- From: Savas Parastatidis <Savas.Parastatidis@newcastle.ac.uk>
- Date: Thu, 27 Jan 2005 10:05:33 -0000
- To: "Rich Salz" <rsalz@datapower.com>, "Francisco Curbera" <curbera@us.ibm.com>
- Cc: "Mark Baker" <distobj@acm.org>, <public-ws-addressing@w3.org>
Hi Rich,
>
> Because people want to only expose a single endpoint, and then
> have their infrastructure dispatch or route based on things like
message
> content. There several reasons for doing this; one of the most
compelling
> is that they do not want to expose *anything* about their internal
> deployment details. It avoids giving attackers any information.
I don't understand why RefParams are needed to do that. Examples...
<soap:Envelope>
<soap:Header>
<wsa:To>urn:bank:deposits</wsa:To>
</soap:Header>
<soap:Body>
<m:deposit-details>
<m:account>urn:bank:account:20</m:account>
<m:amount>1</m:amount>
<m:account>urn:bank:account:30</m:account>
<m:amount>2</m:amount>
<m:deposit-details>
</soap:Body>
</soap:Envelope>
The above message can be routed by the bank service based on its
content. The delivery addresses (the exposed endpoints) may be
http://www.bank.com, smtp:service@bank.com, tel:888-bankcom,
morse:b.a.n.k., pigeon:fly-to-bank, etc. The implementation can do the
dispatching based on the wsa:To URI or the contents of the body (the
bank account numbers). The EPR is unchanged in all cases:
<epr>
<To>urn:bank:deposits</To>
</epr>
Is this the situation you have in mind?
Regards,
.savas.
Received on Thursday, 27 January 2005 10:06:37 UTC