RE: Generic proposal for enganging MTOM in WSDL 2.0

To capture Asir's idea expressed at the end of the last telcon, here's an
updated proposal to allow Canon to indicate MTOM engagement using standard
(or pre-standard) technologies.

Original Proposal (from Paul):

WSDL 2.0 provides an extension attribute which MAY be used to indicate a
WS-Policy attachment is "vanilla", only contains a single set of assertions
and does not contain any compositors, e.g.:

  <Policy wsdli:simpleAssertions="true">
    <wsoma:OptimizedMimeSerialization /> 
    <wsa:UsingAddressing />
  </Policy>

A non-Policy aware processor may process the wrapped assertions as just
another hop in their XPaths, and yet continue to interoperate with
processors which support the whole of WS-Policy.

Updated Proposal (based on Asir's idea):

An implementation may support a subset of Policy that consists of the policy
wrapper appearing as a child of <wsdl:binding> or <wsdl:endpoint> elements,
and containing only the OptimizedMimeSerialization assertion [1] (no
compositors allowed), or indeed any list of assertions supported by the
implementation.  The Policy element can be marked as wsdl:required="true".

  <Policy>
    <wsoma:OptimizedMimeSerialization /> 
  </Policy>

Advantages:
1) No new syntax needed.
2) No changes to specs at all!
3) Profile can be processed by a fully conformant WS-Policy processor.
4) Obviates need to describe the interaction of wsdli:simpleAssertions and
wsdl:required.
5) Can be defined for multiple versions of Policy.
6) Supports any assertions supported by the implementation (just like
WS-Policy.)  This includes a possible future version of an MTOM assertion.
7) Isn't open to unintended side-effects (what evil uses could
wsdli:simpleAssertions be put to?)

Disadvantages;
1) Doesn't provide any standards status to such a profile.
2) Doesn't provide any machine-readable indication that the profile is in
use.

[1]
http://schemas.xmlsoap.org/ws/2004/09/policy/optimizedmimeserialization/opti
mizedmimeserialization-policy.pdf

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, November 30, 2006 3:48 AM
> To: jonathan@wso2.com; ryman@ca.ibm.com
> Cc: www-ws-desc@w3.org
> Subject: RE: Generic proposal for enganging MTOM in WSDL 2.0
> 
> Hi Jonathan,
> 
> > The risk I saw is that you can't take an off-the-shelf WSDL with
> embedded
> > policy and simply add the new annotation and expect it to work with
> Canon's
> > implementation.  You'd have to do a fairly severe restructuring, losing
> the
> > maintainability provided by the indirection.
> 
> I don't think that's the user story ..
> 
> > If that's not the goal, and one would instead be taking an off-the-shelf
> > WSDL with MTOM extension and adding WS-Policy,
> 
> .. I didn't think so ..
> 
> > or crafting a WSDL from
> > scratch that would work with both,
> > then this proposal makes more sense.
> 
> .. yes, that's the use-case I'm thinking of.
> 
> Maybe we can hear from Canon
> given it's them we're trying to help.
> 
> > It is unfortunate that there will be two ways to express an extension -
> >   <my:extension/>
> > and
> >  <foobar wsdli:simpleAssertions="true">
> >    <my:extension/>
> >  </foobar>
> >
> > Are there any ambiguities with this?  Namely, what if I understand both
> > <foobar> and <my:extension>.  Does the processing of
> wsdli:simpleAssertions
> > turn off the "understanding" of foobar?  Another way to ask this is -
> what
> > does the component model look like?  Both from a foobar aware processor
> and
> > a my:extension processor.
> 
> I'd imagine that's precisely the same as saying:
> 
>   <my:extension value="on">
>   <my:extension value="off">
>   <my:extension>
>   ...
> 
> or even:
> 
>   <ex:useTheForce>
>   <wsp11:Policy>
>      <ex:useTheForce/>
>   </wsp11:Policy>
>   <wsp15:policy>
>      <ex:useTheForce/>
>   </wsp15:Policy>
> 
> If you are going to say something twice
> make sure it's the same thing, otherwise you need
> some context dependent rules, and of course Policy is
> pretty good for composition.
> 
> So I don't think WSDL 2.0 needs to say anything new here.
> 
> Paul

Received on Tuesday, 5 December 2006 21:24:12 UTC