[ http://opensource.atlassian.com/projects/hibernate/browse/HSEARCH-178?pag... ] Emmanuel Bernard updated HSEARCH-178: ------------------------------------- Environment: http://forum.hibernate.org/viewtopic.php?p=2381840 -- 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/HSEARCH-178?pag... ] Sanne Grinovero updated HSEARCH-178: ------------------------------------ Attachment: TestCase-HSEARCH-178.patch Attaching a testcase; it fails the test not using transactions. -- 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/HSEARCH-178?pag... ] Matt Sgarlata commented on HSEARCH-178: --------------------------------------- So if I want to use @IndexEmbedded I have to mark the association as eager? Ouch! -- 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/HSEARCH-178?pag... ] Matt Sgarlata commented on HSEARCH-178: --------------------------------------- I see, I thought I had a transaction running but it turns out I didn't. Thanks for your response! -- 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/HSEARCH-178?pag... ] Sanne Grinovero resolved HSEARCH-178. ------------------------------------- Resolution: Fixed resolved in both trunk and 3.1 branch -- 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/HSEARCH-178?pag... ] Emmanuel Bernard updated HSEARCH-178: ------------------------------------- Summary: Out of transaction work causes collection lazy loading to throw AssertionFailure (was: Some collection lazy loading lead to AssertionFailure) -- 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/HSEARCH-178?pag... ] Stephane Epardaud commented on HSEARCH-178: ------------------------------------------- I see from fisheye there are several changes to fix this: it looks like the processing is delayed until after flushing. Unfortunately I'm using Seam which isn't up to date with all the HS 3.1 requirements for hibernate so I can't upgrade, but I've already backported FullTextSession.flushToIndexes() to 3.0 which I'm using for my batch insertion: while(...){ // save some entities session.flush(); // this one first since it triggers the HS event listener session.flushToIndexes(); session.clear(); } This breaks in HS but only when HS's batch_size is smaller than my flush frequency (this is when flush will trigger the insert listener which will process the work instead of batching it for later) which happens easily with @ContainedIn since the queue goes up faster than the number of objects I'm inserting. I'll try backporting your commits and see if that fixes my problem. -- 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/HSEARCH-178?pag... ] Sanne Grinovero commented on HSEARCH-178: ----------------------------------------- Stephane, what do you mean by "not related to transactions for me"? I couldn't reproduce without. Yes you are right, the loading is now postponed until after the flush. Emmanuel I see that even the latest Seam 2.1.2CR2 released some days ago is still coming with hibernate core 3.4.2sp1, using therefore the quite old Search 3.0.1. I wonder if you could push them to revamp all hibernate libraries to the 3.3.x series? -- 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/HSEARCH-178?pag... ] Sanne Grinovero commented on HSEARCH-178: ----------------------------------------- HHH-3225 is not needed, but I'm not sure if this (HSEARCH-178) is also covering your case of flushing inside a transaction; please try it and attach a testcase to a new issue if it doesn't work. -- 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/HSEARCH-178?pag... ] Emmanuel Bernard commented on HSEARCH-178: ------------------------------------------ In a transaction we don't navigate the object graph, we just queue it for later (beforeCommit). Maybe 3.0 was different? -- 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/HSEARCH-178?pag... ] Stephane Epardaud commented on HSEARCH-178: ------------------------------------------- This is using Seam, I am in a Tx, I'm pretty sure... In 3.1 it looks the same though: # I call flush() # the FullTextEntityListener is called for my insertions # The work is added to the BatchedQueueingProcessor.add() # Since my batchSize is > 0 and < than workQueu.size() the DocumentBuilder is invoked # This loads lazy-loaded collections in flush() I'm not sure what would be a workaround, since my insert events cannot be notified to HS before flush is called so calling flushToIndexes before flush is not useful. Note that when doing batch inserts I don't think HS indexing should be deferred before commit, but simply after flush. Unless it could be asynchronous and done outside of my transaction entirely ;) -- 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/HSEARCH-178?pag... ] Emmanuel Bernard commented on HSEARCH-178: ------------------------------------------ ahh ok yes batchSize was a broken concept. In batch indexing, you would have batchSize=0 (deprecated really), then do flush(); flushToIndexes() on a regular basis. -- 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/HSEARCH-178?pag... ] Hardy Ferentschik commented on HSEARCH-178: ------------------------------------------- Well to answer the question regarding running tests. The 3.0.1 version is ant/ivy based, except for the main hibernate3.jar. For this to work you will also need a checkout of Hibernate 3.2.x and you have to run 'ant jar prepare-dir-up'. In your Hibernate Search checkout you then need to specify hibernate-core.home pointing to you checkout. If you have this setup you should be able to run a single testcase by running: ant junitsingle -Dtestname=org.hibernate.search.test.jpa.JPATestCase or ant junit to run them all. You should have a jdbc directory in your checkout though. At least I did when I checked out https://svn.jboss.org/repos/hibernate/search/tags/v3_0_1_GA -- 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