Re: A Working document has been submitted

From: Martin Hill <mchill-at-dial.pipex.com>
Date: Tue, 07 Oct 2003 18:51:22 +0100


Then there's a danger that users get back incorrect results without knowing it, which (I gather) is worse than the query failing.

Tony Linde wrote:

> No, if the user chooses to use UCDs then they take the risk that they mean
> different things in different tables being targetted. So UCDs must be
> included.
> 
> Cheers,
> Tony. 
> 
> 

>>-----Original Message-----
>>From: owner-voql-at-eso.org [mailto:owner-voql-at-eso.org] On
>>Behalf Of Martin Hill
>>Sent: 07 October 2003 18:41
>>To: 'voql'
>>Subject: Re: A Working document has been submitted
>>
>>
>>Errrm - as I understand it, UCDs are *never* uniquely
>>resolvable. ie,
>>even if there is a unique UCD in one table, it doesn't
>>necessarily mean
>>it has the same meaning as a column with the same UCD in
>>another table.
>>
>>That means we need to get user confirmation at every stage
>>where we map
>>between UCDs and table columns at the moment, and wherever we
>>specify an
>>actual applied query, it needs to be resolved to actual table column
>>names.
>>
>>Tony Linde wrote:
>>
>>
>>>I think it is essential that ADQL support the use of *both*
>>
>>UCDs and
>>
>>>column names. It'll be up to the query interface to
>>
>>substitute column
>>
>>>names (by asking the user probably) where UCDs are not uniquely
>>>resolvable.
>>>
>>>Cheers,
>>>Tony.
>>>
>>>
>>>
>>>>-----Original Message-----
>>>>From: owner-voql-at-eso.org [mailto:owner-voql-at-eso.org] On
>>>>Behalf Of Roy Williams
>>>>Sent: 07 October 2003 15:50
>>>>To: Clive Page; voql
>>>>Subject: Re: A Working document has been submitted
>>>>
>>>>
>>>>
>>>>
>>>>> Region ('CIRCLE J2000 19.5 -36.7 0.02')
>>>>
>>>>I have a question in the same style. I would like to start
>>>>using the UCD ontology within VOQL. I am imagining metadata
>>>>queries like
>>>>
>>>>select table where UCDMatch(table, "phys.velocity,
>>
>>src.galaxy") > 0.8
>>
>>>>in order to find tables that have galactic velocities. Can
>>>>VOQL be extended in this way?
>>>>
>>>>Roy
>>>>
>>>
>>>
>>

> Received on 2003-10-07Z19:53:07