FHIR Chat · Slicing by Value Set binding

Stream: conformance

Topic: Slicing by Value Set binding

Giorgio Cangioli (Dec 05 2018 at 17:36):

Is any constraint that I should take in account when an element is sliced by value set binding ?
For example, is it expected that all the bindings used in the slices are "required" bindings, or at least one shall be required or it might be any kind of binding (preferred, example,...).

Richard Townley-O'Neill (Dec 06 2018 at 00:47):

If the binding is not required, there are no syntactic limits to what is valid, so everything is in the slice, and it will overlap with any other slices that you define.

Rob Hausam (Dec 06 2018 at 11:38):

Right. That makes sense and that's what we've observed. I think there are two basic ways to handle this. One is to apply the preferred (or other not required) binding at the root level where the slicing is declared, where it serves de facto as the "default", and then there is a specific named slice with the required binding and an open slicing rule. The second way is similar, but treats the two bindings as siblings - there is again the specific named slice with the required binding and then the additional slice with the preferred (or other not required) binding is the named '@default' slice with a closed slicing rule. Logically I think that both of those should work. What I haven't yet been able to accomplish is getting either of those approaches to be able to successfully recognize the slices and validate example data using them. I keep getting lots of "This element does not match any known slice for the profile" information messages and various combinations of warnings and errors on the slicing validation with the various permutations. If anyone has successfully done this and has the specific discriminator type and path and slice path syntaxes that worked, that would be helpful to see.

Michel Rutten (Dec 10 2018 at 10:57):

Hi @Rob Hausam, interesting, thanks for sharing. I assume that you're using the HAPI Java validator classes - or did you also try the .NET validator by chance?

Rob Hausam (Dec 10 2018 at 13:00):

Thanks, @Michel Rutten. I didn't try the .NET validator, but that's a good suggestion. I'll see about giving that a try in my Windows VM environment (I haven't actually used it before).

Michel Rutten (Dec 10 2018 at 14:15):

FYI Ewout has created a simple GUI for the validator:
https://github.com/ewoutkramer/Furore.Fhir.ValidationDemo
Note: the .NET API is available for FHIR DSTU2 and STU3; However R4 branch is still under development and R4 validator is not yet available.

Rob Hausam (Dec 10 2018 at 15:56):

Yes, I was thinking that R4 isn't yet supported and was going to ask you to verify that for the validator, so thanks. We are using R4, so I guess we'll need to wait a bit for that.