Re: [search-ws] SRU/OASIS: Paramerized Query - draft message

[Farrukh Najmi <farrukh@wellfleetsoftware.com>]

Hi Ray,

Thanks for crafting the issues.

For each such email please make sure that you provide:

    * The issue list URL
    * The issue id
    * The link to the archived thread related to the issue as indicated
      in the issue list

More inline below...

Ray Denenberg, Library of Congress wrote:
> Draft message:
>
> --------------------------------------
> There is a desire among some committee members to incorporate support for
> Parameterized Query.
>
> In Parameterized Query there is a complete query pre-configured on the
> server complete with all supported predicates, and parameter variables.
> Clients invoke the query by  id and supply a subset of supported parameters.
> (The server prunes the query to remove predicate for which parameters  were
> not supplied.)
>
> A typical PQ request might look like:
>
>                       http ://example.com/search?queryId=123&name=japan
>
> Which isn't consistent with SRU/CQL syntax.
>   

Above looks good.

> Alternatively a CQL query could be constructed so the request would look
> like
>
>              http ://example.com/search?query=queryId=123&20and&20name=japan
>
> That is, an index, 'queryId',  would be defined,  and there would be rules
> associated with this index, for example, when a query uses this index,  the
> remainder of the query must be limited to boolean ANDs and predicates where
> the relation is always '=" and the index corresponds to a parameter defined
> by the stored query.
>   

Suggest moving above to later below...

> There has been debate within the committee. Some are comfortable with
> representing PQ as just another CQL query (with the restrictions stated
> above) while other members feel that these  special case semantics and
> constraints are indicative of an unstable design for supporting
> parameterized queries, and would like to see PQ singled out as a special
> mode of operation.
>   

Suggest replacing above paragraph with following:

"

There has been debate within the committee. 

Some feel that a PQ can be represented as just another CQL query using the current syntax.
Here is an example of of the example PQ could look like in this case:

http ://example.com/search?query=queryId=123&20and&20name=japan

That is, an index, 'queryId',  would be defined,  and there would be rules
associated with this index, for example, when a query uses this index,  the
remainder of the query must be limited to boolean ANDs and predicates where
the relation is always '=" and the index corresponds to a parameter defined
by the stored query.


Others feel that the a CQL query can be expressed as just another PQ using the PQ
syntax. The later proposal is described in the following document (which if for discussion only and has no formal standing):

<http://www.oasis-open.org/apps/org/workgroup/search-ws/download.php/25861/search-ws-param-query-1.0-draft1.pdf>

Discussion on this issue may be found in the search-ws TC archives at:

<http://www.oasis-open.org/apps/org/workgroup/search-ws/email/archives/200710/msg00064.html>

"


> SRU implementor feedback is solicited on this issue.
>
>   






-- 
Regards,
Farrukh Najmi

Web: http://www.wellfleetsoftware.com