Stream: conformance
Topic: Binding on type extension
Grahame Grieve (Dec 11 2017 at 20:56):
see GF#13553 - eld-11 allows for a binding to appear when the element type is "Extension" - what's the point of this? There's contention that Forge users are doing this - what are they doing? it doesn't appear to be supported by the validators (certainly not by the Java one).
Grahame Grieve (Dec 11 2017 at 20:56):
FHIR-I proposes to remove this, unless we can clarify why it's allowed
Grahame Grieve (Dec 11 2017 at 20:57):
(note: this would be a breaking change to a FMM5 artifact)
Chris Grenz (Dec 11 2017 at 21:19):
No idea what that would even do?
Grahame Grieve (Dec 11 2017 at 22:11):
well, that's why we propose removing it as an option
Michel Rutten (Dec 12 2017 at 11:06):
If the behavior is unclear or undefined, then yes, let's remove it.
Michel Rutten (Dec 12 2017 at 11:09):
If desired, I could update Forge to disable/hide binding controls for extension elements by default, except when opening a profile that contains such a binding, so the user can see and edit/remove the binding constraints.
Marten Smits (Dec 13 2017 at 10:12):
If desired, I could update Forge to disable/hide binding controls for extension elements by default, except when opening a profile that contains such a binding, so the user can see and edit/remove the binding constraints.
Great idea! Maybe even warn the user if there is an existing binding on an extension, if that's not too much work.
Michel Rutten (Dec 13 2017 at 10:23):
Currently, FHIR (STU3) spec allows bindings on extensions, according to to eld-11. So currently Forge also accepts it.
If FHIR R4 introduces new/updated invariants, then we can also update Forge.
Grahame Grieve (Dec 13 2017 at 10:25):
but you don't have an actual use for it?
Michel Rutten (Dec 13 2017 at 10:42):
I don't author profiles, I just develop a profiling tool.
Maybe we can scan Simplifier db for existing profiles with bindings on extension elements?
Grahame Grieve (Dec 13 2017 at 10:46):
that'd be nice, yes
Ewout Kramer (Dec 13 2017 at 11:55):
Yes, our authors (at least @Ardon Toonstra ) use it, that's why I discovered it could be done in the first place (and why the .NET validator does support it).
Ewout Kramer (Dec 13 2017 at 11:56):
When I discovered it, I asked @Lloyd McKenzie - I think he told me it was indeed a feature, not a mistake, so I implemented it.
Ewout Kramer (Dec 13 2017 at 11:56):
So, in the .NET validator it only works on simple extensions with a bindeable type.
Marten Smits (Dec 13 2017 at 13:09):
Yes, we have used it in the past to define a binding on extension.value (before Forge could do that explicitly). Now we just put it on the value instead of the extension.
Lloyd McKenzie (Dec 13 2017 at 16:03):
I'm fine with it going away
Ewout Kramer (Dec 18 2017 at 14:44):
Well, I added the support for your convenience, so if we change R4 to reflect that, that's fine by me. Don't know whether we need to explicitly disable support in the validator though...
Last updated: Apr 12 2022 at 19:14 UTC
