W3C home > Mailing lists > Public > www-ws-desc@w3.org > January 2007

RE: Operation dispatch when there isn't a SOAP body.

From: Jonathan Marsh <jonathan@wso2.com>
Date: Thu, 4 Jan 2007 11:26:09 -0800
To: <paul.downey@bt.com>, <ryman@ca.ibm.com>
Cc: <www-ws-desc@w3.org>
Message-ID: <003801c73036$31d2e320$3301a8c0@DELLICIOUS>

I'd say yes, we don't want to constrain the ways someone might dispatch.
But we did give some hints in the primer as to common strategies for
operation dispatch.  My proposal is to modestly increase the hints to cover
the important (IMO) case when there is no SOAP envelope or even XML payload
- namely when either the HTTP binding or the SOAP binding is bound to GET.

I want to support REST to the extent of at least noting that sufficiently
unique information in the {http location} facilitates dispatch, rather than
leave the impression that folks simply should avoid GETs altogether.

Jonathan Marsh - http://www.wso2.com - http://auburnmarshes.spaces.live.com
 

> -----Original Message-----
> From: paul.downey@bt.com [mailto:paul.downey@bt.com]
> Sent: Thursday, January 04, 2007 2:58 AM
> To: ryman@ca.ibm.com; jonathan@wso2.com
> Cc: www-ws-desc@w3.org; www-ws-desc-request@w3.org
> Subject: RE: Operation dispatch when there isn't a SOAP body.
> 
> 
> 
> > I am OK with improvements to the Primer.
> 
> me too.
> 
> > The issue of dispatch is somewhat artificial. It was caused by the
> > non-RESTful practice of using a single endpoint for all Web service
> > requests and then relying on the SOAP engine to dispatch the request to
> > the right implementation object. For example, the early Apache SOAP
> > implementation called this endpoint the "router".
> 
> > REST is based on the proper use of well-designed URLs. Web servers,
> > including Java servlet containers, have the ability to map URLs to code,
> > e.g. a servlet can handle a URL pattern.
> 
> > Your proposal for the definition of a default HTTP location for
> > interfaceless bindings is interesting. However, I wonder how useful an
> > interfaceless binding is for REST. I think toolkits would probably
> > implement some strategy for generating HTTP locations based on the input
> > arguments of methods.
> 
> I'm not sure I fully understand the proposal, but I'm reminded
> of the dispatching discussion we had for SOAP, where we were cautious
> to specify GEDs given wsa:Action or indeed other mechanisms might be
> employed, I recall we just require there to be some 'uniqueness' ..
> 
> Given Web frameworks use a wide variety of dispatching methods,
> often as an ordered sequence of regex expressions,
> are we obliged to be similarly vague for the HTTP binding?
> 
> http://httpd.apache.org/docs/2.0/misc/rewriteguide.html
> http://www.djangoproject.com/documentation/url_dispatch/
> 
> Paul
Received on Thursday, 4 January 2007 19:26:13 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 7 December 2009 10:58:45 GMT