> At the IVOA meeting at Cambridge Jonathan McDowell asked me to
> get some discussion going on possible ways of representing transformations
> within the IVOA data model. I made an early attempt, focussing on the use
> of transformations to represent WCS information. This failed to raise much
> response, other than Doug Tody saying that the FITS-WCS papers supply
> everything we need.
"Use FITS" is the default you'll get. We believe, however, that the
FITS approach is too limiting and inflexible. Therefore the document
needs to address that conflict directly, and cite practical examples of
transformations which will not be possible in the FITS model. FITS only
has three transformations in the CTYPE name (the 4-3-3 notation to allow
for a project and one distortion) going from pixel to world
co-ordinates. Yes you can have more than one of these sets just like
AST, but everything has to be in big sticky lumps, not modular.
Modularity gives us that extensibility and flexibility.
To summarise: there has to be an introductory section on why we need
such a set of transformations for the VO.
Malcolm
|