impl-only-use

[hadronized]

This RFC gives Rust the syntax:

use foo::FooTrait as _;

Here, _ means that we want to import the FooTrait trait’s impls but
not the trait symbol directly, because we might declare the same symbol
in the module issuing the use.

---

Rendered

[hadronized]

Just as a reference, Haskell has that:

import Foo.Bar.Zoo () -- imports the instances (impls) but not the typeclasses (traits) directly

[burdges]

We'd want this for inherent methods too, not just traits, right?

[sfackler]

@burdges inherent methods don't need to be imported to be used.

[hadronized]

No need for inherent methods because you never express them in the use position. :)

[sfackler]

"Qualifying"

[hadronized]

Nice catch, thanks!

[Centril]

Neat RFC.

Also related to traits, but perhaps deserving of a separate RFC... How about importing trait methods of the following form as free functions:

trait TheTrait {
    fn notUsingInArgPos2(x: T1, y: T2, ...) -> X<TheTrait>;
}

where X can be something like Result<T, ?>

[egilburg]

Hm, isn't conventionally _ used to describe a concrete thing you just don't care about, rather than a template wildcard?

[hadronized]

Yes, and here it’s the same: a concrete thing (a trait) you don’t care about. :)

[petrochenkov]

use Trait as _ needs to desugar into use Trait as SomeUniqueNameYouCantReferTo (i.e. SomeUniqueNameYouCantReferTo is a "gensym").
With this scheme glob imports/reexports can work properly with such items, i.e. import/reexport them.

mod m {
    pub use Trait as _;

    // `Trait` is in scope
}
use m::*;

// `Trait` is in scope too 

[petrochenkov]

use NonTrait as _ can work the same way, it will just always be reported as unused import.

extern crate my_crate as _ can work in the same way too (this can be useful for linking-only crates).

[hadronized]

Good comments, I’ll have them added! :)

[vitiral]

Oh please this. I could use this in the prelude crate so I don't have to alias io::Write vs fmt::Write.

[arcnmx]

🎉 Every couple months I accidentally do this hoping maybe it got implemented.

A small point brought up previously is that this additionally affects rustdoc, and probably requires some special handling/formatting to display without just showing the generated ident.

[nikomatsakis]

Nominating for discussion in the @rust-lang/lang team meeting.

[withoutboats]

@rfcbot fcp merge

[rfcbot]

Team member @withoutboats has proposed to merge this. The next step is review by the rest of the tagged teams:

• @aturon
• @cramertj
• @eddyb
• @nikomatsakis
• @nrc
• @pnkfelix
• @withoutboats

No concerns currently listed.

Once these reviewers reach consensus, this will enter its final comment period. If you spot a major issue that hasn't been raised at any point in this process, please speak up!

See this document for info about what commands tagged team members can give me.

[rfcbot]

🔔 This is now entering its final comment period, as per the review above. 🔔

[rfcbot]

The final comment period is now complete.

[aturon]

This RFC has been merged!

Tracking issue