Stream: implementers
Topic: StructureDefinition.kind
Mark Kramer (Jun 14 2018 at 12:07):
StructureDefinition.kind gives a choice of primitive-type, complex-type, resource, or logical (pick one). So if I want to include complex types in my logical model, how would I specify that?
Grahame Grieve (Jun 14 2018 at 12:10):
logical is being removed - now there's just primtive, complex, and resource, where:
- only FHIR can define primitives
- anyone can define complex - that's a re-usable set of elements of arbitrary depth but one that has no identity
- anyone can define resources - that's an aribtary re-usable set of elements that has an identity (e.g. it can stand alone)
- the resources we define have a bunch of extra stuff and can be used with the FHIR API
Grahame Grieve (Jun 14 2018 at 12:11):
I'm working on this change tomorrow
Grahame Grieve (Jun 14 2018 at 12:11):
'we define' - that's the namespace in the URL.
Grahame Grieve (Jun 14 2018 at 12:12):
so there's 'the FHIR namespace' = http://hl7.org/fhir and other HL7 namespaces (anything else starting with hl7.org including hl7.org/fhir/xxx . and then there's anything whatever else
Grahame Grieve (Jun 14 2018 at 12:13):
though I expect most of what happens to be in hl7.org namespace. I'm working on http://hl7.org/cda/fhir and http://hl7.org/v2/fhir for instance
Grahame Grieve (Jun 14 2018 at 12:13):
hope this helps
Mark Kramer (Jun 14 2018 at 12:25):
I think so. I think you are saying that logical models are just a parallel universe that can define its own complex types and resources. There's nothing explicitly to mark those models as 'logical'.
Grahame Grieve (Jun 14 2018 at 12:26):
yes. that's what we decided for R4
Claude Nanjo (Jun 14 2018 at 15:51):
That makes sense to me as well.
Robinette Renner (Jun 14 2018 at 20:39):
Based on my understanding of this change to StructureDefinition, for the logical models that I have created using Forge, I should change StructureDefinition.kind from "logical" to "complex." Is this correct?
Grahame Grieve (Jun 14 2018 at 20:51):
depends whether you intend them for them only to be composed into other content, or whether they are identified entities in their own right. From memory, you are doing data element definitions; these would be complex types, yes
Robinette Renner (Jun 14 2018 at 20:59):
I am also using logical models to facilitate our end-user's transition to FHIR. I am using logical models to represent a collection of related data points (can be a combination of things such as diagnosis information, laboratory results, and medication administrations) in a structure that the end-user's are familiar with. I am then using the FHIR mapping language to map the logical model to the related profiles. Would the "complex" kind still work?
Grahame Grieve (Jun 14 2018 at 21:01):
I think so
Lloyd McKenzie (Jun 14 2018 at 23:38):
Forge authors STU3 StructureDefinitions. You don't want to change them unless you're migrating to R4 representation and want to stop using Forge.
Lloyd McKenzie (Jun 14 2018 at 23:39):
The auto-conversion process from R3 to R4 should take care of the change when the time comes.
Robinette Renner (Jun 15 2018 at 10:41):
Perfect. Thanks for the clarification.
Last updated: Apr 12 2022 at 19:14 UTC
