[fitswcs] New WCS conventions posted to the Registry of FITS Conventions

[David Berry]

The same two points I made after the announcement of TPV on fitsbits:

1) I think the description of the convention should also explain that
If FITS reading software encounters a TPV header, then the standard
meanings attached to PV1_0 to PV1_4 (where i is the index of the
longitude axis) are suspended in favour of the meanings attached to
these keywords by the TPV convention. In the published standard these
PV values record information about the native coords of the fiducial
point and celestial pole (see section 2.6 of FITS-WCS paper II), but
in the TPV convention, they record polynomial distortion coefficients.
So in the TPV convention, the native coords of the celestial pole need
to be obtained from the LONPOLE and LATPOLE keywords, and the native
coords of the fiducial point are assumed to take their default values
for a TAN projection.

This is probably implicit in the current description of TPV, but
making it explicit may avoid confusion.

2) In terms of support for TPV, version 5.7-3 of the starlink AST
library supports the TPV convention (AST is the WCS library used by
DS9).

David