Tracking Issue for `range_bounds_is_empty`

[pitaj]

Feature gate: #![feature(range_bounds_is_empty)]

This is a tracking issue for the is_empty method on RangeBounds.

Public API

// core::ops
pub trait RangeBounds<T: ?Sized> {
    ...

    fn is_empty(&self) -> bool
    where
        T: PartialOrd,
    {
        ...
    }
}

Steps / History

• ACP: IntoBounds::clamp and RangeBounds::is_empty libs-team#539
• Implementation: add IntoBounds::intersect and RangeBounds::is_empty #137304
• Final comment period (FCP)¹
• Stabilization PR

Unresolved Questions

• (Excluded(1), Excluded(2)) is considered to be non-empty. This could be a foot gun. Should we add a trait bound to avoid it? Related issues:
  • RangeBounds::is_empty has a nonsensical implementation #149625
  • [ACP] RangeBounds::overlaps libs-team#277 (comment)
  • IntoBounds::clamp and RangeBounds::is_empty libs-team#539

Footnotes

1. https://std-dev-guide.rust-lang.org/feature-lifecycle/stabilization.html ↩

[scottmcm]

This feels like it really wants #[final] so that unsafe code can actually use it.

[pitaj]

Does #[final] allow types in the same crate to still specialize? Would be kinda a shame if the infinite range types couldn't just return true.

[judnich]

Link to comment/discussion: There is a serious correctness bug in the current API design.