TAP issues (fwd)

From: Doug Tody <dtody-at-nrao.edu>
Date: Wed, 2 May 2007 20:23:12 -0600 (MDT)

Hi,

There are two KEY issues to the TAP protocol that I think the Tiger team should be clearly specifying but so far I don't see this is the case. Pat mentioned them today and I think they deserve a lot of attention.

  1. How does the service say which ADQL constructs support and which not?

Considering that the ADQL CORE and EXTENSION concepts don't exist any more and, as far as I know, it is not expected that all services fully support the whole ADQL specification the question remains as

[getCapabilities can probably deal with this, but we will still need to specify someplace what these optional ADQL capabilities or UFNs are. A schema metadata query would be another option for functions. - DT]

2) What is needed to describe a TAP service so users/clients can do decent queries in terms of content and performance?

In essence, how does the service describe tables, columns, indexes and relationships?

I know you have been discussing the INFORMATION_SCHEMA standard and some people don't find it adecuate. I got to agree that INFORMATION_SCHEMA is not the full solution but without a question is the right approach. The old skyNodes have tables describing metadata. The missing parts were the relationships and indexes but columns have information about UCDs, units, etc.

In my modest opinion, specifying the metadata schema is a fundamental area where the tiger experts should focus their efforts. I would say that finding out what is the necessary metadata information come first. Later you can discuss how to make it available and decide whether or not the VOResource schema can hold it.

Cheers

Maria 

-- 
------------------------------------------------
Maria A. Nieto-Santisteban (nieto-at-pha.jhu.edu)
Johns Hopkins University
3400 N. Charles St.
Physics & Astronomy Department
Baltimore, MD 21218 (USA)

Tel: 	1 410 516-7679  Fax: 	1 410 516-5096
Received on 2007-05-03Z04:23:39