On Thu, Dec 18, 2003 at 11:12:49AM +0000, Mike Taylor wrote: > > I would have said yes - a CQL version defines the context set > > version. > > I am uncomfortable about this special-casing, and I don't see how it's > necessary. At the moment this is just _one_ special case defined in > CQL, and that is that unqualified relations and modifiers are taken > from the set whose prefix is "cql" rather than from the currrently > prevailing set used for indexes. Don't forget about result set references too (cannot remember syntax, but it was something like cql.resultsetid=...). This has special semantics tied to CQL (I would have thought). Alan ps: Merry Christmas everyone - I'm off on leave for a while.