Hello, I give my question a second try and give some more details. I did not get an answer on my first try, perhaps because there is not so much experience on Jackrabbit 2.0 around.
We have a similar structure as described in http://n4.nabble.com/Multiple-jcr-deref-in-query-td515143.html#a515143 , like: Root | |+Content | |+DataNodeA | | |@Content | |+DataNodeB | | |@Content | | | |+DocumentNodeA | |+LinkToDataNodeA > my:DataLink | | |@RefToData = DataNodeA | |+LinkToDataNodeB | |@RefToData = DataNodeB | |+Site |+PageA | |+LinkToDocumentNodeA > my:LinkToDoc | | |@RefToDoc = DocumentNodeA | |+PageA1 | |+LinkToDataNodeB > my:DataLink | | |@RefToData = DataNodeB |+PageB |+LinkToDocumentNodeA > my:LinkToDoc |@RefToDoc = DocumentNodeA Now there can be hundreds of page nodes (in a hierarchy) under the site node and depending on the page template, they could use different type names holding the references to the data or document nodes in the content subtree. Now I would like to make fulltextsearch queries on a site (or even more limited scope, like: PageA and its sub-pages) and would like to have as result all pages, where the term is used either in a property of the page node or the nodes referenced by the page. I would already like to use Jackrabbit 2.0 and do the query with either SQL-2 or QOM. Now to solve that query, do I have to make as many joins as there are different property-names, which can have a reference or is there a better - more simple and performant - way to do it (perhaps by even forking the Jackrabbit classes)? Did someone already have such a requirement and how did you solve it ? Regards, Benjamin
