Service model Telecon, August 2, 2001
Updated: August 30, 2001 10:47 - Page Maintained by Allan Doyle

Agenda in blue, discussion during telecon in black, action items in red

Present

Peter Vretanos
Lou Reich
Archie Warnock
Allan Doyle
Mike Adair
Larry Bouzane
Dave O'Mahony
Jeff Lansing
Yonsook Enloe
George Percivall
John Davidson
Ananth Rao
John Evans
Arliss Whiteside

Agenda

1. Review of key happenings, actions, etc.

• Allan -- post of revised term definitions revised by Allan, Peter, & John
  -- any terms flagged for issues?

Action: Aug 4. Allan to finish terms - please help by sending things to Allan.

• Peter - any conclusions from analysis of Jeff Lansing's email on WSDL?

Jeff likes how far Peter has gotten. Peter is going to put up a web site with a WSDL WFS (see next item)

• Peter will send out a URL of a WFS that uses WSDL integrated with Capabilities.

Peter's stuck until his ISP gives him a static IP address. Should happen soon. See item 4 for more info.

• *Lou to post SOAP document via Allan. Lou to resend (or Compusult) [Here it is (.doc, .pdf)]
  

Based on the email to the list, I think these are the distinct threads:

• Terminology

  It seems that we are converging on these definitions:

  - Service: distinct part of the functionality that is provided
         by an entity through interfaces.
         - Interface: named set of operations that characterize the behavior 
         of an entity.
         - Operation: specification of a transformation or query that an 
         object may be called to execute. It has a name and a list of 
         parameters.

  To get some understanding I tried to come up with some
         example services built upon re-useable interfaces.

  George P.

       Service Interface    Functionality of Service
       -------------------   ---------------------   ---------------
       Web Feature Service  Web Feature Interface    Access to features

       Web Catalog Service  Web Feature Interface    Access to metadata

       Gazetteer            Web Feature Interface    Access to related geo terms

       Service Interface    Functionality of Service
       -------------------  --------------------- ---------------
       Web Coverage Service Web Coverage Interface Access to coverages

   Web Sensor Service   Web Coverage Interface Access to sensor

   Web Image service    Web Coverage Interface Image exploitation access

 

• Service Chaining
  • Executive vs. self organizing
  • WSFL, XLANG, www.wfmc.org, DAML-S
• Catalog Paper and Revision
• Use Case Review
• Service Hierarchy
  • Taxonomy
  • Inheritance
  • "Sub functions"
• WSDL experiments

2. Use Case Review telecon results -- yonsook
telecon notes on servicemodel website
Telecon Conclusions :

• Data & services should be presented to user in a bundle of information that can be discovered via a thematic or parameter keyword (e.g. "forest fires")
  • descriptions of data already created
  • descriptions of information that can be created based on user inputs (service invocations)
• Expecting a user to go through a long sequence of steps to get the information desired won't work Shorten the sequence of steps user needs go thru to get information desired (minimize user interactions)
  • If there is a sequence of steps that user must go through to get the information desired, that sequence of steps must be documented in the information description in the bundle.
• Have capability to save the sequence of steps that user needs to go thru that she can invoke again at a later date.

Will get additional use cases:
Ko Fung - for disaster management use case
Skip Reber/Ken McDonald -- atmospheric science user use case

Does the thematic keywording mean that each data item should have keywords attached that get you to services? Not sure. Scientists' point of view is very data centric.

Think of data availability instead of the data itself.

19115 can accomodate the links to service providers.

Terminology issue - in the telecon minutes there was talk of "information products" that may be what we consider "derived products".

We probably face an educational issue.

3. Status report service type hierarchies - Serge/Peter

How does WCS fit into the type hierarchy (John Evans)

next steps for service type hierarchies/object modeling

We have not seen a real start of this picture yet. Lou, John D., Peter, John E. Arliss, willing to participate in a discussion about this. A key issue is "sub function" as we are trying to add Filter to the WMS request.

The terminology has to allow us to express sending a single request to a set of operations that return a single result. We have to figure out how to name the things that we're sending parameters to. A WMS GetMap really is directing some parameters at a "filter" and some at a "render". E.g. BBOX would be directed to the filter, STYLE would be directed to the render.

This kind of discussion leads towards interpreted languages. Consider the eval/apply loop in Lisp. If you find you can't apply something, you eval it, then try to apply again...

Chaining can happen with or without the executive.

Compusult's use of a "layer service" concept seems to contain some insight.

4. WSDL Work Item status

• Integration of WSDL and GetCapabilities - Peter
• WSDL instantiation of WFS -- Peter
• WSDL instantiation of WMS -- Jeff dlb & Ananth

Check email thread of Jeff and Ananth - this is underway

• WSDL instantiation of WCS -- John Evans
• What works? What's easy & what's hard?
• What does WSDL buy us?

GNOME.org has a WSDL interpreter. Look for SOUP. Inside the SOUP tarball, there's a WSDL parser.

WSDL document is strongest on SOAP, weakest on GET

Next step - validate and work out a GET binding.

Next step - use XML Schema to play with inheritance for messages/operations/capabilities - would be nice to come up with some standard fragments that we can reuse/inherit from.

5. Catalogs

• Results from Jeff dlb's discussion about summary/full responses - Jeff
• Comments of Stateless Catalog Profile doc - Lou
• any key issues?
• please send email summary before telecon

Peter to become co-editor of the Catalog paper.

Arliss' email of 2 Aug 2001 09:37:58 to be discussed. Lou to post a synopsis of this via reply to Arliss' email.

Point 13 - BSM work seems to have stopped. Actually, WMS 1.1.0 has a "separable" section that contains BSM kinds of content. It was decided that in order to procedurally move ahead with WMS 1.1.0 we had to keep the material inside the WMS document rather than rely on simultaneously approving a new spec. Perhaps this material should get moved into a "Common Elements for the Web Profiles" spec.

Archie also did a very fine-grained, detailed analysis of the document. One item is the Update Sequence element. We need to play with this at the service model level. DNS was originally used as a discussion point - TTL, Version #, etc. are used there to tell DNS servers how long they can cached values.

Keep sending comments to Lou about this paper!

6. Next steps

Think about what we are doing --

• step 1 - integration of WFS/WMS/WCS using WSDL?
  integrated GetCapabilities - this is a big item. WSDL does not supply this for us.
• step 2 - integrate data and services ?
  how to dynamically bundle data and services into an information bundle
• step 3 - how to add ad-hoc services (services other than WMS/WCS/WFS)
  • We can try to accomodate these via entries in the service catalog.
  • This is currently out of scope (at least being able to invoke them).
• we're taking a look at various elements that will contribute to a service model
  • what are we not covering? - the real question is what do we have?
  • come up with list of uncovered stuff to track as future work/issues
  • system metadata
  • user metadata
  • semantics
• Chaining

Jeff Lansing wants to put together a concrete chaining use case. He's trying to coodinate with the process coordination community to see whether they have anything that will help us. Contributions are welcome and encouraged.

Note that our case is different from supply chains in the business world. We care what's "upstream". i.e. we care about things like which WFS gets queried before the results go to the WMS.

Peter will write something up and send it...

7. Schedule next telecon

- August 16 1100-1300 EDT (1500-1700 GMT)