Is RowMapperResultReader caching result list?

[jgrange]

Hi,

I've just come accross what seems to me, to be rather odd behaviour when using a RowMapperResultReader.

Basically, I'm accessing an oracle stored procedure that returns a cursor as one of it's out parameters. My code is as follows:

public class PositionHistoryFindByPK extends StoredProcedure {

    private static final String SQL = "lv_sys.tracklog_interface_pkg.find_by_primary_key";

    private RowMapperResultReader resultReader = null;

    protected PositionHistoryFindByPK(DataSource dataSource) {
        RowMapper rowMapper = new PositionHistoryRowMapper();
        this.resultReader = new RowMapperResultReader(rowMapper, 1);
        setDataSource(dataSource);
        setSql(SQL);
        declareParameter(new SqlParameter("i_pk", Types.INTEGER));
        declareParameter(new SqlOutParameter("o_result", OracleTypes.CURSOR, this.resultReader));
    }

    public PositionHistoryVO getValueObject(Integer pk) {
            Map params = new HashMap(1);
            params.put("i_pk", pk);
            execute(params);
            return (PositionHistoryVO) DataAccessUtils.uniqueResult(this.resultReader.getResults());            
    }

}

All works fine the first time round. When, however, I try to get another result (with the same parameters), with the same PositionHistoryFindByPk object, it fails at the uniqueResult function, throwing a IncorrectResultSizeDataAccessException. It appears that the result size is equal to the number of calls I have made to getValueObject(pk)

If, however, I change the getValueObject code to read

        synchronized (this.resultReader){
            this.resultReader.getResults().clear();
            Map params = new HashMap(1);
            params.put("i_pk", pk);
            execute(params);
            return (PositionHistoryVO) DataAccessUtils.uniqueResult(this.resultReader.getResults());
        }

then everything works fine - the clearing of the results list between calls cures the problem.

Questions:

1) Is this the desired behaviour of the RowMapperResultReader?
2) If so, please explain why
3) Am I abusing the RowMapperResultReader? If so, how should it be used?

TIA


John Grange

[trisberg]

The JavaDoc states:
Note that a RowMapper object is typically stateless and thus reusable;
* just the RowMapperResultReader adapter is stateful.

I think you would be better off using a plain RowMapper and coding your parameter similar to this:

declareParameter(new SqlReturnResultSet("o_result", new RowMapper() {
	public Object mapRow(ResultSet rs, int rowNum)
		throws SQLException {
		Xxxx x = new Xxxx();
		//x.set .... set your properties here
		return x;
	}
}));

Then you can retrieve the List of returned objects from the Map returned in the execute call.

[jgrange]

Forgive me for being dense, please, but where do i get the list of objects mapped by the rowmapper back?

also, is there some magic I must perform to get this to work with oracle?

TIA

[trisberg]

Sorry, I forgot to include the SQLType - this should work for Oracle:

declareParameter(new SqlReturnResultSet("o_result", OracleTypes.CURSOR, new RowMapper() {
        public Object mapRow(ResultSet rs, int rowNum)
                throws SQLException {
                Xxxx x = new Xxxx();
                //x.set .... set your properties here
                return x;
        }
}));

Now, to get the List of returned objects you retrieve then from the output Map using the parameter name you used when you declared parameter. So in this case it would be:

List l = (List) outputMap.get("o_result");