Forums » #117 - DFDL v1.0 Revision »
Implicit alignment of model group
Added by Michael Beckerle about 9 years ago
Alignment - for model group, should implicit alignment be greatest of the alignments of children, or least common multiple?
Use of 'greatest' may be a holdover from some earlier version of the spec where alignment values were constrained to be powers of 2. In that case greatest is guaranteed to be the same as least-common-multiple.
There is no such constraint on the value of the alignment property, so taking greatest alignment of the children is not conservative. It can result in incorrect alignment.
Suggest change 'greatest' behavior to 'least common multiple'.
Replies (6)
RE: Implicit alignment of model group - Added by Steve Hanson about 9 years ago
Because of the possibility of an initiator on a structure, inferring the alignment of a structure from its children is flawed. Suggest that implicit alignment of complex element, sequence and choice is always '1'.
Action 239 - RE: Implicit alignment of model group - Added by Michael Beckerle almost 9 years ago
This and another public comment about alignment are to be considered as workgroup action 239
Resolved - Action 239 - RE: Implicit alignment of model group - Added by Michael Beckerle almost 9 years ago
Agreed the implicit alignment of a model group is 1. Language about greatest alignments of children are all dropped.
Resolved: Implicit alignment of model group - Added by Steve Hanson over 8 years ago
New errata 4.6 in experience document 1.
DONE - RE: Implicit alignment of model group - Added by Michael Beckerle over 8 years ago
change in draft-gwdrp-dfdl-v1.0.4-r05.docx
(1-6/6)