> > >  >dc1="core-v1"  dc1.title=first and >dc1="core-v1a" dc1.title=second
> > > and dc1.title=third
> > Lets remove the mappings and put in the parens:
> > ((dc1.title=first and dc1.title=second) and dc1.title=third)
> > And put back the mappings:
> > >dc1=v1 ((dc1.title=first and >dc1=v1a dc1.title=second) and
> >           dc1.title=third)

> Oooh, no, you can't do it like that.  The prefix mappings _change_ the
> bindings.

Yeuch.   I've always thought of prefixes as the least important part of
CQL, only there as a (valuable) afterthought to reduce the dependance on
Explain.

> > > The third occurence of dc1 -- is it governed by the first or second
> > > assignment?
> > First, due to the left to right ordering of parens.
> Nope, the last, because it's closest to the operand.
> Have you actually tried this in your own parser?  I bet it does the


Mine produces the correct output for my version of the semantics. ;P

Which admittedly make adding queries together (the only advantage of
prefixes anywhere, to date) not work correctly:
Adding
  >dc=foo a
and
  >dc=bar b or c
becomes:
  >dc=foo (a and >dc=bar b) or c

Which is obviously not what was intended.

Can we put brackets around contexts to limit their scope, I assume?
  (>dc=dc1 a and b) or c

Sigh, back to the drawing board.

Rob

--
      ,'/:.          Dr Robert Sanderson ([log in to unmask])
    ,'-/::::.        http://www.o-r-g.org/~azaroth/
  ,'--/::(@)::.      Special Collections and Archives, extension 3142
,'---/::::::::::.    Nebmedes:  telnet: nebmedes.o-r-g.org 7777
____/:::::::::::::.                WWW: http://nebmedes.o-r-g.org:8000/
I L L U M I N A T I