Stream: Orders and Observation WG
Topic: relevantHistory
John Moehrke (Apr 05 2019 at 14:32):
It was brought to my attention that Task and thus an odd sub-set of resources in R4 include an element relevantHistory that points at Provenance. I am concerned about this trend as it might be a misunderstand how Provenance functions relative to reverse lookups on Provenance.target. Even if this is not a misunderstanding of how Provenance functions, what is the meaning of a Provenance that was recorded, but not listed in relevantHistory? Does that mean it is not-relevant? Does that mean it should not be considered? The concern I have is that not listing should not be seen as an excuse to ignore other Provenance records. This ignoring could result in missed information that truly was relevant. Relevance should not be a static thing, but rather be a 'current' thing.
Lloyd McKenzie (Apr 05 2019 at 15:19):
Yes, it means it's not relevant in the view of the person who updated the resource. The purpose of relevant history is to give a view of key events such as creation, suspension, release, dispense, etc. Some of the Provenances pointed to might not even have a focus of the referencing resource (e.g. the dispense example). Minor edits, transformations, etc. wouldn't be shown. The purpose is to do the same as v2 messages that show "status history" where 'status' is interpreted fairly broadly.
Lloyd McKenzie (Apr 05 2019 at 15:19):
If you actually care about provenance, you'd do so as per usual.
Lloyd McKenzie (Apr 05 2019 at 15:20):
relevantHistory will never include the most recent change - as that provenance record won't exist yet when you make the update.
John Moehrke (Apr 05 2019 at 15:30):
so, how about the consistency across the FHIR specification? Should there be some mention of this mechanism and proper use of the mechanism on the Provenance page? If so, could someone writeup that narrative and propose in a CR?
Lloyd McKenzie (Apr 05 2019 at 15:41):
It's described in the Request pattern
Lloyd McKenzie (Apr 05 2019 at 15:42):
There's actually a profile on Provenance that should be referenced, but the tooling didn't support asserting that constraint
Lloyd McKenzie (Apr 05 2019 at 15:42):
(in a resource definition)
John Moehrke (Apr 09 2019 at 20:13):
started a CR GF#20758 -- welcome specific text recommendation
John Moehrke (Apr 09 2019 at 20:24):
I find a provenance-relevant-history profile, but can't find where this is referenced. It seems floating all alone, only mentioned on the Provenance profile page. I think it is defined in the build with Provenance resource. Right?
Lloyd McKenzie (Apr 09 2019 at 20:45):
Right. The intention was for the 'related history' elements to point to it, but the tooling for the core build doesn't support that right now.
John Moehrke (Apr 09 2019 at 20:55):
okay. I was expecting something to be said on the workflow pages... hoping that I could just clone that onto the Provenance page. I think it would be very important that we make clear how this .relevantHistory is intended to be used. Specifically, why it would not be a copy of all Provenance. It seems very redundant without specific guidance.
Lloyd McKenzie (Apr 09 2019 at 20:58):
It was created as a result of a change request from implementers who very definitely did NOT want to do all provenance. They wanted a very limited subset that would support the use-case of relevant history.
Lloyd McKenzie (Apr 09 2019 at 20:58):
Agree that the workflow pages should point to it at least. Can you submit a change request?
John Moehrke (Apr 09 2019 at 21:30):
done GF#20760
John Moehrke (Apr 16 2019 at 19:11):
@Luis Maas @Peter Van Houte @k connor
Last updated: Apr 12 2022 at 19:14 UTC
