Georg Zotti
Georg Zotti
The LTO flag causes the test to fail? This is weird IMHO.
I am totally ignorant of these. Where to add these options in an MSVS-based environment? Or are they gcc/clang only?
Is there a chance that the Qt test framework has issues with LTO? Why would a unit test that shall compare computation results work without, and fail with it?
This means the problem lies in the processing, not in the algorithm.
https://www.qt.io/blog/2019/01/02/qt-applications-lto discusses LTO. Seems it gains us just a few percent. Interesting, the last comment also mentions a test fail. They also recommend building Qt from source. We should not...
@DarthGandalf the "few precent" were mentioned in the discussion. And yes, of course the last comment did almost certainly not discuss Stellarium, but apparently some other test in Qt's test...
Yes, interesting! Still, the reason for the failed test is not revealed.
That overflow fixed in https://github.com/Stellarium/stellarium/commit/1c1cd4879394ea3057628c91cfd8df4525a23cc1
I added qDebug()
I fear a parallel single-stepping debug session into the algorithms will be required, with and without LTO. Super annoying, and I don't have time for this in these days. Maybe...