This reminds me that I suggested a redesign of ExitStatus back in 2017: Mini-pre-RFC: Redesigning `process::ExitStatus` It doesn't put a distinction between successful and unsuccessful exit into the type system, but I do like that idea. Perhaps start with
pub enum ExitStatus {
SuccessfulExit,
UnsuccessfulExit(NonZeroI32),
FatalSignal(i32)
}
where the FatalSignal case never gets used on Windows. It's supposed to be possible to pass an arbitrary i32 value through exit and recover it in the parent with waitid, so abstractly we can't pack it down any further, even though I'm not aware of any Unix that actually implements that particular POSIX requirement (see the old thread for gory details).