Sequence number different in DB than from assignSequenceNumber()

Question

I am using TopLink 10.1.3 DP4 (I think - inside JDev build 3565). I am attempting to use a sequence in a schema that I DO NOT OWN - I cannot make modifications tot he database sequence because there are other applications in place running against it.
The database sequence is in a 10g database, with an interval of 1 and a cache of 20.
My object uses the value of the sequence as the sole element of its primary key. I am creating a new object, registering it with a UnitOfWork using registerNewObject(obj). It seems that this does not make a clone, since the object returned by this method is the same identity as the object passed in.
My project sequencing policy is set as follows:
<project-sequencing-policy>
<sequencing-policy>
<preallocation-size>1</preallocation-size>
<sequencing-type>Use native sequencing</sequencing-type>
<name-field-handle>
<field-handle/>
</name-field-handle>
<counter-field-handle>
<field-handle/>
</counter-field-handle>
<sequencing-policy-table></sequencing-policy-table>
</sequencing-policy>
</project-sequencing-policy>
Platform is 10g
<platform-name>Oracle10g</platform-name>
The sequence is accessed through a synonym (since JDev couldn't seem to reference a sequence in a different schema from a user's login).
<sequence-number-name>EVENT_SEQ_SYN</sequence-number-name>
<uses-sequencing>true</uses-sequencing>
When I commit my new object, I can see the insert statements in the log claiming a value of 'x' for the primary key. The row in the database actually has a primary key of 'y', where 'x = y - preallocation size' and 'y = sequence last number - preallocation size'. I think I would get the right number if I could set the preallocation size to 0 (performance problems understood) but TopLink chokes on this at run-time.
Is this problem familiar to anyone?
Failing using native sequencing, I tried to write custom SQL for the insert statement, but I couldn't find any decent examples anywere on how to do that. I found in a deep google search an example that showed you could use hash (#) to reference the value of a property, but what if the value that needs to be inserted in the database is held in an indirect referenced object? For example, if I was trying to write an insert sql for a Pet object with an indirect reference (valueHolder) to its owner, and I needed to put the owner id in the row - how would I write that insert statement in the 'custom sql' pane in workbench?
Thanks for your help.
Dave

Answer 1

Very strange case.
I tried to reproduce it using TopLink Employee example - and couldn't.
Here's the code:
    //  to get debug info
    session.setLogLevel(SessionLog.ALL);
    session.login();
    UnitOfWork uow = session.acquireUnitOfWork();
    Employee emp = new Employee();
    emp.setFirstName("sequencingTest");
    uow.registerNewObject(emp);
    uow.assignSequenceNumber(emp);
    uow.commit();
    System.out.println("firstName = "+emp.getFirstName() +"; id="+ emp.getId());And here's the log:
[TopLink Info]: 2006.01.26 04:49:05.734--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])--TopLink, version: Oracle TopLink - 10g Release 3 (10.1.3.0.0) (Build 060116)
[TopLink Config]: 2006.01.26 04:49:06.171--DatabaseSessionImpl(18)--Connection(19)--Thread(Thread[main,5,main])--connecting(DatabaseLogin(
     platform=>Oracle9Platform
     user name=> "test"
     datasource URL=> "jdbc:oracle:thin:@localhost:1521:orcl"
[TopLink Config]: 2006.01.26 04:49:10.140--DatabaseSessionImpl(18)--Connection(39)--Thread(Thread[main,5,main])--Connected: jdbc:oracle:thin:@localhost:1521:orcl
     User: TEST
     Database: Oracle Version: Oracle9i Enterprise Edition Release 9.2.0.1.0 - Production
With the OLAP and Oracle Data Mining options
JServer Release 9.2.0.1.0 - Production
     Driver: Oracle JDBC driver Version: 10.1.0.4.0
[TopLink Finest]: 2006.01.26 04:49:10.625--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])--sequencing connected, state is Preallocation_NoTransaction_State
[TopLink Finest]: 2006.01.26 04:49:10.718--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])--sequence PROJ_SEQ: preallocation size 1
[TopLink Finest]: 2006.01.26 04:49:10.718--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])--sequence ADDRESS_SEQ: preallocation size 1
[TopLink Finest]: 2006.01.26 04:49:10.718--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])--sequence EMP_SEQ: preallocation size 1
[TopLink Info]: 2006.01.26 04:49:11.453--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])-- login successful
[TopLink Finer]: 2006.01.26 04:49:11.703--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])--acquire unit of work: 48
[TopLink Finest]: 2006.01.26 04:49:11.703--UnitOfWork(48)--Thread(Thread[main,5,main])--Register the new container bean Employee: sequencingTest
[TopLink Finest]: 2006.01.26 04:49:11.734--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])--Execute query ValueReadQuery()
[TopLink Fine]: 2006.01.26 04:49:11.750--DatabaseSessionImpl(18)--Connection(39)--Thread(Thread[main,5,main])--SELECT EMP_SEQ.NEXTVAL FROM DUAL
[TopLink Finest]: 2006.01.26 04:49:12.328--DatabaseSessionImpl(18)--Thread(Thread[main,5,main])--sequencing preallocation for EMP_SEQ: objects: 1 , first: 3,469, last: 3,469
[TopLink Finest]: 2006.01.26 04:49:12.328--UnitOfWork(48)--Thread(Thread[main,5,main])--assign sequence to the object (3,469 -> Employee: sequencingTest )
[TopLink Finer]: 2006.01.26 04:49:12.328--UnitOfWork(48)--Thread(Thread[main,5,main])--begin unit of work commit
[TopLink Finer]: 2006.01.26 04:49:12.421--DatabaseSessionImpl(18)--Connection(39)--Thread(Thread[main,5,main])--begin transaction
[TopLink Finest]: 2006.01.26 04:49:12.437--UnitOfWork(48)--Thread(Thread[main,5,main])--Execute query InsertObjectQuery(Employee: sequencingTest )
[TopLink Finest]: 2006.01.26 04:49:12.625--UnitOfWork(48)--Thread(Thread[main,5,main])--Assign return row DatabaseRecord(
     EMPLOYEE.VERSION => 1)
[TopLink Fine]: 2006.01.26 04:49:12.625--UnitOfWork(48)--Connection(39)--Thread(Thread[main,5,main])--INSERT INTO EMPLOYEE (EMP_ID, L_NAME, F_NAME, GENDER, END_DATE, START_DATE, MANAGER_ID, START_TIME, END_TIME, ADDR_ID, VERSION) VALUES (3469, NULL, 'sequencingTest', NULL, NULL, NULL, NULL, {t '09:00:00'}, {t '17:00:00'}, NULL, 1)
[TopLink Fine]: 2006.01.26 04:49:12.640--UnitOfWork(48)--Connection(39)--Thread(Thread[main,5,main])--INSERT INTO SALARY (SALARY, EMP_ID) VALUES (0, 3469)
[TopLink Finer]: 2006.01.26 04:49:12.781--DatabaseSessionImpl(18)--Connection(39)--Thread(Thread[main,5,main])--commit transaction
[TopLink Finer]: 2006.01.26 04:49:12.796--UnitOfWork(48)--Thread(Thread[main,5,main])--end unit of work commit
[TopLink Finer]: 2006.01.26 04:49:12.796--UnitOfWork(48)--Thread(Thread[main,5,main])--release unit of work
firstName = sequencingTest; id=3469
If I understood correctly, the problem is x showing up in insert as a value to be assigned to PK and y actually inserted into the db. If that's the case could there be a BeforeInsert trigger on the table? To test try inserting through jdbc using concrete pk value.