UCDS and DM and "Catalogue" tables

[posuna at iso.vilspa.esa.es]

Dear all,

the fact that I have been one of the authors of the UCD2 draft document
(as a member of the UCD steering committee) made it very bizarre that I
"voted" for the UCD1 option, together with other very few people. 

I think that if we use a Data Model to access metadata, then it is
enough with the UCD1 that was created at the beginning. For me, there is
no need to have more specificity, neither is there a need for matching
functions and other more complex syntactical operations to be performed
with the UCDs. For me, all that should be done through the Data Model
and/or proper VOQL language.

However, I do appreciate that many providers do have their data in
the form of catalogues (what I normally call 2D tables, or X-Y tables,
as catalogue, for me, can be several different things).

I understand that people having data in X-Y tables do NOT want to hear
about data models: they only want to be able to perform operations with
the columns of their tables.

It is in that context (X-Y table handling) where I believe the
need of an overall complex UCD structure appears, as any comparison
operation (or more complex ones as addition, substraction, ...)
they want to do it by comparing (or adding/substracting/...) directly
the columns.

As the UCD2 is just adding more capabilities to the UCD1 without
chopping off any of the existing ones, and as we ("Data Model" oriented
ones) can always use the UCDs in the limited way we want them (i.e.,
just to describe the metadata we give back so that they are
"Universally" understandable) I see no reason to stop the UCD2 to
include more syntactical and operational capabilities: people
"Data-Model oriented" will make use of very limited UCD capabilities (as
they don't need them) and people "X-Y Table" oriented will make use of
as many UCD "functionalities" as possible.

In summary, I'm OK with new funtionalities of the UCDs but I guess I'll
do a very limited use of them in the case that I use the Data Model to
access my data, and hence my vote for the UCD1 "paradigm".

Maybe the UCD document should reflect two main "Areas of Interest",
something like:

- Simple  UCD handling:  Data Model access
- Complex UCD handling:  Syntactical operations on X-Y (catalogue) table
	                 columns


I wait for your comments...

Cheers,
Pedro.











On Mon, 20 Oct 2003 13:55:38 -0400
Tom McGlynn <Thomas.A.McGlynn at nasa.gov> wrote:

> 
> One thing I should have made clearer in the earlier message is that
> the revision I'm sending
> does not reflect any consensus on the correct approach.  Rather it's a
> 
> very detailed alternative to the
> approach presented by Roy at Strasbourg.  It  builds upon the same
> basic elements  that Roy and Sebastien suggested but puts them
> together in -- for me at least --
> a more coherent fashion.    This approach does naturally lend itself
> to the concerns Martin raised, but
> it's certainly not a done deal.
> 
>     Tom
> 
> Tom McGlynn wrote:
> 
> >   Over the
> > past two days I have been writing a revised proposal.
> >
> 


-- 

Pedro Osuna Alcalaya


SOFTWARE Development Group
XMM-Newton Science Archive				
e-mail: Pedro.Osuna at esa.int
Tel + 34 91 8131314  				

European Space Agency
VILLAFRANCA Satellites Tracking Station
P.O. Box 50727 
E-28080 Villafranca del Castillo
MADRID - SPAIN