[TAP] data type for column metadata

[Douglas Tody]

On Tue, 31 Mar 2009, Patrick Dowler wrote:

> I was initially going in this direction and it is roughly in line with the
> library of standard coordinate systems (I think in the appendix of the STC
> document), where one has things like ICRS-TOPO. That would simply this alot,
> and we could get rid of "flavor". For now, I just wanted to express what is
> needed, not necessarily the form it take. ...

The details of what is needed in this area which you and Arnold
suggested are probably about right, but have little to do with TAP
(likewise the details of ucd, utype, unit will all need to be defined
externally).  What we most need to do now is 1) agree that we need
something in this area (I think so), and 2) define a place to put
the information, with some simple examples.

What would be ideal would be if we had something simple like
FRAME which in the simple cases reduced to things like FRAME=ICRS
(for spatial coordinates), or FRAME=UTC or TT (for timestamps),
with more expressive values such as ICRS-TOPO also being possible
in the few cases where this is warranted.  This much would be easy
for the data providers and hence we might actually get them to do it.
It would be very useful for client apps/users to get at least this much
information back in a metadata query, along with other fundamentals
such as whether the unit is degrees or radians, or JD/MJD/UTC etc.
Since this is all optional we need to make the basic facility quite
simple or the data providers will not give us anything at all.

Following up on the comments from Roy or Anita - none of this suggests
that the service need do coordinate or unit conversions (except where
we provide a data model interface such as with POS,SIZE).  We are just
trying to define the basics of what is in the table, which is not at
all difficult.

 	- Doug


On Tue, 31 Mar 2009, Patrick Dowler wrote:

> On 2009-03-31 16:37:08 Douglas Tody wrote:
>> But refsystem/refposition/flavor ("flavor" is awful) would only be
>> used by coordinate-type columns - for many other table columns they
>
> Yes, these are additional metdata to understand coordinate values, but also
> redshifts in catalogues. For the most part they would not be needed.
>
>> would not be used at all.  It might be preferable to compress all
>> this information into a single string attribute, as we do with unit,
>> ucd, utype, etc.  Perhaps something like "wcs", "system", "frame",
>
> I was initially going in this direction and it is roughly in line with the
> library of standard coordinate systems (I think in the appendix of the STC
> document), where one has things like ICRS-TOPO. That would simply this alot,
> and we could get rid of "flavor". For now, I just wanted to express what is
> needed, not necessarily the form it take. Otherwise: yes, this could easily
> be encoded in a single attribute and there are examples of that already in
> STC.
>
> PS-flavor comes directly from STC :-)
>
> -- 
>
> Patrick Dowler
> Tel/Tél: (250) 363-0044
> Canadian Astronomy Data Centre
> National Research Council Canada
> 5071 West Saanich Road
> Victoria, BC V9E 2M7
>
> Centre canadien de donnees astronomiques
> Conseil national de recherches Canada
> 5071, chemin West Saanich
> Victoria (C.-B.) V9E 2M7
>