[LEAPSECS] the big artillery

Zefram zefram at fysh.org
Wed Nov 5 11:27:19 EST 2014

Warner Losh wrote:
>The markers aren't the same.

I was referring to the PPS marks in a time signal, and because TAI and UTC
tick the same seconds the marks work equally well for both.  Taking MSF
as a specific example, the onset of each per-second carrier-suppressed
interval (specifically, the instant when that point in the signal leaves
the transmitter) occurs at the top of a second of UTC(NPL).  UTC is always
an integral number of seconds offset from TAI, and so by construction
UTC(NPL) is always an integral number of seconds offset from TAI(NPL).
Hence each of the marks also occurs at the top of a second of TAI(NPL).

>TAI doesn't actually exist until after the fact.

Nor does UTC.

>Could one create, in real time, a TAI signal, sure. But no one does.

Taking MSF again as a specific example, it's just as much a TAI signal
as it is a UTC signal.  The PPS marks serve equally well for both,
subject to the UTC(NPL)-UTC = TAI(NPL)-TAI realisation error.  The time
code transmitted alongside the marks doesn't directly encode either UTC
or TAI, so decoding to either requires some out-of-band information.
(The broken-down time that is directly encoded is in UTC plus the UK's
current timezone offset, and nothing in the signal says what that offset
is.  There's a bit saying whether the offset includes DST, but there's
nothing saying what the winter-time offset is.  The signal format spec
<http://www.npl.co.uk/upload/pdf/MSF_Time_Date_Code.pdf> is clear that
a change to the UK's winter-time offset wouldn't be detectable from the
post-change signal.)

>So while an outside may see "oh! look! TAI is a nice way to label seconds that
>doesn't suffer from the leap second issue, let's use it!" the insider reacts to that suggestion
>with horror because they know all the behind the scenes machinations.

The same machinations also lie behind UTC.

>So maybe I'm again off in the weeds trying to differentiate between philosophy and a
>notion of the right thing and politics and ideology.

I'm open to there being philosophical differences, and of course
taking UTC as a whole it certainly does have such differences from TAI.
But (for the leap-seconds form of UTC) those differences are only in how
the seconds are labelled.  At the sub-second level, UTC incorporates
TAI's features in their entirety, by reference.  Identical behaviour,
identical philosophy guiding that behaviour.


More information about the LEAPSECS mailing list