Re: [Ntp] Leap second draft

[Warner Losh <[e-mail]>]

On Sat, Mar 30, 2019 at 9:29 AM Kurt Roeckx <[e-mail]> wrote:

> On Sat, Mar 30, 2019 at 06:33:48AM -0700, Steve Allen wrote:
> > Daniel Franke notes that I did not send this and the
> > next message to the list.
> >
> > Greetings Daniel Franke,
> > On Fri 2019-03-29T20:57:42+0100 Daniel Franke hath writ:
> >
> > > Network Working Group                                          D.
> Franke
> > > Internet-Draft
> Akamai
> > > Updates: 5905 (if approved)                               March 29,
> 2019
> > > Intended status: Standards Track
> > > Expires: September 30, 2019
> >
> > >    At their inception in 1972, UTC and TAI were equal.  Since then,
> >
> > No, please look at the second plot on
> > https://www.ucolick.org/~sla/leapsecs/amsci.html
> > and at the third plot on
> > https://www.ucolick.org/~sla/leapsecs/deltat.html
> >
> > As of 1972 TAI - UTC was reset to exactly 10 SI seconds
> > Since then there have been 27 leap seconds.
> [...]
>
> I think this is all not very relevant for NTP where we care about
> the current time, not about what happened in 1972. I'm not sure
> we need all that historic information in the draft.
>

We definitely need that it was exactly 10s in 1972, as opposed to 0. And we
need statements that things were messy before 1972. This information is
'out there' but hard to find and difficult to properly interpret.

And it's less obscure than the crazy notion that leap seconds could happen
any time other than June and December as a practical matter. TF.460 says
those are the primary dates, with March / September being the secondary
dates though at the end of any month is possible. Given the large number of
bugs we hit every single leap second in June and December, March and
September just might work in enough places not to sow chaos, but a leap
second at the end of February would be pure chaos. At least as a practical
matter. Untested code is buggy code.

Warner