我有一个Sping Boot 3.1.1 / Hibernate 6.2应用程序，它使用以下驱动程序连接到Oracle 12 c数据库：

<dependency>
    <groupId>com.oracle.database.jdbc</groupId>
    <artifactId>ojdbc8</artifactId>
    <scope>runtime</scope>
</dependency>

字符串
此应用程序具有以下存储库接口和类：

public interface SomeEntityRepository extends Repository<SomeEntity, SomeEntityId> {
  @Query("""
      FROM SomeEntity
      WHERE id = :id
        AND description is not null
      """)
  Optional<SomeEntity> find(SomeEntityId id);
}

x

@Embeddable
public class SomeEntityId {
    private Long idX;
    private Long idY;
    private Long idZ;
}

@Entity
public class SomeEntity {
    @EmbeddedId
    private SomeEntityId id;
    private String description;
}

的字符串
当执行find查询时，Hibernate 6.2将JPQL转换为以下SQL：

Hibernate: 
    select
        s1_0.idx,
        s1_0.idy,
        s1_0.idz,
        s1_0.description 
    from
        some_entity s1_0 
    where
        (
            s1_0.idx,s1_0.idy,s1_0.idz
        ) in (select
            ?,?,? 
        from
            dual) 
        and s1_0.description is not null

型
我想知道为什么Hibernate 6.2将id = :id转换为子查询，而不是像Hibernate 5那样逐个字段比较复合主键？
我尝试在Sping Boot 2.7.13 / Hibernate 5.6中执行相同的代码，JPQL按预期进行了翻译：

Hibernate: 
    select
        someentity0_.idx as idx1_1_,
        someentity0_.idy as idy2_1_,
        someentity0_.idz as idz3_1_,
        someentity0_.description as description4_1_ 
    from
        some_entity someentity0_ 
    where
        someentity0_.idx=? 
        and someentity0_.idy=? 
        and someentity0_.idz=? 
        and (
            someentity0_.description is not null
        )

型
使用AND和=单独比较复合PK属性，而不是使用IN运算符比较的Hibernate 6.2：

...
        (
            s1_0.idx,s1_0.idy,s1_0.idz
        ) in (select
            ?,?,? 
        from
            dual) 
...

型