For Knowledge Artifacts (Measures, Libraries, PlanDefinition, Questionnaires) test data we can strip anything that can be generated or inferred at runtime. Specifically:
text
data-requirements
parameters
related-artifacts (of type "depends-on")
content of type application/elm+json or application/elm+xml
Basically, the opposite of RefreshIG and RefreshLibrary.
The goal here is to make the test resources as easy to work with as possible. Removing any data that's not related to the functional behavior.
The less data we have, the less there is to parse at runtime, or index in the IDE, or download when you clone the repo. Tidy, small, fast.
The CQL debundling is already a huge win, we're just taking the next steps.
For Knowledge Artifacts (Measures, Libraries, PlanDefinition, Questionnaires) test data we can strip anything that can be generated or inferred at runtime. Specifically: text data-requirements parameters related-artifacts (of type "depends-on") content of type application/elm+json or application/elm+xml
Basically, the opposite of RefreshIG and RefreshLibrary.
The goal here is to make the test resources as easy to work with as possible. Removing any data that's not related to the functional behavior.
The less data we have, the less there is to parse at runtime, or index in the IDE, or download when you clone the repo. Tidy, small, fast.
The CQL debundling is already a huge win, we're just taking the next steps.
A rough guess is that this will half the size of the test data based on how much is narrative and so on. Also, any auto-generated extensions on those resources: http://hl7.org/fhir/us/cqfmeasures/StructureDefinition/cqfm-parameter http://hl7.org/fhir/us/cqfmeasures/StructureDefinition/cqfm-dataRequirement http://hl7.org/fhir/us/cqfmeasures/StructureDefinition/cqfm-directReferenceCode