Mailing-List: contact dev-help@jackrabbit.apache.org; run by ezmlm
Precedence: bulk
Reply-To: dev@jackrabbit.apache.org
Date: Fri, 25 Mar 2011 12:15:05 +0000 (UTC)
From: "Marlis Lamp (JIRA)" <jira@apache.org>
To: dev@jackrabbit.apache.org
Message-ID: 
 <1129332390.11039.1301055305896.JavaMail.tomcat@hel.zones.apache.org>
In-Reply-To: 
 <165457980.8073.1300958825745.JavaMail.tomcat@hel.zones.apache.org>
Subject: [jira] [Commented] (JCR-2930) same named child nodes disappear on
 restore
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Transfer-Encoding: 7bit


    [ https://issues.apache.org/jira/browse/JCR-2930?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13011171#comment-13011171 ] 

Marlis Lamp commented on JCR-2930:
----------------------------------

The error occurs at org.apache.jackrabbit.core.version.VersionManagerImplRestore.internalRestoreFrozen(VersionManagerImpl.java:493) when replacing all existing child nodes on N: 

// if node is present, remove it
state.removeNode(entry.getName(), entry.getIndex());

Let N (node) and F (frozen node) have 2 child nodes.
This removes the correct node from the state object at the first cycle, but at the second, when the first child node has been removed from position 1 and added to position 2 on the state object, the indices in the state object have moved and the childnode that has been added at the first cycle is removed at the second cycle. The second cycle also removes the 2nd child node from the state object in line 505:

                // check for existing
                if (f.getFrozenId() != null) {
                    if (stateMgr.hasItemState(f.getFrozenId())) {
                        NodeStateEx existing = state.getNode(f.getFrozenId());
                        if (removeExisting) {
                            NodeStateEx parent = existing.getParent();
                            parent.removeNode(existing);
                            parent.store(); 
                            ...

Finally the 2nd child node of F is added, resulting in a restored state, where the first child node of the frozen node is missing.

> same named child nodes disappear on restore
> -------------------------------------------
>
>                 Key: JCR-2930
>                 URL: https://issues.apache.org/jira/browse/JCR-2930
>             Project: Jackrabbit Content Repository
>          Issue Type: Bug
>          Components: jackrabbit-core, versioning
>    Affects Versions: 2.2.4
>            Reporter: Marlis Lamp
>            Priority: Critical
>         Attachments: VersionManagerImplRestore.patch
>
>
> When restoring a versionable node which has several (non-versionable) child nodes with the same name, some child nodes disappear. 
>             Node node = session.getRootNode().addNode("myNode");
>             node.addMixin("mix:versionable");
>             for (int i = 1; i < 6; i++) {
>                 Node child = node.addNode("child");
>                 child.setProperty("name", "child_"+i);
>             }
>             session.save();
>             VersionManager versionManager = session.getWorkspace().getVersionManager();
>             versionManager.checkin(node.getPath());
>             System.out.println("number of child nodes: " + node.getNodes().getSize());
>             versionManager.checkout(node.getPath());
>             node.getNode("child").setProperty("name", "modified");
>             session.save();
>             Version baseVersion = versionManager.getBaseVersion(node.getPath());
>             versionManager.restore(baseVersion, true);
>             System.out.println("number of child nodes in restored node: "+node.getNodes().getSize());
> produces the following output:
> number of child nodes: 5
> number of child nodes in restored node: 3
> Giving unique names or adding the mixin versionable to the child nodes solves the problem.

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