disregarding the illegitimate redefinition.

MS