Binary Runner wrote: > Hi, > > I've looked again at the LDAP RDF datasource and outliner and found some fact I'd >like you correct if they're wrong: > I'm ccing peterv and a couple of mozilla newsgroups on this reply, as I think this is of general interest. Hope that's ok.... > 1/ search uri (like ldap://...??sub) doesn't generate hierarchy,all the entries are >placed at the same level that's probably the reason why I was unable to display it as >tree > > > 2/ RDF datasource doesn't make the hierarchy from data acquired from the server > It's true that the RDF representation of a search URI does not generate any hierarchy. In the datasource, what's specifically being represented is a set of assertions about that search URI, not about the hierarchy of the server. That is, the statements being made are of the form "This LDAP search URI returns this directory entry". There's no particular hierarchy associated with that. > 3/ there is no way how to get whole DN of LDAP entry returned, > there is no way how to access objectClass hierarchy for > entry's objectClass, > You're right about the DN. Probably the thing to do is to make the datasource support a predicate (arc) for LDAP DN; I suspect this would need to live in the NC-rdf namespace, as it may be legal in LDAP to have an attribute named "dn". I was under the impression that objectClass was just another LDAP attribute, as far as searches are concerned. Is that not the case? >there is no way how to affect content generation based on matching RDF data against >some >pattern and firing rule only if pattern matches > This is what XUL templates are for, I believe. Can you say more about the specific sort of pattern matching that you want to do? > 4/ If I'm ok in these points it could be good to >implement container based on recursion level in LDAP tree > For the most part, it doesn't really matter whether recursion is represented by containers or arcs, I don't think. I suspect that just extending the datasource to support a special arc for DN (and, if necessary, one for objectClass) would be sufficient. Dan
