grimme-lab
GFN0 sporadically segfaults before printing setup info
Describe the bug Running xtb from this commit (v6.5.1 + a commit to allow compiling with Visual Studio) on Windows or Mac, GFN0 seems to sporadically segfault just before printing Setup info. Strangely, it doesn't always fail; only about half the time does it seem to crash. There doesn't seem to be any dependence on the input molecule (I have seen the failure for H2O, NH3, CH2O, and CH4.
To Reproduce Run xTB with GFN0 on Windows (my specific compilation used ifort and MSVC) or Mac.
Expected behaviour Consistently run a given input. I'm fine with it failing if, say, some maximum molecule size is exceeded if that's the only thing that can be done.
Additional context I do already use an increased stacksize as recommended in a prior issue for the Windows build. I can try to increase it more, but the sporadic nature of the failure makes me wonder if there isn't a real issue beyond just needing a larger stack. On Mac, it should be running with unlimited stack space, so that alone doesn't seem to be the issue.
As we now have a Windows version working again due to #1354, please try again with the latest bleeding-edge version to see if this problem still occurs.
