The fuzzer currently unconditionally considers generating Booleans (atomic_bool and bool) when producing global variables. This means that trying to check the fuzzer's work through Herd simulation is almost impossible.
Since we aren't going to use Herd in most cases, the right solution IMO to this is to make the fuzzer's distribution of types configurable. I'll pop an enhancement issue through soon outlining how I see this working.
The fuzzer currently unconditionally considers generating Booleans (
atomic_boolandbool) when producing global variables. This means that trying to check the fuzzer's work through Herd simulation is almost impossible.Since we aren't going to use Herd in most cases, the right solution IMO to this is to make the fuzzer's distribution of types configurable. I'll pop an enhancement issue through soon outlining how I see this working.