WS Description Teleconference
30 Oct 2003


Present: Erik, David, Allen, Roberto, Glen, Paul, Youenn, Tom, Jacek, Lily, Jonathan, Ingo, Jeff, Dale, Jean-Jacques, Bijan, Igor, Jerry, William, Sanjiva, Umit, Prasad

Regrets: Dietmar, Steve, Sandeep, Philippe, Amelia, Arthur, Jeffrey

Chair: JMarsh

Scribe: Ingo


Scribe: Date: 30 Oct 2003

<dbooth> Agenda: http://lists.w3.org/Archives/Public/www-ws-desc/2003Oct/0254.html

3. Action Items

Scribe: Items 1 to 3 still pending. 4 and 6 will be at F2F.
... Sanjiva email to rational still pending
... Use case for cyclical includes still pending.
... Reformulation of return value RPC rule done.
... wording clarification for fault-replaces-messages pending.
... MTOM communication done.

<dbooth> ACTION: Umit to resend RPC rules, splitting rules into two buckets: conformance and hints

Scribe: Missing Action Item in last minutes (23 Oct).

<umit> http://lists.w3.org/Archives/Public/www-ws-desc/2003Oct/0057.html

4. Administrivia

<umit> This is the list that has the split. There are two buckets.

Scribe: Lily did not resign.
... Bridge requested for F2F.

4.2 Draft Agenda for F2F

<sanjiva> Jonathan: I just updated edtodo to record the action to clarify faults per the action items review; so please remove that pending action item.

<Marsh> swap Tues & Wed morning
... swap RPC and pattern inference

Scribe: Publication draft approval planned for Wed.
... Topic 5 Task Force Status skipped

6. New Issues

Scribe: 150 skipped for today.
... Suggestion on 177: have infault and outfault.
... 2 Proposals: Merge outfould and infault to just fault.

<dbooth> Straw poll question: Should we merge outfault and infault to fault, per Sanjiva's proposal?

Scribe: Sanjiva's proposal in the minority.
... Yes: Sanjiva, Dale, Youenn
... No: Erik, David, Allen, Roberto, Tom, Jacek, Lily, Ingo, Jeff, Jerry, Umit, Prasad
... Abstain: Glen, Paul, Jean-Jacques, Bijan, Igor, William
... It will not be merged.

7. Issue #88

Scribe: Issue will be closed.

8. Optional interface/operation.

Scribe: Proposal accepted: Mark as required

9. RPC-style rules

<umit> proposed two buckets:
... http://lists.w3.org/Archives/Public/www-ws-desc/2003Oct/0057.html
... accepted rules discussed last week:
... http://lists.w3.org/Archives/Public/www-ws-desc/2003Oct/0058.html

Scribe: Idee of Umit suggestion: Order of parameters is indicated by the order
... of the elements in schema.

<dbooth> Umit: "The order of the parameters is indicated by the order of the elements in the schema, given that input parameters will precede output parameters in the method signature."

<Tomj> Jack and Roberto are talking about changing direction.
... We currently have a set of rules that the RPC style implies about the Schema
... And from the rules, processors can generate signatures.
... They are talking about resurrecting something like ParameterOrder, which would explcitly define a signature
... for the RPC crowd.
... This would give tools an explicit hook to get the signature, instead of what we have now which gives the tools some rules/hints as to how
... to create a signature from the Schema
... My feelings: I am opposed to changing direction and support refining the current RPC style to get it right.

<sanjiva> If I don't understand the style=".../rpc" value for style, then I am not going to verify that the schemas are consistent with the semantics of the style.
... However, *IF* I do use that URI to generate stuff, then I MUST verify that the schemas are defined correctly w.r.t. the rules of the style.
... Going beyond that is unacceptable IMO.

<dbooth> Our spec allows someone to answer the question: Given a WSDL document w, is WSDL document w conformant to our spec or not?
... A given WSDL document w also contains message definitions. Our spec also allows someone to answer the question: Given a particular message m, is message m conformant to WSDL w or not?

<sanjiva> I disagree vehemently with what Umit just said. The style URIs do not hvae to be understood. That's why we designed the hints as hints.

<Roberto> +1 to dbooth's comment

Discussion: has a processor check the correctness of the WSDL?

<dbooth> I think it's a mistake to be talking about what a WSDL processor must do. Rather, we should talk about what it means for a WSDL document to be conformant to our spec. What any processor *does* with the information is it's own business.
... The word "required" is very misleading here. Again, we should be talking about *conformance* -- under what conditions should a WSDL document be conformant.

<sanjiva> Interoperability *IS* guaranteed via the schemas - not at the programming model level.

<dbooth> Again, we should not be required a WSDL processor to do ANYTHING. We should only be talking about what it means for a WSDL *document* to be conformant to our spec!

Scribe: People agreed that 56 and 57 are status quo. Questions about conformance.
... Discussion delayed to F2F.
... RPC item left open.
... Deal with syntax about parameter order.

Summary of Action Items

[NEW] ACTION: Umit to resend RPC rules, splitting rules into two buckets: conformance and hints

Minutes formatted by David Booth's perl script: http://dev.w3.org/cvsweb/~checkout~/2002/scribe/
$Date: 2003/10/11 19:07:03 $