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. This _could_ be done by the Registry, if it has the requisite information, but it may not, because not everyone is yet convinced that the fine-grained Registry is the way to go, so not all of them may ingest the necessary data. It can _always_ be translated by the end data centre, since it knows the names of its own columns. If information on the UCDs of each column is not present at the end data centre, then it will almost certainly not be present in the Registry, since where else can it come from? So we concluded that the most reliable way to do the translation is at the data centre. This simplifies the Registry and workflow, without significantly complicating the data centre interface (which has to be handle UCDs and columns anyhow).
> PS: Sorry about the cross-post. This seems to be more of a VOQL question
> now, so shall we continue the discussion on that list ??
That's what I've done. Sorry if Registry folks have to read the VOQL archive to see this posting.
-- Clive Page, Dept of Physics & Astronomy, University of Leicester, Tel +44 116 252 3551 Leicester, LE1 7RH, U.K. Fax +44 116 252 3311Received on 2003-05-20Z18:00:20