Stream: IG creation
Topic: Publisher Issues with Invalid Bundles
Saul Kravitz (Oct 19 2021 at 16:04):
We are getting publication errors on an example bundle in mCODE that appear to trace back to a warning message from invariant in USCore.
See the qa report from a branch.
The error is:
image.png
USCore Observation lab has in invariant:
us-core-4 Observation.value[x] SHOULD use Snomed CT for coded Results
: valueCodeableConcept.coding.system.empty() or valueCodeableConcept.coding.system = 'http://snomed.info/sct'
The instance Bundle-mcode-patient-bundle-jenny-m includes a cancer-genomic-variant-jenny-m that does not satisfy the invariant, since the Cancer Genomic Variant profile uses a LOINC code for Observation.value. The validator does not complain about the example instance, but does complain about its inclusion in the bundle, which is odd.
Since the invariant is a SHOULD (warning), this seems spurious.
Three questions (@Lloyd McKenzie , @Grahame Grieve )
1) do you agree that this is a spurious error
2) if so, can it be addressed?
3) or alternatively, can we get dispensation to publish despite this "error"
Thx
Lloyd McKenzie (Oct 19 2021 at 18:04):
"Error parsing" suggests something weird is going on - which is probably not your fault. @Grahame Grieve ?
Eric Haas (Oct 19 2021 at 18:34):
us-core-4 is a best practice warning and not an error
Lloyd McKenzie (Oct 19 2021 at 18:49):
It suggests something is going wrong with the way the warning's being conveyed through.
Saul Kravitz (Oct 19 2021 at 19:43):
@Lloyd McKenzie Since there are no errors on the instance being included in the bundle, but there is an error on the bundle relating to this instance, I would conclude that validation is being done differently -- more stringently on the bundle.
In fact, the QA report indicates that all of my instances are being validated the same way, but is this correct?
StructureDefinition: Validated against fhir.StructureDefinition
Grahame Grieve (Oct 19 2021 at 20:01):
this is on my todo list, but discriminators can turn warnings into errors
Saul Kravitz (Oct 19 2021 at 21:34):
We will forge ahead....depending on the length of your todo list and the speed of our punch list closeout, we may return with a request for special dispensation for this error at publication time. As always, thanks for your responsiveness and help!
Grahame Grieve (Oct 20 2021 at 17:53):
@Saul Kravitz how do I reproduce? Just build mcode?
Saul Kravitz (Oct 21 2021 at 13:21):
@Grahame Grieve yes.
Saul Kravitz (Oct 21 2021 at 20:39):
@Grahame Grieve I think (or at least hope) that these are at the root of all of my bundle issues.
As far as I can tell, all of the codes are valid.
Terminology_PassThrough_TX_Message
fsh-generated/resources/Bundle-mcode-patient-bundle-jenny-m.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://loinc.org#LA9633-4'
fsh-generated/resources/Bundle-mcode-patient-bundle-jenny-m.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://snomed.info/sct#897275008'
fsh-generated/resources/Bundle-mcode-patient-bundle-jenny-m.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://snomed.info/sct#897275008'
fsh-generated/resources/Bundle-mcode-patient-bundle-jenny-m.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://snomed.info/sct#897275008'
fsh-generated/resources/Bundle-mcode-patient-bundle-jenny-m.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://snomed.info/sct#897275008'
fsh-generated/resources/Bundle-mcode-patient-bundle-jenny-m.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://loinc.org#LA9622-7'
fsh-generated/resources/Bundle-mcode-patient-bundle-jenny-m.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://loinc.org#29308-4'
fsh-generated/resources/Observation-karnofsky-performance-status-normal-activity.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://loinc.org#LA29176-7'
fsh-generated/resources/Observation-us-core-observation-lab-tumor-dcis-jenny-m.json Error parsing response message: This does not appear to be a FHIR resource (wrong namespace '') (@ /) for 'http://loinc.org#29308-4'
Grahame Grieve (Oct 21 2021 at 20:51):
how did you get these?
Saul Kravitz (Oct 21 2021 at 20:57):
Pulled them out of the QA report for the build of the master branch, as they are listed at the end.
All of the SNOMED and LOINC codes appear valid. For the entries in the bundle that fail, they all of such error messages sprinkled in.
The weirdest one is on entry[4] of the bundle. If you click on show reasoning, the error is about entry[3] of the bundle...
image.png
Grahame Grieve (Oct 26 2021 at 22:17):
mcode builds cleanly for me now
Last updated: Apr 12 2022 at 19:14 UTC
