One correction in the revised minutes below: some duplicate text removed from para 2.1. With thanks to Simon for spotting the error.

Ian




Open Grid Forum: Data Format Description Language Working Group

Weekly Working Group Conference Call
17:00 GMT, 7 Nov 2007

Attendees
Mike Beckerle (IBM)
Geoff Judd (IBM)
Steve Hanson (IBM)
Suman Kalia (IBM)
Simon Parker (PolarLake)
Ian Parkinson (IBM)


1. Introduction
The previous meeting left two areas for consideration: the Information Set (InfoSet) proposal, and the work regarding defaulting and nulls. This meeting would cover a review of a revised InfoSet proposal. The defaults/nulls issue has not been progressed.

2. Review of Revised Information Set Proposal
Steve and Simon have collaborated on a revision, v6, of the InfoSet proposal. This has been circulated via the DFDL-WG mailing list.

The meeting reviewed the v6 proposal, highlighting the following points:
1.        There remain two types of information item: Element Information Items and Document Information Items. There are now two categories of Element Information Item: simple, and complex.
2.        The description of Document Information Items incorrectly lists a "root" member. This should be replaced by a "children" member. Geoff pointed out that the associated UML diagram suggests that a Document Information Item may contain multiple children; this will be restricted in the text.
3.        The "dfdl version" member is now used to describe the version of DFDL to which the InfoSet conforms; it does not describe the version of the parser generating the InfoSet. In particular, for an InfoSet to inlcude a dfdl version identifier, there is now no requirement for that InfoSet to have been generated by a DFDL parser.
4.        Mike queried the meaning of "absolute Schema Component Descriptor"; it should identify a schema as a whole. This is a term defined by the SCD specification, and an appropriate citation should be added to the proposal.
5.        The meeting discussed, at length, the "data value" member of Element Information Item, which the proposal lists as being of type String. Mike pointed out that this is insufficient to distinguish between some binary floating-point values, and suggested that it should be of a type defined by the "data type" member. Suman suggested this would also make for a more natural API. Simon remarked on the precedent set by XML Schema where all values are represented as strings. Mike closed the discussion advocating dropping the word "String" from the definition of the "data value" member.
6.        Suman queried the purpose of the document information item. This is an in-memory artifact only, which acts as a container of the InfoSet's data, and has no representation in a serialization of the InfoSet. This should be clarified in the proposal.
7.        The proposal now contains a UML diagram illustrating the object model. The relationship between Parent and Element requires some clarifying: it should be modelled as a composite containment relationship.
8.        Simon highlighted that the "schema" member of Element Information Item may be either a relative or an absolute SCD; while the "schema" member of Document Information Item must be an absolute SCD.
9.        The current version makes no mention of arrays, preferring to treat arrays as repeated children with the same name. The meeting consdered how to denote arrays in the InfoSet, but opted instead to add a note regarding arrays to the specification.

Mike will circulate his notes regarding the agreed changes to the spec. Simon will circulate a new version, v7, incorporating further comments.

3. Other Business
There was no other business.

Meeting closed




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