I’m trying to track down a change that fixed a problem that I’m not sure we knew about. I have a report from a user running macOS Ventura. They are using a Python 3.9 virtual environment running under Rosetta (needed because of some third party tooling that doesn’t yet support M1). They are obtaining CMake through pip. If they use CMake 3.24.0, the cmake binary crashes even doing a simple cmake --version, but it works fine with CMake 3.25.0.
I’m curious if anyone knows of a specific fix in CMake that was made to address this, or if we just happened to have fixed the problem as part of something else. I wasn’t able to find any change that looked like something related to this, but it’s not an easy thing to search for and bisecting isn’t really an option for this scenario (I don’t have access to a similar system, so I can’t test directly). I mostly want to know so that we understand what the problem was and can prevent it from incidentally coming back in a future change.