- From: Henry Lowe <hlowe@omg.org>
- Date: Wed, 22 Nov 2000 11:01:11 -0500
- To: Mark Nottingham <mnot@akamai.com>
- Cc: Henrik Frystyk Nielsen <frystyk@microsoft.com>, rden@loc.gov, XML Distributed Applications List <xml-dist-app@w3.org>
Might I suggest that for interoperability with ebXML and operation with object oriented systems (e.g., DCOM), both the requestor and destination be specified by URIs and they be carried in the header (in case inter- mediaries are involved, they will know where to find the target destination). I don't believe a namespace (if I understand what Henrik is saying) is suffiecient to invoke a DCOM object (but I admit it's been years since I read how DCOM invocations work, so I could well be wrong). In any event, there should be sufficient information for an object invocation regardless of the transport protocol XP is mapped to. In ebXML, we looked at this question and a URI does nicely. Best regards, Henry ------------------------------------------------ At 02:44 PM 11/21/2000 -0800, Mark Nottingham wrote: > >Hmm. Perhaps then it should be a requirement for the protocol binding to >provide a way to determine the request-URI; if there isn't one inherent, it >must be transferred in the envelope. Modules which need greater confidence >in the URI should specify that it be transferred in the envelope in a >pre-arranged fashion (ideally, once for all modules, not once per module). > > >On Tue, Nov 21, 2000 at 02:33:56PM -0800, Henrik Frystyk Nielsen wrote: >> Ah, I see, you are talking about the request-URI so to speak - I thought >> you meant "service" as in "module". >> >> It does make a lot of sense to have a header that knows about the XP (or >> SOAP) message path model and can indicate under what conditions to send >> the message to whom etc. However, I don't think it should be required as >> several protocol bindings (most of the *transfer* protocols like HTTP, >> SMTP etc.) have mechanisms for indicating where to send the message. >> >> Henrik >> >> > Confused - the service URI will most likely *not* be the XMLNS of the >> > message/modules... >> > >-- >Mark Nottingham, Research Scientist >Akamai Technologies (San Mateo, CA) >
Received on Wednesday, 22 November 2000 10:43:18 UTC