Er, I agree, that's why I suggested that processing the form of XML export we have now (without waiting for fancy XSLT transformation scripts) is better idea than attacking CSV.

Allow, however, this 20+ years IT veteran a bit of cynicism: XML is surrounded with too much hype and too little understanding, like any new[1] information technology. It's not the silver bullet. It enables global, transparent B2B applications? What was wrong with using EDI for that?

There are too many competing standards for ancillary functions. Trivial applications are hawked as and-all solutions. There is far too much of reinventing the wheel. But, I guess that cannot be helped and happens elsewhere (for example, it is amusing to see 'Agile programming' reinvent concepts of rapid or prototype-based development of decade ago )

Don't get me wrong: global acceptance of XML does hold a lot of potential. I live in good part from it. It's just that it is not Second Coming.

[1] If we can call decades old technology 'new'. XML is, after all, just simplified SGML. Granted, other techniques that 'surround it' (for storing it, querying, transforming, using it as middleware substrate...) make the difference. SGML's primary raison d'etre was markup.