clarify the role of rustc_const_unstable

[RalfJung]

This came up in rust-lang/rust#129377.

@oli-obk I wonder if it would make sense to say that if a function is #[unstable] but a const fn, that implicitly makes it #[rustc_const_unstable] under the same feature gate?

[oli-obk]

I think the original idea was to force you to add a const_(un)stable attribute. We could also force that to have the same feature gate.

I think there was a reason we couldn't share the regular stability attribute on its own, but I don't remember why

[RalfJung]

I think the original idea was to force you to add a const_(un)stable attribute.

Yeah, we just didn't properly document that it seems, nor actually enforce it in rustc. This fixes the "document" part of that; I didn't find a good place to enforce this in rustc.

For actually changing the rules, let's discuss that in rust-lang/rust#129815.

[oli-obk]

Fairly certain we enforce it for the rustc_stable attribute

[RalfJung]

Fairly certain we enforce it for the rustc_stable attribute

Yes, indeed we do. But stdarch has tons of unstable const fn without any const stability. But I am not sure if it's worth fixing that, I think we want to change the way const stability checking works. But that's a topic for rust-lang/rust#129815. :)