Alex, please have a look. Thanks

Nicole, the logic for this issue is in:
bool vtkMRMLModelSliceDisplayableManager::vtkInternal
::IsVisible(vtkMRMLDisplayNode* displayNode)

I am not sure about the reasoning behind this logic that you added, so I am reassigning to you

My change[1] was to not show model slice intersections if the model wasn't visible.
The problem that's causing this bug is that the volume slice nodes are also MRML model nodes so they're getting checked with the same logic, so if the slice isn't set to be visible (in 3D), don't show the slice plane intersections.
The volume slice model nodes are tricky to distinguish from other model nodes, especially since in the isVisible check it's working soley on the display node (also the same class). I'd have to dig around in the logic (or add my own method) to find a way to determine if I have a volume slice model display node and to not make the extra visibility check in those cases.
Otherwise, I revert the change[1] and model slice intersections are visible when the model is not visible if that flag is on.

[1] http://viewvc.slicer.org/viewvc.cgi/Slicer4/trunk/Libs/MRML/DisplayableManager/vtkMRMLModelSliceDisplayableManager.cxx?r1=23021&r2=23098

vtkMRMLSliceLogic has IsSliceModelNode
So would just need to go from the display node to the model node.

That method is static, you should be able to call it without instantiating the logic:

https://github.com/Slicer/Slicer/blob/916b6e2f1b3742bee09cf08c121ef8c1c5fa3b61/Libs/MRML/Logic/vtkMRMLSliceLogic.h#L340

Also check if it fixes 0003462
Thanks

Working on setting an attribute on the display node, then checking for it in a new method on the slice logic before applying the visibility check.

Make also sure that, when loaded, existing scene will set that attribute.

Otherwise, all existing MRB won't work as expected.

Instead, would it make sense to:

• Get the list of all sliceNode
• .. get the associated displayModelNode
• .. return True if the ID you want to check belong to one of these ?

Fixed in svn 23495:
http://viewvc.slicer.org/viewvc.cgi/Slicer4?view=revision&revision=23495

This doesn't fix 3462.

@Nicole: Could you comment on http://na-mic.org/Mantis/view.php?id=3718#c12283 ? Thanks

JC: didn't see your comment before my commit, have to run, will double check that first thing tomorrow.

@jcfr: the nodes aren't saved with the scene, they're created on the fly when node added events are detected. Testing with legacy mrbs and mrml files shows it's working.

Great. Thanks for checking.

I did notice that the slice intersections weren't updating as I scrolled in one test.
Also: fixed a typo:
http://viewvc.slicer.org/viewvc.cgi/Slicer4?view=revision&revision=23516

@Nicole: when you say "did notice ... weren't" does that mean that the current trunk is broken or that it was broken before your recent fix?

Broken before my fix, when checking bug 3462.

Sorry, but to clarify: are you saying that the issue is now fixed or not (just trying to sort out the what is present tense and what is past tense)

The issue is not fixed, it's still reproducible this morning - I did the steps to reproduce issue 3462, turning on slice intersections and showing the red slice in 3D before running model maker. As soon as model maker finished, the yellow and green slice intersections disappeared from 2d and the yellow and green slices don't show up when I toggle them on in 3d. Slicer is also leaking on exit now.