I'm investigating the following error and output from gdb which I think might be responsible.

warning: Could not load shared library symbols for libvtkSlicerSpatialObjectsModuleLogicPythonD.so.
Do you need "set solib-search-path" or "set sysroot"?
warning: Could not load shared library symbols for libvtkSlicerSpatialObjectsModuleMRMLPythonD.so.

Program received signal SIGINT, Interrupt.
0x00007f3e5c2cfd35 in MetaVesselTube::M_Read (this=0x4b64c60)
at /home/crmullin/Projects/Slicer4-SuperBuild-Debug/ITKv3/Utilities/MetaIO/metaVesselTube.cxx:947
947 while( (c!='\n') && (!m_ReadStream->eof()))

The blame returns this commit where metaVesselTube was added:
https://github.com/Slicer/ITK/commit/e10fc9491dfd32c69fe5427036911b520f0f22e8

Here's the fix by Matt McCormick:
http://itk.org/gitweb?p=ITK.git;a=commitdiff;h=46e05e666e98b2ef38dfaba50333811eedbbc1bd;hp=4c2bfb739d2ec073e1f1477e576d656195edbb11

This patch was already added to ITKv3 in Slicer.

Email sent to Roland:

You need to add the extra condition && c!="\377" at this line https://github.com/Kitware/ITK/blob/dcd655f89cd17b5b864878f0c0e0b77b00b17553/Utilities/MetaIO/metaVesselTube.cxx#L947

@Roland: Could you please retry with a fresh Slicer build and a fresh Extension build? Also, make sure that Slicer is using your built extension by checking the Slicer "Additional module paths". I tried this on that dataset you attached, and it works fine.

There has been a problem with the ExternalProject UPDATE_COMMAND, and your ITK may not have been updated. @JC, FYI, I finally may have made progress with upstream CMake on that front per today:

https://github.com/thewtex/CMake/commit/4dee3a007c640c172330fbab747f4df2aba1e74c#commitcomment-2124450

@JC, Chris, I don't see '\377' in the file that Roland uploaded?

@mccormic:

• I was able to reproduce the problem on my machine. Using ITKv3 (dcd655f), the latest one.

• Using a debugger, Chris and I found out that the reader got stuck in the while loop line 947. The read character was '\377' that is why it was looping forever. By adding the extra condition (obviously a HACK), the file appeared to load successfully. That said, I have no idea where this character is read from ...

  • My system: Ubuntu 10.04, g++ (Ubuntu 4.4.3-4ubuntu5.1) 4.4.3

  • Chris system: Ubuntu 12.04

  • We could schedule a short google hangout if that helps

  • Good news regarding the UPDATE_COMMAND :)

Interesting. A hangout is a good idea -- I'll add a calendar invite for tomorrow at 11 if you can make it.

During our hangout session, we solved the problem. Here are the topics:

• for ITKv4: http://review.source.kitware.com/#/c/8449/
• for ITKv3: https://github.com/jcfr/ITK/tree/fix-metaio-vessel-tube-reader-stuck-while-loop-3.20

While working on the issue, also backported the most recent change of the MetaIO library to ITK3.20. See https://github.com/jcfr/ITK/tree/backport-metaio-itkv4-fixes

Matt> Could you push changes on top of release-3.20 integration branch, and then update the External_ITKv3.cmake in Slicer ? Make the issue as resolved when done. Thanks

Great stuff, JC.

ITKv3 release-3.20 has been merged:

http://itk.org/gitweb?p=ITK.git;a=commit;h=3f5d32015df46591fde9c4656b59842f11efe9e4

Slicer ITKv3 external project has been bumped with r21347.

A test was added for the ITKv4 patch. Will see if someone wants to review there tomorrow, then commit to MetaIO -> ITK/VTK -> Slicer.

Extra. Thanks Matt. Also thank you for writing the test in ITKv4.

Marking the issue as resolved, tomorrow packages should allow Roland to confirm it works and close the issue.

Fixed in r21347
See http://viewvc.slicer.org/viewvc.cgi/Slicer4?view=revision&revision=21347

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