#proc-macro #derive #traits #copy #bounds #generic #clone

macro iderive

Drop-in replacement for derive that doesn't directly depend on generic bounds

6 releases (stable)

1.2.3 Jul 16, 2024
1.1.2 May 13, 2024
1.1.0 Apr 3, 2024
1.0.0 Jun 17, 2022
0.1.0 Jun 28, 2021

#679 in Rust patterns

Download history 12/week @ 2024-09-13 12/week @ 2024-09-20 5/week @ 2024-09-27

377 downloads per month

Zlib OR MIT OR Apache-2.0 OR BSL-1.0

58KB
1.5K SLoC

iderive: Inner Derive

iderive is a drop-in replacement for derive that doesn't directly depend on generic bounds. It only checks the types of a struct's fields when deriving a trait.

Example

#[derive(Clone, Copy)]
struct TaggedIndex<T: ?Sized> {
    index: usize,
    _tag: PhantomData<T>,
}

let a = TaggedIndex::<String> { index: 0, _tag: PhantomData };
let b = a;
let c = a; // Error: Value used after move

This won't work because derive requires that T implements Copy for TaggedIndex to be able to derive it.

In contrast, iderive only checks the struct's fields to determine if a trait can be derived. Because usize and PhantomData<T> implements Copy regardless of the type of T, iderive(Copy) will implement Copy for TaggedIndex:

#[iderive(Clone, Copy)]
struct TaggedIndex<T: ?Sized> {
    index: usize,
    _tag: PhantomData<T>,
}

let a = TaggedIndex::<String> { index: 0, _tag: PhantomData };
let b = a;
let c = a; // Works!

Supported traits

iderive is currently implemented for Clone, Copy, Debug, Default, PartialEq, Eq, PartialOrd, Ord and Hash.

Version history

  • 1.2.3
    • Fix parsing of field visibility, attributes and function trait bounds
  • 1.2.0
    • Rewrite; iderive now has no dependencies
    • Don't use canonical implementations, because this breaks if the other trait fails bounds
  • 1.1.2
    • Remove the non-exhaustive support added in 1.1.1 as it doesn't make sense when all fields are displayed anyway. This matches the output of #[derive(Debug)]
  • 1.1.1
    • Indicate non-exhausiveness in the output of the Debug trait for named structs
    • Guard against redefinitions of the bool type
    • Don't require syn's full feature
    • Add more license options
  • 1.1.0
    • Use canonical implementations of Clone/PartialOrd if Copy/Ord is also derived
    • Update to syn 2.0
  • 1.0.0
    • Remove debug output that was left in by accident
  • 0.1.0
    • First release

No runtime deps