The approach taken here is somewhat slow as the server data gets fetched
into objects first which are then converted to hashes again. Since a
server search will usually deliver the result as a hash the intermediate
object conversion is inefficient.
But as the object classes are able to treat the attributes returned from
the server with custom parsing, this is currently the preferred
method. Especially for large result sets it would be better if this
method would call a static object class function that operate on the
result array returned from the server without using objects.