I am currently trying to build a Spring Boot autoconfiguration starter for Apache Solr and trying to decide what is the correct behaviour (convention over user friendliness). Essentially it seems to be two thought patterns in Spring Data for depedencies. 1.