[SVN revision 24244]
COMP (#4527): fatal message causes a test to fail? Adding outputs for now

[SVN revision 24241]
FIX (#4527): adding string for statemachine pattern in mitkSliceNavigationController instantiation. A pattern name should always be given otherwise the statemachine cannot be initialized.

Analysis of the failed StateMachineTest shows that both EventMapper and StateMachineFactory are instanciated during mitk::GlobalInteraction::Initialize(...) as often as the method is called!
I changed it to delete the previous instance of StateMachineFactory and EventMapper. They are not reused to clear the list of statemachine patterns and event descriptions and not to add the information. The bug didn't show up before, because mitk::GLobalInteraction::Initialize(...) is only called once. In StateMachineFactoryTest I call it several times.

• Restructuring GlonalInteractionTest to also check non singleton

instanciation.

• adding MITK_FATAL Messages in GlobalInteraction and StateMachine
• within initialization of GlobalInteraction: returning false if no startState

recieved by StateMachineFactory. Then it is not initialized (former bug).

• changing itkWarningMacro to MITK_FATAL or MITK_WARN to use same output

strategy

• restucturing of GlobalInteractionTest and using test macros

tests show green. so closing bug

[SVN revision 24372]
FIX (#4527): enabling mitkStateMachineTest again and adding one test case. Tests showed a bug in GlobalInteraction where StateMachineFactory and EventMapper was instantiated each time initialize is called. Adding safety clear in destructor of StateMachineFactory.

Result of the "RFD" session:

So far, but #4527 is solved.

Things done so far without commit:

• Restructuring GlonalInteractionTest to also check non singleton instanciation.
• adding MITK_FATAL Messages in GlobalInteraction and StateMachine
• within initialization of GlobalInteraction: returning false if no startState recieved by StateMachineFactory. Then it is not initialized (former bug).
• changing itkWarningMacro to MITK_FATAL or MITK_WARN to use same output strategy

I will open a new bug for commiting the changes mentioned above and make it dependant on this bug.
Then a service like used for DataStorage (GetDefaultDataStorage) might be the solution for a non static GlobalInteraction.

Adding MITK_FATAL messages is the first step. I have added but not yet commited this.

so I close this bug

changing to party bug

seems as if the changes were only done in one of the current two statemachines! (->Marco, TobiasS). If this is a mistake, then please add to mitk/CoreUI/Bundles/org.mitk.gui.qt.common

[SVN revision 23456]
FIX (#4206): adding return to if block to stop recording once the number of updates are made

another small bug: The first output file contains the specified number of samples, but the following logging data don't. This is due to a missing return statement.

return has been added in the past...

[SVN revision 22695]
FIX (#3953): fixing error message and adding functionality to logg from start to stop button clicked.

I think bug is fixed. Please check and close if so..

closing bug. No changes will be made.

agree, jochen?

I don't really find a proper reference, but very rough to have something to lean on:
http://en.wikipedia.org/wiki/Fiduciary_marker

Well, we basically have the problem, that we have to differentiate between a filter that uses tools / sensors to compensate for patient etc. movement and another filter that uses corresponding positions to register two coordinate spaces.
I would like to call the first one a fiducial...filter and the second one a landmark...filter, because the term landmark is highly used by health professionals for coordinate- or image registration.

[SVN revision 22655]
FIX (#3740): interface change: changing method name CalculateJurisdiction to CanHandleEvent

ups, changing assigned to

Shall I create a Wiki specification page for this change request?
As it is a renaming I asume this is unnecessary.

not MITK 1.0 release relevant, because it only occured on one machine with instable COM port configuration.

I have made my changes to both, but the one in Code/Interactions has a few more patterns (MiddleLClickScroll) i think by Markus E. I have already send a comment in the corresponding bug, but no answer... Should be ok to use Code/Interaction.

It starts to be tedious to have two statemachine.xml... Could you create something like a softlink to one statemachine.xml? Greetz

Tobi S. has worked on a T2413. Due to testing, the xml-file had to be moved. The xml-file in the old directory (interaction) should have been removed and replaced by a softlink. That was oniginally the plan.

[SVN revision 22706]
FIX (#3274): fixed segfault when only one tool is connected. Error message is displayed and the method returns properly.

taking the FR

[SVN revision 21171]
ADD (#3274): Adding IGT bundle that tracks two NDI tools and guides the one to the other

Found a segmentation fault that is caused if only one tool is connected.

[SVN revision 21174]
FIX (#3274): removing warning of unused variable

[SVN revision 21179]
FIX (#3274): removing warning of order of initialization and adding documentation in header and changing opencherry to blueberry in plugin.xml

[SVN revision 21172]
FIX (#3274): removing warning not to use exception variable

not really relevant; nice to have.

[SVN revision 21188]
ENH (#3264): changing icon

Found Bugs:
+ RegistrationWidget: Test "Registration still pending..." not replaced by results of registration
+ Loading SROM files doesn't work according to S. Buccheri
+ Click into Image Fiducials list causes app to crash
+ clicking StartNavigation and then Disconnect leads to an error. StartNavigation->StopNavigation->Disconnect is good.

[SVN revision 21232]
ENH (#3264): Adding Mode3DMean for activation of new fiducial tracking mode.

[SVN revision 21200]
ENH (#3264): adding feature to average (mean) a tracked fiducial position on 100 positions. For the first test 100 is fixed, but is to be made editable via set method. Changes will only take effect if mode is set.

[SVN revision 21248]
ENH (#3264): writing results of registration to gui once finished

(In reply to comment #7)

See T3274 there rendering works because of a timed update. Not really nice but it works yet!

I've discussed this with the NDI developer. The API does not offer to read out which tracking volume currently is in use. One will have to set the volume to know which is currently used.
But this can be implemented in NDITrackingDevice. By default use the first tracking volume and let the user choose if another volume shall be used. Display the default volume (first received volume) and change it if the user chooses a different one.

Only standard volume / cube is implemented, right?

[SVN revision 21831]
FIX (#3248): deleting operation objects if undo is disabled

Xavi,
this should solve the issue, right?

Fixed and bug closed

[SVN revision 21867]
DOC (#3248): adding documentation for enabling and disabling the undo mechanism

[SVN revision 21811]
FIX (#3248): changing clearing lists in Undo mechanism; adding new unit test to check freeing memory in Undo mechanism

[SVN revision 21895]
FIX (#3248): additional reference cleanup (StateEvents and PositionEvents)

Finally I could check the code. The attached patch basically adds lines with "delete doOperation;" to class mitkPointSetInteractor.cpp. Surprisingly this causes no crash after clicking Undo and Redo buttons.

Added wiki specification. Removed memory leak when undo is enabled. Still need to solve the case if undo is disabled.

Didn't check in changes because of issues concerning testing. To be solved soon.

taking the assignment

[SVN revision 21810]
FIX (#3248): fixing bug that caused a runtime error when calling destructor of OperatonEvent while m_Destination was invalid; adding documentation

[SVN revision 21596]
ENH (#3241): Adding Icon for igtToolPairNavigation

Extreme Bugfixing inside mitkNDITrackingDevide::InitializeWiredTools; Wrong
porthandles for PINIT and PENA were sent. Currently during buildup of table,
InitializeWiredTools is called several times. This still has to be nicely
solved.

[SVN revision 21276]
FIX (#3139): severe bugfixing in NDI API methods inside InitializeWiredTools

LoadSROMFileForWiredTools

LoadSROMFileForWiredTools

SROM Files could not be loaded for wired tools yet. Changes as patch due to quickhack with Jochen on phone.

Connected QmitkNDIConfigurationWidget.cpp via dataChanged event from QAbstractItemModel. Method QmitkNDIConfigurationWidget::UpdateTrackerFromToolTable() is called when data changes (during initial setup as well). New Columns 0 (ID) is skipped. Further can be added. Yet, only the name of the first sensor is written into the list (BUG::TODO). Added method mitkNDITrackingDevide::InitializeWiredTools to reinitialize the tracking hardwase (Jochen assisted on the phone).

[SVN revision 21263]
ENH (#3139): adding functionality to add a SROM file to a not flashed wired tool. See T3139 for further documentation.

Nice. I made the same initial changes to test it. Thanks for the work and the commit. So closing the bug.

Please also state the system on which it was tested or on which the bug occured. There is a big difference between Win32 and Win64 systems concerning memory! ;)

needs intensive redesign an no quickhacks, Working so far fine so not v1.0 relevant

splitting all state machine patterns into bundle files will end in many redundant patterns. Many bundles use for example pattern "pointsetinteraction" and multiple initialization would be the consequence.
I would rather recommend to make it optional if a bundle wants to add a special pattern by loading an xml file containing patterns. Then StateMachineFactory would have to check if the pattern already exists and throw a warning if so.

[SVN revision 26730]
ENH (#2820): change name of state machine pattern style from "Powerpoint" to "standard" and removing an unused and double declared event (id 1012, name PositionEvent)

[SVN revision 26741]
ENH (#2820): add and test support to specify additional xml files with state machine patterns and event declarations. See T2820 for further documentation.

I spoke to Markus. An addition if statement will be added that avoids loading the same file again. So loading file A then file B and then file A again will not be caught and warnings will be displayed. Loading file A several times after another will not produce warnings.

I will write an EventMapperTest to also check loading of different event ID declarations and throw errors on parsing events that are already loaded (adding event ids, that are already present).

adding dependency to external bug report / feature request

should be done with care and not in a hurry. Furthermore intensive tests required. so feature request for future releases, not 1.0 in my opinion

[SVN revision 26817]
ENH (#2820): catching if the same file is to be parsed a second time. Test this in EventMapperTest. For further information see T2820.

[SVN revision 26752]
ENH (#2820): adding method to add manually created state machine patterns. Testing in StateMachineFactoryTest. Also correcting public StartElement and EndElement to protected in StateMachineFactory. See T2820 for further documentation.

Current state of EventMapepr:
static New() method, static EventDescriptionVector and static StateEvent instance that is passed to GlobalInteraction.
This makes it pretty static in my opinion!

Advanced StateMachineFactoryTest to check loading of additional xml files. For this I will add two xml files to the testing/data directory and load them in StateMachineFactoryTest.

adding core modification flag (sorry for spamming.....)

fixed, closing bug

Could be, yes.

[SVN revision 19634]
FIX (#2719): removing unreachable states in patterns "drag" and "dragWithCtrl" in statemachine.xml; adding LOG messages in case unreachable states are found; Removing bug that lead to always returning false in RParse(...)

also Log messages are produced now. They show the name of the statemachine in which the error occurs.

[SVN revision 20003]
FIX (#2710): adding property "showVolume" to segmentation node to show volume

Yes and difficult to implement... suggestions?

[SVN revision 20002]
FIX (#2710): twisting labels increase decrease

not practible... but we could think of a mechanism that the main view sends a signal if a different widget was activated. Then the one view could toggle the button and deactivate the interactors.

[SVN revision 19890]
FIX (#2710): changing to QToolBox with bold headlines. For this, the qgroupboxes of subviews had to be removed.

ok, I have a small change that switches interaction of setting the one seedpoint from left mouse button + shift to only left mouse button. But since the statemachine.xml is in the core, I have to wait for a core modification flag.
->tomorrow...