W3C

XML Schema Patterns for Databinding WG F2F

22nd-23rd May 2006

Origo Services, Edinburgh, Scotland, UK

Agenda

Attendees

Present

Jon Calladine (BT)
George Cowe (Origo)
Paul Downey (BT)
Otu Ekanem (BT)
Paul Fremantle (WSO2) (Tues)
Yves Lafon (W3C)
Observers Present for part of meeting:
Mark Seaborne (Origo)
Regrets
Chair
pauld
Scribe
pauld

Contents


 

Administrivia

Chair: Given there are so few of us here and nobody participating remotely, scribing everything that is said is likely to become the primary focus of our attention. I want to have a more informal meeting to try and make progress. The minutes will therefore just capture any decisions we make.

Minutes from 9-May approved http://www.w3.org/2002/ws/databinding/6/5/09-databinding-minutes.html

Action Item Review

Chair: most of our action items surround practical testing of proposed patterns, most of which I've picked up or dumped on Otu:

How We Work

Patterns

Discussion of the risk a Basic pattern may be moved to Advanced subject to CR testing leads to ISSUE-61 being raised. General concern that there is always another toolkit.

Proposal is to mark all Basic patterns as being 'at risk'.

Design Considerations blur this boundary, If we use Design Considerations as warnings (ISSUE-35) we'll need some way of detecting them in a submitted schema. This might not be possible with a schema or XPath statements.

Why a submitted pattern wouldn't be included leads to ISSUE-62 being raised

Overall the WG has developed a good understanding of the deliverables:

Chair:I'm concerned at this point at the lack of patterns being submitted to the WG. Basic patterns may indeed be very Basic!

Issues List Review

Chair: so far we've drilled into each Issue in turn, depth first. I'd like to take a breadth-first review to see if there is a way to make progress

Open discussion on how we make progress - we've a large number of issues stalled on practical testing, no clear method of detecting patterns and our test suite is mostly missing. Finding a way of working, a "steel thread" will help us move forward.

ISSUE-51 - detecting name clashes is a particular challenge for pattern detection, particularly if combined with inheritance.

Origo XForms Demonstration

We're joined by Mark Seaborne of Origo and the XForms WG. Mark gives an excellent demonstration of an XForms application which may be used to validate an instance document against an Origo schema and business rules held as Schematron assertions. The generated XForms may also be used to interactively rectify faults in the input document.

The WG had been considering how to detect patterns in an example schema, either using a schema for schemas or XPath assertions. Based on this work, XPath still looks promising for our purposes. Although very interesting, the WG doesn't see an application for the XForms interaction at present, though this could be layered once we have the rules defined.

Patterns

Rules and Patterns

WG brainstorms on how to make progress. Our documents are a collection of patterns. Each pattern has:

pattern

Yves: Dan Connolly voiced an idea that it would be useful to be able to identify data structures generated from XML Schema using URIs.

WG moves towards transforming the pattern Id into a URI, e.g.:

http://www.w3.org/2002/ws/databinding/patterns/6/05/Collection

spec driven test cases

Detection of Patterns

For the detection of patterns - rules from spec compiled into a processor which generates list of pattern URIs contained in a schema.

Conformance testing looks for schema which only contains patterns.

detecting patterns

Echo Services

To test a databinding toolkit against patterns each example schema may be wrapped into a WSDL containing an 'echo' operation. Sending the example instance SOAP document may then be deserialized and reserialized and the infosets compared by a generic client. It may be possible to generically generate an echo operation in some environments, e.g. using Java or C# reflection.

echo service

Tasks

Tasks

  1. Template WSDL from XSDs wrapped into an 'echo' service
  2. Extract example schema documents (XSDs)from Specs
  3. Extract example XML instance documents (EGs) from Specs
  4. Wrap EGs into example SOAP 1.1 and SOAP 1.2 documents (Messages)
  5. Write Schematron Assertion rules for patterns
  6. Write Schematron Report rules for Design Considerations
  7. Build Schematron Schemas from specifications
  8. Test XSDs/WSDLs/SOAPs with toolkits
  9. Write Patterns for issues not yet in our specs

Work starts on building a "steel thread" using the 'all' example documents:

Yves generates extract.xml which Pauld uses to generate examplexsd.xsd, which George uses to test rules for pattern detection.

WSO2 Tungsten Demonstration

Paul Fremantle demonstrates databinding under the WSO2 Tungsten product. Axis2 supports multiple Java databinding implementations, including:

WSO2 may be able to help build an example 'echo' testpoint for our patterns using their framework.

Origo Patterns

George presents a set of patterns used within Origo schemas: http://lists.w3.org/Archives/Public/public-xsd-databinding/2006May/0046.html

This generates a number of issues:

Wrap-Up

To build upon the progress made during this meeting the Working Group arranges two days to work together on patterns, rules, test cases and practical testing 5-6th June. BT has a room in central London for the Monday, hookup via IRC and phone - no need for Zakim.

We'll have our usual telcon to address Issues raised. Hopefully we'll have evidence to resolve a bunch of them.

WG tries to arrange a F2F in June/July, which descends into a calendar conflict death match.

Chair: I'll arrange a room in London, hosted by BT, last week in July, or first week in August.

Chair: Special thanks to Origo Services for hosting and their great hospitality!

We're ADJOURNED!

Summary of Action Items

[NEW] ACTION-40 ylafon to create an URI id document for each pattern for ISSUE-52
[NEW] ACTION-41 pdowney to organize informal workshop 6-7 June in London
[NEW] ACTION-43 pdowney arrange F2F meeting in London end of July, early August

 
[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.127 (CVS log)
$Date: 2006/06/13 14:59:30 $