Currently, in order for MaybeUninit<T>: FromBytes, we require that T: FromBytes. This was introduced in #308 to deal with the soundness hole described in #299. However, we could relax the bound from T: FromBytes to T: FromZeroes and it would still be sound. It's a bit "asymmetrical", but it's strictly more permissive, so maybe we should do it?
If we eventually implement #251, we should just relax these bounds to NoCell.
Currently, in order for
MaybeUninit<T>: FromBytes, we require thatT: FromBytes. This was introduced in #308 to deal with the soundness hole described in #299. However, we could relax the bound fromT: FromBytestoT: FromZeroesand it would still be sound. It's a bit "asymmetrical", but it's strictly more permissive, so maybe we should do it?If we eventually implement #251, we should just relax these bounds to
NoCell.