> What would be wrong with having a database index in CQL for doing this
> hasn't this been discussed before?)
I'm all for it.
It's been discussed in the ZIG, but never caught on, because Z39.50 allows
multiple databases. But at one point we talked about getting rid of that
feature, as logically it's an artificial construct, just another search
criteria, that could be just as easily treated as an access point. The ZIG
never bought in to this argument and thus the multiple database complexities
remain. I think a database index solves this very neatly.