RE: Issue 71: Additional actors

	> Subject: RE: Issue 71: Additional actors
	> Date: Fri, 31 Aug 2001 17:21:24 -0700
	> From: "Henrik Frystyk Nielsen" <henrikn@microsoft.com>
	> To: "Mark Jones" <jones@research.att.com>, <xml-dist-app@w3.org>,
	>         <mnot@mnot.net>

	> Putting aside the question of why a sender would want to put something
	> in the message that must not be understood, it seems to me to be better
	> dealt with using encapsulation as this does not put us in a situation
	> where we have to redefine what "understand" means. 

Mark Nottingham captured pretty well the very common case that I had
in mind, where you have a block that is referenced by some other
block.  A module that employs such headers would generally be
designed to dispatch off of the 'thisDoesSomething' while simply
referencing the 'whatever' block.  By targeting 'whatever' at an
actor URI that is guaranteed not to match, the module doesn't have
to worry that the final destination may happen to dispatch (possibly
for some other purpose) on a 'whatever' block.

  <s:Envelope xmlns:s="http://www.w3.org/2001/06/soap-envelope">
   <s:Header>
    <a:whatever xmlns:a="http://www.example.org" id="foo"
     s:actor="http:/wherever/this/actor/will/never/match">
       ...
    </a:whatever>
    <b:thisDoesSomething xmlns:b="http://www.example.org">
      <blah ref="#foo"/>
    </b:thisDoesSomething>
   </s:Header>
   <s:Body>
     ...
   </s:Body>
  </s:Envelope>

I think this is a common enough case to warrant a standard
URI, rather than having lots of ad hoc URI's with the same
semantics.

Mark

Received on Monday, 3 September 2001 12:39:17 UTC