Re: Issues relating to simple data and metadata queries

Just passing this on (on behalf of DAL). - Doug

---
Date: Mon, 30 Apr 2007 08:59:33 -0400
From: Robert Hanisch <hanisch-at-stsci.edu>
Subject: Re: Issues relating to simple data and metadata queries

Hi Doug.  I am not expert enough to comment on some of the details about
SQL92 information schema, so let me just make some top-level remarks.

I have never been convinced by the use cases that base table or database
discovery on column names or UCDs.  I just do not think this is how
astronomers decide on what tables or databases they want to use.  Maybe --
emphasis on "may" -- programmers would develop portals for certain types of
research that would make use of data discovery in this way, such as a VO
portal for OH masers or something like that.  But still, discovery of
relevant resources is more likely to happen using the high-level resource
metadata that we already have in the registry.

Users will then explore the available tables and will use and adapt the
information in them.  I doubt that an astronomer would discard a table as
irrelevant if it did not contain a column with a particular UCD.  Rather,
they would cast their queries in ways that make use of the columns that are
there, and then convert or scale the other available quantities into values
that are useful to them.

I think this is even more the case given the difficulties surrounding UCDs.

If registry curators want to cache table column names and UCDs in their
registries, so be it.  The primary/definitive source of such information,
though, should be the table access services themselves.  And the mode of
access to this information, client-side, should be as simple as possible and
as similar to other DAL interfaces as possible.

Bob