Stream: conformance
Topic: Non-resolving bundled/contained reference
Lloyd McKenzie (Jan 08 2018 at 05:40):
If an element is defined as being a Reference with an aggregation mode of "contained" or "bundled", is it legitimate to populate the reference such that it doesn't resolve (e.g. with only an identifier or a display or extension, but no reference)? (If the reference is present, it's an error if it doesn't resolve to a contained or bundled location - as specified by the aggregation mode.)
Grahame Grieve (Jan 08 2018 at 06:14):
I don't know
Lloyd McKenzie (Jan 08 2018 at 06:17):
I have a client who'd like it to be an error, but the spec isn't super clear. The real question is whether the expectation for bundled/contained/referenced comes into play if the element is present or only if the [element].reference is present.
Lloyd McKenzie (Jan 08 2018 at 16:17):
Ewout Kramer (Jan 09 2018 at 15:01):
I am almost ready to rewrite the validator to not have just success/failure but also "undecided". I think this is clearly a case where -from a validation point of view- you cannot say whether it's wrong or right...
John Moehrke (Jan 09 2018 at 15:07):
is that undecided? Or did you get to the point that you know that it is NOT wrong... That is a decision, so is not undecided.
Ewout Kramer (Jan 09 2018 at 15:21):
Luckily, I am still in a stage where I can easily change the term ;-) In three-valued boolean logic I see both "unknown" and "undetermined" being used..
John Moehrke (Jan 09 2018 at 15:27):
not sure how to interpret the difference between those two. How about inconclusive
Ewout Kramer (Jan 09 2018 at 15:38):
I like that.
Lloyd McKenzie (Jan 09 2018 at 15:58):
I don't think "undecided" is useful for a standard. Something is either legal or it's not. Sometimes we have things that "might" be issues - or at least go against good practice and raise warnings rather than errors, but this doesn't seem like a space where that's necessary.
Lloyd McKenzie (Jan 09 2018 at 15:59):
The standard needs to make clear in what context aggregation mode applies - is it when the element is present, when the Reference.reference element is present or when the Reference.reference.value is present.
Lloyd McKenzie (Jan 09 2018 at 15:59):
That's a relatively straight-forward thing for us to clarify - we just need to figure out what we want...
Ewout Kramer (Jan 09 2018 at 16:12):
Lloyd, I agree - that's why I said "from a validation point of view": there are run-time circumstances (unresolvable links, terms servers unreachable) etc that would lead to "inconclusive". Maybe this specific issue is not so much run-time but indeed just plain wrong.
Lloyd McKenzie (Jan 09 2018 at 17:04):
I think inconclusives should be handled as warnings/infos. But this case I think we can tighten down to be an error.
Vadim Peretokin (Jan 11 2018 at 15:35):
I agree with Llyod, an info is a good established term for this.
Last updated: Apr 12 2022 at 19:14 UTC
