[wsia] F2F 11/7 Minutes (Website version)

Thursday, 11/7/02
Roll
Voting Members Company
Stephen Drye (on leave)

Art Technology Group

William Cox

BEA

y
Adrian Fletcher

BEA

Gino Filicetti

Bowstreet

Steven Smith

Capitol College

y
Andre Kramer

Citrix

y
Monica Martin

Drake Certivo

Raj Ramesh

CommerceOne

y
Timothy N. Jones

CrossWeave

Alan Kropp

Epicentric

y
Nigel Ratcliffe

Factiva

Madoka Mitsuoka (on leave)

Fujitsu

Sunit Randhawa

Fujitsu

y
Richard Cieply

IBM

y
Carsten Leue

IBM

y
Thomas Schaeck, chair

IBM

y
Rich Thompson

IBM

y
Charles Wiecha

IBM

y
Eric van Lydegraf

Kinzan

y
Jon Klein

Reed-Elsivier

Adam Nolen

Reed-Elsivier

y
Petr Palas

Moravia IT

Olin Atkinson

Novell

Chris Braun

Novell

y
T.J. Cox

Novell

y
Michael Freedman

Oracle

y
Mike Hillerman

Peoplesoft

Art Machado

Peoplesoft

Ken Pugsley

Peoplesoft

Sasha Aickin

Plumtree

y
Jane Dynin

Plumtree

Joseph Stanko

Plumtree

Michael Young

Plumtree

Amir Blich

SAP

Gennady Shumaker

SAP

Yossi Tamari

SAP

y
Rex Brooks

Starbourne

y
Brian Dirking

Stellent

Alejandro Abdelnur

Sun Microsystems

y
Dave Clegg

Sybase

Joe Rudnicki

U.S. Navy

y
Eilon Reshef (on leave)

WebCollage

Gil Tayar

WebCollage

y
Prospective Members (non-voting)
Dan Machak

Tibco

WSIA Members
(non-voting)
Graeme Riddell

Bowstreet

Bruce Lucas

IBM

y
Ravi Konuru

IBM

Dan Gisolfi

IBM

Rich: We resolved 31 issues yesterday�unfortunately most of them were the easier ones.
Mike F: Make sure everyone realizes that without groupID, must rely on transport level mechanisms (or extension) to support shared sessions.
#140 Internationalization
Inline vs. separate resource. Uses xml:lang, so we're not inventing a new localization mechanism.
Sasha: Set of property descriptions could change?
Rich: Yes.
Sasha: So localization resource would have to be dynamic as well, synchronization issue.
Eilon: Also since description can be filtered, localization resource would have to be similarly filtered.
Rex: Clear performance difference between the two?
Sasha: Yes, although dependent on network considerations�so it's probably a wash.
Rich: One implication of choosing separate structures is a new operation to get localization.
Charlie: "factored" inline is a 3rd approach, which includes localization resource in the same document, without localization URI.
Vote: Inline: 17 Separate: 3 Abstain: 2
Rich: 2nd question: format of the inline for tooling purposes. Won't accept xml:lang with a type attribute. So we need indirection through resources.
Vote: Indirection: Yes: 12 No: 7 Abstain: 4
#149 Should get/setProperty take locales?
Mike F: This opens the question of a user being able to have different preferences based on locale.
Gil: Oppose this layering of properties by locale.
Bruce: Haven't seen this in anyone's notion of properties.
Vote: Yes: 4 No: 14 Abstain: 2
Rich: In the get__Description operations, must take a desiredLocales[] and sendAllLocales.
Andre: Why doesn't user profile carry the user's language?
Rich: The request may not be on behalf of this user.
Eilon: Is there a default Consumer locale?
Rich: This could be the zero-th element of the array (spec should reflect that semantic).
#100 Entity property representation
Keep the reset attribute in the Property list?
Gil: The list has multiple usages, and this attribute wouldn't always make sense. If we want this capability, make it a flag or a separate operation.
Rich: Could create a <ResetProperty> sub-element of PropertyList.
Mike F: Why not a formal method?
Rich: An Edit could comprise several sets and resets, and this gets submitted as a single operation.
Eilon: Restructure PropertyDescription to make some of the attributes into sub-elements, to enhance relatedness of some of them.
#147 MarkupType in MarkupParams should be an array
Eilon: Supported markups for an entity defined?
Rich: In metadata.
Bruce: Like an http accept header?
Rich: Yes.
Vote: Make both MarkupType and Locale are arrays, and the order is by preference. Y: 13 N: 4 A: 3

Sasha: Do we allow "*" like in the accept header? (Vote: Y: 9 N: 3 A: 9) Is it a MUST for the Producer to return a markup that's in this list?
Mike F: Should leave it as mime types right now. Can always extend.
Rich: Doesn't hurt, and it carries forward well-understood mechanism (accept header).
MarkupResponse also needs a markupType and locale.
#144 define bindings for SOAP attachments (#148)
Rich: Benefit is that portions of what is transferred can use different encoding.
Andre: Are we mandating use of attachments if content is sent in a mime type that isn't in the MarkupType array from issue 147?
Mike F: No, not mandating.
May need to hold on this pending standards maturity (SOAP and DIME).
Mike F: No, we should specify bindings for both, and neither spec appears unlikely to proceed. This is very important for our (Oracle's) implementation.
#59 Interface discovery
Mike F: Don't understand the wsdl field at the entity level. Why wouldn't this be an extension?
Carsten: Interface discovery at the entity level�what portTypes are available at the entity level?
Mike F: This is not true for the WSRP portTypes, that can only be discovered at the Producer level. So for an entity that exposes custom interfaces, that should be information that's available in extensions.
Gil: This isn't a WSRP discovery mechanism. It's a SOAP mechanism.
EntityDescription: Drop: Y: 10 N: 2 A: 8
ServiceDescription: Drop: Y: 9 N: 9 A: 3
#85 refHandle/expires MUST/MAY contradiction
Consumer MUST return refHandle when Producer sends one, except in a couple of circumstances.
Gil: Should reword to explain what happens when the Consumer throws away a refHandle.
Mike F: Would be happy with same wording we use for registration. I.e. the Consumer is free to ignore the refHandle, as it is free to ignore a Producer (i.e. never deregisters, or attempts to unsuccessfully). It's important for Producers to be aware that this can occur.
Rich: releaseRefHandle may impact the wording.
#74 Should the Consumer MAY or MUST destroyEntities?
Mike F: Basic question was when the Consumer may consider that a release operation has been successful�i.e. must receive indication of success.
Sasha: There are cases that the operation will never succeed�catastrophic failure, fire, etc.
Rich: The requirement is for the Consumer to attempt to release a handle.
Mike F: Without the burden of going on to specify # of retries, and how long in between, etc.
#57 Unification of session and entity handles (#142)
refHandle.
Refined from 0.7 to 0.8 as entityStateChange proposal evolved.
Mike F: Now that cloneEntity no longer takes a refHandle, it isn't an issue.
Yossi: Combining the handles is a problem in terms of length restrictions.
Sasha: And it's complicated the spec.
Charlie: But they truly are correlated, so it makes sense for them to be combined.
Rich: Related question, that the restriction on handles doesn't need to apply to refHandle as a runtime.
Vote: Keep refHandle: Yes: 10 No: 8 Abstain: 2
Since this (and a couple of other) votes are so close, then we should open an email vote on just these issues.
Gil: There is a problem with the combination of 256 as the max length of a handle, and a vote in favor of keeping the combined handle.
Rich: The choice must be 1. sessionId < 256/split: 11 2. 4k, unified: 7 Abstain: 3
#141 IPR Statements
WebCollage will publish licensing terms.
IBM's statement pending, will probably be similar to ebXML, CPPA.
#145 Should sample implementation only implement JSR 168
Markup encoding vs. message encoding
Attachments (or new MarkupResponse field?)
Sasha: Either an attachment or base64 should solve this.
Should fault codes use our types namespace and drop the leading WSRP?
Yes.
List fault codes with operations
Yes.
Is Extension[] needed everywhere?
Specifically, includes nested structures.
Other metadata?
What does xxxx in xxxx- mean?
Gil: The basic issue is, what is the name of the final spec? This is a joint spec. 'wsrp' is too specific to portals/portlets.
Charlie: One obvious answer: wsrp.

Sasha: That it's a joint spec is unimportant.
Bill: Anything other than 'wsrp' as the choice doesn't reflect the group that created it.
Bruce: Should change the 'P' in WSRP to 'Portlets'?
Adopted
Re-vote on CSS style prefix now that we've selected WSRP: portlet-: 14 fragment-: 4 Abstain: 4
#66 Review interfaces
#104 profileKey
Mike F: Impact of SAML on user identity
Andre: Can we strengthen the "authenticated" aspect of profileKey?
Gil: That's too strong for the spec. Typical usage will be that this is an authenticated key, but that's not always the case. And even if it is authenticated, the spec can't say the Producer can rely absolutely on the Consumer authentication.
Rewording in the UserContext section.
Done with issues!

Next Face-to-Face
SAP, Palo Alto, will host.
Timeline
November 22nd Draft 0.9
December 16th TC provides all editorial comments
Last additional technical issues raised by exception
January 7th Finish draft 0.91 reflecting editorial comments.
Put Draft 0.91 on WSRP web site for public comment
January 27th - 30th Face-to-Face Palo Alto
Finalize 1.0 Specification. Discuss 2.0 direction/concepts and compliance test kit, time permitting.
January 31st Release 1.0
??? OASIS Standard (~March)
WSDL Testing
Rich, Andre, David (Oracle), Richard
Conformance document
Defines requirements for a Consumer and Producer to be in compliance (Gil, lead)
JSR 168 Implementation guide
Thomas: Also an implementation guide for mapping WSRP to JSR 168 (Mike F., lead)
Adjourn

wsia message