[fitswcs] is WCSAXESs necessary?

[Steve Allen]

Perhaps the question is better phrased as
	What exactly is WCSAXESs trying to do and how should a
	standard-conforming FITS processor handle it?

I am writing WCS parsing code and I've encountered a question of
interpretation.

Sections 2.2 thru 2.4 of the 2002-02-06 draft of WCS paper I introduce
WCSAXESs as the keyword to indicate that the dimensionality of the WCS
with version "s" differs (presumably exceeds) the value of NAXIS.

Nevertheless, the text makes it clear that there is a precedent for
FITS WCS dimensionality that exceeds NAXIS.  WCSAXESs is explicitly
stated to be optional and to default to NAXIS or the largest i or j.

If it is to be expected that a conforming FITS WCS reader will look
for instances of i and j that exceed NAXIS, then there is no need
for WCSAXESs and the FITS namespace should not be polluted by
reserving it.

Alternatively, the existence of WCSAXESs in the standard proposed by
paper I could be expressing an intent to deprecate the usage of WCS
dimension greater than NAXIS in the absence of WCSAXESs.  If so,
the paper should say that conforming FITS processors are absolved
of looking for i and j greater than WCSAXES.

Or it could be that the paper is taking a carefully wordsmithed
path trying to avoid the issue.

As things stand it is not clear what happens if there are i and j
(even for insubstantial cards like CUNITis) that exceed the value
of an explicitly-presented WCSAXESs card.

I would like the paper to take a definitive stance, but it seems
likely that the proper stance needs input from the community at large.

--
Steve Allen          UCO/Lick Observatory       Santa Cruz, CA 95064
sla at ucolick.org      Voice: +1 831 459 3046     http://www.ucolick.org/~sla
PGP: 1024/E46978C5   F6 78 D1 10 62 94 8F 2E    49 89 0E FE 26 B4 14 93