See also: IRC log day 1, day 2
<ericP> scribe: Laurent
<ericP> scribeNick: laurenth
jacek: laurent volunteers to scribe the first half day
... scribe this afternoon is laura, carine next morning 31 and joel is last on 31 afternoon
jacek: because of 30 minutes delay, the whole planning is shifted
rama: sgnificant changes sent today on the main spec: examples, clarifying text etc.
jacek: the group should focus on the frozen version
joel: however, a change in a word that impacts the understanding should be taken into account
jacek: ok, apart for that, stick on the frozen version
... ok, agenda accepted
<scribe> ACTION: EricP and JacekK to draft an RDF mapping section for SAWSDL [DONE] [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action01]
<scribe> ACTION: EricP to add text gravy to his annotated Amazon WSDL and put it in Examples document [PENDING] [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action02]
<scribe> ACTION: JacekK to send editorial comments to editors [DONE] [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action03]
<scribe> ACTION: EricP to let XMLCore know about attrExtensions element [PENDING] [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action04]
eric: action 4: they have not replied, maybe a reply in the next two weeks.
jacek: the process should be tracked
<scribe> ACTION: Editors to add text to 2.1.1 that we do not constrain the form of the semantic model for categorization (or for model reference on interface in general) and soften "in particular" to "for example" [DONE] [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action05]
<scribe> ACTION: Rama (with help of Amit) to prepare a UDDI presentation for the 2nd day of the f2f [DONE] [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action06]
jacek: anything substantial should be raised
rama: shuold we mention in the abstract why we do this or compare with others
jacek: we are building a standard. We don't have to say why others don't do that
rama: input to the abstract >>> wsdl lacks semantics, this is why sawsdl exists
thomas: why sawsdl is important statement should be synchronized across the spec and the examples document
jacek: any issues in the text of the introduction?
carlos: a problem with talking about concepts and instances etc. probably this relates to class
jacek: do you have anything better to replace the status quo with?
laurent: isn't this too specific? what about tags,
jacek: no, tags do no fall within the scope, not formal
*section1.1: running example
joel: shoudn't annotations be instrocuced in "runnning example"? and change the name to "example" to use snippets in the rest of the document
jacek: ok, resolved
... change from ean to less localized (UPC?). acceptyed
* section 2
thomas: problem with uris
* section 2.1
rama: problem with properties?
jacek: properties are required for processors that build a representation of the file
... wsdl have decided to have a component model distinct from the xml. So for instance the concrete syntax need not use same quotes as illustrative examples
rama: probably a bit of explanatory tect is required in section 2.1 to clarify this.
jacek: ok, we may view this as an EDITORIAL ISSUE : explain how the WSDL component model is used. Move the paragraph on WSDL 2 components elsewhere better in the text.
thomas: objects that the WSDL standard approach is better
jacek: starting from the xml is more readable, and the WSDL people won't complain about this;
* section 2.1.1
rama: move the whole UDDI discussion to the appendix
... using model references on interfaces to represent preconditions and effects
thomas: ys, and choreographies as well
jacek: our charter says we cannot represent preconditions etc., but we can provide hooks.
rama: this would make jim hendler happy
joel: we don't have examples using UDDI,
rama: yes we we do in the appendix D
joel: in appedix D we have an example of using UDDI. What esle do we have to do with the UDDI issue 20?
jacek: define how SAWSDL fits in UDDI
... discuss the move of UDDI text in section 2.1.1 to when the issue 20 is discussed
... ISSUE: effects and preconditions and behavioral elements (choreos, orchestrations) to be added to the interface?*
eric: the model references should be defined as identifiers, since they can be dereferenced
jacek: can you point to the specific text?
* section 2.1.2
rama: I have an issue: paragraph four: last sentence was decided to be deleted at last telcon.
jacek: ISSUE OK
* section 2.1.3
jacek: editorial note. ISSUE : More text is missing about the meaning. Why do we annotate.
* section 2.1.4
rama: paragraph 4 and 5 contradict
laurent: last sentence of paragraph 4 should be dropped
jacek: ISSUE drop this last sentence to be discussed.
* section 2.1.5 - 2.1.6
jacek: there is no "why?"
rama: this is improved already
* back to erics issue section 2.1 par 4
eric: replace "that reference concepts" to "that identifies concepts"
jacek: everybody ok : issue raised and solved and logged...
* section 2.2
jacek: known isssue: lacks motivatory text
rama: last - 1 paragraph: examples should be inserted here.
jacek: ISSUE: the penultimate paragraph in section 2.2 needs an example. Also, the last two sentences are identical in meaning
* section 3
* section 3.1
jacek: known: coordination with action item relating to xml core
... ISSUE applicability of the attrExtension element: general or specific to WSDL1.1?
* section 3.2
jacek: ISSUE drop the part on faults. Consider the question of annotating faults in sawsdl
... the word "part" is used as a target of annotations. Do we not mean mesage instead?
rama, joel, eric: comments and objections
jacek: ISSUE: we might want to add model references to WSDL1.1 messages constructs as a whole.
* section 3.3
jacek: example translated, no issue from my side.
* section 4
rama what about appendix references?
jacek: should be in "informative references"
* appendix A
holger: do we need both xquery and the rest?
jacek: lets push this to the editors
... anyone having a proposal for xquery can volunteer to fill in the section
* appendix B
rama: table 1 sentence must be changed, and is changed already in the working version
* appendix C
* appendix D
* acknowledgements.
jacek: status of John Miller? ask W3C.
<scribe> ACTION: EricP to check John Miller status [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action07]
<scribe> ACTION: JacekK to put john domingue and jim hendler in bad standing, and not reference them in the acknowledgements [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action08]
* changelog
eric: keep the changelog in the final section
* todo list
jacek: empty at the end
end of first half morning session, and coffee break
jacekK: penultimate paragraph in 2.1.1: moving this into the appendix D
... any objections for moving this paragraph to an appropriate part in appendix D? This solves issue 20
rama: tmodels in uddi is pretty much the same as model references
... when you publish a web service, you can specify the tmodel that applies
jacek: rama to provide an example with UDDI.
... no other UDDI issues?
rama: you must use the categoryBag construct in UDDI
jacek: proposes a complete example with usesTaxonomy
joel: so we would have two versions?
jacek: we could modify the first example accordingly
... chaneg example in appendix D to:
usestaxonomy="uddi:uddi.org:....*
<scribe> ACTION: Rama to change the URI as used in appendix D and add some text detailing how it would look like when mapped to uddi [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action09]
jacek: issue 20 is resolved
joel: maybe Rama should read her motivational text to the group^
rama: this is in section 2.1.6 in the examples document
jacek: make it clear that the concluding text in section 2.1.6 is seen as concluding section 2.1 instead
... do we need this conclusion?
... this could be an introduction, or replaced by a transition at the head of section 2.2
laurent: add a summary subsection at the chapter end?
jacek: transitional text should be insterter at the head of all sections
... the paragraph on data transformation: we should have "client" instead of WS
joel: when secvices communicate, the communication is schema to schema
... a client has a sawsdl that describes the requirements of a service
rama: nothing prevents one to use wsdl to represent the client
jacek: however, the wsdl spec does not describe this
thomas: the text in this section could be moved in the usage document
jacek: the motivatory text would benefit to mention the "hop" from one wsdl to another through semantics, but not to originate from this.
rama: we should say how this lifting lowering is used
jacek: the editors could have a firs paragraph, neutral, that says what the mappings do in the context, and then a second paragraph can show how this can be used
holger: too many assumptions in this little text
joel: in this introduction we should say that this is for solving execution time problems
rama: the spec must tell us why it is so, and what it is for, or it is unreadable
joel: the spec must be rigorous enough
holger: same with the uml: too much text leads to errors
jacek: separating the why and the how is good. how it can be used is in the examples document
claudio: difficult to understand fully a solution. I would like to keep these lines.
jacek: reverse the order of the paragraph, staring by the general text.
... issue 25 is resolved by this editorial issue.
... issue 27 is closed and already implented with minor editorial changes
<Rama> scribe: Laura
starting afternoon session 30/8
Rama: sawdsl does not recommend a language. Saying that we could have precondition and effect , but not how.
LH: we can have a modelRefernce pointing to preconditions,or other workflow....
JK: refer to concepts in ontology, have some part that address semantic description of concept
CB: doubt in insert these features because out of scope of the specs
Conclusion: modelRefernce in interface and operation can also point to behavioural aspects. Add this aspect but not mention preconditions, effects, and other behavioural aspects and other semantic definition (such as non-functional properties, policy ... )
JK: drop the definition, and put it as a plain text in the introduction
JF: semantic annotation covers modelReference and some mapping ....
LH: we need static transformation, and also some mapping
Rama: we also need to redefine Concept
JK: drop Semantics
Rama: does not agree
JK: semantics is used in the document but can be changed
LH: drop last phrase of definition of Semantic Annotation
Conclusion:
Semantics
Semantics in the scope of this specification refers to formal definitions, for instance in terms of concepts in a domain and the relationships between those concepts.
Semantic Annotation
A semantic annotation is additional information in a document that identifies or defines the semantics of a part of that document. In SAWSDL, the semantic annotations are XML attributes added to a WSDL or associated XML Schema document. (??They establish the meaning of elements in WSDL document by referring to a part of a semantic model.: to be discussed tomorrow 31/8, or conference call. Editor to keep in mind the point and see ).
Remove last sentence:
A complex type can be annotated at both the top and member level. Both the high-level type annotation and the specific member element or attribute annotation apply to the member.
Conclusion: there are some examples edited by Rama yet available and sent through e-mail.
And show also example of overriding. UP TO EDITOR.
Drop: It is not valid in WSDL 2.0 documents, where the modelReference attribute should be applied to the operation instead. (JF has notes, about MUST...)
JK: Remove faults from 3.2
LH: annotate also fault for wsdl 2.0
Conclusion: fault support also for wsdl 2.0 (JF has notes)
<ericP> duplicate 2.1.2 into 2.1.3 and s/operatoins/faults/
-do we want annotate message with Model Reference?
NO ACTION
<eric> email
## modelRefernce equal "" is valid?
- let the text 0 or more the text
- modelReference is a list.
- List of URI: list of schema mapping for validation, which is the order to apply?
EP: (Conlusion) Chaning mention of list in set . And also 0 or more.
## A.An xample: the yellow note has to be transformed in text.
## A1.to be completed
## A2.2 to do or to be removed
FOLLOWING TELECON 12-9.
TO HAVE THE REVIEWED VERSION FOR 10/9.
TOMORROW: Semantic Annotations for WSDL - Usage Guide (24/08 version)
<ericP> RDF mapping
<caribouTRN> Scribe: Carine
<caribouTRN> ScribeNick: caribouTRN
JK: proposal for section 4 of the spec
<ericP> RDF mapping
WSDL RDF Mapping is currently in Last Call
JK: the WSDL RDF mapping can handle generic extensions
... as SAWSDL is an extension, it uses the WSDL RDF Mapping + our RDF mapping for the extension itself
LH: I wonder whether we can use the mapping to intantiate a WS ..?
JK: the WSD WG has published the Last Call but will then publish it as a WG Note, not on REC track
... the SAWSDL WG is allowed to take that note and lead it to REC
Joel: I never understood a use case for the RDF mapping
JK: I see it as putting a WSDL doc into SW
EricP: +1
... it's encoding into a different data structure
JK: it may lose some information (e.g. the schema)
... the editors of the RDF mapping for WSDL try to do something useful for somebody's triples
... to be able to do things with RDF tools instead of WSDL-specific tools/XML tools (XQuery)
Rama: why not starting directly in RDF then?
JK: nobody would want an RDF wsdl, nobody wanted RDF RSS
EricP: the RDF syntax is less suitable for human consumption
<ericP> typos in rdf mapping:s/For example, the a graph:/For example, the graph:/
<ericP> also:s/quotedXML#';/quotedXML#'/
<ericP> also:s/4. RDF Mapping/4. RDF Mapping (non-normative)/
EricP: do we need to say that OWL basic knowledge is a prerequisite?
JK: it seems to me that it is obvious
<ericP> SAWSDL Usage Guide
<ericP> (AKA, the examples document)
[discussion around the expression "semantic similarity"]
Rama: it is some relationship inferred from the annotations by a reasoner
... I think I can refine the wording
TV: the word "match" may be better to understand
LH: "to find service descriptions that match a request"
[nobody objects]
JK: anything to put in section 1.1?
[no]
JK: namespaces to add
... section 2 does classification and 3 discovery/composition
... can section 2 be considered as a subtask of 3?
Rama: 2 is publishing, it's a separate task that you do upfront
... if you have a registry like UDDI you can use this information on a WS
... I would not combine this as part of discovery
JK: let's keep it in separate sections
... comment about section 2, does it need a registry out there or you can just annotate your wsdl ?
Rama: we can give some more details about registries just publishing on a website
JK: in example a- in section2, OrderModification does not seem to be a _subclass_ of PurchaseOrderServices
Rama: it's a taxonomy
... proposal is s/with categorization as shown/with taxonomy information as shown/
EricP: URIs could be shortened when we publish
JK: the document will probably end up as a WG Note
... There's a relative URI in the latest code example of section 2
... it has to be real
... we can use xml base, or have entity references in section 1.3 along with the namespaces table
... or make it absolute (it would be more consistent)
... we might want to introduce caption for the code listings to make references from the text more clear
EricP: I can give one of the examples turned into n3
<ericP> @prefix ifce: <http://www.w3.org/2002/ws/sawsdl/spec/ontology/interface#> .
<ericP> @prefix rdfs: <http://www.w3.org/2000/01/rdf-schema#> .
<ericP> @prefix xsd: <http://www.w3.org/2001/XMLSchema#> .
<ericP> ifce:Category rdf:type rdfs:Class .
<ericP> ifce:hasValue rdf:type rdf:Property ;
<ericP> rdfs:domain ifce:Category ;
<ericP> rdfs:range rdfs:Literal .
<ericP> ifce:usesTaxonomy rdf:type rdf:Property ;
<ericP> rdfs:domain ifce:Category ;
<ericP> rdfs:range xsd:anyUri .
JK: Comments on section 3?
... s/ to enable automatic matching and composition of Web services/ to help ....
TV: I'd say help rather than enable (enable suggests "full automation")
typo: s/annoations/annotations
JK: going back to section 2 b- point to the appendix of the spec (instead of repeating the ontology for category)
... correct http://uima.semantica.ibm.com.wsdl/CheckAvailabilityRequestService/ in first piece of code in 3.1
... it will be updated to WSDL 2.0, right?
Rama: Brahmananda will do the translation
JK: deliveryDate and shipDate are not equivalent (a few days difference)
... 3.1 could be rephrased as ontologies and a mapping between them
... if we put a new "3.0-before-renumbering" section with a simple example, 3.2 can continue 3.0 instead of 3.1
Rama: there's an additional level in the figure (EAN added)
LH: why don't you use UML here?
JK: what's "equivalent" in UML?
... "equivalentClass" should have arrows both ways or no arrows at all
EricP: figure 2 is broken
Rama: the examples are not meant to be related. it seems confusing for people here and then for readers
EricP: you have to decide if you model the RDF graph or the OWL
Rama: the latter
EricP: no arrowheads for equivalentClass then
JK: you talk about producing EAN from PartNumber, but you say it's equivalent
Rama: actually the ontologies won't have any relationship, the lookup service is between them
JK: since 3.2 is building on 3.1 it keeps the first "level" of mediation and adds another one
Rama: the mediation is done by an external entity rather than the service
JK: but it keeps the ontology mediation we use in 3.1
... suggestion is
... having 3.0 with both guys using UPC
... introducing partNumber for 3.1 saying it's equivalent to UPC (mapping ontologies)
... and then having 3.2 with EAN and UPC matched through a lookup service
Rama: that would be composition without matching
JK: you can have mapping on other parameters
going further in the document
JK: in section 3.3 sawsdl:modelReference="SampleRetailOntology#PartNumber" "SampleRetailOntology#UPC"/> should be sawsdl:modelReference="SampleRetailOntology#PartNumber SampleRetailOntology#UPC"/>
... (no separate quotes)
... "Best practice recommendation" is about discourage a bad practice
LH: this is actually a best practice, it says "a recommended best practice is to amend the ontology..."
JK: it's not a best practice for annotation
Rama: If I had a schema mapping from FullName to FirstName and LastName, I could use a tool
... this is not necessarily wrong to use FirstName LastName in modelReference
LH: I suggest that we remove that "Best Practice recommendation"
JK: rationale "it's confusing"
... section 3.4...
... shows annotation on complex types at container and child level
... use case?
Rama: I can add a few sentences saying that you can do matching at the complex types level
[agreed]
JK: section 3.3 will be a complete scenario, we don't really want to have 3.4 a complete scenario
... we need to explain in 3.4 how it relates to the preceding subsections
... Section 4...
... Section 5...
... I'd suggest to remove it for the moment
Rama: right
JK: Section 6...
... app A should be "summary of examples"
[discussion on publication of the document as a first draft]
-> http://lists.w3.org/Archives/Member/member-ws-semann/2006Jul/0015.html
JK: title should be more like "Telecom Italia Case Study"
... COnfidentiality?
LF, CV: not sure about the ontology (not in the document)
JK: it would have to be public to be used by this WG
CV: is there a deadline to have something ready?
JK: no
Rama: we can put it in appendix of the usage guide when it's ready
Rama: UDDI is intended to be a public WS registry
... Universal Description, Discovery and Integration
... White pages, yellow pages, green pages
... categorization like NAICS are used for yellow pages
... there's an API with everything to search and publish to registries
... example using UDDI repository and 3 matching services (UML, WSDL, DAML-S)
... a tModel "describedUsing" categorize the kind of description language between those 3
... now the taxonomies are represented as OWL ontologies
... but UDDI is not doing semantic matching natively
EricP: has it been popular enough to be considered a success?
Joel: it's deployed between partners
Rama: there's a renewal in services registries, like UDDI with additional functionalities
... versioning, management...
... green pages = the API you use to find things in the registry, ...
... it was a concept but I don't think it has been fully implemented
... there's no notion of composition actually
JK: anybody knows a registry actually storing WSDL?
[registries and W3C..]
JK: proposal to contact the guy who raised the issue about UDDI
Rama: we can show him how to use the annotations with an example
<JacekK> ACTION: JacekK to respond to Pierre Chatel about our resolution to his issues [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action10]
<scribe> Scribe: Joel
Holger: Need to add my name to minutes of last call
MINUTES APPROVED
<ericP> joel: biggest jobs: change the examples, add lifting and lowering, add XQuery
<ericP> holger: current example is a schemaLifting example
<ericP> ... hard to know how to map from an [arbitrary isomorphism of an] RDF graph to a specified XML format
<ericP> rama: do you want to generate the request?
<ericP> jacek: given an id, amount, ..., this data gets lowered into an XML message
<ericP> joel: how do we get at the semantic data?
Joel: Working on all comments from day one and Rama's editorial comments.
... Work needed on examples, especially in schema mapping
Eric: Can help on XQuery example
Holger: Problems in creating loweringSchemaMapping, how to create an XML message from the semantic data. How to processs the semantic data as input to the mapping
... also not clear on lifting scenario.
Eric: I can provide SPDL to help with this problem.
... choices are SPARQL, XSLT
Jacek: SPARQL seems like it could be combined with XSLT. This is basically data grouding. Might need a new language to define how to combine them.
Rama: Must be a way to use XSLT
Holger: We found 3 varients. Can't be handled in XSLT.
Eric: Another issue - would need nested queries
Jacek: XSLT Foreach can handle.
Eric: Must do big square query
Jacek: Problemm is mutiple serialization of triples
Joel: Can we just see what the serialization is?
Eric: No, can predict.
... First need SPARQL query to get subset of the graph.
... Can report variable bindings in XML or pass graph to another RDF processor.
Jacek: need liftingSchemaMapping on result
... Looks like input is fairly simple. Can assume data we are lowering is one message - one customer number.
... Then see who can get a general solution.
... Showed example. So we need SPARQL + XSLT
Rama: Why is XSLT ugly
Jacek: Many pages of code that you are never sure if is right becuase of multiple serialization.
Rama: And SPARQL
Jacek: ...doesn't deal with serialization, works on the graph
... Language forlowering is slightly extended SPARQL with a transformation extension.
Eric: Better to create an XML format than to extend SPARQL
Rama: can we specify both URI's on schemaMapping
Jacek: No they are alternatives
Holger: I have what I need.
<holger> <OrderRequest>
<holger> <hasLineItems>
<holger> <LineItem>
<holger> <hasProduct>
<holger> <Product>
<holger> <hasProductCode rdf:resource="#018763546736" />
<holger> </Product>
<holger> <Product>
<holger> <hasProductCode rdf:resource="#018763541236" />
<holger> </Product>
<holger> </hasProduct>
<holger> <hasQuantity rdf:datatype="&xsd;float">12</hasQuantity>
<holger> <LineItem>
<holger> <hasProduct>
<holger> </LineItem>
<holger> <Product>
<holger> <hasProductCode rdf:resource="#018763541236" />
<holger> </Product>
<holger> </hasProduct>
<holger> <hasQuantity rdf:datatype="&xsd;float">12</hasQuantity>
<holger> </LineItem>
<holger> </hasLineItems>
<holger> <hasCustomer>
<holger> <Customer>
<holger> <hasID>1234</hasID>
<holger> </Customer>
<holger> </hasCustomer>
<holger> </OrderRequest>
<holger> SELECT ?q WHERE { ?o hasCustomer ?c . ?c hasID ?id;
<holger> hasLineItem ?i . ?i hasQuantity ?q ;
<holger> hasIdentier ?ean }
Carlos: Why are we doing all this
Jacek: Just for example. Will be in appendix, non-normative
Eric: SPARQL is not yet a req.
Jacek: Propose an extension to specify an XSLT to transform output.
<JacekK> TRANSFORM WITH http://example.com/tableToOurSchema.xslt
Joel: Can we have XSLT invoke the SPARQL query and process the output?
Eric: No, but maybe we can use the W3C service for XSLT
Jacek: Can say our solution can be replaced by XML Pipeline.
... Proposal: Move lifting to response element. Use XSLT and SPARQL for the loweringSchemaMapping
<ericP> <sparql xmlns="http://www.w3.org/2005/sparql-results#">
<ericP> <head> <variable name="c"/> <variable name="q"/> <variable name="ean"/> <variable name="id"/> </head>
<ericP> <results ordered="false" distinct="false">
<ericP> <result> <binding name="c">1234</binding>
<ericP> <binding name="q">12^^xsd:float</binding>
<ericP> <binding name="ean">018763541236</binding>
<ericP> <binding name="id">sfawqeqw</binding> </result>
<ericP> <result> <binding name="c">1234</binding>
<ericP> <binding name="q">1^^xsd:float</binding>
<ericP> <binding name="ean">018763541000</binding>
<ericP> <binding name="id">43234h</binding> </result>
<ericP> </results>
<ericP> </sparql>
<JacekK> <lowering><sparql>SELECT ...</sparql><xslt>...</xslt></lowering>
<JacekK> <wsdl><lowering xml:id="a4093"><sparql>SELECT ...</sparql><xslt>...</xslt></lowering> ... <element lowerSchemaMapping="#a4093"/> ... </wsdl>
<ericP> <element name="OrderRequest" sawsdl:loweringSchemaMapping="... pointer to the <lowering> resource>
Rama: No problem with lifting since we pick output RDF XML. Need to describe all this in the example.
Jacek: Say - In our example we combine XSLT and SPARQL. W3C is working on solution (pipeline)
Holger: Have extended OWL, working on mappings
Jacek: Use N3 Turtle for RDF representation
<ericP> http://www.dajobe.org/2004/01/turtle/
HOlger: to long. OUtput must be XML RDF
Jacek: Turtle is easier to read.
... proposed plan, have updated draft, by Joel, by end of Friday. Team reviews on following Tuesday. Holger makes updates from call.
JOel: we should use OWL for our ontology, not abstract OWL syntax.
Jacek: agreed
Holger: Will have all external files completed before leaving on vacation.
Joel: I will do other changes
Jacek: Will shoot for approval in next Telecon
... Check schemas and namespaces
Holger: Do we need the different namespace for examples?
Jacek: Same namespace for all.
Eric: Doesn't seem any namespace conflicts in XML infoset. Will check
Jacek: Update documentation element in schema about attrExtensions, replace version in the spec.
... update namespace document, do be done by Eric
Jacek: Want to do CR in December. Next F2F around 2nd week in November, probably 13th and 14th of November. Need to decide by the call on Sept. 19. Looks like we can do this.
... give reviewers a month, maybe until 20th of October.
... what if we don't have full two days worth of work
<ericP> from http://www.w3.org/2005/10/Process-20051014/tr.html#last-call [[
<ericP> Duration of the review: The announcement begins a review period that SHOULD last at least three weeks but MAY last longer if the technical report is complex or has significant external dependencies.
<ericP> ]]
Joel: Can have presentations by Amit's group.
Carlos: Important thing is to set the dates.
Jacek: I will confirm with Amit by September 19th, but will report to group on the 12th
<JacekK> ACTION: JacekK to contact Amit about having f2f on 13-14 Nov [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action11]
Jacek: we'll try to publish Use case document in September.
Rama: What about WS-MetadataExchange, WSRF?
Jacek: Need to give someone an action to see if there is a relation.
Eric: Volunteer.
Claudio: Was looking at it to implement stateful web services.
<JacekK> ACTION: EricP to investigate relation of SAWSDL to WS-MEX [recorded in http://www.w3.org/2002/ws/sawsdl/minutes/20060830#action12]
Jacek: Next, look at WS-Policy
... don't talk about semantics, but that will not convice anyone
Rama: There is one place they mention it.
Jacek: Is used for non-functional attributes. Does not intend to say more functional information, but to contrain the boundaries.
... we are function, policy is non-functional
Carine: Can have a relation via an ontology
Jacek: But we don't specify any relations between annotations. Policy has "and" and "or".
Rama: but we have looked at annotating WS-Policy expressions, so it is not just functional vs. non-functional
Jacek: Another idea is that our use is matching and composition, theirs is negotiation and usage
Carine: Policy is more self-contained
Eric: All apply is like their "or" but thy also have tree of assertions that can be referenced together.
Joel: We both overlap in that we can both subsume the other, but we deal with intefaces and they deal with implementations
Rama: and we have different requirements
... best practice is to say how each relates to WSDL
Jacek: Like Functional vs non-functional - interface vs implementation
... but we do work on capability
Carine: we talk about meaning of operation, not capability
Jacek: Not our aim to describe capabilities and constraints. What is out of scope for policy? Have been asked by Phillip do find out the relationship.
Rama: People really need to know what the relationship is.
Jacek: WS coordination group has this on the agenda. Jacek and team will use this input and say our charter blocks us from futher investigation
Joel: OUr company didn't see a problem with moveing forward with SAWSDL and Policy
Rama: Should leave it possible to annotate a policy
Jacek: We use URI's and they use elements for assertions. Have a good case for CG if anyone raises the question.
Rama: can use policy while doing composition to limit what can be used.
Claudio: Any coordination with work in OMA?
Eric: There is a liaison. We will find out what coordination might be happening.
Jacek: Claudio should talk to the W3C team.
Jacek: be responsive to editors e-mails over the next week.
... Thanks to Claudio and Laura for a great job in hosting.