Is this fixed with 50b8977b3486?
- Queries
- All Stories
- Search
- Advanced Search
- Transactions
- Transaction Logs
Advanced Search
Oct 23 2018
Oct 17 2018
MITK_PP_ macros were replaced with their original BOOST_PP_ counterparts. Only mitkPPArgCount.h is left as it doesn't originate from Boost.
Pushed new branch T24615-FixBoostRelatedIssues.
Oct 16 2018
Fixed in beta branch. Turned out it wasn't found on Linux and macOS at least since 2013. I just double-checked the 2016.11 installers and storescp was set manually back then.
Pushed new branch T25435-TimeFunctionF10.
Pushed new branch T25526-DoubleClickOnLabelItemCrash.
This error occurs when double-clicking on a label (no need for a second label to be present), or when searching for a label. Both code paths end in a request to update the label's center of the mass which is currently not compiled for four-dimensional images.
Oct 12 2018
Ok, warning level harmonization for Linux and Windows turns out to be impossible at the moment, we have to wait for MSVC updates. The other ideas have been fixed and merged.
Oct 10 2018
Propertylist has more strict behavior. Only none empty strings are allows as property key. Property list now also implements the PropertyOwner interface.
Introduced with T25441: Name DataNodes with DICOM data appropriately by using dicom tags
Oct 5 2018
Found same leak also in mitk::DICOMSegmentationPropertyHelper.
Pushed new branch T25495-MemoryLeak.
Oct 1 2018
I (hopefully) improved our CMake script by hint find_package(Qt5) at Qt default install locations (C:/Qt). Only MSVC for now, Linux and macOS are following tomorrow with /opt/Qt and ~/Qt.
Sep 21 2018
As far as I can see (had to dig through the commits) All changes of this branch, have already been cherry picked to the master at some point. (Don't know if they are also already in the current beta brunch).
Sep 14 2018
The issue still occurs on both Windows and Linux systems. The transfer function widget seems to be initialized correctly. The last MITK method called before diving into VTK internals mentioned above is VtkMapper::MitkRenderVolumetricGeometry.
Sep 6 2018
Sep 5 2018
Aug 30 2018
update: gcc works
Aug 29 2018
Aug 28 2018
I can confirm that in version 15.8.1 this is also happening in Debug, without any global optimization. I have already reported this, hope they will fix it soon.
Aug 27 2018
Pushed fix to beta
Pushed new branch T25298-RemovePATests.
Aug 24 2018
Yes, thank you!
In T23343#162900, @kislinsk wrote:It's not clear to me if this mergable as-is or not?
I guess this also implies that T23343: Dicom editor can not load dicom segmentation objects gets resolved?
It's not clear to me if this mergable as-is or not?
Thank you, good work!
Sure. Thank you!
Pushed new branch T25228-ResolveMatchPointAlgorithmDependencies.
@floca Sorry, but let me grab this task for now as I am currently testing if my assumption above works. :-)
Should be fixed in the beta branch now. I made a commit in your name (rMITK4fb32d1ab856: Fix wrong deletion), hope that's okay for you?
Pushed new branch T25285-EllipsoidCrash.
Aug 23 2018
The difference between Ellipsoid and the other classes is the usage of a vtkSmartPointer in the former case instead of a raw pointer. Unfortunately, the one who changed the raw pointer to a smart pointer forgot to remove the call to Delete(). This makes a difference as SetVtkPolyData() calls TakeReference() to take ownership of the input and then manually calls Register() to increase the reference count to 2. Then, Delete() reduces it to 1 and when the sphere smartpointer goes out of scope the reference count is 0 and the object is deleted but to work as intended it should be 1 after the constructor.
We made some findings today that indicate missing dependencies of the mdra libs in the installer. Adding the ep/src/MatchPoint-build/bin folder to the search path works. So basically it is not enough to have the install rules in mitkInstallRules.cmake. Instead, the mdra-files should be handled like other plugins and passed to fixup_bundle as dynamic runtime libs.
Aug 22 2018
I already downgraded to 15.7.6, sorry! But I still have the error prompt in case that helps. Maybe @aguilera still has 15.8.0 installed and can test the proposed solution?
You can also try to disable global optimizations with a compiler switch and see if if helps (for example, CMAKE_CXX_FLAGS are propagated also to external dependencies).
Yes, ITK and MatchPoint are affected. Both make heavy use of templates which seems to break the new compiler. Did you send an error report? If it wasn't offered, it can be set with a compile option: https://docs.microsoft.com/en-us/cpp/build/reference/errorreport-report-internal-compiler-errors
Aug 20 2018
I can confirm an internal compiler error when building with Visual Studio 15.8.0 and 15.8.1 (I updated today). Python, Numpy, and SimpleITK were not enabled in my case.
Oops, he confused ITK with SimpleITK. The latter still leads to an internal compiler error. I recommend to downgrade to 15.7.x or to change the build configuration to custom and switch off Python, Numpy, and SimpleITK.
Pushed new branch T25269-VisualStudioVersions.
A colleague just told me that there was a second update this week that fixed the internal compiler error. However, we have a CMake script that assembles some variables related to the Visual Studio version, which needs to be updated rather often. 15.8 is missing. I will have a look at this file and try to come up with something generic, future-proof.
Internal compiler errors are usually really errors/bugs in the compiler. Didn't built with 15.8 so far but I'll update my compiler this week.
Aug 16 2018
Pushed new branch T24842-lincenseInformationIcons.
On Windows, the dicom plugin builds and also finds storescp.
Aug 15 2018
Aug 13 2018
@seitela noticed, that with Qt 5.11.1 we also have MSVC 2017 as minimum required compiler as QtWebEngine in Qt 5.11 is only available for MSVC 2017.
Aug 10 2018
Pushed new branch T24502-UserManuals.
Pushed new branch T25238-consoleInvisibleTwoMonitors.
Did a commit on the beta branch, using a commit of my CTK fork due to time pressure. Should point to an official CTK commit as soon as merged by them.
Aug 9 2018
I solved this today and will create a pull request for CTK.
- Linux installers do not work with Qt 5.10.1. QtWebEngineProcess isn't found. It's a bug in Qt 5.10 and fixed in Qt 5.11.1. However, we have a crash with Qt 5.11.1 when closing the workbench related to ctkPluginStorageSQL.
Aug 8 2018
See T24376 about the intial code