Thank you Pedro, I am reassured that what you are doing is to connect
up the different parts of the VO that already exist, rather than making
a whole new way. This is very good.
I like your interpretation of the Data Model as a "view". If I might expand on my understanding a little?
(a) If I have a catalog (RoyCat) with an attribute RoyZ that is
redshift, then I can (obviously) use ADQL like this
select * from RoyCat where RoyZ >4
(b) BUT now I can build this virtual table, this view, called
RoyCatViewedAsSourceDM, which is really the same table, and I can make
ADQL like this
select * from RoyCatViewedAsSourceDM where Redshift>4 where "Redshift" is your semantically-precise utype.
Finally, the story about the source identifier...... I am worried that the data model people will assume that the identifier can be an arbitrary 100-character string, but the database people will assume it is a 32-bit or 64-bit integer. This is my worry, and I hope you will make it clear in your document what is the nature of the identifier.
Roy
California Institute of Technology
626 395 3670
Received on 2005-12-28Z20:02:37