The VTK_USE_QVTK_QTOPENGL option solved the problem on Michi's computer without any modification in the driver configuration.
Maybe there are some things that can influence that, but this variable has definitely something to do.

@Klaus: What's the status?

[SVN revision 20538]
ENH (#2735): Support for a temporary DB added. This was performed to allow the management of removable media. The temporary DB is allways reinited when the application restarts or when a new temporary dir is selected.

[SVN revision 20616]
CHG (#2735): Paths for databases storage where modified according the OpenCherry standards. Instead of putting them direct to .Mitk3M3, they are now put to .Mitk3M3/.metadata/.plugins/org.mitk.gui.qt.dicombrowser/

[SVN revision 20536]
ENH (#2735): Documentation with MainPage.

[SVN revision 20618]
ENH (#2735): Mouse cursor is set to wait modus during image load.

• If the SQL script was not supplied in the command line means that it couldn't be found means that it could not be found in any of the StandardFileLocations. This is an installation issue.

[SVN revision 20370]
ENH (#2735): A few presentation changes were made to the DICOMBrowser. The view does not manage the folder anymore. They are now managed through a preferences page. This allowed the size of the view to be reduced, so it can be placed beneath the standard views. A perspective was created in order to load the view in the mentioned position.
A small change concerning the directory were the DB is placed was also made. The DB is placed in the configuration directory of the application running it, instead of putting it always under .Mitk3M.

[SVN revision 20302]
FIX (#2735): Some bug fixes concerning the stop refresh button. The DicomIndexer process did not stop in windows using QProcess::terminate. It was changed to kill.
Because the process is now being killed and does not comes to its natural end, the lock file is not deleted. Since there is no need for a lock file, the lock file creation and remotion was removed.
Another small bug: after closing the MainApp during refresh, the DicomIndexer process kept running in background. It is (supposedly) fixed by killing the process in the view's destructor.

[SVN revision 20303]
FIX (#2735): See the previous comment.

1. I don't know exactly. It should be in any directory that is searched by

mitk::StandardFileLocations. It also looks in the source directory
"mbi/Modules/DicomIndexer". I think Marco can be of further help in this point.

[SVN revision 20400]
ENH (#2735): Show preferences button was moved to the DicomIndexer widget for better GUI layout.

[SVN revision 20289]
ENH (#2735): Documentation

It does not find the DicomIndexer app in the Linux version. I don't know how to fix this.

[SVN revision 20264]
ENH (#2735): The inclusion of the mitk/bin to the PATH variable worked. LOG_ERROR added, showing the PATH variable and the indexer app name, in case the process do not start.

Patch containing the modifications requested from Marco

[SVN revision 20263]
ENH (#2735): Date formatation now performed with Qt.

This patch contains the modifications and fixes requested from Marco in the previous comment.

[SVN revision 20288]
ENH (#2735): Change in button labels.

@Jan: What is the status?

@Jan: What is the status here?

[SVN revision 19611]
FIX (#2658): Based on the selection in the data manager, the visibility of the other nodes is set accordingly. If a normal image is chosen, all child segmentations are shown, all other images are hidden. If a segmentation is chosen, all other segmentations are hidden.

[SVN revision 22215]
FIX (#2323): Mutex was locked during thread destruction. For this reason, the thread could not be succesfully terminated and hanged on the lock. This happened when StopTracking() was called before m_TrackingFinishedMutex was locked in TrackTools(). Since StopTracking() also locks m_TrackingFinishedMutex, TrackTools() hangs in there for ever. We changed TrackTools(), so it only locks m_TrackingFinishedMutex if StopTracking() was not previously executed.

[SVN revision 21472]
FIX (#2323): Mutex was left locked when a error occurs in the try section.

I am only using the Boost Graph Library (BGL).

I only need the component "graph". No specific options are required.

@Daniel: What is the status?

Importance of the bug adjusted.

In mitk::PolyDataGLMapper2D::Paint, if the node does not have a transform it crashes. A verification if the transform really exists was added.

ImageCropper was modified to use new itkCuboidObjectCutter, which implements a faster version of volume cropping for cuboids. At the moment, it only works fast for cuboid which are not rotated.

Variable not fund bug resolved with commit c18637

[SVN revision 21893]
FIX (#2125): The problem lied on the IsInside method of the Cuboid. It used the WorldToIndex transform of its geometry, which computed at every single call the inverse of the transformation. The inverse was then computed for every single voxel. It was modified to use the VTK transform, which caches the inverse.

[SVN revision 21890]
ENH (#2125): A user complained in the MITK list that the image cropper was still taking long to crop. I asked for a test image, in which he was having trouble. We found out that it was a vector parallelity checking error, for which we gave now a tolerance.

Original version must be restored because of a mistake. We incorrectly assumed that the bounding object could only be a cuboid. For the faster version using cuboid, a new class is being implemented (CuboidObjectCutter), and the bundle (ImageCropper) will be modified to use this new class.

The boost version number in my code is 1.37. I have already modified my code to be compatible if the earlier versions of boost.
I still think though, that the test version of boost has to be upgraded. Version 1.33 is already 4 years old!

I'm sorry. This request is not for the core library but to the modules (I'll modify the component to the correct one).
Currently, MITK is beeing tested with the 1.33 version, which does not have many of the new features. If I commit any of the stuff using the new classes the deshboard will turn red.

The zoom factor global option was set in the global options list (class
QmitkMainTemplate), and the class mitkCameraController was modified in order to
have a zoom factor internal variable.
The link between the global options and the variable set method is still
missing.