Stream: implementers
Topic: R4 JSONSchema
Joe Paquette (May 08 2019 at 21:01):
Looking at the R4 fhir.schema.json schema, I noticed that fields having a primitive type that is also required (i.e., 1..1) has no validation to enforce requiredness. As an example, look at careplan.profile.canonical.json. Note that "id": "CarePlan.status" and "id": "CarePlan.intent" both have 1..1 cardinality. Now, look in R4 fhir.schema.json at lines 10527-10530 and note that the "required" property for careplan only contains "subject" and "resourceType". Using example-json/careplan-example.json and the R4 fhir.schema.json schema on the https://www.jsonschemavalidator.net/ online validator confirms no validation errors when "status" and/or "intent" are deleted from the example.
Primitive fields in the fhir.schema.json have the pattern of defining 2 fields: "propName" (normal field) and "_propName" (extention field). Requiredness must take an either/or approach to these 2 fields since only one of each "status" and "intent" is required.
A straight forward change to the fhir.schema.json can fix this issue.
In fhir.schemn.json, replace lines 10527-10530:
"required": [
"subject",
"resourceType"
]
with the following:
"allOf": [
{
"required": ["subject", "resourceType"]
},
{
"anyOf": [{ "required": ["status"] }, { "required": ["_status"] }],
"not": { "required": ["status", "_status"] }
},
{
"anyOf": [{ "required": ["intent"] }, { "required": ["_intent"] }],
"not": { "required": ["intent", "_intent"] }
}
]
Retesting with example-json/careplan-example.json and the R4 fhir.schema.json schema with the above changes confirms errors for missing any/all of the required fields.
NOTE: This is an issue for all resources having primitive fields that have a 1..1 cardinality!
I'd be happy to create a bug against the generated fhir.schema.json if someone would point me in the right direction.
Lloyd McKenzie (May 08 2019 at 22:34):
To log a bug, click on the "propose a change" link at the bottom of any page in the FHIR spec. You'll need to do a one-time registration, after which you can log as many bugs as you like :)
Grahame Grieve (May 14 2019 at 01:54):
getting back to the issue....
Grahame Grieve (May 14 2019 at 04:16):
returning to this, I do not understand what is going on here
Grahame Grieve (May 14 2019 at 04:17):
{ "anyOf": [{ "required": ["status"] }, { "required": ["_status"] }], "not": { "required": ["status", "_status"] } },
Grahame Grieve (May 14 2019 at 04:17):
the anyOf I understand, but not the not - what's that about?
nicola (RIO/SS) (May 14 2019 at 14:54):
_primitive just was a bad idea :/ - maybe it's not too late fix it :)
Grahame Grieve (May 14 2019 at 19:59):
I agree it was a bad idea and I said so at the time. (and people assured me I just didn't understand JSON at the time). But it is too late to change it now
Last updated: Apr 12 2022 at 19:14 UTC
