This looks looks fine to me. I have one remark, that I mentioned earlier but I do not know to what extent that was agreement or not. I wonder whether it wouldn't be better to leave out the result-set out of the query completely and use this as a separate parameter in the request because the resultset is completely different from indexes. When the resultset is being used in a query, the server has to reconstruct the complete query anyway. It gives the false idee that you can do the same things with a resultset as you can with the other indexes. Especially in distributed searching it would be nicer when queries to different servers can be identical and do not contain elements that are so server specific. Theo >>> [log in to unmask] 29-05-02 20:20 >>> I've (re-)written bnf for CQL based on recent discussion. Not that the discussion has been all that useful in trying to figure out what people really want but it's time to focus on stabalizing this. I've put it up at http://www.loc.gov/z3950/agency/zing/cql.html This is based on Alan's proposal as well as Ralph's earlier work, and the recent discussion. Officially, this is the current state of CQL, so if there are specifics of this draft that you don't like, I'll change it, if you send concrete suggestions. It's time to force the issue and find out how close we are to some consensus. --Ray