Eclipselink JPA在使用生成的主键插入Derby表时出错

当持久化到具有生成主键的表时，EclipseLink似乎错误地将空主键值传递给Derby . 在这种情况下，Derby返回 Attempt to modify an identity column 的错误 . Derby需要一个排除id值的SQL语句 . 我的问题是如何强制EclipseLink发送正确的SQL？细节如下......

我使用Eclipselink映射到最新的Netbeans 7.3beta2下的Derby数据库（v10.8.1.2） .

数据库表主键是自动生成的：

CREATE TABLE STUDENT_BATCH ( 
    ID bigint PRIMARY KEY
        GENERATED ALWAYS AS IDENTITY (START WITH 1, INCREMENT BY 1),
    FILENAME varchar(200) NOT NULL,
    SCHOOLBOARD varchar(100) NOT NULL,
    CREATE_TS timestamp NOT NULL,
    CONTACT_INFO varchar(200),
    NOTES varchar(2000),
    BOARD_NAME varchar(100)
)
;

相应的jpa类指定使用identity策略生成id：

@Entity
@Table(name = "STUDENT_BATCH")
@XmlRootElement
@NamedQueries({
@NamedQuery(name = "StudentBatch.findAll", query = "SELECT s FROM StudentBatch s")})
public class StudentBatch implements Serializable {
    private static final long serialVersionUID = 1L;
    @Id
    @GeneratedValue(strategy = GenerationType.IDENTITY)
    @Column(name = "ID")
    private Long id;
    @Basic(optional = false)
    @Column(name = "FILENAME")
    private String filename;
    @Basic(optional = false)
    @Column(name = "SCHOOLBOARD")
    private String schoolboard;
    @Basic(optional = false)
    @Column(name = "CREATE_TS")
    @Temporal(TemporalType.TIMESTAMP)
    private Date createTs;
    @Column(name = "CONTACT_INFO")
    private String contactInfo;
    @Column(name = "NOTES")
    private String notes;
    @Column(name = "BOARD_NAME")
    private String boardName;
    @OneToMany(cascade = CascadeType.ALL, mappedBy = "studentBatchId")
    private Collection<StudentRecord> studentRecordCollection;

    methods etc follow...

当我去坚持这个课时，我没有指定一个id，所以Derby会提供id .

StudentBatch sb = new StudentBatch();
                sb.setBoardName(meta.get("BOARD NAME"));
                sb.setContactInfo(meta.get("CONTACT INFO"));
                sb.setCreateTs(new Date());
                sb.setFilename(event.getFile().getFileName());
                sb.setNotes(meta.get("NOTES"));
                sb.setSchoolboard(meta.get("SCHOOL BOARD"));
                _logger.debug("persisting batch");
                em.persist(sb);
                _logger.debug("flushing");
                em.flush();
                _logger.debug("flushed");

但是，Eclipselink将id传递给derby为null，Derby提供错误尝试修改标识列：

INFO: DEBUG  11710 27 Nov 2012 18:17:10,558 [http-thread-pool-8080(4)] (FileUploadController.java:75) - persisting batch
INFO: DEBUG  11713 27 Nov 2012 18:17:10,561 [http-thread-pool-8080(4)] (FileUploadController.java:77) - flushing
WARNING: Local Exception Stack: 
Exception [EclipseLink-4002] (Eclipse Persistence Services - 2.3.2.v20111125-r10461): org.eclipse.persistence.exceptions.DatabaseException
Internal Exception: java.sql.SQLSyntaxErrorException: Attempt to modify an identity column 'ID'. 
Error Code: -1
Call: INSERT INTO STUDENT_BATCH (ID, BOARD_NAME, CONTACT_INFO, CREATE_TS, FILENAME, NOTES, SCHOOLBOARD) VALUES (?, ?, ?, ?, ?, ?, ?)
bind => [7 parameters bound]
Query: InsertObjectQuery(ca.ontario.mhltc.studentupload.model.StudentBatch[ id=null ])
at org.eclipse.persistence.exceptions.DatabaseException.sqlException(DatabaseException.java:333)

这对我来说有点意义，因为如果我尝试在指定了id字段的sql命令行上执行插入，我也会收到错误 .

INSERT INTO STUDENT_BATCH (id, BOARD_NAME, CONTACT_INFO, CREATE_TS, FILENAME, NOTES, SCHOOLBOARD)
VALUES (null, 'abc', 'def', current_timestamp, 'aaa', 'aabb', '1234');

给我这个：

Error code -1, SQL state 42Z23: Attempt to modify an identity column 'ID'. 
Line 1, column 1

Execution finished after 0 s, 1 error(s) occurred.

但是，这个insert语句是成功的：

INSERT INTO STUDENT_BATCH (BOARD_NAME, CONTACT_INFO, CREATE_TS, FILENAME, NOTES, SCHOOLBOARD)
VALUES ('abc', 'def', current_timestamp, 'aaa', 'aabb', '1234');

回报

Executed successfully in 0.002 s, 1 rows affected.
Line 1, column 1

Execution finished after 0.002 s, 0 error(s) occurred.

在我看来，Eclipselink应该知道Derby如何处理标识列，并且不应该在insert语句中传递id列 . 有没有解决方法，或者我应该转储Derby并使用其他数据库？