Notes from DFDL call 2005-11-23 by Mike Beckerle


Review of trackers

Arrays - waiting on extensibility

Scoping - not blocked. Design team discussions underway. Shooting for design-team resolution in 7 to 14 days then to WG.

Layers - waiting on (or part of?) extensibility

Null & Default - not blocked. Internal design-team discussions/drafts ongoing.

Property List - not blocked. Suman has done some work recently. Dependent on organization of properties which depends on scoping.

Open Content & Wildcards - not blocked. Distributed discussion doc to WG.

Choice and Guards - not blocked. Distributed discussion doc to WG.

Use Statement - unblocked. Low priority. (note taker's revision - this is actually pending the scoping stuff, not just because I'm working on both, but it is tied up there technically also.)

Extensibility - not blocked. Conf. call held. Need to write up ideas from it. Several trackers depend on this one. Document in formulation by design team members.

Expressions - waiting on scoping. Should study XSL variables in expressions.

Parameters and Keying - doc distributed to WG. Make Priority 2? Other items depend on this. Interacts with choice and guards.

Late Binding - internal discussions in team. Related to parameters and keying.

Parser behavior - place holder for topics we didn't want to drop. Move ownership from Bob McGrath to Steve Hanson and/or Geoff Judd?

Packed bits - place holder to remind us we need to revisit this topic. Low prio for now.

Pointers - place holder. MikeB suggested pointers might not be important for v1.0 of DFDL? Jim Myers indicated there needs to be some way to do them. I.e., someone with a format that has pointers in it still needs to be able to use DFDL even if it doesn't help them with the pointer aspect.

Paramters and Keying discussion

We'll continue discussion on next week's call.

Jim Myers(JM) - overlaps with extensibility. Looks alot like white-box transforms.

Steve Hanson(SH) - yes, that was also observed though we didn't write it up.

JM - are there examples that can't or shouldn't be thought of as transforms?

Bob McGrath (BM) - can variables hold the value of another variable? Lots of papers on language design. Is this proposal a sensible one?

JM - can variables be reset? Can we make sure we know how to order and parse things? I.e., do we need rules like defining variables "above" their point of use?

Geoff Judd (GJ) - Agrees that we need such rules.

BM - relationship to XSL variables? Similarities/differences. Finds XSL variables and their rules complex.

GJ - comment on XSL (didn't catch this in my notes)

Martin Westhead (MW) - XSL variables are single assignment.

SH - need default value in addition to "override" or setting.

JM - we have defaults for data items. Can we use that mechanism and avoid another defaults-resolution means for parameters/variables?

Mike Beckerle (MB) - in declarative languages, variables with assignment can be problematic because the order of execution isn't so clear. The reason to use a declarative language is to allow the implementation to reorder things and so forth. Makes free assignment of variables an issue, so single assignment makes sense.

End Notes
Mike Beckerle
STSM, Architect, Scalable Computing
IBM Software Group
Information Integration Solutions
Westborough, MA 01581
voice and FAX 508-599-7148
home/mobile office 508-915-4767
skype ID mbeckerle