This is an archived snapshot of W3C's public bugzilla bug tracker, decommissioned in April 2019. Please see the home page for more details.
The definition of Interchange Set: The interchange set is the set of documents that constitute the SML model [SML 1.1] to be interchanged. appears to be too strongly stated in the light of other places in the text that speak of interchanging "portions" of the SML model, e.g., sect 1 "To ensure accurate and convenient interchange of the documents that make up an SML model *or a portion of an SML model,* ..." The definition might lead the reader to believe that despite these statements, the interchange set is to be the totality of documents making up the original SML model. I believe our intent was something more like the following: The interchange set is a set of documents to be interchanged that are drawn from the set documents that constitute an SML model. The interchange set may consist of all the documents or a subset of documents from the SML model. Note that the interchange set itself defines an SML model for the purposes of interchange set validation. And mentioning "interchange set validation", I find the phrase "while maintaining" a little disconcerting. Does this imply that the assertions and interrelationships might be otherwise corrupted? Would "based on" or "while observing" be more appropriate?
I agree that there is a problem here, even if it is just a lack of clarity. I believe part of the problem stems from the notion of a "complete model" (which we have abandoned). One could argue that the definition of Interchange Set is correct as a "portion" of an SML model is itself still a model. I don't know that this would be evident though to the average reader.
Section 5.2 states that "The definition and instance documents packaged by an SML-IF document MAY form a valid SML model but it is not required to do so." In addition, we have no concept of a 'complete' model. I think we can still call the content of an SML-IF exchange an 'SML model' - it may not be valid or (complete from the consumer's point of view) but it can still be called an SML model. The definition of interchange set already has a link to the SML spec which defines 'model' and also defines conformance criteria for an SML model. We could make a statement that the SML model contained in the interchange set must form a 'conforming SML model'. We have not stated that as a requirement in the IF spec.
+1 to specifying "conforming SML model"
I'm not sure whether adding "conforming SML model" helps the issue I'm raising. I know we gave up the notion of "complete model"; it's impossible to achieve in reality. But let's assume we have an SML model; it consists of a set of documents. The interchange set may consist of that entire set or a "portion" (subset) of it. In either case the interchange set will represent an SML model, but if the interchange set is a subset of the "original" model, then the SML model represented by the interchange set will (in most likelihood) be a different model. The issue I have tried to raise here is that the definition of interchange set, in saying simply that interchange set "constitutes" the SML model being interchanged does not make it obvious to the reader that we may NOT be talking about the same SML model as the rest of text (namely, the one of which a portion of it may be interchanged).
Resolution on 2/21 conf call is to remove references to "portions of a model". No change to definitions.
Fixed as proposed in comment #5 Sections changed: 1. From: To ensure accurate and convenient interchange of the documents that make up an SML model or a portion of an SML model, it is useful to [..] To: To ensure accurate and convenient interchange of the documents that make up an SML model, it is useful to [..] 2. From: Packaging: The collection of XML documents that make up a model (or model portion) to be interchanged need to be gathered together. To: Packaging: The collection of XML documents that make up a model to be interchanged need to be gathered together.