WSRP-WSIA Joint Interfaces/Metadata Conference Call

June 4, 2002

	William Cox	BEA
	Graeme Riddell	Bowstreet
	Srinivas Vadhri	Commerce One
Y	Monica Martin	Drake Certivo
Y	Alan Kropp	Epicentric (chair)
Y	Charles Wiecha	IBM
Y	Rich Thompson	IBM
Y	Carsten Leue	IBM
Y	Thomas Schaeck	IBM
Y	Rex Brooks	Individual
Y	Joe Rudnicki	Navy
Y	Mike Freedman	Oracle
	Stefan Beck	SAP
Y	Jeffrey C. Broberg	Silverstream
	Suresh Damodaran	Sterling Commerce
	Eilon Reshef	WebCollage
Y	Gil Tayar	WebCollage
	Steve Pruett	Silverstream
	Mike Hillerman	Peoplesoft
	Aditi Karandikar	France Telecom
Y	Ravi Konuru	IBM
Y	Howard Mellman
Y	Angel Diaz	IBM

0.44 Spec

• Bulk of the call was devoted to points of view regarding sessions and entities.
• Session refers to conversational state tied to the end-user through the consumer.

This is by definition transient data.

There is some debate about whether there’s always a 1:1 relationship between a session and a service runtime instance (or entity).

Servlet/Web App environments of today scope state at either the application level (global), or the client level (HTTPSession ID).

Web Service environment, and the portlet environment in particular, may introduce additional levels of scope, particularly around the Portlet Container. Since the consumer potentially aggregates multiple instances of a service (as identified by the persistent entity) in a single presentation or page, it is possible that multiple concurrent invocations occur, but within a single shared session. Two questions here:

1. Is it desirable to model shared session interactions directly in the protocol?

2. Is it possible, in a shared session, to generate the correct markup for the given user, based on where they are in their interaction with the service (e.g., what if there are multiple pages, and User X is on page 1 and User Y is on page 3. If a request arrives telling the service to grab the “next” page, how can the correct markup get rendered?).

The answer to #2 above is likely to be that the session is partitioned by namespace, probably the entity ID (end user ID?).

Action item: Continue this discussion by email. FYI, Eilon has made some interesting comments that are relevant to this topic.

• If we accept that sessions are orthogonal to entities, then producer-consumer interaction becomes a more complicated picture. More than just one interaction style possible.

Action item: Rich, Carsten and Alan will work on diagramming the interactions to make it easier to visualize.

• Support for implicit lifecycle of the producer: A Consumer may simply call getFragment() (a variant of this method, perhaps, since the one in the spec requires an entity handle) in the simplest case, and so the Producer may implicitly create the entity and possibly the session to support that interaction.

• Carsten and Rich have begun merging their respective work on the portlet and joint interfaces specifications. Seems to be a lot of overlap.

Action item: Circulate the draft of this document by the end of the week for comment.