Shiva, implementing separate classes with their own mappings for your special queries should work. You only need to be carefull when changing objects as I described in my previous mail.
Still I think your application will perform better if you use the OID loading demonstrated in PTF as long as your cache is able to hold most of the objects needed. Having said that I use this technic in my applications also. Regards Ralf Shiva P. Kodityala schrieb: > Ralf > > How about this? May be an old idea and not a good one. For each query I make > up a mapping(1:1, 1:m, m:n) involving the columns required. In this case I > get the columns what I need. It may get complicated for very complex queries. > > -----Original Message----- > From: Ralf Joachim [mailto:[EMAIL PROTECTED] > Sent: Wednesday, November 16, 2005 3:36 PM > To: [email protected] > Subject: Re: [castor-user] Select Query > > > Hi Shiva, > > lets assume you have a class AZ with properties A, B, C, .... until Z > and you have a mapping for this class to table AZ. Then you can also > create a class ABC with properties A, B, C only and map this also to > table AZ. But there are some limitations if you use that: > > 1. queries on ABC can only use the properties of ABC so it is not > possible to add a condition for Z. > 2. you have to take care not to be hit by ObjectModifiedExceptions which > happen if a cached object differ from its representaion in the database. > Think of the following situation: > > - You load an object with identity 1 as AZ with A='foo' which causes > this object to be cached. > - Then you load the same object with identity 1 as ABC and set A='bar' > this also causes this object to be cached but with A='bar'. The column > for A in the database now also contains 'bar'. > - If you now call db.load(AZ.class, new Integer(1) you will get the > cached object with A='foo' which is wrong. > - and if you try to set A='foobar' on that AZ instance and try to > persist it you will get that ObjectModifiedException. > > A way to prevent this is to only cache AZ instances and only change AZ > instance while ABC instances are not cached and never changed. > > According to this problems I would not suggest you to use this > workaround until you really know what you are doing. > > If you search for some performance tips you would propably want to have > a look at: > > http://castor.codehaus.org/jdo-best-practice.html > > I also suggest you to study our performance test suite PTF as this also > contains some tricks that could increase load performance dramatically > by optimizing usage of cache. Especially have a look at the tests that > execute pass through queries that only return object identities which > are the then used to db.load() the objects from cache if present there. > Depending on your class hierarchy you can see load time reduced by a > factor of 10. > > Regards > Ralf > > > > Shiva P. Kodityala schrieb: > > >>Thank you, Ralf. >> >>In this case, I may do it. No issue. The problem is when you have so many >>fields in a table and you just want a few of them. Is there any other way to >>do it? Or shall I use plain jdbc? Performance is important when you are >>dealing with huge number of rows. >> >> >> >> >> >> >>-----Original Message----- >>From: Ralf Joachim [mailto:[EMAIL PROTECTED] >>Sent: Wednesday, November 16, 2005 1:31 PM >>To: [email protected] >>Subject: Re: [castor-user] Select Query >> >> >>Hi Shiva, >> >>why not only returning 'p' which holds the properties 'a', 'b' and 'd'. >>Your OQL for this should be: >> >>SELECT p FROM com.scif.test.Table1 p WHERE p.a=$1 >> >>Having said that castor's OQL not supports to return multiple values at >>the moment. >> >>Regards >>Ralf >> >> >>Shiva P. Kodityala schrieb: >> >> >> >> >>>"SELECT p.a , p.b , p.d FROM com.scif.test.Table1 p WHERE a=$1" >>> >>>Is there anything wrong with the query? >>> >>>Getting error: >>>org.exolab.castor.jdo.oql.OQLSyntaxException: An incorrect token type was >>>found >>>near , (found COMMA, but expected KEYWORD_FROM) >>> >>>Can I not select more than one value? Is there any other way to doit? >>> >>>Thanks >>> >>>------------------------------------------------- >>>If you wish to unsubscribe from this list, please >>>send an empty message to the following address: >>> >>>[EMAIL PROTECTED] >>>------------------------------------------------- >>> >>> >>> >>> >>> >> >> >>------------------------------------------------- >>If you wish to unsubscribe from this list, please >>send an empty message to the following address: >> >>[EMAIL PROTECTED] >>------------------------------------------------- >> >> >>------------------------------------------------- >>If you wish to unsubscribe from this list, please >>send an empty message to the following address: >> >>[EMAIL PROTECTED] >>------------------------------------------------- >> >> >> > > > > ------------------------------------------------- > If you wish to unsubscribe from this list, please > send an empty message to the following address: > > [EMAIL PROTECTED] > ------------------------------------------------- > > > ------------------------------------------------- > If you wish to unsubscribe from this list, please > send an empty message to the following address: > > [EMAIL PROTECTED] > ------------------------------------------------- -- Syscon Ingenieurbüro für Meß- und Datentechnik GmbH Ralf Joachim Raiffeisenstraße 11 D-72127 Kusterdingen Germany Tel. +49 7071 3690 52 Mobil: +49 173 9630135 Fax +49 7071 3690 98 Email: [EMAIL PROTECTED] Web: www.syscon-world.de ------------------------------------------------- If you wish to unsubscribe from this list, please send an empty message to the following address: [EMAIL PROTECTED] -------------------------------------------------
