[hibernate-issues] [Hibernate-JIRA] Commented: (HHH-7309) Natural-id not cached in shared cache after insert

[Guenther Demetz (JIRA)]

    [ https://hibernate.onjira.com/browse/HHH-7309?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=46574#comment-46574 ] 

Guenther Demetz commented on HHH-7309:
--------------------------------------

Pull-request is: https://github.com/hibernate/hibernate-orm/pull/331

The proving testcase is: CachedMutableNaturalIdStrictReadWriteTest.testInsertUpdateEntity_NaturalIdCachedAfterTransactionSuccess

> Natural-id not cached in shared cache after insert 
> ---------------------------------------------------
>
>                 Key: HHH-7309
>                 URL: https://hibernate.onjira.com/browse/HHH-7309
>             Project: Hibernate ORM
>          Issue Type: Bug
>          Components: caching (L2), testsuite
>    Affects Versions: 4.1.3
>         Environment: EHCache as 2L-cache implementation (or another 2L-cache implementation with default strict READ_WRITE access strategy)
>            Reporter: Guenther Demetz
>            Priority: Minor
>   Original Estimate: 1h
>  Remaining Estimate: 1h
>
> Problem: 
> when using a 2L-cache implementation with default strict READ_WRITE access strategy,
> then after inserting/updating a new entity and committing, the concerning Natural-id is not cached in shared cache.
> Cause: 
> With strict READ_WRITE access strategy the improvement HHH-7279 "Prevent identical re-cachings in natural-id shared cache",
> fails to prevent identical recachings for in-transaction-created entities, because with strict READ_WRITE the actual put is defered to transaction completition.
> Recaching of identical natural-id-key, finally leaves the shared cache with valid entry.
> Solution description(see following pullrequest):
> To prove this in a hibernate matrix testcase, first I needed to enhance org.hibernate.testing.cache.CachingRegionFactory in manner that it also can return strict READ_WRITE as default access strategy, instead to always non-strict READ_WRITE.
> Going deeper, I saw that org.hibernate.testing.cache.NonstrictReadWriteNaturalIdRegionAccessStrategy under some aspects was reflecting more a strict than a non-strict strategy,
> this I fixed this and had finally convert CachingRegionFactory's default access type to strict READ_WRITE in order to have not a lot of matrix-testcases failing.
> Further detail in following pullrequest...
>  

--
This message is automatically generated by JIRA.
For more information on JIRA, see: http://www.atlassian.com/software/jira