FULL PRODUCT VERSION :
java version "1.6.0-beta2"
Java(TM) SE Runtime Environment (build 1.6.0-beta2-b86)
Java HotSpot(TM) Client VM (build 1.6.0-beta2-b86, mixed mode, sharing)

EXTRA RELEVANT SYSTEM CONFIGURATION :
using Derby database snapshot: db-derby-snapshot-10.2.0.3-412239

A DESCRIPTION OF THE PROBLEM :
The @Select.allColumnsMapped=true annotation value seems to conflict with a data class that uses @ResultColumn to map the table column to the field. When such a combination is used, the returned DataSet<T> does not contain any objects. If allColumnsMapped is false, not specified, or the data class does not have any @ResultColumn-mapped fields, everything works as expected.

I don't believe this is an issue with Derby since it appears that Derby uses the default QueryObjectGenerator implementation (com.sun.sql.QueryObjectGeneratorImpl).

ACTUAL -
The output of the attached program is:

Creating & opening DB
Create person table
Inserting 3 records into person table
Records inserted
Finding person records without 'allColumnsMapped'
DataSet size (should be 3): 3
Displaying all people
Smith
Johnson
Wagner
All people displayed
Finding person records *with* 'allColumnsMapped'
DataSet size (should be 3): 0
Displaying all people
All people displayed

REPRODUCIBILITY :
This bug can be reproduced always.

---------- BEGIN SOURCE ----------
import java.sql.*;

public class QueryTest {
    
    public static void main(String... args) throws Exception {
        
        System.out.println("Creating & opening DB");
        Connection conn = DriverManager.getConnection("jdbc:derby:queryTest;create=true");
        System.out.println("Create person table");
        PreparedStatement ps = conn.prepareStatement("create table person (last_name varchar(50))");
        ps.execute();
        ps.close();
        System.out.println("Inserting 3 records into person table");
        ps = conn.prepareStatement("insert into person values (?)");
        ps.setString(1, "Smith");
        ps.addBatch();
        ps.setString(1, "Johnson");
        ps.addBatch();
        ps.setString(1, "Wagner");
        ps.addBatch();
        ps.executeBatch();
        ps.close();
        System.out.println("Records inserted");
        
        System.out.println("Finding person records without 'allColumnsMapped'");
        PersonQueries queries = conn.createQueryObject(PersonQueries.class);
        DataSet<Person> people = queries.getAllPeople();
        System.out.println("DataSet size (should be 3): " + people.size());
        System.out.println("Displaying all people");
        for (Person p : people) {
            System.out.println(p.lastName);
        }
        System.out.println("All people displayed");
        
        System.out.println("Finding person records *with* 'allColumnsMapped'");
        people = queries.getAllPeopleBroken();
        System.out.println("DataSet size (should be 3): " + people.size());
        System.out.println("Displaying all people");
        for (Person p : people) {
            System.out.println(p.lastName);
        }
        System.out.println("All people displayed");
        
        queries.close();
        conn.close();
    }
    
    public static interface PersonQueries extends BaseQuery {
        
        @Select(sql = "select last_name from person", allColumnsMapped = false)
        DataSet<Person> getAllPeople();
        @Select(sql = "select last_name from person", allColumnsMapped = true)
        DataSet<Person> getAllPeopleBroken();
    }
    
    public static class Person {
        @ResultColumn("last_name") public String lastName;
    }
}

---------- END SOURCE ----------

CUSTOMER SUBMITTED WORKAROUND :
The best workaround I've found is to leave allColumnsMapped = false (the default). The disadvantage is that allColumnsMapped = true would provide more runtime safety in some use cases.