9-10 February 2005 F2F draft Notes for Wednesday morning

["Tony Fletcher" <tony_fletcher@btopenworld.com>]

Title: Message

Dear Colleagues - but especially Martin and Eric,

 

Please find attached draft minutes for the Wednesday morning session.  I have also included a text version as that is what everyone else seem to have downgraded to.

 

Best Regards,

Tony                          

	Tony Fletcher Technical Advisor Choreology Ltd. 68, Lombard Street, London EC3V 9L J   UK
	Phone:	+44 (0) 1473 729537
	Mobile:	+44 (0) 7801 948219
	Fax:	+44 (0) 870 7390077
	Web:	www.choreology.com
Cohesions™
Business transaction management software for application coordination
Work: tony.fletcher@choreology.com
Home: amfletcher@iee.org

 

OASIS WS-CAF  Technical Committee Face-to-Face meeting
 9-10 February 2005

Present:
Mark Little & Kevin Connor (Arjuna), Tony Fletcher (Choreology), Eric Newcomer (IONA), Martin Chapman & Greg Pavlik (Oracle), Doug Bunting (SUN)

Scribes: Tony Fletcher, Choreology, Kevin Connor (Arjuna), Martin Chapman (Oracle) & Doug Bumting (SUN).

Welcome and introductions
The meeting was hosted by Arjuna and Codeworks in Newcastle in the North East of England.

Martin and Eric shared the chairmanship duties.

Roll Call
It was agreed to make Kevin a full member of the TC as he has attended meetings and the interop, although he would only have met the specific defined criteria at the end of this meeting.

Initially the situation was thought to be that we had seven members present out of fourteen total voting members so the meeting was not quite quorate and needed people to dial in.  Subsequent investigation discovered that Jeff Mischkinsky should not be a member having been warned and not then having attended meetings and so there are thirteen members of the TC, and so the meeting was deemed to be quorate.

Appointment of scribes
Tony agreed to scribe for a Wednesday morning, Kevin for the afternoon, Greg for Thursday morning and Doug for Thursday afternoon.  (Note unfortunately both Greg and his daughter were taken ill and so he had to return home ealy and Martin took over his scribe duties.)

Agenda review and additions
Tony suggested taking Context issues earlier than Thursday afternoon.  Agreed to move up to Wednesday morning.

Approval of outstanding minutes
Approval of outstanding minutes for 31 January 2005.  Mark motioned and Greg seconded, and the minutes for 31 January 2005 were accepted with no objection.

Review of Actions
There were no new Actions from the 31 January 2005 meeting.

General and context issues
General
Issue 124  Shall we include in the WSDL a binding for synchronous interaction style?
Martin : leave as is for the moment and wait for WSDL 2.  WSDL 1.1 has portTypes and ports and in and out messages.  Current 1.1 defined bindings (though these are not normative ) say in and out are on the same connection for HTTP and silent on the topic for other bindings.  This is not reliable if there is a long time between the request and response.  Do not want to invent addressing mechanisms to achieve good reliable performance.  Mark motioned and Kevin seconded to close this issue with no action.
It was noted that one can add one's own WSDL to any of the CAF specifications.

The motion was agreed without objection.

Action: Martin / Mark to write up a response to this issue.
Action: Martin / Mark to raise another issue for the primer on the use of WSDL.
	
Issue 128  Some elements in the schemas are unnecessarily extended
Mark motioned and Doug seconded to close this issue with no change, but with a request for Simeon to come back with individual, specific examples.

Issue 143  OASIS namespace
Change namespace for the context CD.  Yes, but need to observe the new OASIS TAB guidelines and the namespace has already been changed in the context CD.
Mark motioned and Greg seconded to close this issue with no action as it is an old issue.  This was agreed without objection.

Issue 224  Mandate WSDL and SOAP?
Related to conformance.  Mark motioned to say that if an implementation supports context / one ways then it must do according to the specification.  Kevin seconded (i.e. optional to implement, but specification is normative if you do implement).  Amendment from Doug: a conformant implementation must implement one ways according to the specification.  Tony suggested putting conformance for the CAF stack in the highest level referencing specification in the CAF family.  Eric said no as will not always have a referencing specification.

Greg: there are different conformance points in the specification. For instance may just comply with the context structure only, or with the structure and by the reference mechanism.  Greg suggested a functional unit approach.  Martin agreed with Tony that high-level specifications can/should add conformance statements for what they used, but also felt each should be usable stand-alone as Eric said.  Doug suggested having a consistent statement on WSDL in all the CAF specifications.

Tony said that he supported what Greg said as well as what he had expressed.

Motion: if you use the facility then must do according to WSDL and text in the specification. 
Agreed unanimously.

It was agreed that people can and should review the resulting text in WS-Context and raise new specific issues if necessary.

Issue 225  Addressing scheme
Use WS-Addressing rather than the open content model?
Martin reported that WS-Addressing has been changed at the last face-to-face and there is no specification to reflect this yet.  Also should allow just use of URLs.  No solid text to consider.  Leave issue open and revisit once WS-Addressing is at Candidate Recommendation or later. 
Tony motioned and Mark seconded to leave open and revisit once WS-Addressing is at CR stage. 

Context 
Issue 209  unknown-context-fault/context-identifier
Why is unknown-context-fault / context-identifier an anyURI, and not contextIdentifierType.  One reason is contextIdentifierType is basically an anyURI though it does have wsu:id added. 
Tony motioned and Martin seconded that we agree to change to contextIdentifierType.  Agreed unanimously 

Issue 214  Correlation ID in AssertionType should be removed
Correlation is now deferred to the addressing scheme and therefore we do not need this parameter any more.  Also need to add text to context to describe this parameter and its use, if we retain.  See note and figure 4 in WS-Context.  Doug suggested making the note mention correlation and removing the word 'handled' plus removing the parameter.  Doug proposed this and Mark seconded and it was passed unanimously.

Issue 218  Remove AssertionType and AssertionWithProtocolURIType
Tony motioned that we remove assertionWithProtocolType and all its contents. 
Doug said that we do need to add text to WS-Context on how messages and contexts are extended. 
Action: editors to add text on extensibility to WS-Context (and WS-CF as well?).
Tony motioned and Mark seconded to remove assertionWithProtocolType, amend the text as Doug suggested, and to add the note on extensibility.  Agreed unanimously 
Action: Doug to raise an issue on adding attribute extensibility, or adding in must understand attribute. 

Issue 219  Remove wsu import and use XML Id instead - or justify
It was explained that wsu:Id is given special meaning in WS-Security and tools will look for wsu: namespace.  WS-Context section 6 paragraph 3 does mention wsu:Id explicitly. 
Mark proposed and Greg seconded to close this issue with no action and this was agreed. 
Action: Tony to look up to use of wsu:Id in WS-Security to check that we have taken the correct action here. 

Issue 220 Describe mandatory services first, as a general principle.
Major editorial.
It was pointed out that in fact both services are optional and the feeling is that the current order reads better than it would be if we were to change the order around.
Mark motioned and Greg seconded that we close this issue with editorial change and this was agreed. 
Remove the optional statement in section 4, first paragraph, as this implies that this service is optional and the lack of such a statement against the other service implies that it might not be optional.  The conformance section makes both optional.
Action: editors to adjust the text of section 4.

Issue 221 Parameters and Context on BEGIN and BEGUN?
It is stated in 5.2 that context may be present - does not say how, but implies in header if using SOAP.  This would be spelt out in the binding. 
Add a table that states which messages must have a context with them and which ones may optionally have a context with them.

Doug pointed out that it was his understanding that there could be several contexts with BEGIN and the service picks the one that corresponds to the type in BEGIN and builds a new context for BEGUN with the selected one.  This raises an issue of should BEGUN echo all the other BEGIN contexts?

There is a related issue as to why 'get' and 'set' have context explicitly embedded in the message.

Current text does not say a context within a context has to be of same type.  It was agreed to log a new issue on multiple contexts on one message, and also to log another new issue as to why 'get' and 'set' have embedded contexts.

State when a context is required and when optional on each operation and clean up the text on types of contexts. 
Tony motioned and Mark seconded this proposal and it was passed unanimously.

Issue 222  Reference to the schema from the text
Martin motioned that the final version of each of the specifications include schema and WSDL as well as a reference to external versions of both the schema and WSDL, and include a statement in the specification that the schema and WSDL in the specification take precedence.  This was seconded by Tony.

Doug agreed that there should be an accurate reference to the schema and WSDL external documents and also with the inclusion of a non normative Appendix, but disagreed with making the machine readable material in a non machine readable specification take precedence over a form that was directly machine readable. 
He amended the motion to make the precedence lie with the referenced schema and WSDL files.  Also need to make the statement that the fragments in the specification do not take precedence over referenced schema and WSDL.

Greg said that next time we should add an info set version. 
Action: Greg to provide information / a proposal on adding an info set version of the schema. 

It was agreed to make a precedence statement and have a clear reference to the specific WSDL and schema files and add an informative Appendix into specification with the WSDL and schema.  It was also noted that the specification text takes precedence over the WSDL and schema files. 
This motion was agreed unanimously 

Issue 223  Consistency of specification approach
Accepted as an editorial to use a consistent style. 
Action: editors to make this so.
Action: Tony to ask Peter to update his table and resubmit (actually Tony will take on this task himself.). 


Need to triages the list of issues submitted by Anish (Arjuna) and put them into Bugzilla, and then triages and address those that are substantive on the calls.
Action: editors to take a first pass and enter the resulting issues from Anish into Bugzilla.

2005_02_9-10_WS-CAF_F2F_Notes_Wed_am.doc