JUTreeBinding - navigation by findByKey() problem

Question

It will not be easy to provide a test case, so I hope that the following questions will clear things up for me.
We have two trees. Both trees have the same structure. Both trees should act as navigation tool.
This is done by using a TreeSelectionListener which somewhat looks like the JUTreeDefaultMouseListener, except:
It uses findByKey() and setCurrentRow() on the target RowSet so this RowSetIterator's currency is set and so that the ViewLink "chain" is activated.
Two problems:
- findByKey() does not return a row.
- both trees are not completely separated.
I did some research on the JUTreeBinding and in my test project, both problems didn't show up, so I am puzzled.
So my questions are:
1.)
Both VOs (the VO the tree node is bound to and the target VO) are two instances of the same VO type.
In the Javadocs, I read "We now use the key build a where-clause for an internal View Object. That where-clause is applied and qualifying rows are retrieved from it to find the requested number of rows." So it should always come up with at least one row, since both VOs are of the same type, right? Is there another thing that could keep findByKey() from finding a row? How does this where-clause look like?
2.)
If I have two trees with the same structure, then JClient creates an internal VO for every node type except the top-level VO. The problem seems to be: When I have two trees then there are two RowSets for each node type, but they are contained within the very same VO. One RowSet for tree A, one for tree B. So, under certain circumstances, if the user clicks a node in tree A (which then turns out to be a leaf) then the parent node of the corresponding node in tree B collapses (and vice versa). Initially both parent nodes collapsed, but I put in the workaround for the bug in JUTreeNodeBinding.updateValuesFromRows() Shailesh mentioned in an earlier posting. Now only the other tree collapses.
It seems to me, that the event (one tree node turning out to be a leaf) is - in that case unnecessarily - forwarded to all RowSets within the VO. So when the user clicks a node in one tree, updateValuesFromRows() is executed in both trees. How can I turn this off?
If this won't help, then I will try to come up with a test case.
One other thing: While I read the source of JUTreeDefaultMouseListener, I found the method setRulesActions(). According to the Javadocs, one can pass a String[][]. According to the sources the method looks like this:
  public void setRulesActions(String [][] rulesactions)
     hm.clear();
  }Strange, isn't it. Can't figure out how this is supposed to work.
Regards and thanks in advance,
Sascha Herrmann

Answer 1

The two trees closing out at random is due to 'different' binding-iterator instances used internally for the node-accessors. We'd be investigating this further!
We would update this post once we have a workaround/bug.