: The dependency graph looks like: : common > solrj > solr > servlet : (while EmbeddedSolrServer depend on solr) : : calling common+solrj "common" is a little strange because it would not be : common to anything anymore: it is the client.
it would be common to clients and servers ... if you call it 'client' you get the same confusion people have right now with the current structure: they don't understand why the solrj client is included in the war ... admitedly the fundemental issue there is that people don't neccessarily realize that Solr servers need to be clients of other Solr servers in order for distributed searching to work, but that's why a name like "client" seems poor. My take on it... "common" code is common to any and all Solr related projects. no matter what you are doing with Solr, if it's in java you need the "common" code. "server" code is what you need if you are running a solr server, either as a webapp, or embeeded in another app. ...but i freely admit that's a subjective opinion. what do other people (besides me and ryan) think? : "core" is bad -- the name is already overloaded, and typically "core" would be : the center, not an edge of the dependency graph. mmmmm... agreed, that is a horrible implication of hte name core. -Hoss
