Hi Steve!

> This defect is in the shipping Axis1.1, yes?

Ayah. :(

I'll fix it the less architecturally-different way (by moving the handler back where it was) ASAP.

> In which case we need to notify Axis-user and soapbuilders 
> that we are 
> being noncompliant with the standard. Interesting that there 
> is no test 
> for it in the interop tests

Well, it's kind of a subtle thing to test.  From the outside, we're perfectly compliant in that we send MU faults when we don't understand MU headers.  The issue is that we do that *after* the service work has been completed, which is the no-no.  In cases where no back-end state is affected by the service invocation (i.e. all the interop tests), it doesn't actually matter.

We should also really be making sure that we have a well-defined way to do service-level rollback if a fault should be produced by a response-chain Handler.  Many services won't want this, but we should be able to do it for those that do.

--Glen