Closed jafeltra closed 6 months ago
bb0dd4c switches back to overwriting the type of a Logical Model but now logs a warning when that type is not an absolute URL. With this change, three of the repos have no actual differences, but do now log warnings about changing the type.
IntelliSOFT-Consulting/ChanjoKe-FHIR-IG#main
and WorldHealthOrganization/ddcc#main
do still have changes where elements from the parent were previously re-added but now they are not added and an error is logged instead. They also have warnings for non-URL types. It looks like the elements that are trying to be re-added are all the same but change cardinality min from 0 to 1. I think both should be able to update to use regular card rules.
Fixes CIMPL-1249.
This PR improves the logic for overwriting the
type
of a Logical Model with a caret value rule to only allow URIs to be set. This seems to line up with the FHIR spec for type. Because you can only overwritetype
with a URI, I also updated the message of the InvalideTypeAccessError to be a little more specific for Logical Models. This also usesthis.pathType
when adding elements so the element id path is properly calculated so that elements that were in the parent aren't re-added when when the logical model overwrites thetype
.This issue came up due to changes in the IntelliSOFT-Consulting/ChanjoKe-FHIR-IG#main repo after PR #1443. With this branch, there are still changes, but I think the changes are correct. Running a full 1-year regression resulted in 5 repos with changes. Looking at the differences, they all resulted from types that were not full URIs, which is not allowed in FHIR and no longer allowed on this branch. These are the 5 repos: