Referenceable Nodes
Overview
The JCR 2.0 specification defines referenceable nodes. Those carry a jcr:uuid property which uniquely identifies a node within a repository workspace. FileVault exports contain this (protected) property jcr:uuid (when present) which is important to keep referential integrity.
The import behavior depends on the used FileVault version.
Import behavior prior FileVault 3.5.2
The ids of referenceable nodes are only kept during import when a node with the same name does not yet exist in the repository. For existing nodes the ids are never updated with the value from the package. They either get a new id or keep their old one (in case the old node was already a referenceable node). In case of conflicts with nodes which are not siblings (i.e. don't share their direct parent node) the newly imported node gets a new id. In case of conflicts among siblings the existing conflicting node is removed but all its references are kept (i.e. they point to a different path after the import) when the existing conflicting node is contained in the filter rules, if the conflicting node is outside the filter rules the to-be imported node is skipped (and import continues with its children as if they were below the existing one).
Import behavior since FileVault 3.5.2
Since version 3.5.2 (JCRVLT-551) FileVault tries to use the jcr:uuid of the node in the package even if a same named node does already exist. In case this cannot be achieved while keeping referential integrity of old and new nodes an exception is thrown. Existing nodes with conflicting identifiers or reference properties towards conflicting identifiers are removed in advance if they are contained in the filter rules to reduce the conflict potential.
Id Handling Policies
The import behavior of packages with can be tweaked with ImportOptions.setIdConflictPolicy(...) per package or the default can be overridden in the OSGi Configuration. For further details refer to the JavaDoc. The old behavior can be achieved with IdConflictPolicy.LEGACY.
Note that CREATE_NEW_ID will enforce creation of new identifiers, even when no node with a conflicting id exists.