Follow-ups to today's call: 3 (SOAP MEPs vs. WSDL MEPs)

The issue mentioned today about separating SOAP MEPs from WSDL MEPs came
up fairly early in the discussions of the async task force (spun off
from WSA and WSDL).  There was some suggestion at first of merging the
two, so that (say) a WSDL in-out always mapped to a SOAP
request-response.  I'm not sure we ever explicitly rejected this
approach, but I believe it's fair to say the discussion moved away from
it fairly early on.

Can someone outline to me the current thinking in the WSDL group about
describing /SOAP/ MEPs for operations, as opposed to just saying (e.g.)
"in-out" and specifying a SOAP binding?