(closed) Last Call (20060928) Issues List of SAWSDL

Accepted all comments as editorial. Will submit our WSDLs to the WSDL test suite.

[meeting minutes]

[email to commenter]

we don't seem to cover the case when modelReference is used both on an
element and on attrExtensions within that element. Something like this:

<wsdl:operation modelReference="a">
  <sawsdl:attrExtensions modelReference="b"/>
</wsdl:operation>

Attributes with the same namespace name and local name MUST NOT appear both on the attrExtensions element and on its parent element.

[meeting minutes]

[email to commenter]

The SAWSDL specification provides only a (probably smaller) part of the necessary specifications for semantic web services. This is probably ok in order to make progress. On the other hand it contains a mapping between XML schema and ontology. This is a quite specific part, which is not needed in every scenario where semantic web services are used. Therefore this part should be a separate specification. It may be even used for completely other purposes than in the context of web services.

Change TOC for WSDL annotation and Schema Annotation, also change title to add XML Schema

[meeting minutes 1, meeting minutes 2]

[email to commenter]

The example ontology purchaseorder is used in the modelReferences of the SAWSDL. It is not contained in the spec and version accessible via http://www.w3.org/2002/ws/sawsdl/spec/ontology/purchaseorder does not contain all used classes.

Editors check and fix.

[meeting minutes]

[email to commenter]

For the schema mapping all together 3 technologies are used in the examples: SPARQL, XSLT, XQuery. This is quite confusing. At least some more positioning - when to use what - is needed.

LC issue 5 will be implemented by the editors to explain the relationships between the mapping technologies in our examples

[meeting minutes]

[email to commenter]

the SAWSDL specification currently provides in its core only the definition of links to the identification of more or less arbitrary documents. I would like to see at least the kind of support that xml schema has in WSDL. That means additionally an embedding of the semantic model definition should be defined. The major additional requirement would be that there is an XML serialization of the semantic models. This would allow in a standard way to transport semantic models within WSDLs.

Move and expand example with additional description to a new section 2.3 Embedding Semantic Models including note above.

[meeting minutes]

[email to commenter]

XML has several possibilities to represent ordered values. The encoding with atomic datatype is very special and somehow outside the XML node mechanisms. Is it really a good idea to use these very special list datatypes?

stick to list of anyURI

[meeting minutes: part 1 and part 2]

[email to commenter]

A general comment to both documents is that while we would expect from [1] to be a formal specification, it still includes the number of examples, which actually should become part of the primer/guide document [2].

No action. Will reply that examples are not use cases. They are for readability.

[meeting minutes]

[email to commenter]

Moreover, the behavioural annotation remains underspecified leaving many options open for describing the Choreography of the Web Service. We believe that WG mean Web Service Choreography when referring to "specify behavioural aspects", but the ordinary reader might not understand the same by behavioural annotations.

we don't want to change it for it would be more confusing for our readers

[meeting minutes]

[email to commenter]

SEE has a Semantic SOA Reference Model which is inclusive of WSMO, where (functional) capabilities are attached not at the level of operations or interfaces, but to whole services, as an abstracted view of the whole functional behaviour. This raises this question of why the SA-WSDL modelReference attribute is restricted in those elements in the WSDL meta-model to which it can be applied, in particular why WSDL services cannot be given model references.

we won't annotate services, it's not necessary for what we need

[meeting minutes]

[email to commenter]

If we take a look at these documents from an industrial perspective, it appears the WG is merely trying to move complexity away. The entire specification is about adding three attribute to WSDL (modelReference, liftingSchemaMapping, loweringSchemaMapping) and opening up a world of semantic magic. However as soon as a reader goes through it, it becomes obvious that the magic (as always) is nowhere to be found. The modelReference attribute itself is just a pointer useful for discovery, but lowering and lifting operations are very complex and should be described using difficult declarative languages (e.g. XSLT is not so well accepted by the industry, especially if XML2XML mappings are required). There would be a need to develop some grounding machine, that would be quite a complex piece of software that could (not should) be feed with declarative descriptions (XSLT, SPARQL, etc.) of the lowering and lifting operations.

SAWSDL should guide the reader that SAWSDL is not the end, it's the beginning, and we expect specs that use sawsdl and specify actual semantics. [from conversation with Emanuele Della Valle]

Address this issue by updating group page by putting SAWSDL in context. Paragraph in beginning of spec saying just beginning and point to WG page. WG page can go further and Usage Guide can include much of the text from WG page.

[meeting minutes]

[email to commenter]

the semantics may be different depending on policy choices on the same endpoint, embedding SAWSDL annotations in policy would enable this [from personal conversation]

Issue renamed to "Using WS-Policy as part of the semantic model" and closed as out of scope for the spec (reopened for the Usage Guide in the main issues list as issue 32).

[meeting minutes]

[email to commenter]

The usual practice in the industry is to include/import schemas into the WSDL documents rather than including them inline. An orgnization may use a variety of schemas that are reused in many webservices with different semantics. This gets complicated further if standard schemas are included (Say from W3C) which can be used for many different purposes with different semantics. One can always find a workaround , the easiest being copying the schema and annotating the copy , but that would mean a considerable effort for porting existing services (I guess the thinking behind the SAWSDL approach is to minimize the effort to port existing applications)

we will add text that recommends RDF or XSLT for external annotations, nothing too normative

[meeting minutes]

[email to commenter]

Many people in the audience felt that they would be using SAWSDL
modelReferences for representing preconditions and effects associated with
Web services and that they would find the spec and usage guide incomplete
without at least a brief discussion on preconditions and effects and a few
examples to show how to do it (sort of like how we show examples for
publishing Web services in registries)

handle this in the Usage Guide, issue moved to old issues list to issue 33

[meeting minutes]

[email to commenter]

When we associate multiple annotations with an element using
'modelReference' (eg. below) people felt that not providing an explicit
relationship (such as 'intersection') could lead to arbitrary
interpretations.

Adopt Rama's proposal (see issue email) -> No action

[meeting minutes]

[email to commenter]

modelReference Vs. lifting and lowering schema mapping extensibility
attributes: Some people felt that they did not appreciate the distinction
between modelReference and lifting and lowering schema mapping
extensibility attributes since all three were meant to serve as URI
pointers. The question was why not just have one extensibility attribute
called modelReference and use it for everything.

model reference and schema mapping URIs point to different things, need to keep them distinct

[meeting minutes]

[email to commenter]

I noticed that our RDF mapping does not specify RDF properties for liftingSchemaMapping and loweringSchemaMapping attributes.

we will declare two RDF properties for lifting and lowering schema mappings

[meeting minutes]

[email to commenter]