Stream: implementers
Topic: Is Bundle.fullUrl required or optional?
Charlie Filkins (Oct 07 2020 at 19:51):
I am using v5.1.16 of the FHIR Validator and I receive this error when an ExplanationOfBenefit entry doesn't have a fullUrl. Isn't this property optional?
Error @ Bundle.entry[1] (line 15, col10) : Bundle entry missing fullUrl
Thanks ... Charlie
David Pyke (Oct 07 2020 at 19:57):
The fullUrl element SHALL have a value except that fullUrl can be empty on a POST (although it does not need to when specifying a temporary id for reference in the bundle)
Charlie Filkins (Oct 07 2020 at 20:40):
Can this be expressed as a constraint, rather than "buried" in documentation? I read that a while ago, but when I looked at the structure definition today all I "saw" was 0..1 and no constraints that would indicate a the larger SHALL conditions.
@David Pyke Thanks for getting back to me, I'll update the rest of my team.
David Pyke (Oct 07 2020 at 20:50):
THat might be possible. Opening a JIRA issue would probably be a good step.
Lloyd McKenzie (Oct 07 2020 at 21:43):
Invariants have to be based on information in the instance - they can't have any awareness of the context in which the instance is being used
Grahame Grieve (Oct 07 2020 at 23:28):
presently, the validator creates that error if there are entries without a full URL unless the bundle is a batch response or a transaction response
Grahame Grieve (Oct 07 2020 at 23:30):
I do not know why, actually. At a minimum it should also be ok for batch and transaction.
Grahame Grieve (Oct 07 2020 at 23:30):
but given this:
Results from operations might involve resources that are not identified.
Grahame Grieve (Oct 07 2020 at 23:30):
it would appear that there's no real capacity for the validator to make any rules here?
Charlie Filkins (Oct 08 2020 at 12:32):
I just want to make sure I understand this. Section 2.36.4 (https://hl7.org/fhir/bundle.html#bundle-unique) provides exceptions only when Bundle.type is a transaction OR batch AND the entry.request.method = POST. I should think this could be expressed as a constraint. If someone can provide a link I'll open a JIRA issue, assuming I have the authority.
If we choose not to persistently identify Resources in a Bundle, e.g., an ExplanationOfBenefit, we SHALL (the text says should) use a UUID, expressed as a URN. However, for this option to be useful we would have to provide absolute references within the ExplanationOfBenefit entries in the Bundle.
Thanks ... Charlie
Lloyd McKenzie (Oct 08 2020 at 13:13):
True. You could submit a tracker item to add the invariant.
Last updated: Apr 12 2022 at 19:14 UTC