negative impls integrated into coherence

[nikomatsakis]

Proposal

Summary and problem statement

@yaahc and I have been investigating the possibility of integrating 'negative impls' into coherence. The precise motivation is to enable

impl !Error for &str { }

and then, in a downstream crate, support

impl From<&str> for Box<dyn Error>> { }
impl<E> From<E> for Box<dyn Error>>
where
    E: Error, { }

Note that this would not permit where T: !Trait syntax, although we could in principle support that, though we would have to be careful about what it means (in particular, it would be relatively easy to support if what it means is "there is a negative impl").

Motivation, use-cases, and solution sketches

• Supporting the project group's work, as described above

Prioritization

Fits under error handling, and is also a kind of core enabler likely to arise elsewhere

Links and related work

• Tracking isssue for negative impls Tracking issue for negative impls rust#68318
• Implementation of negative impls permit negative impls for non-auto traits rust#68004
• Interactions with marker traits and overlapping impls Tracking issue for allowing overlapping implementations for marker trait rust#29864

Initial people involved

nikomatsakis, yaahc

What happens now?

This issue is part of the experimental MCP process described in RFC 2936. Once this issue is filed, a Zulip topic will be opened for discussion, and the lang-team will review open MCPs in its weekly triage meetings. You should receive feedback within a week or two.

This issue is not meant to be used for technical discussion. There is a Zulip stream for that. Use this issue to leave procedural comments, such as volunteering to review, indicating that you second the proposal (or third, etc), or raising a concern that you would like to be addressed.

[rustbot]

This issue is not meant to be used for technical discussion. There is a Zulip stream for that. Use this issue to leave procedural comments, such as volunteering to review, indicating that you second the proposal (or third, etc), or raising a concern that you would like to be addressed.

[scottmcm]

This sounds great!

Related to the !Error for str, it would be cool if this also let us impl !FnOnce for str, instead of needing this #[fundamental]:
https://github.com/rust-lang/rust/blob/7f9ab0300cd66f6f616e03ea90b2d71af474bf28/library/core/src/ops/function.rs#L65-L67

[nikomatsakis]

Ah, interesting, I hadn't considered the idea that we might use it to remove some fundamental, even if we can't remove them all. Yes, that would be good.

[scottmcm]

I don't know if this works or is needed here, but let's find out

@rustbot second

[nikomatsakis]

We agreed to let this work proceed. I will create the relevant tracking issue.

[nikomatsakis]

However, I'd like to discuss this as part of tomorrow's planning meeting. We would need a liaison, I think, and I'm not sure that @yaahc can realistically be the owner here (I think it is realistically me).

[pnkfelix]

(I'm willing to be liason. We should revisit official assignment of liason in future lang team meeting; today's only has two team members present.)

[yoshuawuyts]

As an experiment I implemented async fn main in the compiler using this feature (branch). This uses the following negative trait bounds to allow us to impl Termination for Future<T>:

impl !Future for ! {}
impl !Future for () {}
impl<E: ?Sized> !Future for Result<!, E> {}
impl<E: ?Sized> !Future for Result<(), E> {}

There was figuring out to do regarding dyn traits which might need to be improved; but overall this feature has been incredibly useful and is a missing piece to enable async fn main to be implemented.

[nikomatsakis]

This is basically all set but we just need to create the real tracking issue. @pnkfelix will serve as liaison. There is a repository here for those interested in more details:

https://rust-lang.github.io/negative-impls-initiative/

[yaahc]

This is basically all set but we just need to create the real tracking issue. @pnkfelix will serve as liaison. There is a repository here for those interested in more details:

https://rust-lang.github.io/negative-impls-initiative/

I couldn't find a link to the repo in the mdbook you linked but was able to guess it from the url: https://github.com/rust-lang/negative-impls-initiative

[nikomatsakis]

Closing in favor of https://github.com/rust-lang/lang-team/issues/136