I wanted to say thank you Danilo.
With your tip and a hunk of code from the Lucene project, I was able to
significantly enhance searching capabilities on my site. I need to get a bit
more understanding of how efficient memory wise a RowReader is. I am dealing
with large data sets where people are all searching at the same times, so
memory requirements could get high. This is why i was just trying to get a
distinct selection of service keys.
I am still thinking about the use of anonymous keys. Right now whenever a
person updates their record, the search criteria has to be deleted and re-put
back into the database. At some point (not anytime vaguely soon) I am going
to run out of room for the primary key and I would then have to reindex the
search table. This seems useless since I don't even want the primary key. I
just need to be able to do the 1:N matching. I am concerned about what was
said about anonymous keys not working in clustered environment. I need to
really understand this better.
Thanks again!!! It works!!!
JohnE
----- Original Message -----
From: Danilo Tommasina <[EMAIL PROTECTED]>
Date: Tuesday, November 23, 2004 4:44 am
Subject: Re: Unique query without using primary key?
> A ReportQueryByCriteria should do the job:
>
> something like this:
>
> Criteria crit = new Criteria();
> crit.addLike( "state", "somestate" );
> crit.addLike( "<ref To SearchWords>.searchword", "somesearchword" );
>
> ReportQueryByCriteria query = new ReportQueryByCriteria(
> SearchKeys.class, new String [] { "sid" }, crit, true );
>
> PersistenceBroker pb = null;
> try {
> pb = PersistenceBrokerFactory.defaultPersistenceBroker();
>
> ArrayList results = new ArrayList();
> for ( Iterator resultsIt = pb.getReportQueryIteratorByQuery(
> query ); resultsIt.hasNext(); ) {
> results.add( ((Object []) resultsIt.next())[ 0 ] );
> }
>
> // The results ArrayList now contains all the ids you need
> // Note: always extact the values from the Iterator before
> closing the broker.
>
> } catch ( Exception e ) {
> // Do exception handling
> } finally {
> if ( pb != null ) {
> pb.close();
> }
> }
>
> Replace <ref To SearchWords> with the reference path defined in
> your repository.xml
> I didn't tested the code, so there are probably syntax errors...
>
>
> > I am not getting this. I need a nudge in the right direction.
> >
> > I want to do this query:
> > select distinct sk.sid from searchkeys sk, searchwords sw
> > where sw.searchword="somesearchword"
> > sw.sid = sk.sid
> > sk.state="somestate";
> >
> > table searchkeys:
> > sid <-- unique PK
> > state
> >
> > table searchwords:
> > swid <-- unique PK I wish OJB
> didn't require
> > sid <-- sid in searchkey table
> > searchword
> >
> > searchkeys and searchwords having a 1:n relationship.
> >
> >
> > Right now I both tables mapped according to their
> characteristics which I have done many times before. In the past
> I have always wanted the primary keys and full records.
> >
> > Here I don't want the full records or mapping, but just a
> collection of the unique sid. I would prefer not to get the full
> records and I only require this for read and not write. Imagine
> I am doing a keyword search and I just want to know which services
> match the search criteria.
> >
> > Can anybody nudge me in the right direction? Would be great to
> get past this problem. I am trying to inprove the searching on my
> live OJB used site by including parts of the Lucene project, but
> need to be able to do this for it to work.
> >
> > Thanks lots.
> >
> >
> >
> >
> >
> > -----------------------------------------------------------------
> ----
> > To unsubscribe, e-mail: [EMAIL PROTECTED]
> > For additional commands, e-mail: [EMAIL PROTECTED]
> >
> >
>
>
> -------------------------------------------------------------------
> --
> To unsubscribe, e-mail: [EMAIL PROTECTED]
> For additional commands, e-mail: [EMAIL PROTECTED]
>
>
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
