[SVN revision 19750]
FIX (#2710): checking if member variable m_InputImageNode is not NULL to catch exception when the image was not selected in DataManager

there is a bug in AdaptiveRegionGrowingView: If the image only gets loaded and not selected, the InputImage is not set. I will add a safety check if the member m_InputImageNode is NULL.

Multiple selection for VesselTrees about to be commited.
Need a place to build it into!

[SVN revision 21090]
FIX (#2435): disabling flag MITK_USE_TD_MOUSE to remove compilation error when activated

closing bug

either MITK_USE_TD_MOUSE has to be removed or integration of TD Mouse reimplemented

Yers, that's wright! This happend during the merge from QT3 to QT4 and will be fixed soon.

adding me to cc list

request for core change

Talked to mitk board (MN) and the present implementation was not considered as critical, since the statemachine patterns are static and doesn't occupy much memory space. Still, patterns should be deleted after calling destructor of StateMachineFactory. To do this, StateMachineFactory has to be non static. Adaptions will be made in a second step. For now only the cyclic reference between state->transition->state will be removed by using itkWeakpointer in member m_NextState within class mitkTransition. In destructor of mitkState all added transitions in member in m_Transitions are freed. mitkStateMachineFactory still holds a list of Smartpointers to all created states to enable undoing a statechange.
I will check it with valgrind and then commit the changes.

To check cyclic references in StateMachine, I will adapt mitkStateMachineTest.

tests hopefully all run in revision 19045

request for core change

Checks with valgrind show, that the problem still is singleton StateMachineFactory. This will be solved in the second stage of the commit.
Removing cyclic reference in revision 19015.

Yes, that's what I thought of inside mitkState. There a map of weakpointers onto transitions is held:

StateMachineTest is not the right place, because in the object StateMachine a vector of startState::SmartPointers (for several timesteps) is held.

Yes, after checking mitkState and mitkTransition I see the cyclic reference. The error occured in revision 8798 and 8846 of mitkState.h. The problem can be removed with falling back to c-pointer onto mitkTransitions.

COMP commit (#2352): adapting mitk::GlobalInteraction::GetInstance due to different behavior Linux/Windows of static members

adapted request of a change

adapted request of a change

Second part of changes include a change of class StateMachineFactory. It now is no singleton anymore but is held as member in mitkGlobalInteraction (which is singleton). It is created in constructor of GlobalInteraction and deleted in destructor. In destructor of mitkStateMachineFactory, all instanciated objects of the state machine patterns are deleted. The constructor of GlobalInteraction was changed to contain a second parameter XMLbehaviorFile wich is set to null on default. If NULL, the standard file which contains the information on the state machine patternms is loaded.
Because of the change, one now calls mitk::GlobalInteraction::GetInstance()->GetStartState(const char* statemachineName) instead of mitk::StateMachineFactory::GetStartState(const char* statemachineName)
The only place where changes had to be done because of this interface change was inside mitkStateMachine.

copy-paste from #1419:
From Mathias Seitel 2008-09-04 16:33:49

mitkFocusManager commited in revision 18185

By Jakub:

patch that adds mitkFocusManagerTest

Hi Jakub, (<<sorry for the wrong c in your name>>)
why can't you add a comment in bugtracker?

Hi,
I have initially tested the class mitkFocusManager with mitkFocusManagerTest.cpp (see Attachment). No errors occured!

I'll take the bug.

patch adding mitkFocusManagerTest to mitk/testing

associated bug: when loading two images, the levelwindow slider changes the setting of the first image. To change levelwindow of the second image, one has to change via context menue. But when deleting the first image changes of the levelwindow doesn't take effect on the image, that is left in datatree. Should be changed to use at least the first image in datastorage. Otherwise it looks like it doesn't work...

Found a bug that can be reproduced by clicking maximize button and not leaving the maximized widget. The icons for toggle widgets visualization doesn't display anymore. One has to leave the window and then the icons are visualized in the upper right corner.

well, i will also checkin the changes into depricated qt3 branch (mitk-qt3) because the changes make sense there too.
comitted in revision 17314

patch for changes for Bug #2068 (enabling changing a logo (png-File) during runtime)

Since QT3 implementation will be ended soon and the changes are currently only needed for project internal implementation, I will comitt changes into project internal branch.
I attached the patch (changes in two classes) if anyone needs them.

patch for changes for T2068 (enabling changing a logo (png-File) during runtime)

I was working on the logo-rendering; The problem was, that the path for the png reader inside the logorendering was set initialy but then never again when Logorendering enabled. Changed to be able to set path, then if enabled disable and redraw the logo.

Dafür!!!

assigning to default

I agree. Party bug then?
3M3 relevant?

Bug originally processed by Anne Sauer. Still relevant (no Scene Viewer anymore, Status or Write Project?)

removing keyword mandatory-0.12

Talked to Mathias who has a similar bug.
We will check how to proceed

removing it from target milestone 0.12

assigning anne

closing bug, because currently there is no need for wii support

I will add two lines to cmake file of module WiMote.
Path to WDK / DDK will be searched for and library hid.lib has to be specified. This should resolve the problem of globally adding directories to visual studio for projects with wiimote.

[SVN revision 27388]
ENH (#1543): adding lib and include path in cmake configuration of module WiiMote.

cmake definition for integration of input devices

Cmake integration needs to be restructured.
For this a clear list on what needs to be included is necessary. I will include a patch for so long keep track of current changes into the right direction (hopefully).

cmake definition for integration of input devices

adding marco while working on cmake definition.

currently no recorces for it! Waiting for interested BA / MA or DA developers

assignement was wrong, assigning to Alex

assigning sascha