[hibernate-issues] [Hibernate-JIRA] Commented: (HSEARCH-704) Loading objects leads to 1+n queries when JPA's @Id != @DocumentId property

[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-704?pag... ] Emmanuel Bernard commented on HSEARCH-704: ------------------------------------------ Attaching the discussion between Hardy and me on the subject 11:14 emmanuel ok the crux of the issue is about entities that have their @DocumentId which are not the JPa's @Id hardy: if you look at returnAlreadyLoadedObjectsInCorrectOrder 11:15 hardy i see the FIXME 11:15 emmanuel the idea is that the objects do ahve been loaded and we simply use the HashMap nature of Hibernate to retrieve them But executeLoad does not do that int he case where @Id != @DocumentId do you remember this code? 11:16 hardy it uses a Criteria 11:16 emmanuel right 11:16 hardy well, I remember partly. It was quite some time ago 11:17 emmanuel I think we should try and find a better approach somehow at some point 11:17 hardy for the @Id != @DocumentId case you mean? 11:17 emmanuel Maybe the first criteria query loads the object and project the docmentIdProperty and the entity 11:17 hardy or the whole apporach? 11:17 emmanuel and we keep this aroud yes just for the @Id != @DocumentId Today we literally do 1+ n queries It's not super urgent but we cna't really say that we support this use case until we are somewhat close performance wise. Today it's a bit misleading and I fear people will learn it the hard way 11:20 hardy makes sense. we need to address this 11:20 emmanuel anyways, I wanted to know if you had an idea at the time on how to improve that 11:21 hardy not at the time what do you mean with "Maybe the first criteria query loads the object and project the docmentIdProperty and the entity" 11:22 emmanuel The key is that we want the object by documentId value if we do have that we don't have to do the criteria query again just to order the objects so imagine it's a user with id and a ssn the ssn is the @DocumentId if the first criteria query returns rows that looks like { ssn1, person1 } we have the data in memory and we don't need the other criteria queries makes sense? 11:24 hardy i see where you are going with that an other thing we could try is to tackle the 1 + n problem, by not loading each entity with a separate query 11:26 emmanuel hardy: that's what my proposal aims at 11:26 hardy right 11:26 emmanuel the first query is a criteria query that does where 'ssn in (:listOfSSns)' the n other queries are ssn = :ssn but if we have the ssn, entity tuple in memory from the first query we eliminate the n other queries altogether -- 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