This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
(discussed in 1/22 f2f) 1. "check" is not defined. Need to do so. 2. need to clarify the normative text in 4.4.2.3 Instance Validity Rules, and remove the discrepancy between the normative text and the first row of table 4-2. The discrepancy is the first row, non-reference. The normative text cited does not provide assert anything about the contents of this row. This has no effect on model validity however, because of the way the conformance section is written. One simple fix suggested was to remove the first row of the table.
Since the spirit of the table is to summarize all possible cases, I don't think the first row should be removed as that would leave those cases unaccounted for.
I agree with Jim. For the sake of completeness the first row should remain; otherwise the reader will have to think through the issues that we had to in order to come to the obvious conclusion. We might think about adding a note in order to prevent the discussion that we had at the F2F--perhaps something like the following: Note that if the element declaration *E* is not an SML reference, then the validity rules are considered trivially satisfied in that there is no element that would invalidate the rules. But even that runs the danger of saying "too much". Despite our discussion at the F2F, I now think it the text fine as it now stands.
01/31 meeting resolution is to mark bug editorial and fix as per comment #2; Add Kirk's note comment #2 to the spec, below table. Leave table as is Work on final text by email Refer to the acyclic when fixing this
Editors: see 1/31 telecon minutes for other proposals considered (for those not present) and for a bit more detail on the possible issue with acyclic.
[1] Did not modify table 4-2. [2] Added the following text below the table: Note that, 1. if an element instance is not an SML reference, then the validity rules are considered trivially satisfied for that instance in that there is no element that would invalidate the rules. 2. "Check" in the table above means that the appropriate constraint must be evaluated. [3] I checked the acyclic constraint section. It has similar writing style and the same structure as the target* constraints (i.e., Mapping from Schema/Schema Validity Rules/Instance Validity Rules). I am not sure how it is written quite differently as mentioned in the 1/31 minutes. If a change to the acyclic section is desired, some clarification will help the editors. I have not changed anything in that section at this time.