> Maybe the pragmatic approach is to introduce a new boolean, 
> "with", which is explicitly defined to nothing more or less 
> than an abbreviation of prox/unit=element/distance=0.  Then 
> server implementors can tackle "with" free of Proximity Fear.

I initially liked this (although would have called it "where" myself but
that might be SQL showing through).

On the other hand, putting in an alias for this, does worry me that we
could be opening the floodgates to and whole set of alii? Or perhaps a
whole aliasing mechanism should be a new feature of CQL 2.0 (with
appropriate descriptive support in ZeeRex)?

Matthew