OptionRanged[something]'s is_none and is_some methods take ownership of the value

[babichjacob]

The standard library's Option's is_none and is_some take the value by reference.

I encountered this when using serde:

#[derive(Serialize)]
struct Struct {
    #[serde(skip_serializing_if = "OptionRangedU16::is_none")]
    number: OptionRangedU16<0, 1000>,
}

with the full compilation error:

error[E0308]: mismatched types
    --> src/main.rs:5:10
     |
5    | #[derive(Serialize)]
     |          ^^^^^^^^^ expected `OptionRangedU16<_, _>`, found `&OptionRangedU16<0, 1000>`
6    | struct Struct {
7    |     #[serde(skip_serializing_if = "OptionRangedU16::is_none")]
     |                                   -------------------------- arguments to this function are incorrect
     |
     = note: expected struct `OptionRangedU16<_, _>`
             found reference `&OptionRangedU16<0, 1000>`
note: method defined here
    --> C:\Users\Jacob\scoop\persist\rustup\.cargo\registry\src\index.crates.io-6f17d22bba15001f\deranged-0.3.11\src\lib.rs:1389:1
     |
1389 | / impl_ranged! {
1390 | |     RangedU8 {
1391 | |         mod_name: ranged_u8
1392 | |         internal: u8
...    |
1473 | |     }
1474 | | }
     | |_^
     = note: this error originates in the derive macro `Serialize` which comes from the expansion of the macro `impl_ranged` (in Nightly builds, run with -Z macro-backtrace for more info)

For more information about this error, try `rustc --explain E0308`.

The relevant part of the macro is here:

deranged/src/lib.rs

Lines 958 to 970 in c25ef2c

	/// Returns `true` if the value is the niche value.
	#[inline(always)]
	pub const fn is_none(self) -> bool {
	<$type<MIN, MAX> as $crate::traits::RangeIsValid>::ASSERT;
	self.get().is_none()
	}
	/// Returns `true` if the value is not the niche value.
	#[inline(always)]
	pub const fn is_some(self) -> bool {
	<$type<MIN, MAX> as $crate::traits::RangeIsValid>::ASSERT;
	self.get().is_some()
	}

where changing the self parameter to &self would be trivial, but would be a breaking change.

Working around this on my end without a change to this crate is easy enough.

Thanks for your work on deranged, it's a great crate!

[jhpratt]

I almost certainly assumed it didn't matter because the type is Copy but failed to consider that the function signature itself matters in some contexts.

I can definitely change this. The library isn't widely used (I honestly thought I was one of the only people using it), so having a semver breaking release is essentially a non-issue.

[babichjacob]

It's a great crate! Reflecting the constraints of a number's range in its type when e.g. designing a library for interacting with an API (e.g. a percent is 0 to 100, a color hue is 0 to 360) brings us self-documentation and validation for free (because you and the other handful of contributors did it for us :) ).