OGF DFDL WG Call 2007-10-31

Discussed: Separator semantics via rewriting - Alan Powell (prior to call) raised objections to MikeB's treatment of separator properties semantics in draft of nulls/optionals/defaults memo.

Mike agrees. This stuff is a mess. Needs to be reworked.

Action item: Mike to rework.

Discussed: Boxed data, i.e., sequences with dfdl:length specifications as means of specifying box size surrounding dfdl-described content.

Proposed to drop this from DFDL V1.0.

Action: examine important formats to see if this feature is needed. We know there are formats that use it (IMS blocks), but must they be supported in single schemas or is a two-layer via two schemas approach acceptable?

Resolved: not dropping this. New  simpler examples are needed which make this clearer than the example in the email/memo by Mike.

(Note: this is the whole section on valueCalc properties. This has already been identified as an area needing a write up.)

Discussed: does infoset include things like the representation length of items? This information is needed to express output for formats where one element stores the rep-length of another. Suman pointed out that this information doesn't need to be available in the infoset. The relationships between elements of many kinds cannot be expressed directly in terms of the infoset, so why require this representation-level relationship to be so expressed?

Resolved: infoset does not contain represnetation artifacts like representation length. Mike mentioned that it does contain number of occurrences for arrays. However, later discussion of infoset brings into doubt whether the infoset even knows about arrays.

Discussed: infoset

Metadata link? Suman has advocated for this. It's an optional thing. SCD paths are acceptable to most. Simon maintained that this shouldn't be part of the infoset but really is an implementation consideration.

UML model: Discussed that it should match XML infoset terminology and also the grammar terminology used in the spec. (note: or the grammar terminology be made to conform to the infoset - actually this is more likely) Simon's model has arrays as distinct from sequence groups. We discussed eliminating arrays as a concept and going to a neutral term that captures both sequence groups and arrays (things with children).

Action: Steve H. and Simon P. to work out remaining infoset issues. Provide UML diagram.


Mike Beckerle
STSM, Architect, Scalable Computing
IBM Software Group
Information Platform and Solutions
Westborough, MA 01581
direct: voice and FAX 508-599-7148
assistant: Pam Riordan  
                 priordan@us.ibm.com
                 508-599-7046