As ever what appears to be a simple job is more complicated. One
component of the SURFACEFIT.FIT extension is a component COSYS either
Data or World. Indeed FITSURFACE has COSYS parameter with allowed
values of Data or World. That was a surprise. I thought those had been
replaced in conversion to AST. They're equivalent to AXIS or PIXEL, so
I've translated them into those AST Domains while leaving Parameter
COSYS unchanged. That can alter if we choose to allow fitting to the
current Frame's co-ordinates.
The fit coefficients in SURFIT weren't designed for public viewing. The
co-ordinates are GRID, so I've set the COSYS component to GRID.
There wasn't a SPLINE variant defined in SGP/38, so no SURFACEFIT
extension is created for FITTYPE=Spline. That's partly why FITSURFACE
doesn't yet support SPLINE fitting, although some outline code for that
option is in place.
In SURFIT there's no coefficient variances available for the
SURFACEFIT.FIT extension so I've set those to undefined (bad). Those
could be derived and passed through to the extension, but not tonight.
I've modified MAKESURFACE to recognise GRID, PIXEL, or AXIS values of
COSYS.
Having done all this you're now going to tell me that you want the
coefficients stored in parameters instead.
Malcolm
|