← Back to search
#conformance
Inheritance from parent profile or datatype profile?
14 messages · 2 participants · View on Zulip →

profile inheritance rules

fhir profilesinheritance rulesparent profiledata type profileelement overrideprofile precedencefhir discussion
Ward Weistra Feb 5, 2026, 11:34 AM
Is there any generic rules defined over which inheritance takes precedence when I have both a parent profile and data type profile on an element? Eg. my parent Patient profile has a value X for name.short but the HumanName profile I put on name has a different value Y for short , what should end up in my profile?
Grahame Grieve Feb 5, 2026, 12:01 PM
we've discussed this somewhere before
Grahame Grieve Feb 5, 2026, 12:01 PM
element overrides type, I. think
Ward Weistra Feb 5, 2026, 02:58 PM
Grahame Grieve said : we've discussed this somewhere before That's about what @Ewout Kramer said :laughing:
Ward Weistra Feb 19, 2026, 03:43 PM
In my testing it seems that the Java stack (represented by IG Publisher) and the .NET stack (represented by Simplifier) indeed behave differently. Taking PS-CA Pan-Canadian Patient Summaries - SIMPLIFIER.NET and running it through both tools: IG Publisher gets the short from the parent Immunization resource : { "id" : "Immunization.site" , "path" : "Immunization.site" , "short" : "Body site vaccine was administered" , [ ... ] }, Where Simplifier (Download > Download snapshot as JSON ) gets it from the DataType profile : { "id" : "Immunization.site" , "path" : "Immunization.site" , "short" : "Concept - reference to a terminology or just text" , [ ... ] },
Ward Weistra Feb 19, 2026, 03:43 PM
Ewout made a brave attempt to find back what was decided on this, but no definitive answers yet: FHIR Snapshot - Base selection - Stack Overflow [FHIR-9791] Clarify base for snapshot with profiled types - Jira ("Addressed by other trackers. Withdrawn by Chris") Non normative: Snapshots Determining Refines · chrisgrenz/FHIR-Primer Wiki [FHIR-13402] Clarify snapshot generation rules - Jira [FHIR-14400] Inconsistency in how comments, definitions, etc. are exposed for extensions - Jira
Ward Weistra Feb 19, 2026, 03:43 PM
Regardless of what we decide here, it is going to impact customers. For example: NL Core Patient has comment from nl-core-humanname and not from Patient.name . And that's probably how they'd want it.
Ward Weistra Feb 19, 2026, 03:51 PM
From this ticket Reseach datatype merge behaviour on .NET vs Java · Issue #3432 · FirelyTeam/firely-net-sdk , Copilot made this brave evaluation :slight_smile:
Ward Weistra Feb 19, 2026, 03:52 PM
Perhaps this is best discussed in-person at the May WGM given the potential impact.
Ward Weistra Feb 19, 2026, 03:52 PM
Looping in @Chris Grenz since his name comes up in quite a few of those tickets :slight_smile:
Chris Grenz Feb 19, 2026, 04:06 PM
How does simplifier do it for extensions? Assume that applying the extension profile leads to the short being drawn from the extension profile?
Ward Weistra Feb 19, 2026, 04:31 PM
@Chris Grenz Looking at ACME Canada 2022 | MaintainedDevice - SIMPLIFIER.NET it gets the short ("The total amount of hours the device has operated") from the Extension profile , not the parent Extension element ("Additional content defined by implementations"). CleanShot 2026-02-19 at 17.29.55.png If that's what you're asking.
Chris Grenz Feb 19, 2026, 04:32 PM
Seems strange to me that the application of the extension profile would behave differently than the application of a different data type profile. =)
Grahame Grieve Feb 27, 2026, 12:55 AM
yes I think this is a good subject for an evening meeting in Rotterdam