> Problem is: to offer such a thing with a straight face, we'd have to
> confine ourselves to an Oracle-subset version of SQL. For instance,
> lose the ability to distinguish empty-string from NULL.
I wasn't saying we write it - let Oracle do it :D
Chris
