W3C

WS Description FTF
28 Feb 2006

See also: IRC log

Attendees

Present
Charlton Baretto, Adobe Systems
Allen Brookes, Rogue Wave Software
Glen Daniels, Sonic Software
Youenn Fablet, Canon
Hugo Haas, W3C
Tom Jordahl, Adobe Systems
Anish Karmarkar, Oracle
Jacek Kopecky, DERI Innsbruck at the Leopold-Franzens-Universitšt Innsbruck, Austria
Amelia Lewis, TIBCO
Jonathan Marsh, Co-chair/Microsoft
Jeff Mischkinsky, Oracle
Jean-Jacques Moreau, Canon
Vivek Pandey, Sun Microsystems
Tony Rogers, Co-chair/Computer Associates
Asir Vedamuthu, Microsoft
Umit Yalcinalp, SAP
Regrets
Roberto Chinnici, Sun Microsystems
Paul Downey, British Telecommunications
Kevin Canyang Liu, SAP
Sanjiva Weerawarana, WS02
Chair
Marsh
Scribe
Allen, GlenD

Contents


 

 

<Allen> SCRIBE: Allen

<Arthur> [1] http://lists.w3.org/Archives/Public/www-archive/2006Feb/0038.html

<Arthur> [2] http://cvs.apache.org/dist/ws/woden/milestones/1.0.0M3-incubating/

<Arthur> the current version uses the new namespace

<Arthur> thank you

<Arthur> i'd like to thank all the little people

Woden demo

Arthur: purpose of woden is to provide Java model of WSDL2.0
... Apache incubator project. Need other companies to contribute to get out of inucbator status

<scribe> ... completed milestone 3

Tony handling the demo since Arthur is not local

Arthur explaining demo ...

Two parts to API, document portion, and component model. APIs very close on each model

Glen: How do featrue handlers or other things in my namespace get validated?

Arthur: DOM based implementation, clean separation between API and implementation, hence the factory. By the time you get the description the whole document has been parsed and validated, including your namespace. Not sure how extensions implemented.

Umit: Do you need a separate reader for each document?

Arthur: No you can reuse the reader
... APIs for infoset representation as well

Glen: Can you create a component model and do a serialization?

Arthur: No
... Mostly trying to get enough done to get through CR
... Woden has a suite of JUnit tests. Still 8 failures but that just reflects current state of implementation
... I encourage everyone to get involved, help get us through CR

Tony: Do you need SVN to check out code?

Arthur: Yes, many free versions. If you just want to look at code you can use the web interface

Vivek: Is there a reason to not use generics?

Glen: So it can be used with Java 1.4
... Will probably change in a year

Arthur: Don't want to prevent adoption

WSDL 1.1 to WSDL 2.0 converter

Hugo: Want to see how hard it might be to provide a conversion tool
... XSLT style sheets. Works fairly well on a certain category of WSDLs, those that follow WS-I basic profile
... Message parts with types are difficult

Janathan: Wrote similar converter. Couldn't do everything. Instead flagged parts where user attention was needed.

Hugo: This is a prototype. Big disclaimer to read through results

Umit: How about the mapping into into headers. How is that done?

Hugo: Does handle this.

Vivek: No support for header faults?

Hugo: No

Anish: How do you handle binding of multiple parts to the body?

Hugo: On the limitations page it lists limitations found so far. This is one.
... We need to list WSDL1.1 -> WSDL2.0 gothcas independent of any converter

Jonathan: How much work was this to figure out?

Hugo: Not so much since I knew both specs

Jonathan: We could use this to get a bunch of WSDL 2.0 documents

Asir: Did you find any issues with the spec writing the tool?

Hugo: Did discover a few small issues doing this
... Used Woden as a validator.

Anish: Do you do anything with MIME or MTOM?

Hugo: Multipart related is not supported. If MIME types are returned this is supported.

Jonathan: Do we have a link to this on our page?

Hugo: Yes

Assertion reviews

Glen: Looked at feature and properties assertions. A few left to be done.

Jonathan: Sections assigned have been completed. Look really good.

Jacek: Should we remove MUST where they are pointing to other assertions?

Jonathan: Thought about it but no need

JJM: Mine not done yet

Jonathan: Amy's not yet done

Umit: Haven't done mine yet

Jonathan: No clue on DaveO's
... Risky ones are DaveO's. Should consider reassigning

Arthur: Already reassigned

Charleton: I had some in adjuncts section 5 and did them.

Jonathan: Anything else look risky?

Arthur: For every assertion we need a test case that violates it. Create a document that is schema valid but violates the assertion and then submit it.

Jonathan: One thing we can do to move this along. Give everyone a valid document and have them corrupt it. Let's do this in the last half hour.

Issue 281

Arthur: Test points to a fictitious URL. Test has a catalog that maps the document to a file.
... In the test suite there is an import.

Jonathan: May have already been solved. Just want to see what the status of the issue is.

Arthur: Test case has been corrected.

Jonathan: No longer points to external file?

Arthur: No that is solved with the catalog file. Import has been fixed.

Jonathan: Now there is a dependency on catalog.

Arthur: If we don't use catalog then all imports are for local files.

Jonathan: Can wait and see if people have problems with catalog

Glen: Problem is tooling. If we use relative URIs then we avoid the problem. I would vote for using relative URIs.

Arthur: Other issue raised is location is just a hint. Was going to use the catalog to find wsdl when location is not there.

Glen: Ok for the no location case to use the catalog

Arthur: Their tool doesn't have to support catalog.xml. Test suite uses it.

Jonathan: There will be lots of issues of this type when people start using the test suite. Maybe we should wait till then to solve this.

Glen: This is obvious. Lets fix it now.

Jonathan: Shall we just fix this test case to use a relative URI and close the issue?

RESOLUTION: Issue 281 is fixed

Resuming

I18N

I18N core working group visitors Mary Trumble and Felix Sasaki

I18N has produced 2 documents: Usage Scenarios and Rquirements

Mary: first draft of WS-I18N document
... here because WSD is a web services working group
... looking for additional I18N members
... problem - loosley-coupled system has no locale model
... how do you distribute a locale?
... possible patterns: locale-neutral, service-determined, client-influenced, data/resource-determined
... All are possible, need a way to specify
... have an implementation as a soap header <InternationalizationContext/>
... WS-Policy wants a policy to specify locales

Jacek: Locale is presentation mostly. Web services should transfer data that has nothing specific to presentation so it is up to client.

Jonathan: Not entirely, might want to ship a date off, say to China, and have it formatted.

Glen: Mostly when messages are on the wire should be in a common format and locaize when they reach the node itself.

Jacek: Jonathan's example is a service that does presentation. For such cases the locale should be a parameter not a header.

Glen: Two scenarios, pass a locale with locale as a parameter or send in locale independent form.

Mary: our implementation allows locale to depend on client, server, or data

Jacek: do you have a point to the use cases?

Mary: go to W3C and look for core I18N

Jonathan: What is your timeline

Mary: We were hoping to be done by next month. With loss of team members don't know.

Hugo: to be done do you only need to fix the web services part of you document?

Mary: WS-Policy needs to be done also

Felix: Work needs to be done on locale identifiers

Hugo: If we were to send some comments? Is the TR version the latest version?

Mary: Yes

Hugo: don't want to repeat comments where can I see existing comments?

Mary: There are comments at the official URL for comments
... Comments are welcome

Jonathan: When you make more progress we can put it on our regular agenda to do review

WS-A review

Hugo: sent two emails, some of the language uses could be improved.

Jonathan: We'll forward the comments from the first email on to WS-A

Hugo: email 2, {addressing required} reduces a property with 3 cases to 2

Jonathan: We'll forward this comment
... Arthur's comments ...
... (1) could use component designators instead of new elements.

Jacek: Value in using the elements over the component designators because designators don't really tell the type.
... the approach being used is fine. We don't need to send this comment.

Jonathan: We won't forward this comment
... (2) Should we allow fragment id and not just a WSDL 2.0 document?
... I think we already have in our spec a definition of WSDL 2.0 document can be a fragment of another document.
... (to Jacek) do we not need this comment?

Jacek: yes

Umit: Can't this be any WSDL document, not just WSDL 2.0

Jonathan: Can someone make sure our document refers to documents and fragments?
... (3) Any objections to talking about extension elements not just extensibility elements?

No objectsions

scribe: (4) SEND THIS ONE
... (5) does this add a property to the endpoint?

Jonathan: Will forward this comment.

Break for lunch

<GlenD> Scribe: GlenD

RDF Mapping Issues

Jacek: All current resolutions are already implemented in the current editor's draft

Issue 287, modularization of the ontology

Jacek: Looking for opinions. Bijan thinks this is not necessary.
... Most things in our ontology are in the wsdl-rdf namespace, except where URIs already exist
... If no opinions, I'll leave it as it is... feels a little ad-hoc in places though.

(discussion of the issue definition)

Jonathan: Issue is whether to make the RDF mapping namespaces individual as our spec namespaces are (wsdl-soap, wsdl, etc).

Tony: Are the hashes (#IN, #OUT, etc) related to actual anchors in the document?

Jacek: If the URI points to HTML, then yes, but they don't need to point to HTML. If they don't, then not really.

Glen: So OK, question is whether to create namespaces to match our document namespaces.

Jonathan: Seems reasonable.

Jacek: Suggest editors identify all outstanding URIs, all the places we should define an RDF doc (and try to minimize), and come back

Jonathan: Let's close with no action, go to LC and then see what the community thinks

<scribe> ACTION: Jacek to enumerate all URIs where we should provide RDF representations. [recorded in http://www.w3.org/2006/02/28-ws-desc-minutes.html#action01]

Hugo: how many docs are we talking about here?

Jonathan: that's what Jacek's action is... but it seems like maybe 4 or 5?

Issue 288, coordinating with XMLP WG

Jacek: We asked XMLP for a URI for the concept of SOAP MEPs. They have given us a URI. They will not provide an RDF doc at the end of this URI... this can be reconsidered later, but isn't too harmful.

Jonathan: Don't we need RDF there?

Jacek: RDF is very open-minded. If you don't tell it something, it doesn't care.
... We already say this URI is a class, and it's usable even without extra info (which might still appear later)
... suggest we close this issue with the new URI.

RESOLUTION: Issue 288 is closed, we will use the URI that XMLP provided for us.

Jacek goes through document.

(discussion of component designators being rather verbose)

Jacek: Note that we point to the actual feature/module and not the components.
... Don't support documented or extended feature components - doc should be in the feature itself, and noone is going to use extensions there

(discussion of documentation - it's part of the XML, not the component model)

RDF mapping does not have documentation either

Jacek: Why not just use XML comments?

Jonathan: Structure. Can have script, etc...

(XSLT wrangling ensues)

Jacek: XSLT gets published but not made normative.

Jonathan: Schedule?
... To repub this, we need to mostly just do the work that's already on your plate...

Jacek: Yes, but this isn't Last Call yet
... Completion of mapping tables, polishing, completing the example....
... End of March...?

Jonathan: Let's do heartbeat publication at the end of March with full set of content, still need polishing...

Jacek: Or we could publish next week a partial but much better draft, then do LC

Jonathan: Refresh pub (assertions, etc) soon, this might be good to do at the same time.

<scribe> ACTION: All editors to prepare drafts to consider for republication. Due 03/14. [recorded in http://www.w3.org/2006/02/28-ws-desc-minutes.html#action02]

Jonathan: We'll decide on publication at the call on the 16th

Meeting schedules, etc

Jonathan: No call next week, call on the 16th to decide publications. At that point we'll decide on what future meetings to have.
... No F2Fs currently scheduled.
... Interop events... Sri Lanka?

ADJOURN

Summary of Action Items

[NEW] ACTION: All editors to prepare drafts to consider for republication. Due 03/14. [recorded in http://www.w3.org/2006/02/28-ws-desc-minutes.html#action02]
[NEW] ACTION: Jacek to enumerate all URIs where we should provide RDF representations. [recorded in http://www.w3.org/2006/02/28-ws-desc-minutes.html#action01]
 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2006/02/28 15:42:28 $