RE: Test suite SAWSDL update

Dear Jonathan,
the unclarity of annotation propagation was logged as an issue in [1],
and we resolved to clarify it this way:

The intention is to allow propagation of model references and schema
mappings from global type definitions to global element and attribute
declarations, both in schema and in WSDL component models (only elements
apply to WSDL).

The reasoning is like this: we don't model annotations in the component
model on non-global type definitions and element or attribute
declarations, largely because the WSDL component model only has
top-level components, and we didn't want to go deeper in schema than
necessary for WSDL, at least not in this version. Therefore propagation
to non-global element or attribute declarations would have no effect in
the component model. Further, propagation from local type can be done by
simply placing the annotations on the element (or attribute), not on the
type definition inside the element (or attribute), with no increase in
effort or decrease in reusability. So only propagation from global types
to global elements or attributes is useful here.

This will be clarified in the spec soon. Please let us know if this
clarification is OK with you. 8-)

Best regards,
Jacek

[1] http://www.w3.org/2002/ws/sawsdl/issues/LC-20070410

On Wed, 2007-04-25 at 11:17 -0700, Jonathan Marsh wrote:
> (Moving to SAWSDL list)
> 
> I think you are referring to this text from 4.1.1:
> 
> "A non-empty modelReference on a top-level simple type is represented as
> {model reference} property of the XML Schema Simple Type Definition Schema
> component; the case of an empty modelReference or no modelReference at all
> is represented with an XML Schema Simple Type Definition component that does
> not have a {model reference} property. {model reference} properties are
> propagated from a simple type definition schema component to all attribute
> and element definition schema components that are defined with that simple
> type.
> 
> "A non-empty modelReference on a top-level simple type used in WSDL is
> represented as {model reference} property of the WSDL Element Declaration
> component; the case of an empty modelReference or no modelReference at all
> is represented with a Type Definition component that does not have a {model
> reference} property."
> 
> I confess I am not completely sure what this means.
> 
> The first paragraph appears to be limited to the annotation of the Schema
> component model, and therefore irrelevant to WSDL-XSLT, which only exposes
> WSDL components.  However, I do wonder whether "defined with" is precise
> enough.  For example, does this include derivation by extension and
> restriction?  Are there any ambiguities when substitution or model groups
> are used?
> 
> I find the second paragraph is a bit ambiguous.  It says the {model
> reference} property that corresponds to a top-level simple type (and
> therefore corresponds to a WSDL Type Definition component), but it says the
> property is to appear upon a WSDL Element Declaration component.  Is this
> just a typo?  Should it instead say WSDL Type Definition component?
> 
> If so, I don't see any changes necessary to WSDL-XSLT, as the propagation is
> limited to XML Schema components, despite the implication you make below.  I
> currently don't propagate or roll-up or otherwise combine model reference
> values in WSDL components - what is defined on the Interface, Interface
> Operation, etc. is just what you get in the component model.  Let me know if
> I'm missing something.
> 
> Jonathan Marsh - http://www.wso2.com - http://auburnmarshes.spaces.live.com
>  
> 
> > -----Original Message-----
> > From: Jacek Kopecky [mailto:jacek.kopecky@deri.org]
> > Sent: Tuesday, April 24, 2007 3:24 PM
> > To: Jonathan Marsh
> > Cc: WS-Description WG
> > Subject: Re: Test suite SAWSDL update
> > 
> > Hi Jonathan, thank you for checking. The old namespace is why I wasn't
> > notifying you of the changes. Anyway, as you noted, it doesn't really
> > change anything.
> > 
> > Which reminds me, I'll have to check how WSDL-XSLT propagates the
> > SAWSDL
> > annotations - both model references and schema mappings should be
> > propagated from XML Schema types to the elements (and attributes, not
> > in
> > WSDL) that use them. The propagation is clarified in the LC draft,
> > http://w3.org/tr/sawsdl/
> > 
> > Thank you again, really,
> > Jacek
> > 
> > On Tue, 2007-04-24 at 15:00 -0700, Jonathan Marsh wrote:
> > > I checked for updates to the SAWSDL testcases [1] and found that
> > > SAWSDL-2G has been made obsolete.  Also a new testcase
> > > 05-simpletype-annotation-with-attribute has been added, which I
> > copied
> > > into the test-suite.   I regenerated the results - which didn't
> > change
> > > much since only wsdl-xslt supports sawsdl annotations.
> > >
> > >
> > >
> > > I also note the SAWSDL testsuite is using the old namespace - which I
> > > changed when updating the test cases.
> > >
> > >
> > >
> > > [1] http://www.w3.org/2002/ws/sawsdl/CR/testsuite.html
> > >
> > >
> > >
> > >
> > >
> > > Jonathan Marsh - http://www.wso2.com -
> > > http://auburnmarshes.spaces.live.com
> > >
> > >
> > >
> > >
> 
> 

Received on Sunday, 20 May 2007 20:36:44 UTC