The liver scene had the model hierarchy nodes set to be hidden from editors, it needed to be updated to work with the updated code in the Models module.
New RSNA zip file here:
https://www.dropbox.com/s/8gfuw27tka9iy18/3DVisualization_2013.zip
Working on getting the self test data set uploaded to Midas and the .py file link updated.

LiverData mrb uploaded to Midas and self test updated:
http://viewvc.slicer.org/viewvc.cgi/Slicer4?view=revision&revision=22659

The new LiverSegments_Scene.mrml scene crashes Slicer4.3.1 on MacOS 10.8.

Is this also reproducible using the nightly build ?

I fixed and tested the scene on a nightly debug build on Linux, it works on the nightly on Mac as well.
I just downloaded the Mac 4.3.1 binary, that would have the wrong link in the self test for the LiverData.mrb, but it's loading from disk without re-downloading and it's stuck at 25% on the Loading loading file dialog box.
Downloading the .zip file I linked above and expanding it to load the scene file, it does crash in the same way. Opening that mrml scene file with my development build on my Mac, and it loads fine.

I'll build a debug version of the 4.3 branch on my laptop and try to figure out what's going wrong with loading this scene. Could it be that the stable release on the web site is a bit behind with fixes that have come in since it was uploaded?

@Nicole: Since 4.3.2 has not been released. These are different:

• trunk
• head of Slicer-4.3 branch
• commit of 4.3.1 (stable) release

The scene works with the nightly build on my Mac. However, we were planning to use Slicer4.3.1 for the RSNA hands-on courses.

If 4.3.1 does NOT work but the nightly one does. I propose that after identifying the commit(s) that resolve the problem, we could either (1) do a 4.3.1-1 tweak release or (2) do the 4.3.2 release using the current 4.3 branch.

@Nicole: Do you know which commit is fixing the bug ?

If you don't want to use a different package, you or Nicole will have to modify the scene to work around the issue.

I built the Slicer-4-3 branch on my mac and tested the new LiverData.mrb there, it didn't crash. Opening the scene from the .zip file linked above, also no crash.
I'm going on now to try to localise which commit since the 4.3.1 tag fixed the crash.

Good.

I guess you meant that you will try to localize the commits fixing the crash in the list of commits added to the Slicer-4-3 branch following the 4.3.1 tag.

Building the tagged release on the 4.3 branch and running the RSNA2012Vis Part 3: Liver test I get this error on the command line:

Loading LiverData.mrb...

unpacking to /var/folders/d3/kqchv64d52b96rqq1jw9d0ch0000gn/T/__BundleLoadTemp2013-11-04_14+58+22.058
Traceback (most recent call last):
File "/Users/nicole/Slicer4.3/Slicer43-tag-Superbuild-Debug/Slicer-build/lib/Slicer-4.3/qt-scripted-modules/RSNA2012Vis.py", line 105, in onPart3Liver
tester.test_Part3Liver()
File "/Users/nicole/Slicer4.3/Slicer43-tag-Superbuild-Debug/Slicer-build/lib/Slicer-4.3/qt-scripted-modules/RSNA2012Vis.py", line 482, in test_Part3Liver
loader(filePath)
File "/Users/nicole/Slicer4.3/Slicer43-tag-Superbuild-Debug/Slicer-build/bin/Python/slicer/util.py", line 202, in loadScene
return loadNodeFromFile(filename, filetype, properties, returnNode=False)
File "/Users/nicole/Slicer4.3/Slicer43-tag-Superbuild-Debug/Slicer-build/bin/Python/slicer/util.py", line 153, in loadNodeFromFile
success = app.coreIOManager().loadNodes(filetype, properties)
RuntimeError: std::logic_error: basic_string::compare

@Nicole: If I understand, it means that one of the commit listed in [1] fixes the issue ?

[1] https://github.com/Slicer/Slicer/compare/v4.3.1...master-43

I hope so! I'm having trouble narrowing it down by debugging the crash, will start a search through the commits tomorrow (binary search check out and compile and test).

It's commit svn 22644 [1] / git efd4a37[2] that fixes the scene loading on the 4.3 branch for the scene that I provided:
BUG: 3416. Added generic mechanism for persisting node references in .mrml files.

That makes sense since I created the scene in the current code base where it was using the new generic mechanism, but it's failing to load in a Slicer that doesn't have this functionality.

To fix it, I can fix the model hierarchy hide from editors flags with the original scene open in a 4.3.1 release build and then regenerate the mrbs and zip files using that version.

[1] http://viewvc.slicer.org/viewvc.cgi/Slicer4?view=revision&revision=22644
[2] https://github.com/Slicer/Slicer/commit/efd4a372fccfd67376bdb527d3acada9f721e6b1

Great detective works.

It I understand properly, this mean that non of the scene created with Slicer >= r22644 can be open (confidently) with Slicer < r22644 ?

For RSNA, there are few options:

(1) re-generating the dataset

or

(2) Creating a tweak release 4.3.1-1 including commit r22644

or

(3) Creating a patch release 4.3.2 based on the current state of the Slicer-4-3 branch.

I think (1) or (2) are the "safest" approach.

@Sonia: Is it too late to create a new set of packages for a 4.3.1-1 release ?

@Nicole: How much work would be required to create a new dataset ?

Thinking about this a bit more, I think the backward compatibility needs to be fixed. If you release 4.3.2 built from the current 4.3 branch, scene files saved with it won't load in 4.3.1.

May be we could provide a small converter tool or extension that would be used to convert scene.

Regenerating the data set isn't difficult, but there is a minor change in a .py file that needs to be done so that the self test uses it - right now, in the 4.3.1 binary it's referencing the original LiverData.mrb (midas item 8611) which won't crash but won't show the model hierarchies. For the RSNA tutorial, Sonia will be using a combined zip file with multiple scenes and that can be separate from the self test update link.
The trunk now has the "crashes on 4.3 but shows the model hierarchies" LiverData.mrb of midas item 117033. If a new binary release is made on the 4.3 branch that includes the updated node reference code, it needs to use that one as well.
So, I think that when I generate the new .mrml file that will work with the current 4.3 binary, it should only go into Sonia's tutorial .zip file because that midas id change going at the current head of the 4.3 branch wouldn't be necessary as it would come after the mrml node changes that it's trying to work around.

JC: for the converter, I'd want Alex to do it as he'll have a firmer grasp on the changes.

@Nicole: If I understand, there is currently a crash in Slicer trunk that you identified as "crashes on 4.3 but shows the model hierarchies". Is there an issue to track this ... I guess we can clarify during the coming Dev hangout.

New zip file containing the tutorial files uploaded:

https://www.dropbox.com/s/8gfuw27tka9iy18/3DVisualization_2013.zip

Sonia, let me know if this doesn't work with your version of Slicer.

Yes, the LiverScene from the new zip file works with Slicer4.3.1 release. Thank you so much!

Closing resolved issues that have not been updated in more than 3 months