A u16 that is known not to equal zero.

This enables some memory layout optimization. For example, Option<NonZeroU16> is the same size as u16:

assert_eq!(size_of::<Option<core::num::NonZeroU16>>(), size_of::<u16>());

Layout

NonZeroU16 is guaranteed to have the same layout and bit validity as u16 with the exception that 0 is not a valid instance. Option<NonZeroU16> is guaranteed to be compatible with u16, including in FFI.

Thanks to the null pointer optimization, NonZeroU16 and Option<NonZeroU16> are guaranteed to have the same size and alignment:

use std::num::NonZeroU16;

assert_eq!(size_of::<NonZeroU16>(), size_of::<Option<NonZeroU16>>());
assert_eq!(align_of::<NonZeroU16>(), align_of::<Option<NonZeroU16>>());

Compile-time creation

Since both [Option::unwrap()] and [Option::expect()] are const, it is possible to define a new NonZeroU16 at compile time via:

use std::num::NonZeroU16;

const TEN: NonZeroU16 = NonZeroU16::new(10).expect("ten is non-zero");