BatchFetchQueue.getXXXBatch() cache check impeding batch fetching ----------------------------------------------------------------- Key: HHH-2223 URL: http://opensource.atlassian.com/projects/hibernate/browse/HHH-2223 Project: Hibernate3 Type: Bug Components: core Versions: 3.2.0.ga Environment: 3.2.0GA, Mac OS X/Solaris 10, Sybase 12.5/HSQLDB Reporter: Joel Caplin Priority: Minor Attachments: testcase.tar I have attached a test case which demonstrates the issue at hand by implementing a cheap data model: Pub >---|- Manager -|--< Employee Pub and Employee are uncached; Manager is cached (tried under both Eh and the non-prod HashMap implementation). All associations are lazy. The test case has a default batch size of 10 and uses hsqldb to demonstrate the problem - which was diagnosed against Sybase. When BatchFetchQueue.getXXXBatch() is invoked, it does not load any objects it finds in the second level cache - it merely acknowledges their presence by logging to INFO (isCached()). This in turn means that any child associations in that cached object will not have a proxy created for them. Suppose we want to find all the employees of the pubs in our database. We choose to do this by loading all the Pubs ("FROM Pub") and calling thePub.getManager().getEmployees() iteratively. When the managers are in the second level cache, Hibernate issues 11 select statements: one to load the pubs and one for each distinct manager. When the Managers are not in the cache, Hibernate issues 3 select statements: one to load the pubs, one to load the managers and one to load the employees. A solution might be to load the object from the cache instead of just ignoring it? -- 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