Protocol Versioning [was: Re: I-D ACTION:draft-ietf-sacred-protocol-bss-01.txt]

Stephen,

None of the PDUs described in the current ID contain version information.
I suggest that this may be needed to support future version of SACRED.

I was also wondering if it might be worth placing these attributes and any
other information common to all message types in abstract base types
similar to the RequestAbstractType and ResultAbstract type of XKMS as this
may make the schema simpler.

For example,

  <complexType name="RequestAbstractType" abstract="true">
    <sequence>
      <element ref="sacred:ProcessInfo" minOccurs="0"/>
    </sequence>
    <attribute name="MajorVersion" type="nonNegativeInteger" use="required"
fixed="1"/>
    <attribute name="MinorVersion" type="nonNegativeInteger" use="required"
fixed="0"/>
  </complexType>

   <complexType name="ResponseAbstractType" abstract="true">
    <sequence>
       <element name="Status" type="string" minOccurs="0"/>
    </sequence>
    <attribute name="MajorVersion" type="nonNegativeInteger" use="required"
fixed="1"/>
    <attribute name="MinorVersion" type="nonNegativeInteger" use="required"
fixed="0"/>
  </complexType>


There are a couple of other issues related to schema.

I noticed that you added XML complexity as a major open issue.  Could you
expand on this.

Do we really need to maintain a DTD and a schema? Since it is possible to
generate a DTD from a schema then wouldn't a schema be sufficient? It seems
as is not just unnecessary but also potentially harmful due to possible
discrepancies.