@Csaba: In the TestNodeAssociations test, at line 355 [1], if this helps you to resolve the issue, it returns the subject hierarchy node (vtkMRMLSubjectHierarchyNode7) whereas the associated node (vtkMRMLModelHierarchyNode1) is probably expected:

8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----
276: vtkMRMLSubjectHierarchyNode (0x282b980)
276: Debug: Off
276: Modified Time: 9661
276: Reference Count: 2
276: Registered Events: (none)
276: ID: vtkMRMLSubjectHierarchyNode7
276: Indent: 0
276: Name: Model21_SubjectHierarchy
276: Description: (none)
276: SingletonTag: (none)
276: HideFromEditors: 0
276: Selectable: 1
276: Selected: 0
276: AssociatedNodeID: vtkMRMLModelHierarchyNode1
276: ParentNodeID: vtkMRMLSubjectHierarchyNode3
276: SortingValue: 13
276: AllowMultipleChildren: true
276: Level="Series
276: OwnerPluginName="NULL
276: OwnerPluginAutoSearch="true
276: UIDs="DICOM; MODEL21:"
8<----8<----8<----8<----8<----8<----8<----8<----8<----8<----

[1] https://github.com/Slicer/Slicer/blob/260df3d2c91465dbcb5f1f5a97730dd31b8e378a/Modules/Loadable/SubjectHierarchy/Testing/Cxx/vtkSlicerSubjectHierarchyModuleLogicTest.cxx#L355

@Csaba: This is an attempt to fix the failing test ... that brings me further but now I have issue with hardening transform.

See https://github.com/jcfr/Slicer/compare/3715-attempt-to-fix-associated-hierarchy-node-getter

Strange. The test worked for me a few days ago, and nothing changed in the SH MRML or logic since then.
Is this with VTK5 or 6?

The test is also failing using the branch [1] built against VTKv5 and also failing on the nightly dashboard for the Linux, MacOSX and Windows

In case of nested hierarchy, what is suppose to return the function:

static vtkMRMLHierarchyNode* GetAssociatedHierarchyNode(...)

For example, looking at the example of PopulateScene [2], given the node "model21", should we expect the direct parent (1) "SubjectHierarchyNode" or the grand parent (2) "ModelHierarchyNode"

If we expect (2), the hack I am doing in [3] is along the right way ...

[1] https://github.com/cpinter/Slicer/tree/subject-hierarchy-integration

[2] https://github.com/Slicer/Slicer/blob/64e40c5175634d5f423226498a67b96cb82e9558/Modules/Loadable/SubjectHierarchy/Testing/Cxx/vtkSlicerSubjectHierarchyModuleLogicTest.cxx#L91-108

[3] https://github.com/jcfr/Slicer/commit/d83c46bcede04a80de934ae0d351bc48fefc2481

Additionally, checking the comment in [1], I can read:

Note: This must be used instead of vtkMRMLHierarchyNode::GetAssociatedHierarchyNode, because nested associations have been introduced to avoid conflicts.

[1] https://github.com/Slicer/Slicer/blob/64e40c5175634d5f423226498a67b96cb82e9558/Modules/Loadable/SubjectHierarchy/MRML/vtkMRMLSubjectHierarchyNode.h#L75

It seems that the PopulateScene method is incorrectly creating the nested association:

ModelHierarchyNode -- SubjectHierarchyNode -- ModelNode (model22)

instead of

SubjectHierarchyNode -- ModelHierarchyNode -- ModelNode (model22)

Exactly. The nested association should look like
SubjectHierarchyNode -- ModelHierarchyNode -- ModelNode (model22)
This should be the issue then.
Thanks a lot!

Arfff ... not sure to fully understand the intricacy of nested association. Would be great if you could have a look.

Thanks

In fact, the code is doing the right thing .. the comment in the test was just misleading.

There was an issue with vtkMRMLHierarchyNode::GetAssociatedHierarchyNode, namely that when the UpdateAssociatedToHierarchyMap function was called before getting the associated node, subject hierarchy's GetAssociatedNode was called, not the function from the base class, which caused the nested associations not to work properly. Fixed in rev23370.

Also it was discovered that the harden transform function did not work properly, thus this test failed, because it also tested the harden transform feature of subject hierarchy. Fixed by Alex Y. in rev23391