You can do what you want :) in practice it's really hard to bring down the erlang VM in zig. I'm also going to experiment with OTP managed OS threads, which I

You can do what you want :) in practice it's really hard to bring down the erlang VM in zig. I'm also going to experiment with OTP managed OS threads, which I think should allow you to safely run code that panics (but maybe not segfaults), and will clean up the memory resources after you. I think in general though you should be using zig not rust, because zig is aggressively unopinionated about memory allocators, so when you allocate memory in your code, you'll use the allocator that zigler provides, which makes the BEAM alloc conformant to zig's allocator interface. This means your code will be more performant, because there are fewer trips to the kernel on allocate and destroy, and your code will respect the BEAM's memory fragmentation, cache optimization, etc.