MyUCDs & Registry
Tony Linde
ael at star.le.ac.uk
Tue May 20 14:28:37 PDT 2003
Tom,
> - query to the registry to find resources of interest
> - explosion of a 'generic' query to selected resources
> - translation of the generic query to local query syntax
This is one scenario.
> Here the registry query may involve terms separate from the
> generic query. E.g., this might be 'Find me all the catalogs
> with proper motions'. The syntax of this query would need to
> be expressible in VOQL and the basic registry service is to
> satisfy such queries.
The question of whether it makes sense for the registry query to be the same
format as the more specific VOQL is still open. Your example is one
difference between them. Another is that the registry query will need to
specify the metadata formats (subsets of metadata) to be returned, though
this could be in a separate region of the xml query document from the actual
registry query. Does VOQL need to cater for image queries etc?
> The explosion of the generic query to the selected sites
> is the responsibility of whatever
Yes. In AstroGrid it will be the responsibility of one of the components of
the job controller (to which the workflow engine submits jobs).
> of SQL constraints on the coordinates. So for a generic
> query, the user needn't know of the actual column names until
> getting the results back.
>
> Presumably VOQL also needs to support queries where the
> specific column names are specified too -- but these would be
> 'manual' queries where the user already knows the structure
But the VOQL should also support the creation of hybrid queries to satisfy
the scenario I described.
> "Find me the proper motion of all stars in the Hyades region."
I think it is safer to say this was the astronomer's original intention. In
order to understand the intention, the astronomer will have to go through
whatever interface his/her portal of choice puts up: in the case of
AstroGrid, it'll be drop-down lists, checkboxes etc. I think we're some way
off being able to type in the above phrase - next project, I hope :)
Cheers,
Tony.
> -----Original Message-----
> From: Tom McGlynn [mailto:Thomas.A.McGlynn at nasa.gov]
> Sent: 20 May 2003 19:13
> To: voql at ivoa.net
> Subject: Re: MyUCDs & Registry
>
>
>
>
> Clive Page wrote:
> > On Tue, 20 May 2003, Tony Linde wrote:
> >
> >
> >>The column names have to be inserted into the query before
> it is sent
> >>out. It may start out as generic but, in order to be
> resolved at the
> >>dataset end, it must have some specificity added.
> >
> >
> > Tony et al
> >
> > This was discussed at some length in the VOQL group (and
> maybe also in
> > the registry group?).
> >
> > We are assuming that we have some generic queries, such as a cone
> > search with added constraints (e.g. V_magnitude > 20) which can be
> > sent out meaningfully to multiple sites. Clearly the only way to
> > specify these is to use UCDs, since column names will vary.
> Somewhere
> > the UCD has to be translated into a specific column name.
>
> ...
>
> Clive,
>
> This sounds right to me. If I understand it we have a sequence of
>
> - query to the registry to find resources of interest
> - explosion of a 'generic' query to selected resources
> - translation of the generic query to local query syntax
>
> Here the registry query may involve terms separate from the
> generic query. E.g., this might be 'Find me all the catalogs
> with proper motions'. The syntax of this query would need to
> be expressible in VOQL and the basic registry service is to
> satisfy such queries.
>
> The explosion of the generic query to the selected sites
> is the responsibility of whatever
> VO service the ultimate user is communicating with. I don't
> think I would define it as a registry function, though I can
> certainly imagine building services that combine registries
> and exploders. The generic query might be something like a
> cone search in the Hyades region.
>
> The generic query is sent to each of the selected services
> and translated into local terms separately at each of them.
> Our example cone search might be translated into a nasty bit
> of SQL constraints on the coordinates. So for a generic
> query, the user needn't know of the actual column names until
> getting the results back.
>
> Presumably VOQL also needs to support queries where the
> specific column names are specified too -- but these would be
> 'manual' queries where the user already knows the structure
> of the table to be queried. So getting these names from the
> registry doesn't seem particularly vital -- though I'd want
> the protocol for returning data from the registry to define
> how column names are returned if a registry chooses to include them.
>
>
>
> The astronomers original query may have been:
>
> "Find me the proper motion of all stars in the Hyades region."
>
> The specification of the desired outputs defines distinct
> elements of the registry query. We'd probably also want to
> pass along the elements of the generic query too, since the
> registry may be fine-grained enough to use them. E.g., the
> registry can filter out proper motion catalogs that don't
> cover the Hyades region. However a given registry may choose
> to ignore any or all of the generic query constraints.
>
>
> Tom
>
>
>
More information about the voql
mailing list