Re: Query message format

From: Cui Chenzhou <ccz-at-bao.ac.cn>
Date: Wed, 19 Feb 2003 17:49:41 +0800

Tony Linde wrote:
> Hi Ray,
>
> I'll clear up one thing first:
>
> .....
>

>>What is less 
>>clear is whether any can address the problem of converting 
>>queries containing standard metadata into native metadata (in 
>>the case where a user interacts with many databases 
>>uniformily).  

>
>
> Good point.
>
> To generalise, the steps in a query are:
>
> 1. user constructs query, probably using interactive form
>
> 2. user submits query to workflow engine which constructs voql-type
> query from form fields
>
> 3. query is 'parsed' to determine both the sources addressed and the
> UCDs (for want of a better term) used
>
> 4. query is deconstructed into a plan (including voql sub-queries on
> individual data sources and subsequent joins, requeries, etc)
>
> 5. individual voql sub-queries are translated to that database's
> language
>
> 6. individual db queries have columns mapped from UCDs to internal
> schema names
>

Should columns be mapped from UCDs to internal schema names first before its individual querie is translated to the database's language?

> 7. plan is executed (includes steps for reverse mapping of individual
> results back to UCD form)
>
> 8. results returned to user (in VOTable form)
>
> Wow! Big job whatever form we choose for voql. Hands up all those with
> knowledge of constructing languages, parsers etc :)
>
> Cheers,
> Tony.
>

-- 
============================================================
Chenzhou Cui  (Ph.D. Candidate, LAMOST project)
National Astronomical Observatory | Tel: (8610)64877703-1320
Chinese Academy of Sciences       | FAX: (8610)64878240
Datun Road 20A, Chaoyang District | Email: ccz-at-bao.ac.cn
Beijing 100012, China             | WWW: www.lamost.org/~cb
============================================================

Received on 2003-02-19Z10:53:54