New Draft

All,
 
    Please find attached a ZIP of the revised spec together with a new
schema.
 
    You will find that the request/response and message processing section
is very much more comprehensive than previously. In particular the handling
of pending requests is much improved and there is a new 2 phase protocol for
preventing request replay attacks. These mechanisms are all designed to
allow embedding at the SOAP layer at a future date.
 
    The issue of QNames is somewhat fuzzy, I suggest that we have a concall
to discuss the issue. In particular how extensible is the QNames mechanism?
In appears to me that if we enumerate the QName types we end up in the same
situation as if we had used string since the QName enumeration is not
extensible. 
 
    I have as a pro-tem measure specified the QName type and an enumeration
that could serve if we wanted strong typing. We can choose which to use at
the F2F.
 
    The main outstanding issue is the matter of handling the hash of the
request. It seems to me that some sort of attribute on the response
signature is appropriate.
 
    The examples are starting to show their age, we could do to generate
some new ones from the new schema.
 
    I have removed more of the Word formatting and I am working on an XSL
script to generate the TOC.
 
    Irving and Steve will note that even more of Xbulk has been sucked into
XKMS in this iteration
 
        Phill