This applies only when using a JDOM tree as the source.
When duplicate nodes returned by a path expression need
to be eliminated (or when nodes need to be sorted into
document order), this is not being done correctly in
the case where the node-set to be sorted contains one
node that is an ancestor of another. The effect is that
the node-set contains duplicates and is not correctly
Applies to 7.4 and (untested) to 6.5.2.
Test case: axes100, when run in JDOM mode
Source cleared in both code branches.