Mike:

XML would basically be the syntax. You can still write OOP code in XML, and if you think about it, XML might even be better suited for marking up complex class design with metadata than traditional C/Basic style syntax that was useful when writing procedural code.

As a conceptual exercise, I agree. But are you going to write your own OOP language using XML syntax? It would take quite a lot of time before you were ready to do real work with it. Or is there an XML syntax OOP development language out there that I don't know about?
Mike,

If the essential question is "Can it be done"? Then the anwer is "of course". Oh, it has lots of flexibility -- you could create extensive metadata -- hell, drive everyting with data. And I can immediately think of some wonderful advantages like,

• Absolute platform independence (finally!) -- in fact I can see applications so divorced from the platform that they never know and do not care what box they are being run on -- its all mediated;
• Instant documentation -- comments in "code" could become documentation -- in fact, I think the distinction between documentation and code would essentially disappear -- the code would be self documenting -- or the documentation would in fact be the code, however one wants to look at it;
• Very flexible and adaptable data definitions -- run-time adjustable data structures -- or maybe no pre-defined structures at all, let the application define its own structures based on how the user uses it -- all from meta-data. A very exciting concept.

Currently we create UML diagrams and find someway to generate code from them. If we wrote the diagrams in XML, and gradually added enough information and business logic to the diagrams so that the documentation could become the app, I think that better applications could be built faster.

But who's going to write a framework for all this? I don't have either the time or resources. Do you? I think creating an object that would respond to events out of what is still essentially a markup language would be a huge undertaking. In fact, I cannot see immediately how it could be done at all given the current limitations of XML. Of course I may be overlooking something obvious.

So, I think it boils down to, you say its theoretically possible. I agree, but say it is not practical. You say "not yet, but it will be -- maybe". I agree with that too. If someone decides to develop the necessary framework. But in the meantime, although you can enhance and expand a lot of what we do with XML, it is not a panacea for all current ills.

Regards,