Ray Denenberg, Library of Congress wrote:

> Sorry for my loss of memory here. In Mike's (excellent) CQL tutorial
> http://zing.z3950.org/cql/intro.html there is the example:
>
> >dc="http://www.loc.gov/srw/index-sets/dc"
>    dc.title=dinosaur and dc.author=farlow
> showing how to associate a url with an index prefix.
>
> And I actually do (vaguely) recall that we agreed to this, after a
> very long discussion and one of the hottest debates we've had  (so
> long and hot that I can't validate agreement from the archive).
>
> But if we did agree, it's not in the bnf.
>
> Did we agree and if so why was it never added to the bnf?
>
> --Ray
>

IIRC Ray suggested for indexes in the beginning of a query. Others such
including myself wouldn't mind having indexes everywhere. This is the
grammar we
use (and is very close to the one I suggested on the ZNG list, 13 Nov 2002):

search-clause ::= "(" cql-query ")" | [index-name relation] term
                          | ">" [identifier "="] term cql-query

Examples of equivalent CQLs:

   > p a  ==  >p (a)  ==  (>p a)
   > p a and b    ==    > p (a and b)
   > p1 > p2 a and b   ==   >p1 ( >p2 (a and b))
   > p1 a and >p2 b   ==  >p1 (a and >p2 b)

-- Adam

>