Stream: implementers
Topic: Handling STU3 Extension canonical URLs
Ken Sinn (Jan 18 2019 at 21:12):
What are the recommended guidelines for handling/using STU3 extensions that are no longer available in R4?
For example, our STU3 profile use http://hl7.org/fhir/STU3/extension-observation-eventtiming.html, but our profile structure definition references the canonical URL -- http://hl7.org/fhir/StructureDefinition/observation-eventTiming -- which no longer resolves after the R4 publication.
The STU3 profile itself declares its FHIR version to be 3.0.1 -- is that sufficiently obvious that someone would need to muck with the canonical URL declaration (http://hl7.org/fhir/StructureDefinition/observation-eventTiming) in order to get at the actual resolvable page (http://hl7.org/fhir/STU3/extension-observation-eventtiming.html)? Or should the STU3 profile reference the version-specific extension, via http://hl7.org/fhir/STU3/extension-observation-eventtiming.html or http://hl7.org/fhir/StructureDefinition/observation-eventTiming|3.0.1? (and is there a preference for one URL format or the other?
Grahame Grieve (Jan 19 2019 at 03:03):
actually, the formal version specific canonical URL is http://hl7.org/fhir/3.0/StructureDefinition/observation-eventTiming
Grahame Grieve (Jan 19 2019 at 03:04):
the only way to know that from an R3 perspective is magic, but it is documented formally as part of R4
Grahame Grieve (Jan 19 2019 at 03:05):
http://hl7.org/fhir/StructureDefinition/observation-eventTiming|3.0.1
that's something else - which we haven't yet worked on for the spec - the R4 definition of the R3 version of the extension - but that doesn't exist anyway
Ken Sinn (Jan 21 2019 at 17:52):
A related question: when defining a FHIR Profile and binding an element to a FHIR version-specific valueset, which of the following should we use for the canonical URL?
A) http://hl7.org/fhir/stu3/ValueSet/document-relationship-type
B) http://hl7.org/fhir/ValueSet/document-relationship-type|3.0.1
C) http://hl7.org/fhir/3.0/ValueSet/document-relationship-type
Lloyd McKenzie (Jan 21 2019 at 18:00):
B
Lloyd McKenzie (Jan 21 2019 at 18:00):
The other two resolve, but they're not the canonical URL.
Ken Sinn (Jan 21 2019 at 18:02):
And in defining profiles, the valueset bindings should always be the canonical URL, not simply a resolvable URL?
Whereas the extension canonical URLs are differently-structured, i.e. to Graham's point above?
Lloyd McKenzie (Jan 21 2019 at 21:59):
Elements of type "canonical" must always point to the canonical URL of the resource being referenced (possibly with the business version appended after '|'). The extension URLs he's defined have different canonical URL roots because they're intended by design to be version-specific
Ken Sinn (Jan 21 2019 at 22:46):
For the valuesets defined by FHIR, is the business version the same as the FHIR version? I notice that in the R4 publication, version is explicitly stated as 4.0.0 for all valuesets, but this isn't defined in stu3 or earlier. For example, https://www.hl7.org/fhir/stu3/valueset-name-use.html is assumed to be business version 3.0.1?
Lloyd McKenzie (Jan 22 2019 at 01:01):
It should be, yes. The tooling didn't do that though. I believe we're planning a technical correction to STU3 that will fix this. I don't know that it'll be addressed for DSTU2 though. @Grahame Grieve ?
Grahame Grieve (Jan 22 2019 at 20:46):
we're not planning a technical correction for R2
Last updated: Apr 12 2022 at 19:14 UTC
