I think it make sense and we should change it as you suggested. We already faced a similar situation with Lucene and Neo4j