Stream: IG creation
Topic: Snapshot rendering of sliced polymorphic element
Rob Eastwood (Sep 04 2019 at 01:16):
The HL7 AU AU Base Medication Statement profile has sliced the polymorphic element medication[x], with Australian flavours of the 2 choices medicationCodeableConcept and medicationReference.
The differential table renders as expected, with just the 2 slice choices; like so
pasted image
However, the snapshot view has what appears to be a duplication of the 2 choices; like this:
pasted image
Wondering if that is by design, or something unexpected.
@Grahame Grieve - this is one of the things we recently chatted about
Grahame Grieve (Sep 04 2019 at 01:21):
it's... not a bug, but I think we can do something less confusing for readers of the spec
Michel Rutten (Sep 04 2019 at 08:53):
@Grahame Grieve I wouldn't expect duplicate type slicing constraints?
This is how Forge now renders the this profile:
pasted image
Note that this is a rendering from the Forge development branch, including all the recent updates/corrections to the SnapshotGenerator.
Michel Rutten (Sep 04 2019 at 08:58):
Note that I have also updated the rendering of slices, in sync with HL7 build. Named slices are now displayed using the format elementName:sliceName (repeating the element name), for example coding:pbs. This also includes the rendering of extensions, e.g. extension:medicationClass. However I decided to make an exception for type slices, as value[x}:valueCoding seems redundant/noisy to me.
Grahame Grieve (Sep 04 2019 at 20:29):
great.
Grahame Grieve (Sep 04 2019 at 20:30):
it's not really duplicate type slicing constraints, though it certainly looks like it. It's showing the base unsliced thing, with it's information, and then it shows the slices. But it sure looks, if you're not the person who generates it, as if it's showing the slices twice.
Rob Eastwood (Sep 04 2019 at 23:47):
Thanks @Grahame Grieve and @Michel Rutten
Michel Rutten (Sep 05 2019 at 12:55):
@Grahame Grieve
I still don't understand the type slice duplication in the the generated snapshot of https://build.fhir.org/ig/hl7au/au-fhir-base/StructureDefinition-au-medicationstatement.htmlx
Per our discussion in Redmond, all type slices in the snapshot should be normalized to elementName[x]:sliceName representation. Renamed elements can only appear in a differential, as an optional shortcut notation.
Choices
1. no slicing on snapshots - renaming only
2. renaming in differentials, slicing on snapshots
3. renaming or slicing in differentials, slicing in snapshots
decision - #3
The current (R4 develop) version of the .NET SnapshotGenerator is still capable of processing snapshots with renamed elements, for compatibility reasons. But a snapshot that includes both representations of the same type slice constraint (e.g. valueCoding and value[x]:valueCoding) seems ambiguous.
Grahame Grieve (Sep 06 2019 at 21:14):
are you asking the presentation of the snapshot? or about the snapshot itself?
Michel Rutten (Sep 09 2019 at 09:22):
I'm confused about the snapshot itself.
Last updated: Apr 12 2022 at 19:14 UTC
