ietf-nntp Section 11.5 - NEWNEWS

Clive D.W. Feather clive at demon.net
Sat Nov 18 07:46:25 PST 2000


Charles Lindsey said:
> May I suggest
> 
> + The date and time are given in the server's approximation to UTC
> + (commonly referred to as GMT).

UTC is *not* GMT. Or anywhere near it. This is the source of the problem.

> And will somebosy please explain to me what "TAI" is?

Okay.

There are two common ways to measure time.

(1) Watch the sun cross the meridian and note how long it takes to come
back to the meridian the next day. Call the result "1 day". Divide this
by 86400 and you get "1 second".

This period varies through the year, but if you take the mean value you get
something that's approximately constant from year to year. If done by
measurements at Greenwich, it's "Greenwich Mean Time" (GMT). Other
equivalent methods yield "Universal Time 0" (UT0 or UT).

Applying certain systematic corrections to UT0 gives UT1 or UT2. The
difference between these three is unimportant for this topic.

* UT tracks the real day-night cycle.
* There are always 86400 seconds in a UT day.

(2) Use an atomic clock. This produces a period whose length never changes,
which we call "1 second". Multiply this by 86400 and call the result "1 day".

This system is called TAI (International Atomic Time, only in French).

* TAI seconds are all the same length.
* There are always 86400 seconds in a TAI day.

Now, note that UT seconds aren't all the same length, because the day is
slowly getting longer. And TAI days don't match the day-night cycle for the
same reason. At present the difference is increasing at about 0.7 seconds
per year.

So there's a comprise timescale called UTC (Universal Time (Coordinated)).
This uses TAI seconds but is designed to track the real day-night cycle.
To do this, whenever |UTC-UT| is getting near to 0.9 seconds, a second is
added to or subtracted from UTC to reduce the error (these are called leap
seconds).

* UTC seconds are all the same length.
* UTC tracks the real day-night cycle.

But, the number of seconds in a UTC day varies - introducing a programming
problem - and that variation cannot be predicted in the long term - leap
seconds are only announced a month or two ahead.

Clear ?

[The Austin Group has had a long argument over what "time" means in POSIX.
I don't think it's resolved yet.]

-- 
Clive D.W. Feather  | Work:  <clive at demon.net>   | Tel:  +44 20 8371 1138
Internet Expert     | Home:  <clive at davros.org>  | Fax:  +44 20 8371 1037
Demon Internet      | WWW: http://www.davros.org | DFax: +44 20 8371 4037
Thus plc            |                            | Mobile: +44 7973 377646 



More information about the ietf-nntp mailing list