W3C home > Mailing lists > Public > www-ws-desc@w3.org > October 2003

Re: PROPOSAL: Drop interface/operation/(input|output)/@headers

From: Glen Daniels <gdaniels@sonicsoftware.com>
Date: Thu, 30 Oct 2003 00:31:15 -0500
To: Jeffrey Schlimmer <jeffsch@windows.microsoft.com>, Roberto Chinnici <Roberto.Chinnici@Sun.COM>, Sanjiva Weerawarana <sanjiva@watson.ibm.com>
Cc: www-ws-desc@w3.org
Message-id: <00ea01c39ea7$0a06f120$7b00a8c0@AURORA>

Hi Jeff!

I think that the question Dave Orchard just asked bears strongly on the
answer to this with respect to role and relay.  Until we have some
reasonable model with which to deal with intermediaries, I'm not sure how
much we can really talk about them - I agree with Dave that we should think
about it.  That said, however, the extension I'm imagining really just fills
a bag with data for anyone to read - so the header in question probably
should be targeted at the "none" actor.

With respect to mustUnderstand, I'm not sure it makes sense to mark the
individual "cookions" as MU.  For example, one of the ones people seem to
talk about a lot is a "dispatch key" - some value that you put in a header
to allow an endpoint to uniquely target the correct piece of handler code,
even in the face of multiple operations sharing the same body data.  Why
would you mark such a thing MU?  You use MU="true" to indicate extensions
which the *sender* requires the *receiver* to understand - in this case,
it's the receiver (the server) who needs the data, and they're going to
throw a fault if it's missing, but they don't care if it's marked MU or not.

Do you have other examples of this kind of data that you feel needs the MU


----- Original Message ----- 
From: "Jeffrey Schlimmer" <jeffsch@windows.microsoft.com>
To: "Glen Daniels" <gdaniels@sonicsoftware.com>; "Roberto Chinnici"
<Roberto.Chinnici@Sun.COM>; "Sanjiva Weerawarana" <sanjiva@watson.ibm.com>
Cc: <www-ws-desc@w3.org>
Sent: Tuesday, October 28, 2003 8:47 PM
Subject: RE: PROPOSAL: Drop interface/operation/(input|output)/@headers

> So I think it's a pretty simple matter to define a "sideband data"
> module which simply takes a property consisting of a set of elements,
> inserts them as SOAP headers.

Do you see any value in allowing the WSDL to specify actor/role,
mustUnderstand, and/or relay for such 'cookie-esque' header blocks?

Received on Thursday, 30 October 2003 00:30:58 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:54:45 UTC