- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
All Stories
Dec 20 2019
Pushed new branch T26930-FixPreloading.
CI
Pushed new branch T26942-ImproveCESTPrivateTagParsing.
Despite all of the obvious flaws of the parsing code around the highlighted line linked above, the particular reason for the crash is a missing check against npos regarding the results of the find() calls. Even without a more comprehensive fix, the hex number string extraction and conversion to ASCII characters should be enhanced by checking against non-ASCII hex values.
@k656s Is the example data available on our network drives?
Pushed new branch T26944-ClearBorderOfRenderWindowMenuButtons.
Pushed new branch T26943-FixShowCrosshair.
Dec 19 2019
Pushed new branch T26941-WorkaroundCMake3.16.
Dec 18 2019
Okay, this is somehow unsatisfying but I noticed that a lambda function I connected in addition to the original connection always triggered as expected. So I used it as indirection to call the actual member function.
Pushed new branch T26938-FixExtensionPointContextMenuActionTrigger.
Extension point actions are connected like QAction::triggered() QmitkDataNodeContextMenu::OnExtensionPointActionTriggered(). The second time, the slot isn't called anymore.
Dec 17 2019
Pushed new branch T26936-MigrateUiStdMultiWidgetNames.
Only partly fixed so far. Logo can be disabled in the Overlay Manager View.
Pushed new branch T26934-FixScreenshotMaker.
Pushed new branch T26936-FixStdMultiWidgetRenderWindowNames.
Pushed new branch T26935-DoNotResetCMAKE_DEBUG_POSTFIX.
If MITK_USE_Python3 is on, CMAKE_DEBUG_POSTFIX is reset. ❗
Which version / commit did you use?
Dec 16 2019
Thanks for the explanation. It happens on two freshly installed Ubuntu 18.04.3 PC's. An older setup seems to magically work fine
This was supposed to be solved in T26150: Build tree version of MitkWorkbench crashes twice before eventually starting on Ubuntu with GCC. Long story short, it has probably something to do with a messed up linkage of mixed superbuild/system dependencies. After some months of silence I think you are not the first one to encounter this bug again so we have to reinvestigate. As far as I remember using Clang does not lead to this issue in the first place.
FYI: VTK integrated this fix: