[fitsbits] Re: leap second alert
Patrick Wallace
ptw at star.rl.ac.uk
Fri Dec 17 17:59:16 EST 1999
On Fri, 17 Dec 1999, Steve Allen wrote:
> I can't write a program for my computer which will cause it to emit a
> reliable value of TAI or TT -- at least, not unless I make it a rather
> savvy network-aware program which can several times annually navigate
> its way through a maze of firewalls, surf the supposedly-reliable URIs
> that give leap second information, and then update its own database.
If your system clock is set by a local time service there's clearly no
problem - you can choose whatever timescale you like. But if it's set by
NTP (in which case it's a bit iffy, especially just after a leap second,
something not all data acquisition applications can tolerate) then, in
the absence of accessible information from the net, there needs to be a
place in your application where you can manually set the current UTC-TAI.
If your application has an configuration file of some kind, and I would
guess many big FITS-writing data acquisition programs do, that's the
place to put it. The Gemini TCS, for example, doesn't attempt to get
leap-second information from the net. The numbers are stored in an
initialization file (in a way so the update can be done as soon as the
leap second is announced and, months later, will be handled properly).
The file also contains the UT1-UTC and polar motion stuff, which is
updated nightly (not that you need to).
> I'll wager they're rather glad that the DATE keyword was specified as
> UTC.
Or perhaps this is a chicken waiting to come home to roost. I certainly
agree we don't want to start creating examples where date and time are
in different timescales.
Patrick Wallace
____________________________________________________________________________
Starlink/HMNAO Internet: ptw at star.rl.ac.uk
Rutherford Appleton Laboratory Tel: +44-1235-445372
Chilton, Didcot, Oxon OX11 0QX, UK Fax: +44-1235-446667
____________________________________________________________________________
More information about the fitsbits
mailing list