RE: Duplicate @binding, @address on endpoint

The purpose of a specification, at the structural level, is to define
constraints.  We start out with the constraint that WSDL is an XML language,
and so must be well-formed.  That limits the set of documents that can be
called WSDL.  Furthermore we have a constraint that the XML document must be
valid wrt the schema we define, which further constrains the set.
Additional constraints, not expressible through XML Schema, are listed in
the specification as assertions.  A profile would presumably add additional
constraints (e.g. XML Schema 1.0 MUST be used as the type system) which
would further reduce the set of possibilities in the interest of
interoperability.

 

For instance, if it turned out many users were defining duplicate endpoints,
and this caused interoperability problems in practice, then an
interoperability profile could be developed which contained such an
assertion.  Those who conformed to that profile would have to follow that
assertion.

 

The WSDL spec tries to include useful constraints as assertions, but there
are clearly many possible assertions which are not codified, as they seem to
have limited impact on interoperability at this point.  I expect
implementation experience may result in additional assertions being added
through errata, or through an external profile.

 

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

 

  _____  

From: [email protected] [mailto:[email protected]] On
Behalf Of Ramkumar Menon
Sent: Thursday, February 08, 2007 9:39 AM
To: Amelia A Lewis
Cc: [email protected]
Subject: Re: Duplicate @binding, @address on endpoint

 

Hi Amelie,

 

I agree with you on both points - 100%. I think both seemed to be silly
questions to ask :-)

I guess I am misinterpreting the meaning and purpose of Assertions. 

 

Apologies,

Ram
 

On 2/8/07, Amelia A Lewis <[email protected]> wrote: 

Dear Ram,

The first of these strikes me as hand-holding.  Sure, one can do silly
things with WSDL, and we don't prevent it.  I don't think that we need to 
issue even a warning about this, frankly.  (If we do, could we recommend
that processors issue warnings in Python-speak or Seussian verse?)

 

As to the second, the assertions about best practices in another
specification developed by a different organization with no ties to W3C 
seems to me entirely out of scope for WSDL 2.0.  WSI will be able to
profile WSDL 2.0 itself, when/if it wants to.  Definitely not our job.

 

Amy!
(speaking for herself/her company, not the working group)
--On February 7, 2007 5:41:50 PM -0800 Ramkumar Menon 
< <mailto:[email protected]>  [email protected]> wrote:

>
> Gurus,
>
> Two questions.
> 1) Is it a valid use-case [even if its possible to model] to have a WSDL 
> 2.0 document that has two endpoints that possess identical values for 
> "binding" and "address" attributes [but with different names] ? If not,
> we could have an "SHOULD" assertion that covers this. 
>
> 2) Is it possible for the User to model WSDL 2.0 documents that are not
> WS-I BP compliant ?
>    If so, does it make sense for the Validator to emit warnings on
> incompatibility ?
>
> I am interested in knowing your thoughts on these points. 
>
> rgds,
> Ram



--
Amelia A. Lewis
Senior Architect
TIBCO/Extensibility, Inc.
[email protected]

 




-- 
Shift to the left, shift to the right!
Pop up, push down, byte, byte, byte!

-Ramkumar Menon
A typical Macroprocessor