Clarify docs for "`!` and traits"

[camelid]

The section "! and traits" says:

When writing your own traits, ! should have an impl whenever there is an obvious impl which doesn't panic!.

Yet it doesn't explain why you should impl your traits for !. I think it would be good for someone who understands ! well to write up a sentence or two about why you should impl your traits for !. I myself don't quite understand what the reason for doing so is ;)

@rustbot modify labels: T-doc C-enhancement

[jyn514]

I myself don't quite understand what the reason for doing so is ;)

rust/src/librustc_interface/util.rs

Lines 575 to 582 in a888b02

	// FIXME: Currently the `everybody_loops` transformation is not applied to:
	// * `const fn`, due to issue #43636 that `loop` is not supported for const evaluation. We are
	// waiting for miri to fix that.
	// * `impl Trait`, due to issue #43869 that functions returning impl Trait cannot be diverging.
	// Solving this may require `!` to implement every trait, which relies on the an even more
	// ambitious form of the closed RFC #1637. See also [#34511].
	//
	// [#34511]: https://github.com/rust-lang/rust/issues/34511#issuecomment-322340401

might shed some light, especially the link to rust-lang/rfcs#1637

[camelid]

So does that mean that ! coerces to any concrete type, but not to an impl Trait?

[jyn514]

Correct. ! does not implement all traits, and if it does not implement Trait, then unimplemented!() as the body of a function returning impl Trait will be a type error.

[jyn514]

See also #65863 and related issues linked from there.