[TAP] data type for column metadata - units

[Anita M. S. Richards]

> The simple way is NOT to implement astronomical time in TAP!
....
> These automated coordinate transformations are scary enough with Time. Is the 
> intention now for the TAP engine to do spatial transformations? For example: 
> "Select points from the X catalog (in B1950 as seen from Jupiter) which are 
> in the Y region (in heliocentric supergalactic coordinates)".
>
> Will it be "allowed" by the IVOA to have a TAP engine that does not have full 
> support for coordinate transformations?
> Roy
>

Friends,

We need to distinguish between the need to specify units, and to convert 
them.  A lot of problems arose in the early VO implementations because we 
decided that we would mandate that e.g. search radii were in decimal 
degrees and omit the unit, and then people published data in decimal 
arcsec....

There is a Units model under development in the Data Model working group 
and also a EuroVO AIDA workpackage to provide (modest) effort for this. 
This will bring a proposal for units to be adopted internally by the IVOA. 
The scope of this package is purely linear or self-contained conversions 
(e.g. m to km,  arcsec to deg in the same system not spatial, temporal 
coordinate frames).  This would allow a query to ask for JD from a data 
set in MJD.  However, if no unit was specified or the data set does not 
have a valid unit, you get back the original data unit (so some warning 
should also be present if the unit is ambiguous or not what was 
requested).

So Roy's simplest case could be supported.  On the other hand, a frequent 
user complaint is problems querying catalogues (or formulating queries) in 
Galactic coordinates, or wanting data in a certain time range.  This is 
outside the scope of the Units model but there are excellent libraries 
(used bY SIMBAD, TopCat etc.) which do temporal and spatial coordinates to 
high accuracy and which have been successfully wrapped for VO 
applications.  There is undoubtedly a demand for this, the question is 
whether we want to write  generic VO services which the user would hook 
in, or ideally the service provider.  I think that this would be a great 
deal better use of effort than many things, and if Pat's model enables 
this (without disallowing services which can't or for whom it is 
irrelevant), so much the better.

cheers
a

- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -

Dr. A.M.S. Richards, UK ARC Node/AstroGrid,
Jodrell Bank Centre for Astrophysics, Alan Turing Building, 
University of Manchester, M13 9PL
+44 (0)161 275 4124
and
MERLIN/VLBI National Facility, Jodrell Bank Observatory, 
Cheshire SK11 9DL, U.K. +44 (0)1477 571321 (tel) 571618 (fax)

"Socialism or barbarism?" Rosa Luxemburg (1915)