That's pretty much what I had thought when I read the XLL spec. Personally,
in evaluating the 80/20 mix for a query language, I would think that boolean
operators, boolean functions, and precedence would be pretty important.
Another significant limitation of XPointers as a query language is that each
term specifies *one* location, if I understand the spec correctly. It
doesn't seem to be set up to allow result sets, e.g. the set of patient
records that satisfy a particular requirement, the set of catalog entries
that specify a particular requirement, etc. I would think that result sets
are pretty important for query languages.
I really like the simplicity, readability, and design cohesiveness of XLL,
and I do think that the functionality it contains should be present in a
query language for SGML/XML documents. It is not clear to me whether there
is a good, orthogonal way to add in some of this other functionality to XLL;
if so, XLL could be used as the basis for a query language. Using the same
primitives would be nice, since anybody working with XML is going to have to
learn XLL, and we don't want every poor schmo to have to learn 50 different
ways to do a query.
Jonathan
________________________________
Jonathan Robie
Email: jonathan@texcel.no
Texcel Research, Inc. ("http://www.texcel.no")