Daria Sukhonina
Daria Sukhonina
So i ran `cargo miri test --test quickchecks ...`, and it have spuriously failed on `alloc_strs` and `can_allocate_big_values` tests, only indicating that one of its spawned `cargo miri` process exited...
For now i made it possible to not enable `quickcheck` via similarly named feature in #168.
I think it could be better to not switch to `cargo-fuzz` if we would run 1 test in quickcheck. This would allow to check UB in the `quickcheck!` code. However...
Fixed on main + current nightly
This is mostly for ergonomics, for example to eliminate creation of arbitrary lifetimes for `from_raw*` functions.
> I doubt that lifetime variance has any effect on optimizations. Of course, it's just that there's currently no nonaliased covariant pointer for rust, so the only option is maybe...
Basically `deref_mut` happens inside of the `Vec::drop`, so the `&mut [char]` is created to uninitialized memory.
> I don't see anyone in the linked that confirms that it is safer either. It is mentioned on the top, that fields of a struct (`&mut T` there) must...
> I missed this whole conversation, but we pass in the stack size at compile time, do we not? (we do for other languages)... can't that just be compared against...
> Couldn't you also do this with a sentinel tag at the end of the stack area? And just detect if it's overwritten each frame (it should never be touched)?...