[fitswcs] Keywords for other FITS image representations

[Mark Calabretta]

On Mon 1998/08/10 14:58:58 +1000, Mark Calabretta wrote
in a message to: William Pence <pence at tetra.gsfc.nasa.gov>
and copied to: fitswcs at fits.cv.nrao.edu

>Consider the following scheme:

Apparently people are still recovering from this suggestion - it wasn't meant
to reduce the list to stunned silence!

In general terms, the basic idea is that with a with a few minor changes to
what was already proposed plus two simple rules that bintable FITS writers
have to follow we can get images with up to 99 axes in columns 1-99 of a
binary table.

The rules are

   1) Fixed two-digit parameter numbers "kk" in PVkk_iiv (need apply only to
      bintable and pixlist).

   2) Doubling "i" and "j" together in Ciijjvnn versus Cijvnnn (applicable to
      bintable only).

All other indexing parameters are still variable length.  The parser for most
of these keywords should be simple enough, especially if you have access to
unix regular expressions.

Even if it's felt or can be shown to be unworkable you still can't ignore the
requirement that LONGPOLE, LATPOLE, EQUINOX, RADECSYS, and MJD-OBS must now
include an axis number as well as a version number which means that their
names must change.  It is not an accident that the original CTYPE, CRPIX,
CRVAL, CDELT and CROTA were limited to five characters and so fit this scheme
since the limit on NAXIS stated in Wells, Greisen & Harten was 999 (although
we're now effectively reducing that limit in particular cases).

>If not too much of a shock to the system I suggest the "v" version letter be
>lower-case to distinguish it from the fixed part of the keyword.  Gasp!

Or maybe this was what did it!?

Cheers, Mark