See org.hibernate.search.backend.lucene.document.impl.LuceneNestedObjectDocumentBuilder#contribute: the two fields are only used when indexing. Operations on nested documents, for example org.hibernate.search.backend.lucene.search.predicate.impl.LuceneNestedPredicateBuilder#doBuild, rely on the nested document path and ignore anything else: joins rely on the fact that the root document and nested documents are indexed in the same "block", so there's no need to Note we will still need to index the root ID somewhere as explained in HSEARCH-3834 Open . |