Reality check on interpreted XML process specs

[Bob Haugen <linkage@interaccess.com>]

Help me out here.  I'm probably missing something.

In ebXML we will have these XML business process specifications
that can be interpreted by runtime software.

Is this relatively new stuff for a B2B standard?

Do RosettaNet PIPs usually work this way?  Everything I've heard
is that software for each PIP is hard-coded, so the business 
service interface would more-or-less look up the PIP number
and dispatch to the code that handles that PIP.

A new PIP requires new code.  Correct?

For ebXML business process specifications, that should not
be required.  As long as you can produce the XML process
spec, when you design a new business process, the existing
ebXML-compliant BSI software should be able to interpret it.

Correct?

I know of proprietary code that works this way, but do any
deployed open standard B2B transaction models work this way?

(I know about BPML, and it does as far as I can tell, but
they're not deployed yet (and of course neither is ebXML).
Likewise Microsoft's XLang and process orchestrator.)

My point here is not to be competitive, although that's fun,
but to get a feel for the state of practice.  

Thanks,
Bob Haugen