W3C

MEP Task Force
2 Apr 2003

See also: IRC log

Attendees

Present: Gudge, Philippe, dbooth, Umit

Regrets:

Chair: dbooth

Scribe: Gudge

Contents


Scribe: The Agenda:

<dbooth> document structure

Scribe: use cases
... binding specifics

Document Structure

<dbooth> http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl12/meps-vs-iops/meps-vs-iops_clean.htm

Scribe: dbooth asks whether people ( specifically Umit and Gudge ) are comfortable with the structure of the document whose URI is above
... Umit is comfortable
... Gudge is comfortable

Use Cases

Scribe: Gudge observes that use-cases might subsume Amy's action to collect description for given patterns
... dbooth intent here is to accentuate the diff between IOP vs MEP approaches

dbooth: Central question is 'should interface/porttype be modelling IOPs or MEPs' ?
... If modelling IOPs then additional info that MEPs provide would be specified in bindings

plh: Question: Does not say, who is supposed to use the abstract i/f
... and therefore we don't know what it is supposed to represent
... differing views on what the relationships are

umit: is what we're trying to decide whether IOPs lose info? and how hard it is to regenerate that info?

dbooth: different people have had different understanding of what was being described at the abstract level
... some people thought IOP, some MEP
... MEP contains additional info
... my understanding is that if we model porttype level as IOPs then additional info would go in the binding

umit: Specific info will be somewhere in the WSDL ( either at porttype or binding )

dbooth: could also be external to WSDL

umit: if we think about abstract vs concrete people may not care about bindings ( may only be describing at porttype level )

dbooth: i think that some people think IOPs are enough
... others would argue that more that IOPs ( MEPs ) is needed
... want to collect use cases to show why IOPs are, or are not, better than MEPs

Scribe: looking at UC1 in http://dev.w3.org/cvsweb/~checkout~/2002/ws/desc/wsdl12/meps-vs-iops/meps-vs-iops_clean.htm

dbooth: wants to ensure that interactions with services are uniform

umit: I would say it is different from the perspective of the programmer or service provider
... would know that requests come in and response go back to the requester

<dbooth> The BigCo Client App programmer writes the following code:
... Services services = lookupServices(...);
... Price bestPrice = infinity;
... foreach s in services
... {
... Price p = s.getPartPrice(); // Request-response pattern.
... // s will handle transport details depending on binding style.
... if p < bestPrice { bestPrice = p }
... }
... print "Best price from all suppliers: " bestPrice

dbooth: codegen could generate the code (the generic parts)

Summary of Action Items


Minutes formatted by David Booth's perl script: http://dev.w3.org/cvsweb/~checkout~/2002/scribe/
$Date: 2002/02/19 16:35:31 $