For example slice based segmentation: A message box asking if the segmentation should really be deleted remains open until the test runs into a timeout. Thus, tests are running over the whole day on my computer...
Description
Description
Event Timeline
Comment Actions
Aside this behaviour, until the test reaches SliceBasedSegmentation on eigg, MainApp uses 1.17 GB! (started with Pic3D as an image)
I would sya, that's way too much...
Comment Actions
One try to get rid of the message boxes:
Now a timer is started before clicking "Delete segmentation". After 1 second the program tries to find a message box and close it. Only once! Now I changed the behaviour that it checks for messageboxes every second, UNTIL one is closed successfully.
BTW I couldn't reproduce the behaviour on eigg manually.
Comment Actions
This bug has been marked as mandatory for the MITK-0.9.2 release targeted for wednesday, 30.01.08.
Don't move its target milestone without talking to Ivo!
Comment Actions
No clue, where this behaviour comes from. I saw those message boxes with Max on eigg:
It seems like all message boxes that ever popped up, where never closed, but the test continued anyway. There was no obvious trace of a main application belonging to the message boxes. No MainApp process either. No process that reminded us of anything that could open such boxes...
To find out more, I add much more debug code to QmitkFunctionalityTesting, QmitkGraphCutSegmentation and QmitkMessageBoxHelper. Perhaps this will tell us more tomorrow.
Comment Actions
We decided (in one of the last Reliver meetings) that this is a test setup problem. Closing of messageboxes works fine when tests are run manually, problems only occur within the ctest environment or Xvfb.
The test runs successfully on several Windows machines and on makalu, so we'll disable the machines where above mentioned problems occur.