[ http://opensource.atlassian.com/projects/hibernate/browse/HHH-3930?page=c... ] Stuart Douglas commented on HHH-3930: ------------------------------------- I have run into this issue, and it is really hurting performance. I think the way to solve it is to add a new Map to PersistenceContext than maps foreign key values to the owner entity, with the key being a tupple of (EntityType,PropertyName,foreignKeyValue). Then when a lazy fetch is triggered from the non owning side of a relationship this map can be checked first to see if the object is already in the PeristenceContext thus avoiding the extra SQL. -- This message is automatically generated by JIRA. - If you think it was sent incorrectly contact one of the administrators: http://opensource.atlassian.com/projects/hibernate/secure/Administrators.... - For more information on JIRA, see: http://www.atlassian.com/software/jira

[ http://opensource.atlassian.com/projects/hibernate/browse/HHH-3930?page=c... ] Strong Liu updated HHH-3930: ---------------------------- Description: {code} @Entity public class Address { @Id @GeneratedValue(strategy = GenerationType.AUTO) private Long id; @OneToOne(mappedBy = "address") private Customer customer; } @Entity public class Customer { @Id @GeneratedValue(strategy = GenerationType.AUTO) private Long id; @OneToOne(fetch = FetchType.EAGER, cascade = CascadeType.ALL) private Address address = new Address(); public Long getId() { return id; } public Address getAddress() { return address; } } {code} This mapping causes 2 instead of the expected 1 query to retrieve a Customer and its Address from the db: select * from Customer customer0_ left outer join Address address1_ on customer0_.address_id=address1_.id where customer0_.id=? select * from Customer customer0_ left outer join Address address1_ on customer0_.address_id=address1_.id where customer0_.address_id=? Changing the mapping to a LAZY fetch type: @OneToOne(fetch = FetchType.LAZY, cascade = CascadeType.ALL) private Address address = new Address(); Causes 3 select queries instead of the expected 2 queries to retrieve a Customer (and its Address) from the db: select * from Customer customer0_ where customer0_.id=? select * from Address address0_ left outer join Customer customer1_ on address0_.id=customer1_.address_id where address0_.id=? select * from Customer customer0_ where customer0_.address_id=? The third select is superfluous because the relationship is already completely known: you already have the customer, so why not just set it on the address entity? Making the address field in Customer a @ManyToOne doesn't make a difference. Making the customer field in Address a @OneToMany does remove the extra select, but forces our model to change the relationship from Customer to List<Customer> where we *know* there'll be only 1 element. Apparently Hibernate can figure out the reverse relationship with a @ManyToOne - @OneToMany without the need for additional queries, can't this be extended to @OneToOne bidirectional relationships as well? was: @Entity public class Address { @Id @GeneratedValue(strategy = GenerationType.AUTO) private Long id; @OneToOne(mappedBy = "address") private Customer customer; } @Entity public class Customer { @Id @GeneratedValue(strategy = GenerationType.AUTO) private Long id; @OneToOne(fetch = FetchType.EAGER, cascade = CascadeType.ALL) private Address address = new Address(); public Long getId() { return id; } public Address getAddress() { return address; } } This mapping causes 2 instead of the expected 1 query to retrieve a Customer and its Address from the db: select * from Customer customer0_ left outer join Address address1_ on customer0_.address_id=address1_.id where customer0_.id=? select * from Customer customer0_ left outer join Address address1_ on customer0_.address_id=address1_.id where customer0_.address_id=? Changing the mapping to a LAZY fetch type: @OneToOne(fetch = FetchType.LAZY, cascade = CascadeType.ALL) private Address address = new Address(); Causes 3 select queries instead of the expected 2 queries to retrieve a Customer (and its Address) from the db: select * from Customer customer0_ where customer0_.id=? select * from Address address0_ left outer join Customer customer1_ on address0_.id=customer1_.address_id where address0_.id=? select * from Customer customer0_ where customer0_.address_id=? The third select is superfluous because the relationship is already completely known: you already have the customer, so why not just set it on the address entity? Making the address field in Customer a @ManyToOne doesn't make a difference. Making the customer field in Address a @OneToMany does remove the extra select, but forces our model to change the relationship from Customer to List<Customer> where we *know* there'll be only 1 element. Apparently Hibernate can figure out the reverse relationship with a @ManyToOne - @OneToMany without the need for additional queries, can't this be extended to @OneToOne bidirectional relationships as well? -- This message is automatically generated by JIRA. For more information on JIRA, see: http://www.atlassian.com/software/jira