Re: Suggested enhancements to ADQL

Dear Clive,

Regarding the LIMIT n, which I find a very useful feature to add, I have to say that the (at least Sybase) behaviour is a little different from what you describe in the quoted example:

> when finding the top N in some sorted order, e.g. the top N faintest
galaxies, or N
> stars with largest proper motion: in this case LIMIT n would be used
on conjunction
> with an ORDER BY by clause.

The Sybase behaviour goes like this:

 SET ROWCOUNT n

will make the server !first! return the first n rows,

then the subsequent order by will sort those rows.

That is, you do not get back the n faintest sources, but you get back the first n sources (in whatever internal sybase order) which are then sorted by brightness. (The example will work if and only if the table is already organised by brightness (e.g. using a clustered index on that column)).

But for data exploration I find it very useful (even though the results could be biased by the behaviour I mentioned above).

---

SELECT into,

Absolutely!, people want to create sample, and to store them.
The problem is that not every data center will be ready to accept
anything else than a "read-only" query, ie., a select from (without into).
There must be a way to know whether a given centre offers such capability
or not.

---

DESCRIPTION table

The Sybase way is:  sp_help table

Whether superfluous or not,
I use it all the time. I can't leave without it.
Sorry about the stupid question, but I have not followed that much
the VOQL so far. How would this metadata query work?

Thanks,

Alberto