Just to bring everyone up to speed, this is Norman's inital email on the
TimeFrame.
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
---------- Forwarded message ----------
Date: Fri, 11 Apr 2003 17:14:06 +0100 (BST)
From: Norman Gray <[log in to unmask]>
To: David Berry <[log in to unmask]>
Cc: Malcolm J. Currie <[log in to unmask]>,
Tim Jenness <[log in to unmask]>
Subject: TimeFrame proposals
David,
I think I'm now reasonably au fait with time systems, enough to start
thinking of what to support in an AST TimeFrame. David, Malcolm and Tim:
have I missed any astronomically important timescale, do you think?
There are three groups of timescales. Transformations within the
groups are pretty straightforward, transformations between them
potentially more involved. I therefore plan to pick a representative
for each group, define the inter-group transformations using that one,
and the intra-group ones easily follow. I haven't worked this out in
final detail, so there may still be semi-obvious gotchas implied here.
Angles (ie, sidereal measure): Greenwich sidereal time, from which
follow LAST, LMST, GMST and UT0/1 (possibly UT should be the basic one,
from which I derive GST). LAST and LMST would need an Observatory
Longitude attribute on the frame.
Duration (ie, atomic measure): TAI, from which UTC follows (though
this is more logical, it might be more practical to have these the
other way around).
Coordinate (ie, ephemeris times): TT, from which TDB, TCG, TCB.
JD and MJD (and Greenwich sidereal date) don't fit into any category
necessarily, but should be easy.
The conversion TAI->UTC needs a table updated every 6 or 12 months, and
given that it is desirable to produce local civil time in the Format and
Unformat procedures (ie, BST or times in other time zones), that would
need external and updatable table support. Parsing general dates is
a pain in the neck, so Unformat might have to be rather restricted --
it would cope with at least FITS and ISO 8601 times, though.
Conversion to and from Mayan and French Revolutionary calendars I'll
probably leave for the moment.
What time support is there in NDFs? Looking quickly through the NDF
documentation, I realise that there doesn't appear to be any, beyond the
time stored in the history and any times in FITS extensions. I suppose
that makes sense, since there's rather little metadata in NDFs, but it's
still, now I come to think about it, rather surprising. Am I missing
something embarassing?
As regards popping down to visit you, Wednesday would probably be best
for me. Can we leave this provisional until next week, though? That'll
give me a chance to look through the AST code and see if I really need
to come down, since it might be that a chat on the phone would be enough.
Have a good weekend. See you,
Norman
--
---------------------------------------------------------------------------
Norman Gray http://www.astro.gla.ac.uk/users/norman/
Physics and Astronomy, University of Glasgow, UK [log in to unmask]
|