Open Grid Forum: Data Format Description Language Working Group

Weekly Working Group Conference Call
17:00 GMT, 06 Feb 2008

Attendees
Steve Hanson (IBM)
Geoff Judd (IBM)
Simon Parker (PolarLake)
Ian Parkinson (IBM)
Alan Powell (IBM)

Apologies
Mike Beckerle (Oco)
Suman Kalia (IBM)

1. Review of the draft DFDL Schema Abstract Model

Simon has distributed, via the DFDL working group mailing list, a draft of the UML description of the DFDL schema components, with diagrams built from scratch but using concepts from a number of different, public, UML descriptions of XML schema. The meeting reviewed and discussed this draft.

'Particle' diagram

Steve noted that the diagram has changed some of the object names. Simon agreed that he expected some of these to revert to standard XML nomenclature - but that some concepts don't have an existing naming convention.
The diagram allows for "all" groups - which are not supported in DFDL. This is as specified in section 5.1 of the present DFDL specification draft.
The diagram shows "Group" as a subclass of "GroupDefinition". The XML schema specification does appear to allow this, but the meeting agreed this should be removed from the diagram.
Steve and Geoff suggested that GlobalElement should be modelled as a subclass of ElementDeclaration.

'Type' diagram

The diagram distinguishes between GlobalType and TypeDefinition, to allow GlobalTypes to be named while TypeDefinitions remain anonymous.
The 'base' relationship should be between TypeReference and SimpleType, not between TypeReference and TypeDefinition. Complex type inheritence is not allowed by DFDL.
After consulting the schema specification, Steve withdrew his suggestion that ComplexType should contain a ModelGroup rather than a Particle. Simon pointed out that XML Schema 1.1 will allow particles to take annotations, which may prove useful for DFDL.

'DFDL Annotations' diagram

Hidden elements are wrapped inside DFDL annotations. This allows for a DFDL schema to be mapped to an XSD schema by stripping all DFDL annotations; the resulting XSD schema describing a DFDL InfoSet for the DFDL schema. This is a goal of DFDL.

Alan asked about the purpose of these digrams. They are intended to model and communicate the valid subset of XML schema, and also show where DFDL annotations can be placed in a DFDL schema. We could also use these diagrams to model which XSD objects each DFDL property affects, as well as showing where the properties may be defined - however Simon felt that as a property is not itself a DFDL annotation, this information is better documented elsewhere.

Steve felt that the diagrams included in this draft are overcomplicated, and wondered if we could reduce the number of boxes in the diagrams, for example by removing the distinction between named and anonymous objects. Simon observed that this is an important distinction for DFDL as, for example, named types may be reused in different contexts, and this distinction allows us to describe how properties apply differently to named and anonymous types. With this in mind, the diagrams are as simple as possible.

2. Other business

Alan will incorporate the UML diagrams, along with the other planned specification updates, into a new draft of the specification to be distributed on the 7th Feb.

Meeting closed, 18:00 GMT

Ian Parkinson
WebSphere ESB Development
Mail Point 211, Hursley Park, Hursley, Winchester, SO21 2JN, UK

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU