After some unsuccessful attempts with manifests, I ended up solving the problem by simply modifying the python lib and dll name to python27_mitk. Clearly, no conflict is present anymore. Let me know if you want the patch. Not sure I can separate it from my changes allowing debug config being built.

This _might_ also be the reason for this bug:
http://bugs.mitk.org/show_bug.cgi?id=18197

Happy to help :)

Well, the UpdateArchive is conceptually already in my pull request - the code that compares the time stamps essentially does that. However, I agree there might be a better approach to store the "state" (the time stamps and file name of the scene file). Given that this is basically per-application-instance information, we could use either a temporary external file instead of storing this state in the SceneIO itself, or, even better, use berry Preferences to store it.

Daniel, I agree with most of what you are saying, but here are some thoughts.

Here's a discussion that we had with Miklos Espak some time ago regarding the auto-saving and what problems might arise when implementing it. I think it's meaningful to revisit this before going to implementation.

Yes, of course. Forgot that again, apologies :)

Yeah, my problem disappeared with those fixes. And I also found it strange that MitkWorkbench itself is working fine. Furthermore if, for example, I loaded a project and then closed it - no crash happened upon closing my app. But as you said, the fix is simple and reasonable, so big thanks for merging it in (one less difference for my MITK fork :)).

Dangling pointer fix

I honestly am not sure what is going on so deep in blueberry, but seemingly fixing the first problem unveiled another one. The second patch fixes that one.

Could you please make a change - I think it's easier in this case.

Also, in release mode the test also passes fine on my machine (actually, on a different one - this time using Win7, the rest being similar).

I added the comments you asked for, hope they will clear some things up for the future maintainer.

Sounds good!

Commented source

Your renaming patch works fine if applied after my changes as well. The last thing to do after that is to rename IntersectIntervals to SubtractIntervals.

Added smaller threshold to image comparison function

iteration 4

iteration 4

Ok, hopefully the final iteration.
Instead of always using plane geometry itself to determine the bounds, which is definitely different from what it was before, I finally made the version that is consistent with the 3D plane geometry data mapper:

Added smaller threshold to image comparison function

Ref image should go into MITK-Data\RenderingTestData\ReferenceScreenshots

I created the test and also added some changes to the mapper itself.

Hi Caspar,

undesired gaps illustration

undesired gaps illustration

Do you mean creating a rendering test?

Patches and reference image

patch 2/2

(In reply to Caspar Jonas Goch from comment #8)

Do you happen to have example data you could share?

patch 1/2

patch 1/2

Well, I've written my own implementation for interval arithmetic. Applying the following two patches should bring it to the good state without using boost.

patch 2/2

patch

I understand. I actually tested this on Ubuntu 14 machine and it definitely fixed the z-buffer issue that I had there - I'm not sure which GPU is there though. Similarly, it fixed the issue in Ubuntu running in virtual box.

I believe this is related to this:
https://bugreports.qt-project.org/browse/QTBUG-42360

http://public.kitware.com/pipermail/vtk-developers/2014-May/014998.html

Actually it's better also to request qt.conf file, so the call is like
install_qt5_executable("bin/${MY_APP_NAME}${CMAKE_EXECUTABLE_SUFFIX}" "${Qt5Gui_PLUGINS};${Qt5Sql_PLUGINS}" "" "" "plugins", TRUE)

Comma is of course redundant
install_qt5_executable("bin/${MY_APP_NAME}${CMAKE_EXECUTABLE_SUFFIX}" "${Qt5Gui_PLUGINS};${Qt5Sql_PLUGINS}" "" "" "plugins" TRUE)

Also, for SVG icons to work, it is necessary to install svg plugin as well, so the line should be:

Modified deployment script

Replace QVTKWidget with QVTKWidget2 patch.

For example, this is what I modified the ApplyLookuptable function to look like (see attachment).

My users did not report any problems since I've changed this in my fork.

ApplyLookuptable

ApplyLookuptable

patch

I think this is related: when loading a vti/vtk image with active scalar array not set, the application crashes. Attached is a simple patch. Consider it signed-off if you like to integrate it.

I think this is related: when loading a vti/vtk image with active scalar array not set, the application crashes. Attached is a simple patch. Consider it signed-off if you like to integrate it.

patch

Patch with fix.

An additional patch to resolve the name clash of debug and release PythonQt libraries.

Here's a patch that contains some changes I made to build the debug version. I didn't check OpenCV so perhaps some more things need to be done, but the general gist should be clear. The patch is not based on the master, but I think resolving the conflicts should be very easy.

patch part 2

patch

Guys, please refer to this e-mail of mine.
https://www.mail-archive.com/mitk-users@lists.sourceforge.net/msg05149.html

Fixed: https://github.com/MITK/MITK/pull/76

Fixed: https://github.com/MITK/MITK/pull/76

Hi, the bug with deleting/reordering nodes was fixed and integrated into master - see https://github.com/MITK/MITK/blob/master/Plugins/org.mitk.gui.qt.datamanager/src/QmitkDataManagerView.cpp - lines 901-906.

https://github.com/MITK/MITK/pull/74

In case anyone has similar needs, one more thing could be added. Property to instruct mitkImageVTKMapper2D to render gradient magnitude instead of an actual image without preparing full 3D gradient magnitude image. I have implementation that uses vtkImageGradientMagnitude filter and vtkImageHistogramStatistics to set the gradient magnitude to color mapping.

In case anyone has similar needs, one more thing could be added. Property to instruct mitkImageVTKMapper2D to render gradient magnitude instead of an actual image without preparing full 3D gradient magnitude image. I have implementation that uses vtkImageGradientMagnitude filter and vtkImageHistogramStatistics to set the gradient magnitude to color mapping.

Great! Thank you!

Also, yes, I need the closed curve, so for now it would be nice to have the subdivision polygon functionality.

Yes, that is completely OK. You probably look at a different user (I have two forks under "khlebnikov" and "rkhlebnikov" - there was a reason for that, but I will soon delete the fork under "khlebnikov"). So if you look at the "rkhlebnikov" fork - the branch should be there. Plus, I'm pretty sure that you can merge the pull request (as opposed to a branch): https://help.github.com/articles/merging-a-pull-request

This bug is present in the current master branch.

I will try that out and write a comment about the outcome here.

Icon in PNG and SVG

Bug is fixed in the current release

Ellipse icon in SVG and PNG

I actually use it for the cases where manual segmentation of 2D contour of a vessel is needed. Probably I could use the recently added Bezier curve planar figure for the same purpose.

Ellipse icon in SVG and PNG

Pull request submitted:
https://github.com/MITK/MITK/pull/57

Also, made an icon for ellipse.

I have successfully imported a local directory with the DICOM editor - if this works as any kind of DICOM functionality test.

Pull request submitted: https://github.com/MITK/MITK/pull/47

Hi Marco,

Pull request submitted
https://github.com/MITK/MITK/pull/58

Pull request submitted:
http://github.com/MITK/MITK/pull/56