[time-nuts] Re: MJD 60000 this week

[Magnus Danielson]

Hi,

On 2023-02-20 04:21, Steve Allen via time-nuts wrote:
> On Sun 2023-02-19T16:11:03-0800 Tom Van Baak via time-nuts hath writ:
>> All linear time scales have an origin date (aka epoch).
>> Here are some examples:
>> 01-Jan-1958 (JD 2436204.5) is TAI epoch  (MJD 36204)
> There are so many "yes but"s in that.
>
> The epoch of TAI is 1961-01-01T20:00:00 UT2(OP).
> The scale A3 which became TAI was not named until then.
>
> https://www.ucolick.org/~sla/leapsecs/taiepoch.html
>
> Early atomic time had many hiccups.
>
> https://www.ucolick.org/~sla/leapsecs/earlyAT.html
>
> TAI did not exist until 1971, and even ignoring all the early hiccups
> TAI is not linear because
>
> The rate of TAI was stepped on 1977-01-01.
> TAI suffered from a notable seasonal rate variation into the early 1980s.

Also, there is no "counter start at zero" type of epoch for TAI, a 
common misunderstanding by especially programmers that seems to think 
"epoch" has that meaning, always.

There is TAI-like time-scales, such as GPS and PTP, where counters was 
set to zero.

PTP starts a bit into 1969 for it to be ticking in SI seconds such that 
it has integer offset as UTC starts with leap seconds, essentially a 
variant of the UNIX epoch/time but which align to integer relationship 
after 1 Jan 1972, so that the leap second count adds to make UTC. UTC 
did not have SI seconds prior to 1 Jan 1972, so the UNIX/POSIX type of 
clock starting from 1 Jan 1970 and count SI-seconds ends up not being 
UTC-aligned.

Cheers,
Magnus