>>Point taken. I know about this, and I oversimplified my answer, pointing at the current VFP implementation.

You mean, CURSORTOXML() etc? Well that's possible.

It's only recently that Wintel/MS-tool-oriented folks have had easy access to what I would consider to be robust processing of XML.

But even in the DOM implementation, MS has come a long way in recent versions to improve ability to handle large data sets. So there's hope <g>.

Don't sell XML short. In some ways, Martin, it really is the silver bullet. Certainly there are scenarios in which it does not belong. (like anything else!)

Also, people invest the schemas commonly available with some boundaries of "what XML can do", and this is needless. Standards are good and it's great that there are known schemas that are well understood.

But sometimes for large datasets you shouldn't be thinking in terms of the standard schemas, only in terms of the most efficent representation. You can carry some metadata in a separate (or in line) XSD that describes the use of the data elements much more efficiently.

This is (one place) where XSLT comes in. One of the great strengths of XML is in moving data between unlike systems, and often that means between unlike schemas. This is perfectly fine, it is not a reason to despair <s>.

Martin thanks for welcoming me here but, as I've told other people, I don't know how much time I'm going to be able to spend on UT. IAC, look: if people are in doubt about this, I'd like to help.

>L<