jackrabbit-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Stefan Guggisberg <stefan.guggisb...@gmail.com>
Subject Re: Problem removing some nodes with SameNameSiblings
Date Tue, 25 Oct 2005 16:30:08 GMT
hi javier
yep, that's a bug. i created a jira issue:
http://issues.apache.org/jira/browse/JCR-261

it's an issue with the CachingHierarchyManager which doesn't update/flush its
path cache correctly on removal of same-name sibling entries. please note that
there was never the risk of unintenionally removing the root node, the
msg of the
exception is misleading.

btw: if you comment all "dump(topNode);" stmts in your test case the problem
doesn't occur anymore ;)

thanks for finding and reporting this bug!

cheers
stefan

On 10/25/05, Javier Bermejo <jbermejo@germinus.com> wrote:
> Hi,
>
> I'm trying to make a content management system using JSR170 and the
> Jackrabbit implementation using ORM-persistence connecting to a PostgreSQL.
>
> I think I have discovered a bug in jackrabbit when trying to delete
> several nodes using SameNameSibling.
>
> I have developed a test you can place in:
>
> src/test/org/apache/jackrabbit/test/api/
>
> and add the testCase to the suite in:
>
> src/test/org/apache/jackrabbit/test/api/TestAll.java
>  > suite.addTestSuite(RemoveSameNameSiblingNodesTest.class);
>
> The structure of the test is the following one:
> 1º I add a first topNode under the rootNode.
> 2º I add 4 nodes with the same path (with no properties to avoid
> complexity of the test)
> 3º Later, I get a NodeIterator to get the 4 previous added nodes:
>  > NodeIterator it = topNode.getNodes(myNodePath);
> 4º Now, this is the suspect BUG, inside the while loop getting the
> nodes, I make a call to the remove() item method. If I save each change
> on the parent node, you can see that every remove operation have a
> strange behaviour and when you are removing the third item it tries to
> delete the root node!!!! (You can see the dump output of the test)
>
> You can avoid this bug saving the session or the rootnode outside the
> while loop.
>
> Best regards,
>   Javier
>
>
>

Mime
View raw message