develooper Front page | perl.perl6.language.datetime | Postings from August 2000

Re: TAI time

Tim Jenness
August 18, 2000 16:07
Re: TAI time
Message ID:
On 18 Aug 2000, Russ Allbery wrote:

> Mark-Jason Dominus <> writes:
> > TAI is an international time standard.  It has a number of technical
> > advantages over UTC.  One of these advantages is that it doesn't have
> > any silly truck with leap seconds.

libtai looks like a great idea to me.

> One point to be aware of with TAI, however, is that Unix system clocks
> don't normally keep TAI so you can end up with a 20-second differential or
> so due to the difference in leap seconds.  Dan's libraries use an external
> table to track leap seconds and adjust.  This won't matter for most
> applications (the TAI64 formats are independent of whether you're keeping
> true TAI or the bastardized Unix UTC-like TAI), but it's something to be
> aware of.

I think this is a bit of a red herring. Presuambly the time on your
computer depends on where you are synching from (if at all) - if you
are synching to a TAI time server then you are fine. In the worst case you
are synching to the sys admins watch so the minor actual difference
between UTC and TAI may be irrelevant in most practical applications (our
telescope synchronizes with the clock on the GPS satellites - a 1 second
error in our clock means we can miss our target by 15 arcseconds (and
that's a lot).

Tim Jenness
JCMT software engineer/Support scientist Perl Programming lists via nntp and http.
Comments to Ask Bjørn Hansen at | Group listing | About