Stream: IG creation
Topic: Version mismatches
Grahame Grieve (Jul 23 2020 at 20:56):
Working on implementing some previously published versions of IGs, I've discovered that the IG publisher allows an IG to depend on IGs that are based on different FHIR versions to the FHIR version the IG itself is based on. I basically overlooked this; there'll be a message in the log pointing out that this is unpredictable but trying to go ahead. But since there's nothing in qa.html about it, people haven't really paid attention.
Also, I haven't paid enough attention to not referencing current in the IGs.
So, for example hl7.fhir.us.davinci-deqm#1.0.0 (3.0) depends on hl7.fhir.us.cqfmeasures#1.0.0 (4.0), and hl7.fhir.us.cqfmeasures#1.0.0 depends on hl7.fhir.us.qicore#current
Grahame Grieve (Jul 23 2020 at 20:57):
both should be regarded as fatal. I'll be adding a check to qa.html for both
Yan Heras (Feb 14 2022 at 00:21):
I have a different version mismatches question. I updated the Risk Adjustment IG to dependsOn USCore 3.1.1 (version that is named in regulations). The RA IG also dependsOn DEQM 3.0.0 because we reuse an Extension defined there. However, the DEQM 3.0.0 dependsOn USCore 3.1.0. In a profile (e.g., https://build.fhir.org/ig/HL7/davinci-ra/StructureDefinition-ra-measurereport.html) where we reference USCore Patient, it looks like the IG publisher chose to point to the version of USCore Patient from the USCore 3.1.0 instead of 3.1.1. Is this by design of the publisher to choose the older version if there are mismatches? Is there a way to set at a profile level to indicate which version of profiles to reference in this scenario?
Grahame Grieve (Feb 14 2022 at 00:45):
no it's random which it will point to, I think
Yan Heras (Feb 14 2022 at 02:03):
I built multiple times, so far it points to the same 3.1.0 each time, not sure if it's random. Are we able to specify which version to point to somehow?
Grahame Grieve (Feb 14 2022 at 02:19):
I'm not sure. I've never considered the question, though I obviously should've. by random, I don't mean, it will change all the time, just that there's no control over how ti resolve
Last updated: Apr 12 2022 at 19:14 UTC
