W3C home > Mailing lists > Public > public-ws-addressing@w3.org > December 2004

Re: i022: Relationship to the SOAP Binding Framework: a SOAP Request MEP and OneWay HTTP Binding

From: Mark Baker <distobj@acm.org>
Date: Thu, 23 Dec 2004 11:10:48 -0500
To: David Orchard <dorchard@bea.com>
Cc: public-ws-addressing@w3.org
Message-ID: <20041223161048.GQ6495@markbaker.ca>

Nice work, David ...

FWIW, I've juggled the order of these in order to respond...

On Fri, Dec 17, 2004 at 02:06:04PM -0800, David Orchard wrote:
> - Relationship to media type.  I think this be covered by the soap
> media-type, but I'm not 100% sure.  I'm not sure about the case of
> whether a soap+xml is good enough for this mep+binding, I sure hope so
> though.

Seems fine to me.

> - Webmethod support: I said POST only

Ok, I guess.  But it doesn't seem a big deal to me to support other
methods by reusing the WebMethod feature.

> - Streaming: I consistently said that requesting SOAP nodes must avoid
> deadlock by accepting binding-specific response messages

Yeah! 8-)

> - I removed "receiving" state from the next state tables.

Why was that?

> - The identification of the mep in use can't be gleaned from the
> information in the binding, unlike the SOAP HTTP Binding

How exactly do you expect responses using this MEP & binding to differ
semantically from responses using, say, the SOAP 1.2 default HTTP
binding & req/resp MEP?  What I mean is, that I assume that if I want to
submit an XQuery document and receive the results of that query in the
response, that I shouldn't use this MEP & binding, right?  But a
response should be returned, since this is HTTP.  So of all the possible
semantics that an HTTP response message can have, what subset of those
are permitted by this binding?  Right now 200 is returned which doesn't
tell me much, but I wonder if you don't want 202?  Also, is there an
expectation on behalf of the client that a subset is in play when it
sends a message?

> - SOAP faults cannot come back over the http response.  For
> request-response bound to 2 http requests, life sucks.
...
> - I kept the HTTP status code at 200

Is the latter a direct consequence of the former?

But I'm wondering if this is really what you want to do.  If response
semantics are limited to ack like semantics (202), then what's wrong
with using SOAP faults as nacks?

> - There is an optional binding specific response in the one-way MEP.

How does this relate to the forced 200?  I notice that the recipient has
to handle non-200 response codes anyhow.  I reckon this is just the same
issue above about nacks & faults.

> - The binding can allow an empty body, especially for cases where the
> action is sufficient.

Ok.

> I would also like to mention that I found this exercise very
> informative.  I think that SOAP has provided an excellent framework for
> creating interoperable meps and bindings as it forced me to think about
> many hard issues.

I personally think it's overkill, but some guidance is certainly better
than none.

Cheers,

Mark.
-- 
Mark Baker.   Ottawa, Ontario, CANADA.        http://www.markbaker.ca
Received on Thursday, 23 December 2004 16:11:11 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 2 June 2009 18:35:00 GMT