8221RE: [soapbuilders] Re: SOAP 1.2 GET binding
- Jul 24, 2002Graham,
I don't believe the GET MEP is thought of as experimental. It is
necessary if we're binding SOAP to HTTP and want to allow
web-friendly SOAP web services. Of course, if the reality shows
that web-friendly SOAP web services are impractical or if just
nobody creates them, then of course the next version of SOAP may
bring us a standard binding which won't have the HTTP
(application protocol) issues about it and which will be simple
and usable for the (then) current way of creating web services.
Such binding would probably be a TCP binding. HTTP binding may
then become unnecessary and even dropped from SOAP altogether, as
SOAP and HTTP are not a really great match.
This evolution can be helped greatly if somebody (soapbuilders?
any big player?) comes up with a TCP binding (MS has DIME, but
it's not really a protocol, it's just encapsulation, a protocol
would have to be specified on top of it).
Senior Architect, Systinet Corporation
On Thu, 4 Jul 2002, graham glass wrote:
> one associated question is: given the relatively experimental nature of MEP,
> should it really
> be in the SOAP 1.2 specification at all? perhaps it would be better to have
> a separate
> document that contains descriptions of promising experimental features that
> if additional
> thought prove compelling, move into the official specification. my fear is
> that things like MEP
> sound cool to begin with, but end up introducing an avalanche of
> unanticipated complexity.
> we already went through enough of that pain with the 1.1 specification, and
> many features
> such as sparse arrays ended up getting swept under the carpet.
> -----Original Message-----
> From: Jacek Kopecky [mailto:jacek@...]
> Sent: Thursday, July 04, 2002 5:39 PM
> To: firstname.lastname@example.org
> Subject: Re: [soapbuilders] Re: SOAP 1.2 GET binding
> Hi all, 8-)
> I think I agree with Sanjiva that the use of SOAP together with
> WSDL seems to diminish the usefulness of the SOAP Response MEP
> over HTTP GET.
> Since I believe the SOAP HTTP binding is not going away (even
> though many think it's suboptimal at best) the WSDL group will
> have to think hard on how to make SOAP Response MEP over HTTP GET
> describable in WSDL and how to describe RESTish applications in
> WSDL so that they are recognized as such in order to be able to
> use HTTP effectively.
> For soapbuilders interop testing, unless we see a great number
> of implementations with converging views of how the new MEP over
> GET is to be implemented, we should concentrate on what we
> already do - headers, faults, attachments.
> Best regards
> Jacek Kopecky
> Senior Architect, Systinet Corporation
> P.S: I'm off for a week's vacation starting 7/5.
> On Thu, 4 Jul 2002, Sanjiva Weerawarana wrote:
> > "cdthoughtpost" <chris@...> writes:
> > > Where does the WSDL HTTP GET binding come into play here? Should
> > > that be used as the basis for support of this feature? If so, then
> > > WSDL seems to assume either a replacement scheme or encoding as the
> > > query string, defined in the GET binding.
> > The WSDL 1.1 HTTP GET binding is a different beast from the new
> > SOAP MEP. WSDL 1.2 will need to have an additional binding for
> > request-response operations to support the new MEP: HTTP GET for
> > the request and SOAP envelope in the response of that GET. That
> > binding would have to define a way to generate the URL from the
> > message parts - probably in a manner similar to the current GET
> > binding does.
> > The funny thing to me is this: (if and) when WSDL has a binding
> > for the new MEP and (if and) when toolkits support that binding,
> > the use of the new MEP will be indistinguishable from the use
> > of the "old" MEP of the request too being in SOAP via HTTP POST.
> > That is, a stub generated from these WSDLs would behave exactly the
> > same way! Thus, the use of GET would become an implementation detail
> > that users would not care about, just like they don't care about
> > POST being used right now (or in fact that HTTP is being used).
> > This seems to defeat the main objective of this new MEP: to allow
> > application developers (i.e., people who call the stub) to recognize
> > that an operation is a "safe" operation and hence can be "linked"
> > to by that application. The app developer (in most cases) probably
> > didn't even read the WSDL and hence probably never knew that GET
> > was used instead of POST!
> > Sanjiva.
> Yahoo! Groups Sponsor
> This group is a forum for builders of SOAP implementations to discuss
> implementation and interoperability issues. Please stay on-topic.
> To unsubscribe from this group, send an email to:
> Your use of Yahoo! Groups is subject to the Yahoo! Terms of Service.
- << Previous post in topic