Re: Providing a short name for single-request-response MEP

Mark Baker writes:

>>Well, HTTP won't have a problem with a 3-month duration
>>transaction  if  the service  uses  the 202  (Accepted)
>>response code.
>>
>>And that brings up an issue with the "request-response"
>>name; when using the default binding, the response can
>>actually be a response to the "acceptance" of the
>>message.  I don't think that means we need a new name,
>>but I do think that we should point out that a
>>"response" isn't always the result of processing.

The SOAP MEP makes very clear that the SOAP response is
just that, the response to SOAP processing.  I believe
that our HTTP binding only handles the case where the
response is returned reasonably promptly, on the
still-open connection. 

If we had a separate MEP called "long running
req/resp", then we could implement that via a different
or enhanced HTTP binding (or email, or various business
message queuing systems, etc.).  We could use 202 as
you suggest for http, but would also have to specify
how the eventual real response would get back 3 months
later.

------------------------------------------------------------------
Noah Mendelsohn                              Voice: 1-617-693-4036
IBM Corporation                                Fax: 1-617-693-8676
One Rogers Street
Cambridge, MA 02142
------------------------------------------------------------------

Received on Tuesday, 23 April 2002 21:56:33 UTC