Stream: implementers
Topic: Logical Model and slicing for terminology
Alexander Henket (Apr 17 2017 at 07:05):
When an element in a needs multiple terminology bindings then we sort of agreed that we needed to create slices (on coding), because actual multiple bindings were out of the question.
But when the StrucDef is a logical model, there is an invariant in the way: sdf-1: Element paths must be unique unless the structure is a constraint (expression : derivation = 'constraint' or snapshot.element.select(path).isDistinct())
Is this limitation by design or by accident? Or should I just add derivation = 'constraint' to my logical model even though I don't have a base, since my model *is* the base?
Alexander Henket (Apr 17 2017 at 07:12):
In general you'd say that the snapshot paths shall always be unique, except when slicing. In which case invariant sdf1 needs reformulation to reflect that.
Alexander Henket (Apr 17 2017 at 08:24):
Added GF#13207
Grahame Grieve (Apr 17 2017 at 08:54):
I don't knowabout this with logical models. I'd have expected that you'd have different elements, not slicing in the base model
Alexander Henket (Apr 17 2017 at 15:40):
I'm auto converting from an ART-DECOR model that supports multiple bindings. I cannot just add elements, nor can I just create the ValueSet that rules them all in the conversion process. So I either need multiple bindings, or slicing. Not having both means that auto conversion is impossible.
Alexander Henket (Apr 17 2017 at 15:42):
Adding elements is changing the ART-DECOR model that remains the normative model. Creating a runtime ValueSet that doesn't exist in ART-DECOR gives maintenance headaches
Alexander Henket (Apr 17 2017 at 15:43):
Adding it to the project is again a change to the underlying project I cannot do just because I'm converting to FHIR.
Grahame Grieve (Apr 17 2017 at 19:04):
ok, I understand the situation. Will have to think about it some more
Last updated: Apr 12 2022 at 19:14 UTC
