Stream: terminology
Topic: Valueset without compose nor expansion
Yunwei Wang (Jul 09 2018 at 13:33):
When I test _summary on some FHIR Servers, I noticed that the Value sets in the result bundle don't have compose nor expansion. According to 4.8.4.2 Constrains Rule vsd-5, "Value set SHALL contain at least one of a compose or an expansion element (expression : compose.exists() or expansion.exists()). Should constrains rules apply to summary search also?
Yunwei Wang (Jul 11 2018 at 13:24):
I have checked three servers: hapi, onto and test.fhir. None of them returns compose/expansion component with searching with ValueSet?_summary=true.
The constrain VSD-5 could cause problem for search with large number of extensional value sets.
Is there anywhere in the FHIR spec says that constrain rule does NOT apply to _summary search? If not, I would suggest to remove VSD-5.
@Rob Hausam @Grahame Grieve
Grahame Grieve (Jul 11 2018 at 22:39):
@Yunwei Wang expansion is not included in a summary
Grahame Grieve (Jul 11 2018 at 22:39):
VSD-5 is a problem for _summary = true
Grahame Grieve (Jul 11 2018 at 22:40):
do you want to make a tak to remove VSD-5?
Brian Postlethwaite (Jul 12 2018 at 00:55):
Wondering if a resource needs to pass invariant validations when returned from a summary call?
I can imagine many profiles/rules where this would not be the case.
And I don't see the problem with that.
Brian Postlethwaite (Jul 12 2018 at 00:56):
The caller has asked for the _summary (or elements) and they are told that they are subsetted...
Yunwei Wang (Jul 12 2018 at 01:51):
@Brian Postlethwaite If the community agree that _summary result doesn't need to pass invariant validation, then this need to be specified somewhere in the spec. Since the current spec doesn't make such exception, all resources generated from _summary call shall be valid.
Brian Postlethwaite (Jul 12 2018 at 02:04):
Agree with you there Yunwie, the _summary does include the meta profiles, which would then permit the resource to indicate that it claims conforms to the provided profile, however it might not pass.
Grahame Grieve (Jul 16 2018 at 00:12):
_summary of resource still has to be a valid resource. Tooling checks cardinalities, but can't check constraints like this one
Brian Postlethwaite (Jul 16 2018 at 02:34):
Or really most profile constraints, and profile that introduces a minimum cardinality of >=1 will not be done... and most invariants.
Brian Postlethwaite (Jul 16 2018 at 02:35):
The .net serializer wouldn't know to include those props, and expect that the java RI would be the same.
Brian Postlethwaite (Jul 16 2018 at 02:36):
(or slicing would be tough to check with minimal content)
Brian Postlethwaite (Jul 16 2018 at 02:36):
Should this issue really be taken back to the main implementer chat?
(as isn't really a terminology specific issue?)
Grahame Grieve (Jul 16 2018 at 02:53):
the tooling doesn't check profiles, no. It's up to the profile author to decide whether allowing summaries is relevant for the profile
Michael Lawley (Jul 16 2018 at 22:16):
FWIW Ontoserver gets its _summary, _elements, and paging (_count) behaviour from the HAPI framework
Michael Lawley (Jul 16 2018 at 22:20):
So does the profil author also need to decide something about _elements and validity? I would think not, but what is the line between _summary and _elements?
Last updated: Apr 12 2022 at 19:14 UTC
