Votable/FITS fits all

[Doug Tody]

> NB I don't advocate stopping development effort on VOTable just now. 
> However I agree with Tony that we ought to note in the standards doc 
> that the VOTable format is not meant to be a one-size-fits-all for the 
> future.  I want to avoid seeing a whole lot of VO effort disappear into 
> trying to shoehorn all astronomical formats into VOTable, and building 
> tools, stylesheets, parsers etc to handle this VO-specific format, when 
> this effort would be better spent creating things astronomers will find 
> useful.

VOTable is actually two things: a general table mechanism (a resource of
type table) and a general XML-formatted container capable of containing
an abitrary collection of "resources" (stuctured elements expressed in
XML) of different types.  We need both things.  To describe arbitrary
datasets we need things which haven't been defined yet, to fit into the
non-table resource elements.  For stuff which is natually tabular data,
there are advantages to using a standard table format.  Of course it is
wrong to try to stuff everything into the "table" data model, but it is
a mistake to think of VOTable as only a table mechanism.

Within VO we probably have at least two classes of data: stuff which is
internal to the framework, which is probably most natually expressed in
"native" XML, and data-centric stuff, which should ideally be expressed
in a largely framework and technology independent data model, with a
clean separation between data model and representation.  For the latter,
the framework can deliver the data, but data-centric (and framework and
technology independent) code will understand it.

	- Doug