| sitter | |
| dfaure |
| Frameworks |
Allows us keep the message provided by QQuickWindow::sceneGraphError to later have on our bug reports.
https://doc.qt.io/qt-5/qquickwindow.html#sceneGraphError
These are generally problems at the OpenGL level that the user will usually be unable to take action on.
CCBUG: 375913
See D27985
| Automatic diff as part of commit; lint not applicable. |
| Automatic diff as part of commit; unit tests not applicable. |
Unit test? (:
I also wonder if maybe this should be lazy allocated. The vast majority of applications probably will never set it, so allocating the memory on first use should be better for the great majority of applications while making no difference for the gdb side of crash reporting.
| src/kcrash.cpp | ||
|---|---|---|
| 992 | & on the wrong side of the space | |
Paging @dfaure for expert input.
Is that global static qbytearray safe? Should we make it a q_global_static to comply with our library policy? Any objections to the diff?
Every other static variable here is a char* so that there's no global constructor being called at application startup.
Why not do the same? Just strdup the result of the toUtf8() call.
BTW: why the 1024 limit?
| src/kcrash.cpp | ||
|---|---|---|
| 107 | Still needs to be set to nullptr. | |
The following is notice that I intend to revert this change, along with the corresponding commits that make use of this functionality in Dr Konqi, as they cause a FTBFS on both FreeBSD and Windows which has not been addressed.
This regression is over a week old at this point.
Please see https://build.kde.org/view/Failing/job/Plasma/job/drkonqi/job/kf5-qt5%20FreeBSDQt5.13/lastFailedBuild/ and https://build.kde.org/view/Failing/job/Plasma/job/drkonqi/job/kf5-qt5%20WindowsMSVCQt5.14/lastFailedBuild/ for more information.
The FreeBSD issue at least should be fixed by D28095.
EDIT: the windows one is probably the same issue, though I couldn't discern the actual error from the build log.