On Tue 2007-08-21T14:05:26 +0200, LC's NoSpam Newsreading account hath writ:
But even with such conventions, we are still left with the problem of
what a generic reader should do with (older or not) files not following
any convention.

That is true for more than just keywords, and is inherent to FITS.
There is no mechanism for a FITS file to communicate how it is expected
to be used or what any of the meanings are for "additional" keywords,
i.e., keywords neither "mandatory" nor "reserved", nor for how to
use any appended images and tables.

Getting back to the issue of dupes, our spectrographs produce
duplicate keywords, where the different values are of different data
types. Without delving into the current code and doing archaeology to
find the source for old versions which are still in use I can't even
write out a recipe for interpreting whether those duplicated keywords
are indicating that the system was in a normal or abnormal state when
the image was acquired. I do not expect that anyone else should care
how they would properly be interpreted, but I am sure that an explicit
directive in the FITS standard would be misleading.

I don't think this particular problem can be solved other than by
explicitly indicating that the standard does not assign any meaning.
But because I think there are a lot of other similar "problems"
I'm not sure that the FITS standard benefits by making it explicit.
This is the sort of thing that does belong in a user guide.

--
Steve Allen WGS-84 (GPS)
UCO/Lick Observatory Natural Sciences II, Room 165 Lat +36.99855
University of California Voice: +1 831 459 3046 Lng -122.06015
Santa Cruz, CA 95064 http://www.ucolick.org/~sla/ Hgt +250 m